Radiation Therapy Dosimetry

Radiation Therapy Dosimetry
A Practical Handbook
Radiation Therapy Dosimetry
A Practical Handbook
Edited By
Arash Darafsheh
First edition published 2021
by CRC Press
6000 Broken Sound Parkway NW, Suite 300, Boca Raton, FL 33487-2742
and by CRC Press
2 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN
© 2021 Taylor & Francis Group, LLC
CRC Press is an imprint of Taylor & Francis Group, LLC
Reasonable efforts have been made to publish reliable data and information, but the author and publisher cannot assume responsibility for the validity of
all materials or the consequences of their use. The authors and publishers have attempted to trace the copyright holders of all material reproduced in this
publication and apologize to copyright holders if permission to publish in this form has not been obtained. If any copyright material has not been acknowledged
please write and let us know so we may rectify in any future reprint.
Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmitted, or utilized in any form by any electronic,
mechanical, or other means, now known or hereafter invented, including photocopying, microfilming, and recording, or in any information storage or retrieval
system, without written permission from the publishers.
For permission to photocopy or use material electronically from this work, access www.copyright.com or contact the Copyright Clearance Center, Inc. (CCC),
222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. For works that are not available on CCC please contact [email protected]
Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without
intent to infringe.
Library of Congress Cataloging‑in‑Publication Data
Names: Darafsheh, Arash, editor. Title: Radiation therapy dosimetry : a

practical handbook / edited by Arash Darafsheh. Description: First
edition. | Boca Raton : CRC Press, 2021. | Includes bibliographical
references and index. Identifiers: LCCN 2020040810 (print) | LCCN
2020040811 (ebook) | ISBN 9781138543973 (hardback) | ISBN
9781351005388 (ebook) Subjects: LCSH: Radiotherapy. |
Radiation dosimetry. Classification: LCC RM849 .R325 2021 (print) | LCC RM849
(ebook) | DDC 615.8/42--dc23 LC record available at https://
lccn.loc.gov/2020040810LC ebook record available at https://
lccn.loc.gov/2020040811
ISBN: 9781138543973 (hbk)

ISBN: 9781351005388 (ebk)
Typeset in Minion
by KnowledgeWorks Global Ltd
Contents
Preface ix
About the Editor, xi
Contributors, xiii
Part I Radiation Dosimeters and Dosimetry Techniques
Chapter 1 ◾ Fundamentals of Radiation Physics and Dosimetry 3

Blake R. Smith and Larry A. DeWerd
Chapter 2 ◾ Ionization Chamber Instrumentation 19

Larry A. DeWerd and Blake R. Smith
Chapter 3 ◾ Calorimetry 31
Larry A. DeWerd and Blake R. Smith
Chapter 4 ◾ Semiconductor Dosimeters 39

Giordano Biasi, Nicholas Hardcastle and Anatoly B. Rosenfeld
Chapter 5 ◾ Film Dosimetry 61

Sina Mossahebi, Nazanin Hoshyar, R ao Khan and Arash Darafsheh
Chapter 6 ◾ Thermoluminescence Dosimetry 75

Tomas Kron and Peta Lonski
Chapter 7 ◾ Optically Stimulated Luminescence Dosimeters in Clinical Practice 97

Stephen F. Kry and Jennifer O’Daniel
Chapter 8 ◾ EPID-Based Dosimetry 109

Brayden Schott, Thomas Dvergsten, R aman Caleb and Baozhou Sun
Chapter 9 ◾ Scintillation Fiber Optic Dosimetry 123

Arash Darafsheh
v
vi ◾ Contents
Chapter 10 ◾ Cherenkov and Scintillation Imaging Dosimetry 139

R achael L. Hachadorian, Irwin I. Tendler and Brian W. Pogue
Chapter 11 ◾ Clinical Considerations and Dosimeters for In Vivo Dosimetry 151

Douglas Bollinger and Arash Darafsheh
Chapter 12 ◾ Dosimeters and Devices for IMRT QA 173

Nesrin Dogan, Matthew T. Studenski and Perry B. Johnson
Chapter 13 ◾ Area and Individual Radiation Monitoring 185

Nisy Elizabeth Ipe
Chapter 14 ◾ Monte Carlo Techniques in Medical Physics 211

Ruirui Liu, Tianyu Zhao and Milad Baradaran-Ghahfarokhi
Part II Brachytherapy
Chapter 15 ◾ Brachytherapy Dosimetry 231

Christopher L. Deufel, Wesley S. Culberson, Mark J. Rivard and Firas Mourtada
Part III External Beam Radiation Therapy
Chapter 16 ◾ Photon Beam Dosimetry of Conventional Medical Linear Accelerators 255

Francisco J. Reynoso
Chapter 17 ◾ Dosimetric Considerations with Flattening Filter-Free Beams 267

Jessica Lye, Stephen F. Kry and Joerg Lehmann
Chapter 18 ◾ Linac-Based SRS/SBRT Dosimetry 277

K aren Chin Snyder, Ning Wen and Manju Liu
Chapter 19 ◾ CyberKnife and ZAP-X Dosimetry 305

Sonja Dieterich, Georg Weidlich and Christoph Fuerweger
Chapter 20 ◾ Dosimetry in the Presence of Magnetic Fields 315

Carri Glide-Hurst, Hermann Fuchs, Dietmar Georg and Dongsu Du
Chapter 21 ◾ Helical Tomotherapy Treatment and Dosimetry 325

Reza Taleei and Sarah Boswell
Chapter 22 ◾ Gamma Knife Dosimetry 343

Nels C. Knutson
Chapter 23 ◾ Kilovoltage X-Ray Beam Dosimetry 355

C. M. Charlie Ma
Contents ◾ vii
Chapter 24 ◾ Electron Dosimetry 375

John A. Antolak
Chapter 25 ◾ Proton Therapy Dosimetry 393

Michele M. Kim and Eric S. Diffenderfer
Chapter 26 ◾ Ion Range and Dose Monitoring with Positron Emission Tomography 413
K atia Parodi
Chapter 27 ◾ Prompt Gamma Detection for Proton Range Verification 427

Paulo Magalhaes Martins, Riccardo Dal Bello and Joao Seco
Chapter 28 ◾ Acoustic-Based Proton Range Verification 443

Kevin C. Jones
Chapter 29 ◾ Proton Radiography and Proton Computed Tomography 457

Xinyuan Chen and Tianyu Zhao
Part IV Imaging Modalities
Chapter 30 ◾ Dosimetry of Imaging Modalities in Radiotherapy 467

George X. Ding
INDEX, 479
Preface
This book is intended to cover the everyday use and underlying principles of radiation dosimeters used in radiation
oncology clinics. It provides an up-to-date reference spanning the full range of current modalities with emphasis
on practical know-how. The main audience is medical physicists, radiation oncology physics residents, and medical
physics graduate students.
This book contains 30 chapters in four main parts. The chapters in Part I deal with the fundamentals of radiation
dosimetry, basic physics of various dosimeters, practical dosimetry considerations, and Monte Carlo applications in
medical physics. Part II covers the brachytherapy dosimetry. The chapters in Part III deal with external beam radia-
tion therapy dosimetry; dosimetry of cutting edge techniques from radiosurgery to MR-guided systems to small
fields and proton therapy as well as advanced range verification techniques in proton therapy are addressed in Part
III. Part IV is on the dosimetry of imaging modalities in radiotherapy.
I sincerely thank all of the authors for their outstanding contributions to this book; their efforts have brought
all of the chapters to a very high standard. I am grateful to my mentors, mentees, and colleagues in Washington
University in St. Louis and University of Pennsylvania. I am also grateful to the publishing team for their assistance,
guidance, and advice throughout the project.
Arash Darafsheh
Washington University in St. Louis, St. Louis, Missouri
ix
About the Editor
Arash Darafsheh, Ph.D., is an associate professor of Radiation Oncology, a certified medical physicist by the
American Board of Radiology (ABR), and the PI of the Optical Imaging and Dosimetry Lab at the Department of
Radiation Oncology at the Washington University School of Medicine in St. Louis. He holds Ph.D. and M.Sc. in
Optical Science and Engineering, and an M.Sc. in Radiation Medicine Engineering. His current research interests
include optical methods in medical physics, detector development for radiotherapy, ultra-high dose rate FLASH
radiotherapy, photodynamic therapy, and super-resolution microscopy. He has served as a mentor for many gradu-
ate students, postdoctoral research fellows, and clinical residents. He has published over 90 journal and conference
papers, six book chapters, and one patent. He has been awarded research grants from the National Institutes of
Health (NIH) and the American Association of Physicists in Medicine (AAPM). He is a member of AAPM and
senior member of the Optical Society of America (OSA) and SPIE-the international society for optics and photon-
ics. He has served as an associate editor for Medical Physics and as a reviewer for numerous scientific journals.
xi
Contributors
John A. Antolak Xinyuan Chen

Mayo Clinic Department of Radiation Oncology
Rochester, Minnesota Washington University School of Medicine
St. Louis, Missouri
Milad Baradaran-Ghahfarokhi
Department of Radiation Oncology Wesley S. Culberson
Vanderbilt University Medical Center Department of Medical Physics
Nashville, Tennessee School of Medicine and Public Health
University of Wisconsin – Madison
Riccardo Dal Bello Madison, Wisconsin
German Cancer Research Center - DKFZ
Division of Biomedical Physics in Radiation Oncology Arash Darafsheh
University of Heidelberg Department of Radiation Oncology
Department of Physics and Astronomy Washington University School of Medicine
Heidelberg, Germany St. Louis, Missouri
Giordano Biasi Christopher L. Deufel

Centre for Medical Radiation Physics Radiation Oncology Department
School of Physics Mayo Clinic
University of Wollongong Rochester, Minnesota
Wollongong, NSW, Australia
Peter MacCallum Cancer Centre Larry A. DeWerd
Melbourne, VIC, Australia Department of Medical Physics
School of Medicine and Public Health
Douglas Bollinger University of Wisconsin-Madison
Department of Radiation Oncology Madison, Wisconsin
University of Pennsylvania
Philadelphia, Pennsylvania Sonja Dieterich
University of California Davis
Sarah Boswell Department of Radiation Oncology
Accuray Incorporated Sacramento, California
Madison, Wisconsin
Eric Diffenderfer
Raman Caleb Department of Radiation Oncology
Department of Radiation Oncology University of Pennsylvania
Washington University School of Medicine Philadelphia, Pennsylvania
St. Louis, Missouri
xiii
xiv ◾ Contributors
George X. Ding Nicholas Hardcastle

Department of Radiation Oncology Centre for Medical Radiation Physics
Vanderbilt University School of Medicine University of Wollongong
Nashville, Tennessee Wollongong, NSW, Australia
and
Nesrin Dogan Peter MacCallum Cancer Centre
Department of Radiation Oncology Melbourne, VIC, Australia
University of Miami
Miami, Florida Nazanin Hoshyar
New York-Presbyterian/Queens Department of
Dongsu Du Radiation Oncology
Department of Radiation Oncology Flushing, New York
City of Hope Medical Center
Duarte, California
Nisy Elizabeth Ipe
Consultant, Shielding Design, Dosimetry and
Thomas Dvergsten
Radiation Protection
Department of Radiation Oncology
San Carlos, California
Washington University School of Medicine
St. Louis, Missouri
Perry B. Johnson
Hermann Fuchs University of Florida Health Proton
Department of Radiation Oncology Therapy Institute
Division of Medical Radiation Physics Jacksonville, Florida
Medical University Vienna
Wien, Austria Kevin Jones
Christoph Fuerweger Rush University Medical Center
European Cyberknife Center Munich Chicago, Illinois
Munich, Germany
Rao Khan
Dietmar Georg Department of Radiation Oncology
Department of Radiation Oncology Washington University School of Medicine
Division of Medical Radiation Physics St. Louis, Missouri
Wien, Austria Michele Kim
Carri Glide-Hurst University of Pennsylvania
Department of Human Oncology Philadelphia, Pennsylvania
University of Wisconsin-Madison
Madison, Wisconsin
Nels Knutson
Rachael L. Hachadorian
Thayer School of Engineering
St. Louis, Missouri
Dartmouth College
Hanover, New Hampshire
Contributors ◾ xv
Tomas Kron Jessica Lye

Department of Physical Sciences Australian Clinical Dosimetry Service
Peter MacCallum Cancer Centre Australian Radiation Protection and Nuclear Safety
Melbourne, VIC, Australia Agency
and Olivia Newton-John Cancer Wellness and Research
Sir Peter MacCallum Depatment of Oncology Centre, Austin Health
University of Melbourne Melbourne, Australia
Parkville, VIC, Australia
and C.-M. Charlie Ma
Centre for Medical Radiation Physics Department of Radiation Oncology
University of Wollongong Fox Chase Cancer Center
Wollongong, NSW, Australia Philadelphia, Pennsylvania
Stephen F. Kry Paulo Magalhaes Martins

Department of Radiation Physics German Cancer Research Center - DKFZ
The University of Texas MD Anderson Cancer Center Division of Biomedical Physics in Radiation Oncology
Houston, Texas Heidelberg, Germany
and
Instituto de Biofísica e Engenharia Biomédica - IBEB
Joerg Lehmann
Faculty of Sciences of the University of Lisbon
Lisbon, Portugal
Calvary Mater Newcastle
Waratah, NSW, Australia
Sina Mossahebi
and
Institute of Medical Physics
University of Maryland School of Medicine
University of Sydney
Baltimore, Maryland
Sydney, NSW, Australia
and
Firas Mourtada
School of Mathematical and Physical Sciences
Radiation Oncology
University of Newcastle
Helen F. Graham Cancer Center and Research Institute
Newcastle, NSW, Australia
Christiana Care Health System
Newark, Delaware
Manju Liu
Department of Radiation Oncology Jennifer O’Daniel
William Beaumont Hospital Department of Radiation Oncology
Royal Oak, Michigan Duke University School of Medicine
Durham, North Carolina
Ruirui Liu
Department of Radiation Oncology Katia Parodi
Washington University School of Medicine Department of Experimental Medical Physics
St. Louis, Missouri Ludwig-Maximilians-Universität München
Garching b. München, Germany
Peta Lonski
Brian W. Pogue
Peter MacCallum Cancer Centre
Thayer School of Engineering
Melbourne, VIC, Australia
Dartmouth College
xvi ◾ Contributors
Francisco J. Reynoso Matthew T. Studenski

Department of Radiation Oncology Department of Radiation Oncology
Washington University School of Medicine University of Miami
St. Louis, Missouri Miami, Florida
Mark J. Rivard Baozhou Sun

Brown Medical School Department of Radiation Oncology
Rhode Island Hospital Washington University School of Medicine
Providence, Rhode Island St. Louis, Missouri
Anatoly B. Rosenfeld Reza Taleei

Centre for Medical Radiation Physics Department of Radiation Oncology
School of Physics Sidney Kimmel Medical College at Thomas Jefferson
University of Wollongong University
Wollongong, NSW, Australia Philadelphia, Pennsylvania
Joao Seco Irwin I. Tendler

German Cancer Research Center - DKFZ Thayer School of Engineering
Division of Biomedical Physics in Radiation Dartmouth College
Oncology Hanover, New Hampshire
University of Heidelberg
Department of Physics and Astronomy Georg Weidlich
Heidelberg, Germany Zap Surgical
Brayden Schott
Department of Radiation Oncology Ning Wen
Washington University School of Medicine Department of Radiation Oncology
St. Louis, Missouri Henry Ford Health System
Detroit, Michigan
Blake R. Smith
Department of Radiation Oncology Tianyu Zhao
University of Iowa Department of Radiation Oncology
Iowa City, Iowa Washington University School of Medicine
St. Louis, Missouri
Karen Chin Snyder
Henry Ford Health System
Detroit, Michigan
I
Radiation Dosimeters and Dosimetry Techniques
1
Chapter 1
Fundamentals of Radiation
Physics and Dosimetry
Blake R. Smith
University of Iowa
Iowa City, Iowa
Larry A. DeWerd
University of Wisconsin
Madison, Wisconsin
CONTENTS
1.1 Absorbed Dose 3
1.2 Charged Particle Transport 5
1.3 Photon Interactions 7
1.4 Quantities Used to Describe Ionizing Radiation 8
1.5 Radiation Dosimetry 10
1.5.1 Cavity Theory 11
1.5.2 Overview of TG-21 13
1.5.3 Overview of TG-51 15
1.6 Conclusion 16
References 17
1.1 ABSORBED DOSE In a theoretical sense, this point quantity refers to the
One of the most important quantities that concerns the energy deposited within an infinitesimal amount of mass
practice of medical physics is dose. This quantity is pre- (and thus volume). Realistically, energy deposition from
scribed by physicians to treat tumors, set by radiation atomic and subatomic events is discrete with respect to
safety officers as exposure limits to workers, and refer- an infinitesimal volume for which the definition of dose
enced by radiobiologists while performing cell irradia- is better represented as an average over a specified space
tion studies. This quantity, however, is rather arbitrary leading to the adjacent expression in Equation 1.1. For
and ill-defined by itself. Dose, D, is defined as a point example, a whole-body dose of 4 Gy has about a 50%
quantity from the fundamental quantities of energy and chance of killing an adult human in 60 days [1]. On
mass as: the other hand, patients undergoing radiation therapy
often receive 40 Gy or more to their tumors, but in this
dE ∆E  J  context the dose is fairly localized with a large amount
D= → , (1.1)
dm ∆m  kg  of energy deposited within a smaller volume.
3
4 ◾ Radiation Therapy Dosimetry
It is also important to specify the medium that the 1  ∆E   ∆E 

→D= ⋅  = φ⋅  for multiparticle scenario
dose is referred to. Dose to water versus dose to air A  ρ∆x   ρ∆x 
imply subtle, but extremely important, differences in
the amount of energy expended in the medium as well (1.3)
as how the energy was transferred. While different, both
where the normalization of the particle’s stopping power
quantities are reported using the same units of Gray (Gy)
to the density of the medium, ρ, is referred to as the mass
which is defined as 1 Gy = 1 J kg−1 = 6.24 × 109 MeV g−1. The
stopping power. Dose is rarely defined or calculated from a
subject of absorbed dose differences and dependences is
single particle as a single particle’s trajectory can vary sub-
paramount to the understanding of detector response,
stantially if the particle were to travel from the same initial
which will be a subject of further discussion later on.
conditions. The energy loss on this scale is largely stochas-
Calculating dose during discrete particle transport
tic in nature. The progression of Equation 1.3 illustrates
through a medium can be difficult where only a hand-
how the calculation and definition of dose change from
ful of scenarios exist that can be solved analytically.
the single-particle scenario to a more familiar scenario of
However, more complicated problems can be solved
an incident fluence of charged particles, φ, upon an inci-
using Monte Carlo methods where the path a particle
dent area, A, and volume, V , of the medium. The above
experiences is simulated discretely. Dose is only deliv-
example is referred to as the thin-film approximation,
ered to matter through charged particles. Uncharged
which subtly assumes no changes in the rate of energy loss
particles, such as photons and neutrons, will traverse
of the particle as it traverses through the medium. While
through a medium unimpeded until either an elastic
fine for our conceptual discussion of dose, more consider-
or inelastic interaction occurs. During inelastic interac-
ation is necessary to comprehensively describe the energy
tions, energy is released from the uncharged particles
loss of charged particles. For further discussion, consider a
to the medium potentially transferring kinetic energy
general relation between dose and the transport of charged
to charged particles. Those liberated charged particles
particles,
expend their kinetic energy to the surrounding medium
putting other charged particles into motion or produc- Emax
 dE 
ing uncharged particles and the process repeats. Solving
these radiation transport problems requires an in-depth
D=
∫ φ′( E ) ρdx  dE
0 c
(1.4)
understanding of the types of interaction that can occur,

the probability of their occurrence, and the kinematics For the calculation of dose, φ′ ( E ) in Equation 1.4 is used
following these interactions. for the differential fluence with respect to the energy
Let us consider the definition of dose a little more ( )
of the charged particles and ρdEdx is the portion of the
c
closely. Of concern is the transfer of energy to matter stopping power responsible for collisional energy losses
within a medium from charged particles. Following the from the incident charged particle undergoing multiple
definition of dose provided in Equation 1.1, dose to a frag- Coulombic scattering with the surrounding orbital elec-
ment of matter within the medium can be determined trons in the medium. Another assumption necessary
from the kinetic energy loss, ∆E, that a particle experiences to allow us to calculate dose accurately is to limit our
across a fragment of the medium with mass, m. If we model energy fluence spectrum from energetic, knock-on elec-
the kinetic energy loss of the particle in discrete, straight- trons, known as δ-rays, which are produced within our
line steps of length ∆x , then we can relate the total path, s, region of interest and deposit their energy elsewhere.
that the particle travels within the fragment of matter to This restriction is also referred to as charged particle
the kinetic energy lost by that particle and imparted to the equilibrium (CPE) and is necessary for the calculation
medium. This is referred to as the stopping power of the of dose. Simply, CPE requires that the charged particle
medium, ( ∆∆Ex ) , which is fundamentally related to the force fluence which enters our region of interest also leaves
acted on by the medium to slow the particle down. This the region. In a sense, CPE is a spatial assumption that
loss of energy is then related to dose by the charged particle fluence remains constant through-
out the region of interest and the point that dose is
s  ∆E  s  ∆E  defined at. The problem becomes more complicated to
D= ⋅ = ⋅ (1.2)
m  ∆x  V  ρ∆x  determine the differential fluence spectrum at a point,
Fundamentals of Radiation Physics and Dosimetry ◾ 5
requiring a more complete understanding of charged power, respectively. The mass stopping powers defined
particle transport through matter. in Equation 1.5 are conventionally given in units of
MeV cm2 g−1. The portion of radiative energy loss is
1.2 CHARGED PARTICLE TRANSPORT carried away from the region of interest in the form of
Elastic and inelastic collisions are the two fundamental photons. Therefore, in our calculation of dose, we are
types of interactions that occur in a two-body system primarily concerned with the contributions of soft and
consisting of an incident projectile and target particle. hard collisional stopping powers. Derived by Bethe [2],
While the products and participants can vary widely assuming that the kinetic energy is much greater than
among all atomic and subatomic particles, the general the electrons orbital energy of an atom (referred to as
trends are consistent. Elastic collisions are character- the Born approximation), the soft collisional stopping
ized by a preservation of the total kinetic energy and power of an incident charged particle with charge ze
momentum of the system before and after the collision. and the surrounding medium with an effective atomic
A corollary to this assumption is that the products after number of Z is:
the collision must remain identical to the constituents
before the collision. Inelastic collisions do not preserve  dE  2
N A Z   z 2    2me c 2β 2  
 ρdx  = 2πr0 me c  A   β 2   ln  I 2 (1 − β 2 ) H  − β 
2 2
the equality of products before and after the collision. s  
This often leads to the loss of kinetic energy of the sys- (1.6)
tem to the form of intrinsic energy, which includes the
production of daughter particles, excitation of atomic
or subatomic particles, and the emission of bremsstrah-
( )
where NAA Z is effectively the number of electrons per
unit mass in the medium defined from Avogadro’s con-
lung. Inelastic interactions can be further categorized by stant, N A , and the mass number of the medium, A. The
their distance from the central atom from which they
occur, known as their impact parameter, b. Soft colli-
( )
2
quantity βz 2 describes the ratio of the atomic number of
the incident charged particle, z , and the Lorentz factor,
sions, when b  atomic radius, result in small energy
β = v /c, associated with the incident charged particle’s
losses of the primary particle traversing through the
velocity, v, relative to the speed of light, c. The product
medium. The majority of these energy losses are a result
of the classical electron radius, mass of an electron, me ,
of the atomic excitation or ionization of valence electrons
and Avogadro’s constant is sometimes condensed as a
from the atoms within the medium. Hard collisions
single constant where 2πr02me c 2 N A ≈ 0.1535 MeV cm2 g−1.
occur near the atom’s atomic radius, b ≈ atomic radius,
The variable H is an arbitrary cutoff used to separate
and result in the liberation of inner-shell orbiting elec-
the evaluation of hard and soft collisions. The right
trons. Nuclear electric field interactions are the least
portion of the equation, contained in square brackets,
common of the three and occur when the incident
is an approximation of the change in kinetic energy
charged particle is within the nuclear radius of the atom,
experienced by the traversing particle from the force
b  atomic radius. The production of bremsstrahlung is
acted upon it from the electric field of nearby orbital
classified as a nuclear electric field interaction.
electrons. This is classically integrated as a function of
The mass stopping power, initially utilized in
distance from the electrons, assuming they are popu-
Equation 1.2, can be divided into soft and hard colli-
lated uniformly from the traversing particle. An over-
sional losses in addition to a radiative loss component
view of the derivation of this approximation is beyond
the scope of this discussion but can be found in most
dE  dE   dE 
= + modern graduate physics textbooks [2–4]. An important
ρdx  ρdx  c  ρdx  r
term that arises within this component of Equation 1.6
(1.5) is the mean excitation potential, I , which is the mean of
 dE   dE    dE 
=   +   +  all excitation and ionization potentials in the medium.
 ρdx  s  ρdx  h c  ρdx  r A rule of thumb is that I ∼ 10 Z , except in cases for
mediums with low atomic number. The component of
where the subscripts s, h, and r correspond to the the collisional stopping power resulting from hard col-
soft, hard, and radiative components of the stopping lisions is given by
 dE  Z z 2   Emax  2  and 15, respectively. Note that the above expression is

 ρdx  = 2πr0 me c N A A β 2  ln  H  − β  (1.7)
2 2
only valid for electrons and positrons. The generation
h
of Bremsstrahlung from heavy charges particles is neg-
where Emax in the hard collision portion of the total mass ligible as the intensity of the emitted Bremsstrahlung
collisional stopping power is the maximum energy that varies inversely with the square of the incident particle’s
can be transferred in a head-on collision with an atomic mass. For instance, while a carbon ion would have a
electron. It can be approximated for a heavy charged charge number 12 times that of an electron, the amount
particle (mparticle  me ) traversing the medium as of Bremsstrahlung produced is still minuscule relative
to an electron since the mass of a carbon nucleus is over
 β2  22,000 times larger than the electron’s mass.
Emax = 2me c 2  (1.8) Our current treatment of stopping power has ignored
 1 − β 2 
the presence of δ-rays produced from hard collisions. If
Accounting for hard collisions, the mass collisional we have δ-ray CPE, then this is not an issue. However,
stopping power for a heavy charged particle is given by since the condition of δ-ray CPE is not commonly sat-
isfied, our treatment and calculation of dose to small
 dE  NAZ   z 2  volumes must account for the energy carried away from
2
 ρdx  = 2πr0 me c  A   β 2 
2
the region of interest. The restricted stopping power is
c
the portion of the total collision stopping power which
   2m c 2β 2   includes all of the soft collision energy losses but only
×   ln  2 e 2 H  − β 2  hard collision losses resulting in delta rays of kinetic
   I (1 − β )   energy less than ∆. This assumes that any δ-rays pro-
duced below this threshold deposit their energy locally
 E   within the defined region of interest and are related to
+  ln  max  − β 2  ,
  H   the unrestricted collisional stopping power as
and simplifying,
 dE   dE 
lim ∆→ EK0  = (1.11)
 ρdx  ∆  ρdx  c

 dE  2
N A Z   z    2me c β  2 
2 2 2
 ρdx  = 4 πr0 me c  A   β 2   ln  I (1 − β 2 )  − β 
2
c   Restructuring Equation 1.9 to accommodate a

(1.9) restricted stopping power prohibits expressing Emax
using Equation 1.8 as it must be replaced with the
While not shown above, there are further corrections which energy cutoff, ∆
have been derived for this expression of stopping power to
account for relativistic scaling, shell correction to account for  dE  N A Z   z 2    2me c 2β 2 ∆  
2
 ρdx  = 2πr0 me c  A   β 2   ln  I 2 (1 − β 2 )  − 2β 
2 2
decreased interaction with K-shell electrons at lower particle
∆  
velocities, and polarization effects of dense media.
The final component of the total stopping power is the (1.12)
mass radiative stopping power. According to Podgorsak A direct application of stopping powers outside of deter-
[4], this expression for electrons or positrons is mining dose is predicting range. An estimate of a par-
ticle’s range, in terms of mass thickness (ρ× l in units of
 dE  1 2 2 NA
(
 ρdx  = 137 r0 Z A EK0 + me c Brad
2
) (1.10) g cm−2), in matter can be determined from
r
EK0 −1
 dE 
where Brad is a slow-changing function slightly depen-
dent on the atomic number, and EK0 is the initial total
CSDA =
∫0
 ρdx  dE (1.13)
kinetic energy of the light charged particle. For an ini-

tial kinetic energy of 1 MeV, 10 MeV, and 100 MeV, the Equation 1.13 is referred to as the continuous slowing
corresponding values of Brad are approximately 6, 10, down approximation (CSDA) of a particle’s range, which
is the average length traveled by a charged particle when is related to the size of the dart and the size of the bull’s
it slows down from its initial total kinetic energy to a eye relative to the dart board. This is analogous to a flu-
final kinetic energy close to zero. This is often taken as ence of uncharged radiation, ψ (cm−2), of particles inci-
a conservative estimate which overpredicts the range. dent upon a slab of matter composed of many potential
For heavy charged particles, CSDA is a fairly indicative atomic “targets” each with a little cross-sectional area,
and close approximation. However, this metric fails to σ a (cm2). The number of targets within a unit volume of
adequately predict the observed range for lighter ions, the matter can be estimated as N A ⋅ρ/A from the mass
such as electrons and positrons, since the particle’s path density, ρ, and mass number of the substance, A. The
is assumed straight and CSDA does not take into account rate that the initial fluence reduces in depth due to the
the influence of scatter on the particle’s range. number of interactions that occur is
1.3 PHOTON INTERACTIONS dψ σ a ψN A ρ

− = = µψ (1.14)
Our treatment of dose to this point has focused upon dl A
the kinetic energy loss of charged particles traversing
If the target for the reaction is instead an electron bound
through a medium. However, the question remains on
(or loosely bound) to an atom, the atomic cross section is
how these charged particles were initially set in motion.
replaced with the combined cross section of all the elec-
Some medical applications require the use of a beam of
trons bound to the atom, σ e ⋅ Z = σ a . The reduction of
charged particles, namely, electrons or protons, although
fluence, or attenuation, can be modeled as a single vari-
some heaver ion beams have been used albeit less fre-
able, µ, the total attenuation coefficient in units of cm−1
quently. For calculations of dose from these directly
or the unit probability per depth an interaction occurs.
ionizing beams of charged particle radiation, our pre-
This quantity is also expressed as the mass attenua-
vious discussion of dose is sufficient. However, a large
tion coefficient by normalizing µ to the density of the
majority of radiation applications in medicine rely upon
medium. This simple differential equation can be solved
the use of indirectly ionizing radiation, namely, photons
to obtain the uncollided fluence at a distance l in the
produced from Bremsstrahlung, called x-rays, or from
medium
the radioactive decay of nuclei, denoted as γ -rays. The
calculation of dose from indirectly ionizing radiation is
ψ ( l ) = ψ 0e −µl = ψ 0e −µ /ρ⋅Σ (1.15)
a two-step process: a photon must first interact within
a medium and then impart kinetic energy to a charged where ψ 0 is the initial fluence and Σ is used to define the
particle placing it in motion. The kinetic energy is then mass thickness through a material. The total mass atten-
transferred from the charged particle to the surround- uation coefficient can be broken down into components
ing medium following the mechanics discussed in the for each of the three aforementioned interaction types
previous section. For the purposes of this discussion, we
will focus solely on photon–electron interactions as they µ τ σ κ
are the most prominent interactions concerned within = + + (1.16)
ρ ρ ρ ρ
the practice of medical physics dosimetry.
There are four primary photon–electron interactions where ρτ , σρ , and κρ represent the mass attenuation coef-
that our discussion of dose is concerned with and they ficients for photoelectric, Compton, and pair and triplet
are Rayleigh, photoelectric, Compton, and pair and production, respectively.
triplet production interactions. However, only the latter The photoelectric effect absorption coefficient is
three emit charged particles after the interaction takes dominant at low energies, around the rest mass of an
place and dominate within the ranges of photon energies electron. As such, this particular interaction is respon-
encountered in the practice of medical physics. Each of sible for much of the contrast observed in radiological
these interactions contributes to a loss of incident pho- imaging systems. For this interaction to occur, the inci-
ton fluence through a medium. The occurrence of an dent photon must coherently interact with the entire
interaction taking place can be thought of as a dart hit- atom and impart enough energy to an orbital election
ting a binary dart board – either the dart hits the bull’s to overcome its binding energy to the atom. The interac-
eye or it misses. The probability that the bull’s eye is hit tion results in a fluorescence photon in addition to at
least one electron, possibly more if an Auger electron incidents energies, a form factor function is employed to
is emitted. The work of Heitler [5] approximated the account for the binding energy of the electron. However,
K-shell component to the photoelectric effect absorption the impact that this correction factor has upon our esti-
coefficient as mate of dose is not very large since the photoelectric
effect dominates at such low energies.
7/2
8πr02  Z 5   me c 2  Pair production is a manifestation of Einstein’s mass–
τK = 4 2 (1.17)
3  137 4   hv  energy relationship where a positron and an electron are
produced from a photon interaction within an atom’s
As indicated from Equation 1.17, the interaction cross electric field. In the event of triplet production, two
section is largest at the binding energy of the k-shell, electrons and a positron are produced. The minimum
2
EB =  nth Z=1 ⋅137  ⋅ me2c . This implies that the cross section threshold photon energy for this interaction to occur is
2
 ( level ) 
spikes at the shell-binding energies and also suggests  2me c 2 if pair production
that the medium is transparent to its own fluorescence hν ≥  (1.22)
4me c 2 if triplet production
photons. 
The Compton scattering effect occurs between an
incident photon of energy, hν, and an unbound, station- Bethe and Heitler [7] initially derived the atomic cross
ary electron resulting in a scatter photon at angle θ γ , section assuming that the interaction is far from the
and energy nucleus, thereby ignoring the effects of the nuclear elec-
tric field.
hν
hν′ = (1.18) r02 Z 2
1+
hν
me c 2
( ( ))
1 − cos θ γ κ pair =
137
P (1.23)
The scattered electron is then emitted with an energy of where P ≈ 28

9 ln ( )−
2 hυ
me c 2
218
27
. The cross section for triplet
Ee = hν − hν′ at an angle of production is similar to that of pair production but
requires an extra term, C, to account for election exchange
1
tan ( θe ) = (1.19) effects of the form
(
(1 + α 0 ) tan θγ /2 )
κ pair
where α 0 = mheνc 2 . The interaction cross section for the κ triplet = (1.24)
C ⋅Z
Compton effect was initially derived by Klein and
Nishina [6] where C is a slowly changing value ranging between
1.6 for 5 MeV < hυ < 20 MeV and 1.1 for 20 MeV < hυ <
 1 + α 0   2 (1 + α 0 ) ln (1 + 2α 0 )  100 MeV. The total triplet and pair production cross sec-
σ e = 2πr02  2 
−  tion is
 α 0   1 + 2α 0 α0
r02  1
ln (1 + 2α 0 ) 1 + 3α 0  κ = κ pair + κ triplet = ZP  Z + 

(1.25)
+ −  (1.20) 137 C
2α 0 (1 + 2α 0 )2 
1.4 QUANTITIES USED TO DESCRIBE
and IONIZING RADIATION
N A Zρ Up until this point, we have treated the interaction of
σ = σe (1.21) indirectly ionizing radiation stochastically. The inter-
A
action cross sections recently discussed describe the
For incident photon energies larger than 1 MeV, the probability of one particular event occurring and the
assumption of an unbound electron is very good, espe- consequential attenuation of the primary beam. While
cially since the scattered electron tends to receive most these quantities are useful for the understanding of radi-
of the incident photon’s energy. However, for lower ation transport through matter, they are not so useful
to practically calculate dose. In our previous discus- coefficient is also defined by the constituents of each
sion, the emphasis was placed primarily on attenuation interaction.
of the primary beam. However, as we progress through
our understanding of dose, it is practical to define non- µ tr τ tr σ tr κ tr
= + + (1.28)
stochastic quantities to predict the expectation value of ρ ρ ρ ρ
energy loss in the indirectly ionizing beam of radiation.
For the purposes of this discussion, we will discuss these Since the energy transfer coefficient is defined as the
quantities with respect to photons representing the expected portion of energy initially transferred to
energy fluence of the beam as Ψ ( E ) = ψ ( E ) ⋅ E . However, charged particles, we can also define it as
analogs exist for other indirectly ionizing radiations
such as neutron beams. µ tr µ Etr
= ⋅ (1.29)
The first nonstochastic quantity that describes the ρ ρ hυ
interaction of indirectly ionizing radiation with mat-
ter is Terma, T , which is defined as the quantity of total where Etr is the average energy transferred to charged
energy transferred to matter. This quantity reflects the particles from an incident photon of energy, hυ.
first stage of energy deposition from indirectly ionizing Let’s consider photoelectric interactions. Upon an event’s
radiation with the surrounding medium and describes occurrence, we know that the incident photon’s energy will
the rate of attenuation of the primary beam. Using our be divided among an ionized L- or K-shell electron, a fluo-
definitions of the interaction cross sections in the pre- rescent photon, and potentially an Auger electron. Thus,
vious section, we can define Terma from the incident the fraction of the energy imparted to the electrons is the
energy spectrum, up to the maximum energy, Emax difference between the incident photon energy and the
mathematically as portion lost due to fluorescence. The mass energy transfer
coefficient can then be determined by
Emax
 µ µ τ tr τ  hυ − PKYK hυK − (1 − PK ) PLYL hυL 
T=
∫
0
 ρ  ⋅ Ψ ′ ( E ) dE → T ≡ ρ Ψ , if monoenergetic
E =
ρ ρ  hυ  (1.30)
(1.26)
where PKYK hυK describes the expected fraction of energy
which has units of J kg−1. Following an interaction loss from K-shell fluorescence of photon energy hυK with
between the beam of indirectly ionizing radiation and a yield of YK occurring with a probability of PK in addition
matter, a portion of energy from the primary beam may to the energy losses should the interaction take place with
be transferred to charge particles in the medium. This an L-shell electron instead of a K-shell electron.
quantity is referred to as kerma, K, and is defined for- For the case of Compton interactions, the mean energy
mally as transferred to an electron can be determined from weight-
Emax
ing the scattered electron energy following Equation 1.18
 µ tr  µ tr with the respective cross section value calculated using
K=
0
∫  ρ  ⋅ Ψ ′ ( E ) dE → K ≡ ρ Ψ , if monoenergetic Equation 1.20 while normalizing to the integral of the
E
cross section over all scattered electron energies.
(1.27)
Emax
Once an interaction occurs, the amount of energy trans- dσ e E
ferred to the medium’s charged particles is stochastic
in nature. The equations we used to specify the mass
σ e, tr =
∫
0
⋅ dE
dE hυ
(1.31)
attenuation cross sections for the three types of photon

interactions with matter, Equations 1.17, 1.20, and 1.25, It follows that the Compton mass energy transfer coef-
can be adjusted to reflect the portion of energy loss from ficient is
the primary beam and are referred to as the respec-
tive mass energy transfer coefficients. Analogous to the σ tr N Z E σ
= σ e ,tr A = tr ⋅ . (1.32)
total attenuation coefficient, the mass energy transfer ρ A hν ρ
Pair and triplet production mass energy transfer coef- exist outside of well-controlled experiments. However,
ficient is fairly straightforward as the energy initially this is the basis from which modern dosimetry protocols
transferred to the charged particles is the residual rely on to provide accurate, traceable standards of radia-
amount of energy left after the creation of the electron(s) tion dose and kerma.
and positron
1.5 RADIATION DOSIMETRY
κ tr κ  hν − 2me c 
2
Dose is arguably one of the most important radiologi-
=   (1.33)
ρ ρ hν cal metrics utilized in the treatment of cancer. It is also
used to benchmark the constancy of the output from a
Kerma should not be confused with dose. While the medical linear accelerator. Intercomparisons of clinical
two quantities are similar, even defined by the same radiation treatment outcomes reference delivered dose
unit of Gy, dose maintains an element of locality where and dose rates. Radiation safety limits are set from stud-
energy must be deposited, whereas kerma simply states ies that have reported in analogs of dose. Most pertinent
the initial kinetic energy transferred to charge particles to the practice of therapy physics is the measurement
at an interaction point. As the charge particle traverses of absorbed dose to water from a medical linear accel-
through the medium, a portion of its transferred kinetic erator. A direct measurement of dose to a medium from
energy is expended by both collisional and radiative loss fundamental quantities, while possible, is very difficult
and susceptible to large uncertainty unless great care is
K = Kc + Kr (1.34) taken to minimize errors in setup and fully characterize
the assumptions made within the measurement. If every
The portion of the energy not radiated away is defined as clinic was responsible for this measurement, large dis-
the collision kerma, K c , crepancies could arise between departments. Instead, it
is safer and more practical for one entity to maintain
Emax
 µ en  µ en the standard device, which measures a reference quan-
Kc =
∫  ρ  Ψ ′ ( E ) dE → K c ≡ ρ Ψ if monoenergetic
E
tity. This standard can then be disseminated to several
0 clinics using a precise instrument that can be easily cali-
(1.35) brated with the standard. For example, an ionization
chamber is both a practical and precise instrument that
The mass energy absorption coefficient, ( ) , can be
µen
ρ has been used to transfer the standard of absorbed dose
stated generally as to water measured from a calorimeter to a clinic.
Ionization chambers are the most common instru-
µ en τen σ en κ en µ
= + + = (1 − g ) tr (1.36) ment to measure dose to water, which is the standard
ρ ρ ρ ρ ρ quantity used to calibrate the output of medical linear
accelerators. However, depending on the primary quan-
where the term 1 − g denotes the fraction of the initial tity used for chamber calibration, the determination of
charged particle’s kinetic energy that wasn’t radiated dose to water by the physicist changes. Over the years,
away by photons. By the definition of collision kerma, this primary quantity has changed historically from air
we know the amount of energy transferred and kept by kerma to absorbed dose to water from a 60Co beam of
the charged particles from the incident uncharged radi- radiation with a near-monoenergetic photon spectrum
ation until they come to rest. Conceptually, if CPE exists center around an average energy of 1.25 MeV. Ionization
within a defined volume, then we would know that any chambers are conceptually simple consisting of an air-
particle that left the specified volume with some kinetic filled cavity that collects the charge from ionizing radia-
energy would be replaced with another particle with the tion interacting with the gas molecules residing in the
same kinetic energy that the first particle left with. Thus, cavity. Further discussion of ionization chambers and
in this special set of circumstances, we can state that their applications are discussed elsewhere [8], but for the
dose is equivalent to the collision kerma. Much of the context of this chapter, most portable chambers consist of
practice in modern dosimetry rests upon the application a wall encapsulating a collection volume that is charged
and understanding of these quantities. Rarely does CPE to establish a voltage potential with a collection electrode.
The current challenge at hand is to relate the measured  dE 

signal in the chamber to the dose to water at the center of ∫
Dp in a = φa′ ( E ) 
 ρdx  E ,a
dE (1.38)
the chamber as if the chamber was not present to begin
with. This exact problem has been the primary focus of
Let’s assume now the following:
several published reports. Today, the United States rec-
ognizes the task group report numbers 21 (TG-21) [9]
1. The thickness, t, of the layer is much smaller than
and 51 (TG-51) [10] from the American Association of
the range of charge particles traversing the layer.
Physicists in Medicine (AAPM) as the primary proto-
Therefore, no particles either start or stop in the
cols to determine the dose to water using an ionization
layer.
chamber that was either calibrated to a known air kerma
or absorbed dose to water, respectively. 2. The change in the layer’s material from the sur-
rounding medium does not substantially change
1.5.1 Cavity Theory the fluence at p.
The determination of dose within one medium as if it
were occupied by a different material relies on the use These assumptions are commonly referred to as the
of cavity theory, and the basis of much of the TG-21 and Bragg–Gray conditions and imply that
TG-51 formalisms is grounded in these applications.
The foundations of cavity theory are based on Bragg– φ′a ( E ) = φ′w ( E ) = φ′ ( E ) (1.39)
Gray theory [11]. Consider a plane-parallel field of pho-
ton radiation as shown in Figure 1.1. The final result provides a general relation that relates
If the spectrum of charged particles, φ, was known, the dose in the two media
the dose at a point, p, in the medium, w, can be deter-
mined from the spectrum’s differential fluence, φ′  dE 
Dp in a ∫ φ′( E ) ρdx  E ,a
dE
 dE  = (1.40)
∫
Dp in w = φ′w ( E ) 
 ρdx  E ,w
dE (1.37) Dp in w
∫
 dE 
φ′ ( E ) 
 ρdx  E ,w
dE
If now the region surrounding point p was changed to and upon normalizing, the result with the total fluence
air, the dose at p is then equivalent to in both the numerator and denominator
 dE 
Dp in a ∫ φ′ ( E ) 
 ρdx  E ,a ∫
dE / φ′ ( E ) dE
= (1.41)
Dp in w  dE 
∫ φ′ ( E ) 
 ρdx  E ,w ∫
dE / φ′ ( E ) dE
reveals that the doses in the two media are related by the
ratio of their mean stopping powers, S, denoted by both
a superscript and a subscript nomenclature
Dp in a ( dE /ρdx )a Sa
= = = Swa (1.42)
Dp in w ( dE /ρdx )w Sw
FIGURE 1.1 Simple slab geometry illustrating the principles
of Bragg–Gray cavity theory. A uniform fluence of charged
particles, φ, within an area, A, traverses a thin slab of mate- While great as a first-order approximation, Bragg–Gray
rial, a, sandwiched between material slabs composed of w . theory falls short for large discrepancies in the atomic
The dose at point, p, can be evaluated for both materials a and numbers and density between the two media. This sug-
w assuming the Bragg–Gray conditions. gests that the two Bragg–Gray assumptions were not
well satisfied. The work of Spencer and Attix [3, 12]

sought to improve upon the basis of Bragg–Gray theory
by accounting for the generation and influence of δ-rays
by including restricted stopping powers in the absorbed
dose determination at a point
E0
∫
D = φe ,δ ⋅ S ( E , ∆ ) dE
∆
(1.43)
FIGURE 1.2 Burlin cavity theory discriminates charges pro-
duced within the wall (w) and cavity air (g) as they contrib-
where φe ,δ is the equilibrium fluence accounting for the ute to the total dose within the chamber. Considering each
of these contributions separately bridges the gap between the
existence of delta rays and the restricted stopping pow-
assumptions made between very small chambers, which uti-
ers, S ( E , ∆ ), are evaluated from an energy ∆, which is
lize the Bragg–Gray assumptions, and very large cavity vol-
just large enough for the charged particles to traverse umes where the influence of the wall is negligible.
the cavity. Any particles with energy less than ∆ are
assumed to neither be able to enter the cavity nor trans-
port energy. Using the Spencer–Attix cavity theory, the most commonly achieved with large-cavity chambers
relationship between the doses in two media is therefore using the subscript–superscript ratio notation presented
simply the ratio of the mean restricted mass stopping in Equation 1.42. A subtle but important note to take
powers, L∆. away is the difference between large-cavity and Bragg–
Gray/Spencer–Attix theory: one relates the expected
Dp in a L∆ ,a conditions of energy transfer with respect to the charged
= = La∆ ,w (1.44)
Dp in w L∆ ,w particle field verses the incident uncharged radiation
field. The merger of these two theories can be extended
Bragg–Gray and Spencer–Attix cavity theories build to our chamber measurement as shown in Figure 1.2
from assumptions on the charge particle fluence since most chambers do not behave completely within
between contiguous media. The relationship of dose the confines of one of these theories due to the presence
between materials reflects the change in energy loss of a wall.
from the charged particle field traversing through the This merger is known as Burlin cavity theory. If the
medium. However, a similar relation can be made with volume of the cavity is very large in comparison to the
respect to the uncharged radiation field from our defini- wall thickness, then the fluence of electrons from the wall
tions of collision kerma. Similar to CPE, radiation equi- material will exponentially decrease as they are attenu-
librium (RE) is used to describe a spatial quality of point ated in the gas volume. However, the fluence of elec-
where the radiation fluence entering a specified volume trons generated within the gas volume will exponentially
is the same as the radiation fluence exiting the volume. increase leading to the conditions governing large cavity
Because the charged particle fluence is a direct conse- theory. On the other hand, if the chamber volume is very
quence of the incident uncharged radiation field, it is small, the majority of electrons traversing the cavity orig-
conceptually simple to understand that if RE exists, then inate from the wall material. The Burlin cavity relation
CPE exists. However, the inverse of this statement is not can be expressed as
true. It follows from our earlier discussion of collision
g
kerma that if RE exists within a volume of space, then Dg µ 
the relationship in absorbed dose between two media is = d ⋅ Swg + (1 − d ) ⋅  en  (1.46)
Dw  ρ w
Dp in a K c,p in a ( µ en /ρ)a  µ en 
a
= = = (1.45) where d is the portion of electrons generated from the

Dp in w K c,p in w ( µ en /ρ)w  ρ  w wall that contribute to the equilibrium fluence in the
cavity, and 1 − d is the portion of electrons from the cav-
The relation shown in Equation 1.45 reflects the ity that contribute to the equilibrium fluence in the cav-
uncharged radiation field under the conditions of RE ity. These four theories build the foundation from which
TABLE 1.1 List of the Four Primary Theories Covered in this Chapter and Their Suitable Applications
Chamber Theory Domain of Validity Assumptions
Bragg–Gray ρgt g  ρwt w CPE, the majority of charges are produced in the chamber wall
Spencer–Attix ρgt g  ρwt w The majority of charges are produced in the chamber wall
Large Chamber ρgt g  ρwt w RE → CPE is dominated by the charges produced in the chamber cavity
Burlin Any CPE, charges produced in the cavity replaces those attenuated from the wall
Note: These theories share an implicit set of assumptions, which include a homogeneous wall, w, and cavity gas, g, material,
uniform radiation field across the entire chamber, little-to-no attenuation of the incident radiation field throughout
the chamber volume, and a charge particle equilibrium spectra that is shared between the wall and cavity air.
the current protocols in radiation dosimetry are derived charge collected. They must be calibrated at a specified
and practiced and are summarized in Table 1.1. point with an instrument, called a standard, which can
directly measure the intended primary quantity. For the
1.5.2 Overview of TG-21 TG-21 protocol, the intended quantity that chambers are
The AAPM TG-21 established a protocol to determine calibrated in is the kerma in air from a 60Co radiation
dose to a medium of water following an exposure cali- field. At a specified point one meter from the source, the
bration of a chamber in a 60Co beam of radiation [9]. air kerma that has been measured by the physicist can
While a full discussion of chamber calibration and stan- be used to determine the dose to a point from the mea-
dards is outside the scope of this chapter, a brief discus- sured quantity of exposure. Exposure is defined as the
sion will be discussed to provide context of the problem charge released per unit mass from a beam of radiation.
to the reader. Chapter 2 will handle ionization chambers
in detail. W
Kc = X ⋅ k ⋅   (1.47)
Seeing how humans are considered to be equivalent to  e  gas
water, the quantity of absorbed dose to water is a desir-
able metric to reference. It is the concern of the medical describes the relationship for the measured exposure, X,
physicist to accurately and precisely quantify how much in units of Roentgen (R) to collision kerma in units of
dose is delivered in a variety of scenarios. However, it is Gy. The constant k converts from the unit of Roentgen to
good experimental practice to always maintain a well- the charges produced per unit mass of air, k = 2.58 × 10−4
established reference that can benchmark all other mea-
surements. This reference is universally recognized as
( )
C kg−1 R−1, and the quantity We gas, approximately 33.97 J
C−1, is the energy required to create a charge pair in dry
the dose to water given: air. The standard used for 60Co energies at the US stan-
dards lab was a known-volume graphite chamber with a
1. Measurements are performed with a reference- well-known collection volume. Based on the measured
class chamber (i.e., a very stable chamber whose charge from ionized gas in the cavity, the exposure can
design mitigates any influence of charges probe calculated from
duced outside the intended collection volume and
wall) in a water tank with dimensions of at least chamber wall air
Q  L  µ en 
30 cm × 30 cm × 30 cm . X=
k ⋅ mchamber gas  ρ  chamber gas  ρ  chamber wall
β −wall
1
(Πi Ki−1 )
2. Depth of chamber is placed at the depth cor-
(1.48)
responding to the peak dose, as this is the only
location along a depth profile that CPE is (nearly)
Radiation equilibrium is assumed within the cham-
established.
ber’s collection volume and Spencer–Attix cavity the-
3. The source position relative to the surface of the ory allows us to relate the collision kerma from the air
water is 100 cm. inside the cavity to the surrounding cavity wall. If we
assume that the photon energy fluence is unperturbed
Furthermore, most chambers cannot accurately deter- by the chamber wall, and that the majority of charges
mine dose solely from the fundamental quantity of contributing the wall kerma are from the wall than the
surrounding air, large-chamber cavity theory can be factors Aion and Awall correct for the collection efficiency
applied to determine the kerma to the room air from and attenuation and scatter in the wall and buildup cap,
the kerma in the wall and Spencer–Attix cavity theory respectively. Note that these quantities are only well-
can be used to determine the dose relationship between defined for a 60Co radiation field. The calculation of the
the wall and air inside the chamber cavity. The correc- dose to the gas volume in the chamber can simply be
tion factor β accounts for the ratio of dose to collision determined from
kerma since true CPE is not fully established, but still
very close (within 0.5%). The remaining correction fac- Dgas in chamber = M ⋅ N chamber gas (1.51)
tors, K i−1, are employed to account for electrometer cali-
bration, chamber ion collection efficiency, water vapor However, this is of little use if we wish to calculate dose
content in the air, stem scatter, and the influence of wall to the water with our chamber. Working backward, the
thickness on the collision kerma. Since the air tempera- value of Dwater can then be calculated from Dgas in chamber
ture and pressure in the room will influence the charge from
collected from the chamber, a standard lab corrects the
medium
raw charge reading to a standard temperature and pres-  L
sure. The product of Equation 1.48 is the determination Dmedium = M raw N chamber gas   Pion Prepl Pwall PTP
 ρ  chamber gas
of exposure from air kerma in the room air at the center
of the chamber from the measured ionized charges that (1.52)
are a result of the air kerma within the chamber.
where N chamber gas is now considered a chamber-specific
The response, N X , of any chamber can be related to
calibration coefficient as defined in Equation 1.50, Pion is
the known radiation field by
the inverse of the ionization collection efficiency of the
X chamber, Prepl is the replacement correction for the pertur-
NX = (1.49) bation in the fluence due to the placement of the chamber
M
in the medium, and Pwall is the wall correction when the
where M is the measured charge normalized to stan- chamber wall and phantom composition are different,
dard temperature and pressure from the ionization and the air density inside the chamber. Using Burlin cav-
chamber for a known delivered exposure, X. However, ity theory, Pwall can be analytically determined as
this only relates an arbitrary response to the radiation
d ( L /ρ)wall (µen /ρ)medium + (1 − d )( L /ρ)chamber gas 
medium
field established at a lab. What we need is a transferrable  chamber gas wall 
Pwall =
quantity unique to the chamber that reflects the dose to ( )chamber gas
L /ρ
medium
the chamber. Using Burlin cavity theory, we can work

(1.53)
back from the exposure response of the chamber to the
dose response of the chamber in air. Inserting the expression for Pwall from Equation 1.53 into
Equation 1.52 to determine the absorbed dose to the
N chamber gas =
medium is given as
W
k   Aion Awallβ wall
 e    L  wall medium
NX chamber wall air chamber cap air  µ en 
 L  µ en   L  µ en  Dmedium = M raw N gas Pion Prepl PTP  d  
d  + (1 − d )     ρ  chamber gas  ρ  wall
 ρ  chamber gas  ρ  chamber wall  ρ  chamber gas  ρ  chamber cap
(1.50)
 L
medium

+ (1 − d )  
where d in this case is the ratio of charges that con-  ρ  chamber gas 
tribute to the charged particle fluence in the chamber’s (1.54)
collection volume from the wall and the surrounding
medium. The parameter d is equal to unity in the spe- Alternatively, this expression of Dmedium can be derived
cial case when the material composition of the chamber directly from the expression of Dgas in chamber using Burlin
wall and the buildup cap are the same. The correction cavity theory. It should be noted that the stopping
powers and mass absorption attenuation coefficients are calculation by using Monte Carlo-derived correction
now defined for the beam used at the clinic, no longer factors [10]. These correction factors are simulated to
the 60Co environment at the calibration lab. Further cor- determine chamber perturbation factors between beam
rections must be made to M raw at the clinic that are also quality and chamber components. For the context of
referenced to the same conditions of standard tempera- this report, beam quality, Q, refers to a type of radiation
ture and pressure, collection efficiency, polarity effects, beam in a sense that it uniquely describes the emitted
and electrometer response that were considered during spectrum. Following the TG-51 formalism, the dose to
the chamber’s calibration. In addition to the ionization a medium of water for a beam quality, Q, and chamber
chamber, electrometers are also calibrated to traceable charge measurement, M , is
standards of resistance and capacitance and carry their
own calibration coefficients that must be applied to the DwQ = M ⋅ N DQ,w (1.55)
charge reading. Specifics about these calibration fac-
tors and how they are determined will be discussed in where N DQ,w is the absorbed dose to water calibration fac-
Chapter 2. tor for an ionization chamber in the beam quality, Q. The
beam quality expressed in Equation 1.55 is the radiation
1.5.3 Overview of TG-51 field of the Medical Physicist’s medical linear accelera-
The TG-21 report provides an analytical method to tor. However, the calibration of the ionization chamber
determine the dose to water with an ionization chamber is still performed in a 60Co beam quality, Q = 60Co pro-
Q = 60 Co
from an exposure-based calibration in air. Conceptually, viding an N D ,w . To convert between the calibration
TG-21 is a process outlining the theory of how to relate beam quality and the user’s beam quality, a kQ factor is
the kerma at a specified point in air to the charges col- employed
lected within an ionization chamber’s gas cavity to the
60 Co
expected quantity of dose in a medium of water. While N DQ,w = kQ ⋅ N D ,w (1.56)
this report offers a fundamental insight into the phys-
ics behind modern dosimetry, it is an outdated protocol. In parallel to the TG-21 protocol, a substantial amount
There are several assumptions that were made, which of information and physics is tied into the single kQ fac-
limits its achievable accuracy in the determination of tor. Factors including the absorbed dose differences,
absorbed dose to water to around 3–4%. However, bet- fluence perturbation from the presence of the chamber,
ter accuracies are desired in the practice of modern volume averaging of the chamber, and differences in
radiotherapy dosimetry of around or less than 2%. For radiation beam type have been calculated or measured
example, TG-21 does not provide specific corrections for several clinical chambers [13–15]. The analytical
for the influence of the central electrode on the inci- approximations that were made using cavity theory are
dent fluence, and there is a string disconnect between replaced with a Monte-Carlo-derived correction factor,
the beams of radiation used to calibrate the chamber kQ , of the form
response and the measurement of dose at the clinic. At a
calibration lab, the chamber response is calibrated from N DQ,w DwQ /MQ D Q M 60Co
a radiation field of 1.17 MeV and 1.33 MeV 60Co γ -rays, kQ = 60 Co = 60 Co = 60wCo ⋅ (1.57)
N D ,w Dw /M 60Co Dw MQ
whereas a medical linear accelerator at a clinic produces
a Bremsstrahlung distribution of 6 MV or higher energy
x-rays. Finally, TG-21 allows mediums other than water Equation 1.57 is composed of four elements. It is com-
to perform the measurement in, such as water-mimick- monly described in literature using the central form in
ing plastic phantoms. Equation 1.57 but is perhaps better described pedagogi-
The current absorbed dose-to-water protocol that is cally as the rightmost expression in Equation 1.57 using
practiced clinically is TG-51. This protocol is different a series of ratios. The first ratio relates the dose to water at
than TG-21 as TG-51 is completely water based, water- the point of measurement between both beam qualities.
mimicking plastics are not allowed. Ionization chambers The second ratio describes the response of the chamber,
are calibrated to an absorbed dose to water, not expo- M , which can be easily measured or simulated by tally-
sure, while also simplifying the absorbed dose-to-water ing the dose to the sensitive collection volume of air in
the chamber. Thus, each kQ factor is unique to a chamber additional correction term, PgrQ , is necessary to account
and beam. Large tables exist that have tabulated vari- for the nonuniform gradient across the chamber’s col-
ous kQ factors for specific ionization chamber make and lection volume. PgrQ is unity for plane-parallel chambers.
models used within a particular beam of radiation. kR50 is a product of two other factors considered uniquely
Beam quality is unique for each medical linear accel- for electron beams
erator as inherent difference exists among different
manufacturers of medical linear accelerators, the nomi- kR50 = kR′ 50 ⋅ kecal (1.59)
nal manufacturing dimensions of components, and the
room scatter conditions. Therefore, each linac’s beam Conceptually, the factor kecal is a photon-to-electron
quality must be experimentally measured based on conversion factor unique for each ionization chamber
the applied peak voltage and exiting particle type (i.e., model. The factor of kR′ 50 then corrects the product of
a photon or electron beam and the respective energy). kecal ⋅ N DQ,w to a specific electron beam quality. The depth
The beam quality for a photon beam with any medical of where an electron beam falls to 50% requires two steps.
linear accelerator is defined at the percent depth dose The initial depth measured with an ionization chamber
of 10 cm in water, also referred to as %dd (10 )x . Prior is referred to as the 50% depth ionization, I50 . This point
to measuring the percent depth dose of a photon beam, is adjusted to account for gradient effects in the beam to
the relative depth placement of the chamber must be achieve the depth dose at 50%, R50. Considering these
adjusted to implicitly correct the chamber’s inherent additional factors, the absorbed dose to water from an
perturbation of the charged particles produced in the electron beam is expressed as
water. This shift for a cylindrical chamber with radius
60 Co
rcav is approximated as 0.6 ⋅ rcav upstream for photon beam DwQelectron = MPgrQelectron kR′ 50 kecal N D ,w (1.60)
qualities – thus either the chamber should be displaced
downward before measurement or the depth dose curve The TG-51 addendum [16] estimates that the uncertainty
shifted upward after the measurement. in the measurement of %dd (10 )x by a clinical physicist is
The determination of a beam’s kQ factor using a spe- about 2%. This, in turn, corresponds to a 0.4% error in
cific ionization chamber from the measured %dd (10 )x the measured dose. Incorporating the uncertainties in the
will reflect the incident beam’s average energy. The Monte stopping powers and cross sections used for the Monte-
Carlo work of Muir and Rogers [13, 15] has investigated Carlo-derived kQ factors and expected uncertainties in
the relationship between the absorbed dose response of the chamber measurement and charge correction factors
several chambers among a multitude of medical linear results with a final uncertainty within the absorbed dose
accelerator brands and their producible beam energies. of less than 2% at the 95% confidence level.
Since the beam’s energy will have an influence on the
chamber’s response relative to its calibration conditions 1.6 CONCLUSION
in a 60Co beam, as shown in Equation 1.57, the physi- A substantial amount of work and effort has gone into
cist can correct this response using an analytical fit to developing the current absorbed dose-to-water protocol
the Monte Carlo data supplied in the TG-51 addendum that is based in a rich historical context reaching as far
[16] that has been normalized to the expected chamber’s back as the 1960s. What the modern-day medical physi-
response in 60Co. cist has inherited is a streamline process for determin-
Electron beams are also defined by a unique kQ facing a very complex and arbitrary quantity that is a major
tor, but the determination of an electron beam’s kQ is focus of their practice in medical physics. However, the
slightly different than a photon beam kQ factor. An elec- physicist should be warned that the simplicity and elo-
tron beam quality factor is determined from quence of the TG-51 formalism can be misleading as
so much of the fundamental physics is hidden. A pru-
kQ = PgrQ ⋅ kR50 (1.58) dent reader will observe that the TG-51 formalism is
not exhaustive; it is only well defined for a single set of
where kR50 is a chamber-specific factor defined by the reference conditions. Any changes to these specified
range that beam’s distal portion of its depth dose profile conditions will require us to look critically at how our
falls to 50%. If a cylindrical chamber is used, then an absorbed dose-to-water estimates could be affected from
the assumptions we have made, forcing us to rethink the 7. H. Bethe and W. Heitler, “On the stopping of fast particle
way we determine dose at the physical level. These influ- and on the creation of positive electrons,” Proceedings of
the Royal Society of London, A 146 (1934).
ences may manifest from changes from CPE conditions,
8. L. A. DeWerd and B. R. Smith, “Ionization Chamber
volume averaging effects, and how our detectors behave Instrumentation,” in Radiation Therapy Dosimetry:
from our initial calibration conditions, which will be A Practical Handbook, edited by A. Darafsheh (CRC
the focus of our attention later on. The modern role of Press, Boca Raton, FL, 2021), Chapter 2, pp. 19–30.
the practicing medical physicist is far from being con- 9. R. Schulz, P. Almond, J. Cunningham, J. Holt, R.
sidered ad nauseam. As the practice of medical phys- Loevinger, N. Suntharalingam, K. Wright, R. Nath,
G. Lempert, “Task Group 21, Radiation Therapy
ics expands, so must the knowledge and abilities of the Committee: A protocol for the determination of
physicist to solve tomorrow’s problems. absorbed dose from high-energy photon and electron
beams,” Medical Physics 10, 741–771 (1983).
REFERENCES 10. P. Almond, P. Biggs, B. Coursey, W. Hanson, M. Huq, R.
Nath, D. Rogers, “AAPM’s TG-51 protocol for clinical
1. R. Cox, J. Hendry, A. Kellerer, C. Land, C. Muirhead, reference dosimetry of high-energy photon and electron
D. Preston, J. Preston, E. Ron, K. Sanlaranarayanan, beams,” Medical Physics 26, 1847–1870 (1999).
R. Shor, R. Ullrich, Biological and Epidemiological 11. L. Gray, “An ionization method for the absolute mea-
Information on Health Risks Attributable to Ionizing surement of γ-ray energy,” Proceedings of the Royal
Radiation: A Summary of Judgements for the Purposes Society of London, A 156, 578–596 (1936).
of Radiological Protection of Humans. International 12. L. Spencer and F. Attix, “A theory of cavity ionization,”
Commission on Radiological Protection, Committee 1 Radiation Research, 3, 239–254 (1955).
Task Group Report: C1 Foundation Document (2005). 13. B. Muir and D. Rogers, “Monte Carlo calculations of
2. H. Staub, H. Bethe, J. Ashkin, N. Ramsey, Experimental kQ, the beam quality conversion factor,” Medical Physics
Nuclear Physics (John Wiley and Sons, Inc., New York, 37(11), 5939–5950 (2010).
NY, 1953). 14. M. McEwen, “Measurement of ionization chamber
3. F. Attix, Introduction to Radiological Physics and absorbed dose kQ factors in megavoltage photon beams,”
Radiation Dosimetry (WILEY-VCH Verlag GmbH & Medical Physics 37(5), 2179–2193 (2010).
Co., KGaA, 2004). 15. B. Muir, M. McEwen, D. Rogers, “Measured and Monte
4. E. Podgorsak, Radiation Physics for Medical Physicists Carlo calculated kQ factors: Accuracy and comparison,”
(Springer International Publishing, New York, NY, 2016). Medical Physics 38(8), 4600–4609 (2011).
5. W. Heitler, The Quantum Theory of Radiation (DOVER 16. M. McEwen, L. DeWerd, G. Ibbott, D. Followill, D.
Publications, INC., New York, NY, 1954). Rogers, S. Seltzer, J. Seuntjens, “Addendum to the
6. O. Klein and Y. Nishina, “On the scattering of radia- AAPM’s TG-51 protocol for clinical reference dosim-
tion by free electrons according to the new relativistic etry of high-energy photon beams,” Medical Physics 41,
quantum dynamics of Dirac,” Zeitschrift für Physik 52, 1–20 (2014).
853–868 (1929).
Chapter 2
Ionization Chamber Instrumentation

Larry A. DeWerd
Madison, Wisconsin
Blake R. Smith
University of Iowa
Iowa City, Iowa
CONTENTS
2.1 Introduction to Ionization Chambers 19
2.1.1 Ionization Chambers 20
2.1.2 Types of Ionization Chambers 20
2.2 Basic Components of Ionization Chambers 21
2.2.1 Thimble Chambers 22
2.2.2 Parallel-Plate Chambers 22
2.2.3 Spherical Chambers 23
2.2.4 Brachytherapy Well or Re-Entrant Chambers 23
2.3 Other Components of the Dosimeter System 23
2.3.1 The Triax Cable 23
2.3.2 Electrometers 24
2.4 Characteristics of Ionization Chambers 25
2.4.1 Beam-Quality Dependence (Energy Dependence) 26
2.4.2 Pressure/Temperature (Air Density) Effects 26
2.4.3 Recombination Effects 27
2.4.4 Polarity Effects 28
2.4.5 Leakage Currents 28
2.5 Conclusion 28
Acknowledgments 29
References 29
2.1 INTRODUCTION TO there is variation among individuals, this was a subjec-

IONIZATION CHAMBERS tive measurement technique. The erythema method
It was recognized shortly after the discovery of x-rays also involved risk to the patient if there was an excessive
that there was a need for a method to measure the expo- exposure. Today, radiation fields are usually quantified
sure from radiation. Initially, the measurement was done by the amount of ionization generated in air or in water
by observing radiation-induced erythema (reddening of under standard conditions. The application of theoreti-
the skin) on the individual receiving the exposure. Since cal principles describing the absorption of the radiation
19
with a series of correction factors allows the measured

radiation field to be related to the dose in human tis-
sue. This radiation measurement is the scientific field of
dosimetry and is the basis for much of medical physicists’
contribution to the use of radiation, especially in radia-
tion therapy. The process begins with the measurement
by an ionization chamber, which is a precise instrument.
An ionization chamber with a known volume may also
be accurate enough to be used as a standard in a primary
laboratory. Even without a known volume, the ioniza-
tion chamber is an essential device for the measurement
of radiation. The ionization chamber can be calibrated
to give an accurate and precise measurement. There are
three components involved for this type of measure-
ment device. The ionization chamber together with the
electrometer and connecting (triax) cable makes up the FIGURE 2.1 Voltage curve for a typical ionization chamber.
dosimetry system. Useful reviews on this topic can be
found in works by Attix, Podgorsak, and Andreo [1–3]. chamber, and the spherical chamber. These chambers
range in size to accommodate various radiation field sizes
2.1.1 Ionization Chambers and dose rates. Each chamber requires calibration to a pri-
Ionization chambers are a generalized class of instru- mary standard for good accuracy. Primary standards are
ments that also include proportional counters and used to calibrate ionization chambers for measurements
Geiger–Mueller (GM) counters. These instruments are of x-ray beams as well as megavoltage electron and pho-
primarily classified by the voltage applied to the cham- ton beams. Calorimetric standards are further explained
ber. As shown in Figure 2.1, the response of a generalized in Chapter 3 of this book. For medium- and low-energy
chamber is dependent on the voltage applied that gener- x-ray beams (<300 kVp), the primary standard is a free-air
ates an electric field to collect the ionization. The ioniza- chamber. For megavoltage photon and electron beams,
tion chamber region that is the topic of interest for the the quantity measured is absorbed dose to water. Within
rest of this chapter is the ion chamber region or the satu- the United States, a 60Co beam, air kerma, and absorbed-
ration region, or the range of voltages such that the initial dose standards are available. Air kerma is measured with
charges released in the chamber are collected with near a chamber that has a known volume and wall composi-
100% efficiency (the plateau in Figure 2.1). As the voltage tion (e.g., graphite). For absorbed dose to water, the stan-
increases, the proportional counting region is reached. dard is a water calorimeter.
At voltages in and above this level, specialized gases are The ionization chambers commonly used clini-
generally used. In the proportional counting region, the cally are cavity ionization chambers, which are well-
initial charge released is accelerated enough to create characterized by Bragg–Gray/Spencer–Attix cavity
additional ionization, but the total charge collected at a theory. These chambers consist of a known solid mate-
given voltage is still proportional to the incident radia- rial, called the wall, or shell, containing an air volume.
tion dose. As the voltage is continually increased, the Standard materials used for the wall are C552 (air equiv-
characteristics change to the point of continuous dis- alent material) or graphite. Thus, the effect of the wall
charge, or a pulse of current, for each radiation event; necessitates a correction factor known as Kwall for in-
this is called the GM region. There are a number of char- air measurements or Pwall for measurements made in a
acteristics involved in each particular chamber that need phantom. Generally, Kwall is determined via Monte Carlo
to be considered for a dosimetry system. computational methods, since experimental techniques
have been shown to produce incorrect results [4]. If the
2.1.2 Types of Ionization Chambers chamber is used in air, the walls need to be thick enough
There are three main ionization chamber designs for to provide full buildup, which is a thickness of mate-
radiation dosimetry, the thimble chamber, parallel-plate rial equal to the range of an electron for a given energy.
Ionization Chamber Instrumentation ◾ 21
In most cases, a build-up cap is placed on the chamber, a thimble chamber connected to an electrometer via
and the cap combined with the chamber wall provides a triaxial cable (triax) is shown in Figure 2.2. An elec-
full buildup. The build-up cap has the added advantage tric field is established between the ionization chamber
of protecting the thimble. components in order to stabilize the ionization collected
Manufacturer tolerances can affect variations in the in the air of the cavity. The electric field is typically gen-
volume of the cavity, and thus the response of the cham- erated by a high-voltage power supply or battery within
bers is affected. The size of the signal received for a given the readout device (electrometer). The schematic of
radiation dose is related to the volume of the cavity. Figure 2.2 is a charge measuring setup; for current mea-
Therefore, a given chamber can be precise in its response, surement, the capacitor is replaced by a resistor. Further
but the signal magnitude can vary among a given details on the electrometer and the triaxial cable are
model. Thus, each chamber needs calibration related to given later. Electrical insulation is provided between
a primary standard. The correction of the readout, via each of the ionization chamber components by materi-
a calibration coefficient, relates the measurement value als such as polyethylene or PTFE®. The purpose of these
to the conventional true value of a primary standard. insulators is to support the ionization chamber com-
The calibration coefficient relates the charge reading, ponents structurally and to prevent leakage currents.
in coulombs, to the radiological quantity, air kerma or However, even when effective insulators are used, leak-
absorbed dose to water. This calibration coefficient is age currents may occur when dirt, a hair, or some other
determined at an American Association of Physicists in type of fiber bridges the insulator. Skin oil is a significant
Medicine Accredited Dosimetry Calibration Laboratory cause of leakage; thus, touching an insulator should be
(AAPM ADCL) by an intercomparison of a reference avoided. In general, the signal to noise for a given cham-
standard chamber with the chamber being calibrated. ber system always needs to be considered. Ideally, the
In places other than North America, chambers are cali- signal-to-noise ratio should be ≥1000.
brated by a Secondary Standards Dosimetry Laboratory Many of the less expensive electrometers, and those
(SSDL) or a primary laboratory, and Technical Reports no longer meeting original specifications, can have high
Series (TRS) 398 [5] is used. The density of the contained leakage around 10 −13 A [6]. Thus, if the signal is from
air is always corrected to a standard temperature and a small volume chamber, then the ideal signal-to-noise
pressure. ratio of 1000 may not be achievable. The leakage should
be subtracted, or the system should be repaired when
2.2 BASIC COMPONENTS OF the noise becomes a significant portion of the measured
IONIZATION CHAMBERS signal.
The components of an ionization chamber are the shell The guard of a typical ionization chamber has two
(also called the wall, thimble, or window), the guard, main purposes: it electrically shields the collector cur-
and the collecting electrode (collector). A schematic of rent through the triaxial cable to the electrometer, and
FIGURE 2.2 Schematic of a thimble chamber connected to a charge reading electrometer (a feedback coulombmeter).
it defines the shape of electric field in the ionization guard, and the outer shell or thimble) are connected
chamber collecting volume. In addition, since the guard through the triaxial cable to an electrometer. The col-
is at the same potential as the high-impedance collector, lector of the chamber delivers the current resulting from
it provides a low-impedance path for leakage current, ionization of the air volume in the chamber to an elec-
thus reducing the leakage current collected. Improved trometer device. The electrometer registers the current
electric field uniformity in the sensitive volume of the or the charge collected in a given time period. The elec-
chamber is also provided by the guard. Good field trometer also supplies the high voltage that holds the col-
uniformity can minimize fluence perturbations when lector at this potential. When a standard voltage of 300 V
measurements are performed in electron beams or is applied between the guard and the thimble, the electric
pulsed beams. The volume is determined by the guard field lines remain uniform across the chamber, defining
and is more uniformly defined if the chamber is “fully the collecting volume. The thimble wall contains the air
guarded.” A chamber is fully guarded when the termi- volume. Any ionization occurring in the collecting vol-
nation of the guard extends into the air volume of the ume of the chamber results in the positive and negative
chamber. A partial or minimal guard has its terminus at ions being pulled either to the collector or to the thimble
the beginning of the air volume or prior to the air cav- wall. Most often, the collector voltage is positive with
ity. The extent and quality of the guard may cause a dif- respect to the outer shell and results in the collection of
ference between the physical and effective volumes of a negative charge. There are corrections necessary for the
chamber. Voltage differences between guard and collec- operation of ionization chambers. A typical volume used
tor [7] or dead spots can occur in the field near the cor- for Farmer chambers is 0.6 cm3, although there is a range
ners where the electric field lines from the collector and of volumes as small as 0.007 cm3.
guard come together. Dead spots also can occur where
the cone meets the cylinder in the classic, Farmer-type 2.2.2 Parallel-Plate Chambers
chamber. Thus, the design of the fully guarded cham- In a parallel-plate ionization chamber, an electric field
ber becomes important for constancy of the volume and is generated between the window and the guard and the
determination of the charge collected. A rule of thumb window and the collector. The cylindrical separation of
is that the width of the guard should be greater than the collector and the window defines a collecting volume
or equal to three times the width of the gap for a fully that is filled with air. Figure 2.3 is a schematic of this
guarded parallel-plate chamber. type of chamber. The guard electrode, which provides a
Although all ionization chambers have the same uniform electric field and a defined volume, is a donut-
essential components, chambers are readily defined by shaped ring in the plane of the collector with a high
their geometric differences. Whether thimble, parallel- voltage applied between the guard and the inner surface
plate, or spherical, the size of the chamber signal for of the window. Thus, the collector will collect charge
a given radiation dose is related to the volume of the from all the ions that are generated in the gas between
cavity. Thus, the volumes of the different geometrical the plates. Generally, these chambers have a thin win-
shapes become important since the volume of the col- dow usually composed of a few microns of conductive
lecting region in the chamber generally determines the
response of the chamber.
2.2.1 Thimble Chambers

The thimble-type chamber can be used as a model to
describe the essential elements of an ionization cham-
ber as well as the characteristics of chambers. Figure 2.2
shows a schematic of a thimble chamber connected to
the electrometer through a triaxial cable. The thimble
chamber was developed by Farmer [8] (Aird and Farmer,
1972) and is commonly called a Farmer chamber. It has
been a standard ionization chamber used by many clin- FIGURE 2.3 Schematic of parallel-plate chamber with field
ics. The electrodes of the chamber (the collector, the lines shown.
Mylar®, Kapton®, or other thin, conductive material. The beams in the United States. (A known volume cylindri-
window would be oriented toward the beam. Most mea- cal chamber is also used.) Historically, spherical cham-
surement protocols recommend parallel-plate ioniza- bers have been used as air-kerma reference chambers in
tion chambers for electron beams with energies below the clinic. Similar characteristics as given for thimble
10 MeV [9]. The plate separation for most of the par- chambers also apply to spherical chambers. A typical
allel-plate ionization chambers used for electrons falls volume for spherical chambers is 3.6 cm3. These cham-
between 1 and 2 mm, resulting in a negligible change bers can become very large, up to 15,700 cm3.
in beam intensity across the sensitive volume. The small
plate separation gives better spatial and depth resolution 2.2.4 Brachytherapy Well or Re-Entrant Chambers
than cylindrical chambers in beams with a large gradi- Brachytherapy is the use of radioactive sources inserted
ent intensity along the beam axis. Commonly, a paral- into body cavities or into body tissue. Radiation from
lel-plate chamber is also the type of ionization chamber brachytherapy sources is measured using a well cham-
used for diagnostic exposure measurements. However, ber that has its collector between two ground plates in a
the volume of the chamber in this case must be larger cylindrical configuration and a guard at the bottom of
since the dose rate is lower. Parallel-plate ionization the chamber. This type of chamber is elaborated upon
chambers are calibrated with their plates oriented per- in the AAPM 2009 summer school [10]. The radioactive
pendicular to the beam axis and their window toward source is inserted into the center of the cylinder, referred
the x-ray beam. For diagnostic radiological applications, to as a well. The strength of a brachytherapy source is
the electrometer reading is converted from coulombs specified as the air-kerma rate that the source would
to an air kerma value measured in units of gray (Gy) produce at one meter in vacuum, called the air-kerma
[joules per kilogram (J/kg)], roentgens (R), or coulombs strength, SK. A complete discussion on the clinical appli-
per kilogram (C/kg). Again for parallel-plate chambers cation of these chambers for brachytherapy dosimetry is
there is a range of volumes, with 0.62 cm3 being used provided in Chapter 15 of this book [11].
frequently, and a range of chamber diameters, with
20 mm as an average. 2.3 OTHER COMPONENTS OF
THE DOSIMETER SYSTEM
2.2.3 Spherical Chambers Besides the ionization chamber, the system is composed
Spherical chambers (such as the one shown in Figure 2.4) of the electrometer and the triax cable. Each of these
with precisely known volumes are used for the primary components has its own characteristics that affect the
standard determination of air kerma for 60Co or 137Cs readout.
2.3.1 The Triax Cable

The ionization chamber current is transferred to the
electrometer for measurement via a low-noise triaxial
cable. A schematic of a triaxial cable is shown in Figure 2.5.
A low-noise triaxial cable has insulating qualities that
reduce electrical noise from triboelectric and piezo-
electric effects in the cable. Triboelectric currents are
generated by charges that are created at the interface
FIGURE 2.4 Schematic of a spherical chamber with field

lines shown. FIGURE 2.5 Schematic of a triaxial cable.
between a conductor and an insulator due to friction process to the feedback capacitor and is related by the
when the cable is bent or flexed. Piezoelectric currents capacitance definition shown in Equation 2.1:
are generated when mechanical stress is applied to cer-
tain insulating materials. Minimizing the extraneous T
1
small signals is of the greatest importance while main-
taining applicability. Flexibility of the cable is one of the
Q = CV or V =
C ∫
I (t )dt = I × T /C , if constant current
0
properties required for convenience. The cable should (2.1)
be positioned in a relaxed state, avoiding twisted coils
and sharp bends that induce mechanical stress. Trauma where
can break the delicate collecting electrode or degrade Q = charge in coulombs
the insulating qualities. The guard connector provides C = capacitance of feedback element in farads
a contiguous guard throughout the length of the cable. V = voltage across feedback capacitor = VOUT
Connections for the cable have to be secure with good I = ionization current in amperes
insulation between the layers. Also, the cleanliness of T = integration time in seconds (exposure time)
the connectors is of extreme importance since one of the
major causes of leakage is a dirty connector. Generally, This coulombmeter will require feedback capacitor val-
good-quality cables and connectors have very low leak- ues from 0.001 to 1.0 μF depending on the magnitude
age (≤10 −15 A, when 300 V is applied). A low capacitance of charge. Different capacitor values are connected into
per meter is also desirable for a good triaxial cable. the circuit by a rotary switch or by an external plug-in
Low-capacitance cables allow ionization chambers to module.
have a fast equilibration time following any change in For current measurements, an operational amplifier
an applied voltage. This fast equilibration time prevents is configured as an inverting amplifier with a feedback
continually increasing or decreasing readings after a resistor. So, replacing the capacitor of Figure 2.2 with
change in voltage. a resistor, the amplifier output voltage is a measure of
input current related by Ohm’s law:
2.3.2 Electrometers
The current produced by ionization chambers for medi- V = IR (2.2)
cal physics applications is typically between 0.5 pA and
800 nA. To measure and display such a low-magnitude where
current, or to integrate the current over time to dis- V = voltage across feedback resistor = VOUT
play a charge collected, requires an instrument with an I = ionization current in amperes
extremely high input impedance. Starting in the 1970s, R = resistance of feedback resistor in ohms
electrometers began to be manufactured to meet the
requirements of the medical physicist and have been This ammeter will require feedback resistors from 107
refined to include more capabilities and features. As to 1012 ohms, depending on the magnitude of current.
mentioned, an electrometer is a high impedance mea- Different resistor values are connected into the circuit
suring device for which low currents or charges are by a rotary switch, relay, or external plug-in modules.
detected such that the device itself does not affect the The particular feedback element that is selected in either
measurement. There are two basic circuits, one or both of the two circuits represents a specific “scale” or range
of which are found in the preamplifier (front end) of of measurement values. Zero drift current and leakage
every electrometer. Both circuits are impedance concurrent values are generally measured only for a charge
verters in that a high-impedance input is converted to scale calibration of an analog electrometer (coulombme-
a low-impedance output. Low-level measuring tech- ter). Most electrometers have the capability of nulling
niques are described in the Low Level Handbook [12]. To out display offsets caused by the background current.
measure charge, an operational amplifier is configured This nulling, or zeroing, process should be performed
as an inverting amplifier with a feedback capacitor, as with the cable and chamber attached, the HV (high-
illustrated in Figure 2.2. The amplifier output voltage voltage) bias on, and the display stabilized. All elec-
is a measure of charge transferred from the ionization trometers need a “warm-up” time when they are first
turned on, so do not zero until after the warm-up has 300 V is the typical high-voltage bias magnitude. In
completed. Zero drift current is obtained by first zeroing Figure 2.2, the guard (and collector) is +300 V with
the charge scale display and then observing the change respect to the outer shell. This perspective is commonly
in reading during a measurement interval. Electrometer referred to as the “collecting electrode potential.” Some
leakage current is determined by injecting a half-scale electrometer manufacturers reverse the reference point,
charge value. The source of the charge is then removed, indicating the outer shell potential with respect to the
and the decay rate of the reading is determined. A good guard. This perspective is commonly referred to as the
electrometer should have less than 0.1% change in the “thimble bias.” Thus, the thimble bias shown in Figure 2.2 is
reading over the period of one minute. High humidity −300 V. The high-voltage bias configuration in Figure 2.2
may increase leakage in some electrometers. will produce a negative charge collection. Unfortunately,
In what can be classified as a “digital” electrometer, the the manufacturers do not always label the bias polarity
preamplifier voltage is digitally manipulated. For exam- selector switch to indicate whether it is the collecting
ple, a “charge digitizer” electrometer has a coulombme- electrode potential or thimble bias. There may even be
ter preamplifier and digitizes the signal by producing an inconsistency between models from a particular manu-
output pulse for a specific value of input charge (charge facturer. In addition, some electrometer models display
bucket). Each time VOUT reaches a certain level, the ampli- the opposite polarity of the measured charge or cur-
fier is reset, and a pulse is counted. The unit quantifies the rent. Generally, the high-voltage bias is measured at the
charge collected by counting the number of pulses and input connector and reported as the guard voltage with
multiplying by the charge bucket size. The charge mea- respect to the outer shell for each corresponding bias
surement resolution is determined by the bucket size. The selector setting of the electrometer.
digital electrometer measures the current by dividing the The scales and lowest reading capability of an elec-
measured charge by the time interval. trometer are of great importance. Scale selection should
Another type of digital electrometer is a “numerical ideally provide a reading of four significant digits so that
current integrator” that incorporates the transresistance the reading precision is always within 0.1%. Therefore,
amplifier configuration. Current is measured by sam- if the reading is a picoampere, the last digit should be
pling and averaging the preamplifier output. Collected a femtoampere. The lowest end of the measuring range
charge is calculated by a microprocessor, which numeri- should be at least 200 times larger than the digital res-
cally integrates the current. Scale selection determines olution, which is the smallest significant increment of
the feedback resistance and the computing algorithm. the reading. An electrometer may not display the “true”
Note that each scale of this type of digital electrometer charge or current value. Thus, a “scale-specific” calibra-
is rate limited. Thus, for a given charge scale, the maxi- tion coefficient must be applied to obtain the correct
mum current, rather than the maximum charge read- measurement value. A charge scale electrometer cali-
ing, is usually the limiting factor. bration coefficient (Pelec) has units of coulomb/reading
The electrometer also applies the high-voltage bias to (C/rdg), where “rdg” represents the units of the read-
the ionization chamber through the triaxial cable. As out display. A current scale Pelec has units of ampere/
illustrated in Figure 2.2, the high-voltage supply is con- reading (A/rdg). Although the calibration coefficient
nected between the outer shield of the triaxial connector is often 1.000 C/rdg or 1.000 A/rdg, it should not be
and the guard. With the inverting feedback configura- assumed a priori to be unity.
tion of the preamplifier, the inverting input (collector)
is actively forced to maintain equal potential with the 2.4 CHARACTERISTICS OF
noninverting input (guard). Thus, the collector, guard, IONIZATION CHAMBERS
and all other circuitry “common” to the guard “float” Ionization chambers require some corrections that are
at the high-voltage bias level with respect to the outer related to the basic nature of the chamber. These cor-
shield. If the collecting electrode and the outer shield are rections are made to the measurement value. The ion-
shorted together, the input circuitry of the electrometer ization chamber should be calibrated by an ADCL to
can be destroyed. determine a calibration coefficient, ND,w . The equation
Properly defined voltage requires an indication of used to determine dose for therapy applications (DwQ) is
the magnitude and the polarity. In the United States, given in Task Group 51 (TG-51) [9, 13]. The TG-51 dose
equation is given in Equation 2.3 and Mraw is listed in of beam quality (energy) [14], since a measurement of
Equation 2.4, with all of the corrections required for the half value layer can be affected by the variation in air-
ionization chamber. In addition, the electrometer should kerma sensitivity [15].
also be calibrated and is assigned a calibration coefficient,
Pelec. This factor, Pelec, is usually close to 1.000 C/rdg, but it 2.4.2 Pressure/Temperature (Air Density) Effects
can vary from 0.995 to 1.005, or more, from unity. In some The number of ions collected, or the rate of their collec-
countries, chamber and electrometers are calibrated as a tion in the gas of ionization chambers, is the signal that is
single system. For diagnostic radiology measurements, recorded, and this depends on the amount of gas, usually
the electrometer and chamber are generally calibrated as air, in the volume. The calibration coefficient is stated at
a system to read out directly in air kerma or exposure. a standard temperature and pressure (usually 101.3 kPa
The TG-51 dose-to-water formalism is (760 mm Hg) for pressure and 22°C in North America
but 20°C in most of Europe). The density of the air in the
60 Co
DwQ = MkQ N D ,w , and (2.3) chamber must be corrected to these standard conditions
during the measurement, since the calibration coefficient
M = Mraw PT , P Pion Ppol Pelec Pleak PrP (2.4) assigned to an ionization chamber specifies the charge
collected in the chamber per unit quantity of radiation
where in absence of the chamber. The collecting volume of the
kQ is the conversion from cobalt calibration energy to chamber is often open to the outside atmosphere. Thus,
the energy of the linac the mass of air is dependent on conditions of tempera-
60 Co
N D ,w is the chamber calibration coefficient deter- ture, pressure, and humidity. The temperature and pres-
mined at cobalt energy sure should be measured in the clinic without reliance on
Mraw is the raw reading of the chamber outside values. The paper by Rogers and Ross [16] deals
PTP is the temperature – pressure or gas density cor- with humidity effects and shows that the humidity effect
rection factor can be considered constant within ±0.15% over a relative
Pion is the ion recombination correction factor humidity range of 30–80%. On the other hand, tempera-
Ppol is the polarity correction factor ture and pressure effects can be significant since the mass
Pelec is the electrometer correction factor of air in the chamber depends on these atmospheric con-
Pleak is the correction factor from any contribution to ditions. The equation (Equation 2.5) expressing this cor-
the measured reading that is not due to ioniza- rection, PTP, is derived from the ideal gas law (see [17]).
tion in the chambers collecting volume The temperature and pressure must be obtained at the
PrP is the correction factor accounting for any off-axis time of measurement. The correction for temperature and
variation in the intensity profile of the radiation pressure is given by:
field over the sensitive volume of the ionization
chamber. For further discussions of this correc-  T +T   P 
PTP =  a 0   S  (2.5)
tion factor, the reader is referred to the TG-51  TS + T0   Pa 
addendum [13].
where
2.4.1 Beam-Quality Dependence Pa is the ambient pressure in kPa
(Energy Dependence) PS is the standard pressure (101.3 kPa)
The response of ionization chambers can vary with the Ta is the ambient temperature in °C
energy of the beam, which is accounted for by the beam T0 is the temperature in kelvin at 0°C (273.15 K)
quality correction, kQ. Values for the kQ are given in the TS is the standard temperature (22°C in North
addendum [13] or in the TG-51 protocol [9]. This energy America)
dependence is also important for electron beams, ortho-
voltage and superficial beams (x-ray beams), and for Higher altitudes have a greater PTP correction due to the
diagnostic x-ray energies. The major requirement for lower pressure. It is common to have corrections of more
some diagnostic applications is that the air-kerma sensi- than 10% in mountainous areas. When a calibrated
tivity of the chamber varies by less than 3% as a function device is used, the physicist must correct the reading
using Equation 2.5 to that which would be obtained at rate, but the occurrence of general recombination will
standard temperature and pressure. However, for cer- increase with increasing dose rate.
tain chambers with non–air-equivalent components, Scott and Greening [28] demonstrated that for a
the standard correction breaks down for low-energy continuous radiation beam (e.g., 60Co or kilovoltage
radiation [17–20]. This breakdown occurs when the x-rays), a linear relationship exists between the inverse
ranges of the electrons entering the chamber cavity of the collected charge and the inverse of the electric
are short with respect to the dimensions of the cavity, field strength squared when general recombination
and when the photon cross sections differ between the was the dominant recombination process. Boag and
chamber wall and air. La Russa and Rogers [17] dem- Currant [23] presented a “two-voltage” technique for
onstrated such an effect for thimble-type chambers in determining the general recombination for pulsed
kilovoltage x-rays using Monte Carlo calculations and beams (e.g., linear accelerators), and Weinhous and
later confirmed the results with an experimental inves- Meli [29] presented a numerical method to evaluate
tigation [20]. The investigators found a smaller devia- the general recombination correction for these beams
tion than observed for well-type ionization chambers, using the two-voltage technique. For measurements
with a measured 2.5% effect at 0.7 atm for an unfil- in pulsed beams close to saturation, the relationship
tered 60 kV x-ray beam with a graphite-walled thimble between the inverse of the collected charge versus the
chamber. An effect of 13% was measured at an air den- inverse of the electric field strength is close to linear,
sity typical of Mexico City (0.76 atm) for a modified so the correction from Weinhous and Meli was simpli-
thimble chamber with an aluminum wall, using the fied in the TG-51 report [9] to using a half voltage. The
same 60 kV x-ray beam. A chamber with an aluminum underlying assumption for the two-voltage technique
wall is not common and was used only to demonstrate is that the chamber is behaving according to Boag
the effect. theory [23]. Equation 2.6 shows the relation as recom-
mended in TG-51,
2.4.3 Recombination Effects
Ion recombination effects occur in ionization cham- V 
1−  H 
bers when ions of opposite charge recombine before  VL 
they reach the collecting electrode or chamber wall. Pion (VH ) = (2.6)
 M raw   VH 
H
This leads to an incomplete charge collection, and a  M L  −  V 
raw L
correction should be applied for it. Early research in
this area was performed by Boag [21, 22], and it has where VH and VL are the high and low voltages, respec-
been an area of continued interest for clinical radiation tively, and M raw
H and M L are the readings at the high
raw
dosimetry (e.g., [7, 23–26]). The recombination effects and low voltages, respectively. The TG-51 report states
can affect small-volume chambers in a significant way that for a voltage ratio of 2 and a value of Pion less than
[7]. Three mechanisms are involved in ion recombina- 1.05, Equation 2.6 should be within 0.2% of the exact
tion, and all of these have a dependence on the elec- equation for pulsed beams and within 0.4% for pulsed
tric field strength in the ionization chamber. The first swept beams. Since the measured charge is related to the
is initial recombination, in which oppositely charged square of the electric field (or voltage) for continuous
ions from the same charged particle track meet and beams, the general recombination correction for those
recombine. This mechanism is generally only impor- beams is
tant for high linear energy transfer (LET) radiation
such as alpha particles or for gases at high pressure [1, 2
V 
21]. The second mechanism is ionic diffusion, in which 1−  H 
 VL 
ions diffuse to the collecting electrode or chamber wall Pion (VH ) = 2 . (2.7)
H  V 
 M raw
against the electric field [27]. The third mechanism is H
general (or volume) recombination where oppositely  M L  −  V 
raw L
charged ions arising from different charged particle
tracks meet and recombine. Initial recombination Ionization chambers used for radiation therapy usually
and ionic diffusion have no dependence on the dose use an applied voltage of ±300 V, and the value of Pion
can be up to 5% depending on the type of chamber and

Ppol =
( Mraw raw )
+ − M−
(2.8)
dose rate. Larger values of Pion can indicate a malfunc- 2 M raw
tioning ionization chamber and/or electrometer.
Recombination effects can be reduced by increasing
the rate of collection of the ionization products pro- where
duced in the chamber. This collection rate is dependent M raw
+ is the reading when positive charge is collected
on the mobility of the ionization products and the elec- M raw

− is the reading when negative charge is collected
tric field strength. A common method to reduce recom- Mraw is the reading corresponding to that polarity
bination effects is to increase the applied potential so generally used in the clinic
the chamber operates nearer to the saturation region
(Figure 2.1). The chamber will asymptotically approach The type of charge normally collected should be the
full saturation (complete charge collection) with same as for the chamber calibration. Typical polarity
increasing applied potential, but eventually this can lead corrections should be between 0.997 and 1.003 (i.e.,
to electrical breakdown of the chamber components or within ±0.3% of unity). The TG-51 protocol recom-
gas multiplication in the air. Sometimes, depending on mends that if users measure a polarity correction out-
the chamber volume, the voltage should be reduced to side this range for a photon beam of 6 MV or lower
be able to operate in the ion chamber region. It is always energy, they should determine the polarity correction
important to measure at a number of voltages from low for the 60Co calibration beam and apply the appropri-
to high to be sure the operation of the chamber is in the ate correction.
plateau of the ion chamber region.
2.4.5 Leakage Currents
2.4.4 Polarity Effects The leakage current for the system is measured using the
Ionization chamber polarity effects refer to the differ- same measurement setup with the accelerator but with-
ence observed when negative charge versus positive out the beam on. Although leakage may originate from
charge is collected. Polarity differences can occur when any component (chamber, electrometer, or cable), leak-
radiation interactions in the collecting electrode of the age generally originates in the cable. If a high leakage is
ion chamber add or subtract from the measured signal. measured, it is suggested to change cables and check it
This effect is known as the Compton current. Current again. The leakage should contribute to less than 0.1%
arising from ionizations outside the defined collecting of the signal. At this level or less, leakage is generally
volume, known as extracameral current, can also lead ignored and Pleak is set to 1.000.
to polarity effects. Kim et al. [30] have associated polar-
ity effects with electric field distortions due to poten- 2.5 CONCLUSION
tial differences between the guard electrode and the As has been explained in this chapter, a good under-
collecting electrode. In general, this effect is larger for standing of ionization chambers and the components
parallel-plate chambers than for thimble chambers and involved in the dosimetry system allows the clinical
depends on the type of irradiation (i.e., photon beam or physicist to use them correctly to determine the dose
electron beam) as well as the depth of measurement [31]. delivered to the patient. Calibration of the ionization
A correction for this effect is determined by perform- chambers and electrometers should be done every
ing measurements with positive and negative voltages 2 years so that they are traceable to a primary laboratory;
equal in magnitude. The correction is then determined this could be through an AAPM-accredited ADCL. The
using the equation from the TG-51 protocol [9]. If suf- clinical physicist should always be aware of the condi-
ficient time for the chamber to reach equilibrium after tion of their cable. Special care needs to be taken so that
changing voltage is not allowed, the polarity correction the cable is not kinked, stepped on, or met with other
would be greater than the chamber actually has, leading trauma. Such trauma is the major source of leakage cur-
to an incorrect measurement. The first polarizing volt- rent. Various protocols have been developed and should
age measurement should be repeated to correct for any be understood for determinations of dose as well. This
chamber drift. The polarizing correction is then calcu- chapter has followed TG-51 for the most part, but TG-51
lated by Equation 2.8: is similar to the IAEA TRS 398. The protocols spell out
all the corrections that should be applied for the cham- 12. Keithley Instruments, Inc., 2016. Low Level Measurements
ber use. Handbook: Precision DC Current, Voltage, and
Resistance Measurements, 7th edition (Keithley/
Tektronix, Cleveland, OH, 2016).
ACKNOWLEDGMENTS 13. M. McEwen, L. DeWerd, G. Ibbott, D. Followill, D.
The authors would like to thank members of the O. Rogers, S. Seltzer, J. Seuntjens, “Addendum to the
Medical Radiation Research Center at the University AAPM’s TG-51 protocol for clinical reference dosim-
of Wisconsin for their support and the customers of etry of high energy photon beams,” Medical Physics 41,
041501-1–041501-20 (2014).
the center who provided research support for gradu- 14. L. A. DeWerd and L. K. Wagner, “Characteristics of
ate students. L.A. DeWerd also has a partial interest in radiation detectors for diagnostic radiology,” Applied
Standard Imaging, Inc. Radiation and Isotopes 50, 125–136 (1999).
15. L. A. DeWerd, J. A. Micka, R. W. Laird, D. W. Pearson,
REFERENCES M. O’Brien, P. Lamperti. “The effect of spectra on cali-
bration and measurement with mammographic ioniza-
1. F. H. Attix, Introduction to Radiological Physics and tion chambers,” Medical Physics 29, 2649–2654 (2002).
Radiation Dosimetry (John Wiley and Sons, Inc., New 16. D.W. O. Rogers and C. K. Ross, “The role of humid-
York, NY, 1986). ity and other correction factors in the AAPM TG-21
2. E. B. Podgorsak (ed), Radiation Oncology Physics: A dosimetry protocol,” Medical Physics 15, 40–48 (1988).
Handbook for Teachers and Students (International 17. D. J. La Russa and D. W. O. Rogers. (2006). “An EGSnrc
Atomic Energy Agency, Vienna, Austria, 2005). investigation of the PTP correction factor for ion
3. P. Andreo, D. T. Burns, A. E. Nahum, J. Seuntjens, F.H. chambers in kilovoltage x rays,” Medical Physics 33,
Attix, Fundamentals of Ionizing Radiation Dosimetry 4590–4599.
(Wiley-VCH, Germany, 2017). 18. S. L. Griffin, L. A. DeWerd, J. A., Micka, T. D. Bohm,
4. D. W. O. Rogers and A. F. Bielajew, “Wall attenuation “The effect of ambient pressure on well chamber
and scatter corrections for ion chambers: Measurements response: Experimental results with empirical correc-
versus calculations,” Physics in Medicine & Biology 35, tion factors,” Medical Physics 32(3), 700–709 (2005).
1065–1078 (1990). 19. T. D. Bohm, S. L. Griffin, P. M. DeLuca, Jr., L. A. DeWerd,
5. IAEA, “Absorbed Dose Determination in External “The effect of ambient pressure on well chamber response:
Beam Radiotherapy,” Technical Reports Series No. 398 Monte Carlo calculated results for the HDR 1000 Plus,”
(International Atomic Energy Agency, Vienna, Austria, Medical Physics 32(4), 1103–1114 (2005).
2000). 20. D. J. La Russa, M. McEwen, D. W. O. Rogers. (2007).
6. L. DeWerd and R. Mackie, “Comment on Comparison “An experimental and computational investigation of
of ionization chambers of various volumes for IMRT the standard temperature-pressure correction factor for
absolute dose verification [Med Phys 30: 119–123 (2003). ion chambers and kilovoltage x rays,” Medical Physics
Letter to the Editor],” Medical Physics 30, 2264 (2003). 34, 4690–4699.
7. J. Miller, R. B. D. Hooten, J. A. Micka, L. A. DeWerd, 21. J. W. Boag, “Ionization measurements at very high
Polarity effects and apparent ion recombination in intensities,” The British Journal of Radiology 23, 601–611
microionization chambers, Medical Physics 43, 2141– (1950).
2152 (2016). 22. J. W. Boag and T. Wilson, “The saturation curve at high
8. E. G. A. Aird and F. T. Farmer, “The design of a thimble ionization intensity,” British Journal of Applied Physics
chamber for the Farmer dosimeter,” Physics in Medicine 3, 222–229 (1952).
& Biology 17, 169–174 (1972). 23. J. W. Boag and J. Currant, “Current collection and ionic
9. P. R. Almond, P. J. Biggs, B. M. Coursey, W. F. Hanson, recombination in small cylindrical ionization cham-
M. S. Huq, R. Nath, R. D. W. O. Rogers, “AAPM’s bers exposed to pulsed radiation,” The British Journal of
TG-51 protocol for clinical reference dosimetry of high Radiology 53, 471–478 (1980).
energy photon and electron beams,” Medical Physics 24. J. W. Boag, E. Hochhäuser, O. A. Balk, “The effect of
26(9), 1847–1870 (1999). Also available as AAPM free-electron collection on the recombination correc-
Report No. 67. tion to ionization measurements of pulsed radiation,”
10. D. W. O. Rogers and J.E. Cygler (ed), Clinical Dosimetry Physics in Medicine & Biology 41, 885–897 (1996).
Measurements in Radiotherapy, American Association 25. C. Zankowski and E. B. Podgorsak, “Determination of
of Physicists in Medicine, Monograph 34, DeWerd, et al. saturation charge and collection efficiency for ioniza-
Chapter 6, pp. 181–204 (2009). tion chambers in continuous beams,” Medical Physics
11. C. L. Deufel, W. S. Culberson, M. J. Rivard, F. Mourtada, 25, 908–915 (1998).
“Brachytherapy dosimetry,” in Radiation Therapy 26. F. DeBlois, C. Zankowski, E. Podgorsak, “Saturation cur-
Dosimetry: A Practical Handbook, edited by A. Darafsheh rent and collection efficiency for ionization chambers in
(CRC Press, Boca Raton, FL, 2021), Chap. 15, pp. 231–252. pulsed beams,” Medical Physics 27, 1146–1155 (2000).
27. J. Böhm, “Saturation corrections of plane parallel ion- 30. Y.-K. Kim, S.-H. Park, H.-S. Kim, S.-M. Kang, J.-H, Ha,
ization chambers,” Physics in Medicine & Biology 21, C.-E. Chung, S.-Y, Cho, J. K. Kim, “Polarity effect of the
754–759 (1976). thimble-type ionization chamber at a low dose rate,”
28. P. B. Scott and J. R. Greening, “The determination of Physics in Medicine & Biology 50, 4995–5003 (2005).
saturation currents in free-air ionization chambers by 31. B. J. Gerbi and F. M. Khan, “The polarity effect for com-
extrapolation methods,” Physics in Medicine & Biology mercially available plane parallel ionization chambers,”
8, 51–57 (1963). Medical Physics 14, 210–215 (1987).
29. M. S. Weinhous and J. A. Meli, “Determining Pion, the
correction factor for recombination losses in an ioniza-
tion chamber,” Medical Physics 11, 846–849 (1984).
Chapter 3
Calorimetry
Larry A. DeWerd
Madison, Wisconsin
Blake R. Smith
University of Iowa
Iowa City, Iowa
CONTENTS
3.1 Introduction 31
3.2 Graphite Calorimeters 33
3.3 Water Calorimeters 34
3.4 Applications of Calorimeters in Particle Dosimetry 36
3.5 Conclusion 37
References 37
3.1 INTRODUCTION ∆E = ∆m c p ∆T (3.1)

Calorimetry is used to realize the primary standard for
absorbed dose to water. A primary standard is a device that where cp is the specific heat capacity in J kg−1 K−1 of the
measures a physical quantity directly from the fundamental material and ΔT is the change in temperature (K) caused
quantities of substance quantity (mol), length (meter), mass by the absorbed radiation. The dose to the material, Dm,
(kilogram), time (second), charge or current (Coulomb or is then determined from the known mass, ∆m, of the
Ampere), temperature (Kelvin), and light intensity (can- absorbing medium.
dela). The basis of calorimetry is simple, but its practice is ∆E
complicated. A good summary is given in Chapter 15 of Dm = = c p ∆T (3.2)
∆m
the AAPM 2009 summer school [1]. Historically, in 1904,
Curie and Dewar developed the first cryogenic calorim- The assumption for Equation 3.2 is that all of the energy
eter and the history continues to the present day at various that is deposited is expressed in terms of heat (tempera-
temperatures. Calorimetry is the determination of energy ture rise) with no heat defect (discussed later). If a rate of
absorbed, generally expressed in a temperature rise. For change in temperature is measured, then the net power
water, the increase in temperature is approximately 2.4 × (in J s−1 or Watt) is expressed by the equation
10−4 K/Gy or about 0.24 mK/Gy. There are complicating
dT
factors involved in water calorimetry that will be addressed Pnet = ∆m c p (3.3)
dt
later. The quantity of absorbed dose, Dm, for a material, m,
is determined from a change in energy, ∆E, that is mea- Generally, the temperature is measured with a therm-
sured from the material’s temperature change istor, connected via a Wheatstone bridge illustrated in
31
change in the system’s temperature will result in corre-

sponding changes to their dependent (i.e., temperature)
variable intercept on the respective axis. Adiabatic tech-
niques fix the absorber and the surrounding material at
the same temperature, thereby accounting for any ther-
mal disequilibrium between the core and the surround-
ing material. The last method, which is arguably the most
difficult, is an isothermal technique. Isothermal calorime-
ters electronically cool the core to balance the heat energy
imparted. The power used to keep the temperature stable
is the measure of the heat energy imparted by the radia-
tion. Graphical illustrations of the calorimeter response
are shown in Figure 3.2.
FIGURE 3.1 Diagram of a Wheatstone bridge circuitry com-

monly used to read out calorimeters. The voltage measured
across the bridge provides a measure of the change in resis-
tance from a thermistor placed on the core.
Figure 3.1. A typical bridge output signal is shown in
Figure 3.2, with the calorimeter being insulated. The
design of all calorimeters is to maximize the ratio of
intended signal to the background noise while main-
taining simplicity. The thermistor is probably the limit-
ing factor due to self-heating effects. However, control
of the environment is also very important; this results
in the phantoms and the calorimeter being as isolated
as possible, surrounded by well-controlled, insulated
phantom materials.
A typical calorimeter consists of three fundamental
components: the core, the jacket, and the shield. The core of
a calorimeter is the sensitive portion of the detector that the
measured change in temperature is referenced. The jacket
surrounds and thermally isolates the core from the outside
environment. Several thermometers and heaters are placed
on the core and jacket to help regulate quasi-adiabatic con-
ditions so that the temperature rise of the core is propor-
tional to the energy absorbed. The shield helps to thermally
isolate the jacket and core system. Conventionally, all three
components are constructed from the same material and
are thermally isolated from one another.
There are three different ways a radiometric calorim-
eter can operate. The most common method is heat flow
in which the change in temperature from the absorber
FIGURE 3.2 Schematic of typical bridge output signal for
to the jacket is monitored. Temperature changes are not an adiabatic (a), isothermal (b), and heat flow (c) system. The
expected to be constant and unchanging before or after time frame that the beam of radiation is turned on and off,
the irradiation due to the convection of heat from the core and its impact on the system, is defined into three regions for
to surround material. Specifically, if the system’s tempera- each case. The measure of temperature change, ΔT, is then
ture is linearly changing at some constant rate, then the used to calculate the dose to the medium.
Calorimetry ◾ 33
The two types of calorimeters used in the past by pri- of temperature change in the core is determined rela-
mary laboratories have been a graphite calorimeter or a tive to the shell that is held at a constant temperature,
water calorimeter. Of course, the preferred quantity is thus acting in an adiabatic fashion previously discussed.
absorbed dose to water, but the procedure for water cal- Temperature changes are measured with a set of thermo-
orimetry is more involved. Water calorimetry has a heat couples. This is in contrast to water-based calorimeters,
defect that must be taken into account. A heat defect which often rely on resistance changes in an electronic
arises when some of the energy deposited in the mate- system, rather than voltage changes, to detect changes
rial is not liberated as heat because of chemical reactions in temperature. To determine dose also necessitates an
or other effects, exothermic or endothermic [2]. Water accurate measurement of the core’s mass. Although this
can have a significant heat defect, which depends on the primary standard is used in many primary laboratories,
impurities and/or dissolved gases. Any perturbations the desirable calorimeter is a water calorimeter as no
from the surrounding materials must also be taken into material absorbed dose correction, and thus assumption
account. If graphite is used, there needs to be correction upon the primary beam of radiation, is required. This
of the absorbed energy from graphite to water. Graphite type is further explained in the next section.
does not have a significant heat defect [2]. Graphite calorimeters have also taken on more com-
pact forms and streamline operation. Contemporary
3.2 GRAPHITE CALORIMETERS graphite and water calorimeters are very fragile, cum-
The modern era of graphite calorimeters can be traced bersome, and require specialized equipment and expe-
back to Domen and Lamperti [3]. A number of primary rience to perform a successful measurement. Probe
laboratories use this type of calorimeter as their primary calorimeters have been proposed and shown as a poten-
standard. This type of calorimeter measures the temper- tial integration of absolute graphite dosimetry, which
ature change of the core and determines the quantity of can be performed at a clinic using the resources conven-
dose to water through a series of analytical conversions tionally available to a medical physicist [5]. The proposed
that relate the energy deposition behavior between the instrument has been operated using both adiabatic and
core material and the liquid water. This dose can then be isothermal modes. A Wheatstone bridge is built into the
used to calibrate ionization chambers. An illustration of detector electronics, which relates a voltage difference
a graphite calorimeter initially proposed by Petree and measured using a voltmeter. An illustration of a graph-
Lamperti [4] is shown in Figure 3.3. ite probe for MV dosimetry is shown in Figure 3.4.
The graphite core is surrounded by a graphite phan- While these compact devices show promise for per-
tom, and this entire arrangement must be thermally sonalized, absolute dosimetry, they still face several
isolated to eliminate any extraneous effects. A measure challenges. The sensitive regions (shield, insulation,
FIGURE 3.4 Diagram schematic of a graphite calorimeter

probe proposed by Renaud et al. [5]. The size and sensitive
detection volume is close to that of a farmer-type ionization
FIGURE 3.3 A cross-sectional diagram of a spherical graph- chamber. Several thermistors are placed through the shield,
ite calorimeter with a solid core. The initial calorimeter jacket, and core to measure relative temperature changes due
design by Petree and Lamperti [4] was 2.05 cm in diameter. to the incident radiation.
and jacket) are layered in insulative materials, epoxy, Alternatively, the glass vessels are filled with well-
and thermistors that act as impurities which result in known quantities of pure, dissolved gases in addition
thermal conduction. Accurate measurement of the core to high-purity water. In this way, the influences of the
mass can also lead to an offset for absolute measure- impurities within the medium are known and can be
ments. Additionally, the expected accuracy of these accurately corrected. A caveat, though, to sealed vessels
devices is estimated to be on par with a calibrated ion- is their complex design and introduction of material
ization chamber rather than their standard calorimeter inhomogeneities within the system.
counterparts due to the limited thermal isolation among The temperature of the surrounding phantom is
components, which can be achieved within these com- maintained at 4°C to minimize the possibility of con-
pact devices [5]. vection as liquid water is densest at this temperature.
The water phantom dimensions are 30 × 30 × 30 cm3
3.3 WATER CALORIMETERS containing the vessel at a reference depth. The dose to
Water calorimeters are the primary standard for Task water is then determined by
Group 51 (TG-51) calibrations. There have been a num-
ber of varieties built in prior years. At the National cw ∆T
Dw = k (3.4)
Institute of Standards and Technology (NIST), the 1 − kHD
calorimeter in use is the sealed water calorimeter [6].
This calorimeter uses a sealed glass vessel surrounded And
by water in the phantom. Sealed water calorimeters are
advantageous in that they can reduce impurities and gas Ea − Eh
kHD = (3.5)
exchange with the water, as the medium from which Ea
temperature changes are measured is independent of
the surrounding calorimeter. As such, a very tight qual- where kHD (which can be positive or negative) is the
ity control over the water and volume of water within heat defect; 1/(1 − kHD) is the correction factor. Ea is
the vessel can be achieved. An example of a glass-vessel the energy absorbed by the irradiated water, and Eh is
water calorimeter is shown in Figure 3.5. the energy which appears as heat. The other k is the
“foreign mass” correction and represents various cor-
rection factors such as the effect of the glass in contact
with the water [7]. With research about the heat defect,
impurities in the water and gasses are found to affect
the value of kHD [2]. Water saturated with pure hydro-
gen or nitrogen gas has a heat defect that is well known
and has also been found to reduce the amount of heat
defect [2]. This heat defect remains one of the areas for
a greater uncertainty in the measurement.
The components composing a water calorimeter are
not truly isolated from one another. As such, conduction
occurs among the components including the thermis-
tor probe, containment vessel, and the surrounding
medium with respect to the relative dose distribution
heating the surrounding water at different rates. The
thermistor probe is a metallic solid within the calorim-
eter, which acts as a source of unwanted heat. Likewise,
water has a high thermal capacity and will transfer heat
energy to any surrounding material with a lower heat
FIGURE 3.5 A sealed-vessel water calorimeter design from capacity. For example, a glass containment vessel has
Domen [6] that measures the change in temperature from the about a 1/5th lower heat capacity than water. The water
change in resistance between two thermistors. by itself also experiences some conduction due to the
Calorimetry ◾ 35
physical dose distribution of the beam. However, the

temperature probes for modern calorimeters are able to
discriminate dose to a very small region that is nearly
negligible relative to the dose gradients resulting from
therapeutic x-ray radiation beams but could be larger
for nonstandard beams and small fields. Holistically
speaking, a calorimeter requires some time to achieve
thermal equilibrium due to the heat dissipation, which
is dependent upon the location of materials and their
abundances relative to one another.
Convection is the tendency for an aqueous material
to migrate due to changes in temperature; hot material
rises, and cold material sinks with respect to the direc-
tion of gravity. The gradients in the dose distribution
from the primary beam, both laterally and distally,
will cause a gradient in the heat energy imparted to
the medium and induce convective currents for fluid
media. One of the simplest methods to minimize the FIGURE 3.6 An ultrasonic calorimeter setup similar to the
effects of convection is to direct the incident radiation one published by Malyarenko et al. [8]. A 5 MHz transducer
is acoustically coupled to the side of the water tank wall. The
from the top of calorimeter’s water tank. However,
transducer tracks changes in the emitted frequency in high-
there are instances where this may not be ideal, such as purity water due to temperature changes in the water and
for interferometry-based calorimetry. An alternative, wall from the incident radiation.
albeit more involved, method is to maintain a water
temperature at 4°C. At this temperature, water reaches
where the time of flight, τ , is equal to the time it takes
its maximum density and the influences of convection
the ultrasound pulse to travel an acoustic pathlength,
are negligible.
2 L , at an average velocity of v in water. Assuming the
In addition to thermistor and thermocouple water
water and enclosure are in thermal equilibrium, the
calorimeters, ultrasound and interferometer-based cal-
increase in temperature will cause an expansion of
orimetry can also be used to measure the heat changes
the walls and increase the acoustic path length by ∆L
in water from radiation. These alternative calorimeters
and also change the acoustic impedance of the water
have an advantage over solid-state calorimeters in that
resulting in a change in the speed of the ultrasound
their detector components do not interact within the
pulse by ∆v . From Equation 3.6, the change in fre-
primary beam. This removes some of the issues regard-
quency can be related to the changes in path length
ing component convection previously discussed. An
and velocity by
acoustic calorimeter measures the change in the speed
of sound wave propagation in water due to the temper- ∆ϕ ∆L ∆v
ature changes in the water. Time-of-flight techniques = − (3.7)
ϕ L v
have been used successfully to measure these heat
changes by emitting a tone-burst pulse with a known The increase in temperature, which caused the changes
frequency and comparing the phase change measured in path length and velocity, can be approximated by [9],
with a high-precision ultrasonic phase detector [8]. An
illustration of this type of water calorimeter is shown ∆L /L ∆Twall
≈ 0.036 (3.8)
in Figure 3.6. ∆v /v ∆Twater
A tone burst with a well-known phase, ϕ , is related to
the central frequency, f , by Due to the velocity of sound in water, subsequent mea-
surements can be performed on the order of a millisec-
2L ond, which allows near real-time measurement of dose
ϕ = 2π fτ = 2π f (3.6)
v imparted to water.
FIGURE 3.7 An interferometer calorimeter design from Flores-Martinez et al. [11]. A He-Ne laser with an optical wavelength
of 632.8 nm is split between a stationary reference and measurement arms. The measurement arm was translated using a
piezoelectric transducer to create a phase distribution observed using a photodiode.
An alternative, noninvasive calorimetry technique to improve their accuracy. For example, these calorimeters
ultrasonic methods is interferometry. The interferome- are very sensitive to vibrations and thermal instabili-
ter was initially proposed by Michelson and Morley [10] ties. For interferometer calorimeters in particular, the
in an attempt to detect the luminiferous ether. An inter- stability of the laser is a limiting factor in the accuracy
ferometer measures the change in phase from an inter- of the measurement as it is assumed that the change in
ference pattern observed at the intersection of a single intensity measured with the photodiode is due exclu-
laser split along a fixed and variable-position measure- sively to the changes in the refractive index of the water.
ment arm. A diagram for an interferometer-based calo- Additionally, these calorimetry techniques do not tech-
rimeter is shown in Figure 3.7. nically measure dose to a point (or small region) like a
The refractive index of a material is related to its thermistor-based calorimeter. Instead, their response is
temperature. An interferometer can therefore be used integrated over the entire beam path. As a result, ther-
to measure the dose to a sample of water by measur- mal instabilities within the water phantom contribute
ing the phase shifts, which occur as the water is heated substantially to the overall uncertainty. Tomographic
from the incident radiation. There are several empirical techniques have been suggested to reconstruct abso-
and theoretical mathematical models, which exist that lute dose distributions but have yet to be demonstrated
predict water’s refractive index dependence on tem- successfully.
perature which vary in complexity. One such model was
proposed by Abbate [12], who measured the change in 3.4 APPLICATIONS OF CALORIMETERS
refractive index and inferred the following relation: IN PARTICLE DOSIMETRY
Calorimetry holds a lot of promise as a primary stan-
−dnw   T − T0   dard for proton and heavy ion beams. Today, clini-
= B 1 − exp  −  (3.9)
dT   Tk   cal ionization chambers used for proton and heavy
ion dosimetry are indirectly calibrated in a 60Co beam
where B = 26.2 × 10−5 K −1, T0 = 2.0°C , and Tk = 48.5°C . quality but require several large conversion factors to
While these noninvasive water calorimetry tech- account for the response of the chamber in a particle
niques show promise and offer several advantages, they beam [13]. These calibrations and conversions are rather
are currently not as accurate as thermistor-based calo- arduous and susceptible to large uncertainties from the
rimeters and have been the focus of several works to conversion factors. A direct measure of dose to water
Calorimetry ◾ 37
from a primary standard in particle beams is challeng- 3. S. R. Domen and P. J. Lamperti, “A heat-loss-compen-
ing. Similar to water calorimetry in photon beams, con- sated calorimeter: Theory, design, and performance,”
Journal of Research of the National Bureau of Standards
duction can induce additional sources of uncertainty,
A. Physics and Chemistry 78A(5), 595–610 (1974).
especially for spot-scanning beams [14]. Accounting for 4. B. Petree and P. Lamperti, “Comparison of absorbed
heat defect is also difficult to quantify due to an appar- dose determinations in graphite by cavity ionization
ent dependence in the ionization density induced along measurements and by calorimetry,” Journal of research
the particle track [14]. This is particularly important to of the National Bureau of Standards – C. Engineering
consider for particle beams as the ionization density and Instrumentation 71C(1), 19–27 (1966).
5. J. Renaud, A. Sarfehnia, J. Bancheri, J. Seuntjens, “Aerrow:
changes as the beam of radiation traverses through mat- A probe-format graphite calorimeter for absolute dosim-
ter. Graphite calorimetry may be a better alternative etry of high-energy photon beams in the clinical envi-
to water-based calorimetry for particle beams. While ronment,” Medical Physics, 45(1), 414–428 (2018).
graphite calorimetry is not influenced by conduction, 6. S. R. Domen, “A sealed water calorimeter for measur-
there are radiative losses, which must be considered [14]. ing absorbed dose,” Journal of Research of the National
Institute of Standards and Technology 99, 121–141 (1994).
Finally, the conversion between water and graphite car-
7. C. K. Ross and N. V. Klassen, “Water calorimetry for
ries its own sources of uncertainty from the stopping radiation dosimetry,” Physics in Medicine and Biology
powers between water and graphite. As it stands, the 41, 1–29 (1996).
k = 1 uncertainty for the absorbed dose-to-water mea- 8. E. Malyarenko, J. Heyman, H. Chen-Meyer, R. Tosh,
surement with the transferred standard to a clinical ion- “High-resolution ultrasonic thermometer for radiation
dosimetry,” Journal of the Acoustical Society of America
ization chamber is around 2% and 3% for proton and
124(6), 3481–3490 (2008).
carbon-ion beams, respectively, relative to about 1% for 9. H. Chandler, R. Bowen, G. Paffenbarger, “Physical prop-
modern medical mega-voltage x-ray beams [13]. erties of a radiopaque denture base material,” Journal of
Biomedical Materials Research 5, 335–357 (1971).
3.5 CONCLUSION 10. A. Michelson and W. Morley, “On the relative motion
A calorimeter thus is the primary standard used to of the Earth and the Luminiferous Ether,” American
Journal of Science 34(203), 333–345 (1887).
establish absorbed dose to water. Since it is measured in 11. E. Flores-Martinez, M. Malin, L. DeWerd, “Development
a 60Co beam, there needs to be corrections, kQ, for cham- and characterization of an interferometer for calo-
bers when measuring an accelerator beam. The calorimeter-based absorbed dose to water measurements
rimeter could be used in an accelerator beam, but few in a medical linear accelerator,” Review of Scientific
primary laboratories have a linac. Thus, the standard Instruments 87(11), (2016).
12. G. Abbate, U. Bernini, E. Ragozzino, F. Somma, “The
used is a 60Co source with corrections to the ionization temperature dependence of the refractive index of water,”
chamber. A water calorimeter is generally the preferred Journal of Physics D: Applied Physics, 11(8), 1167 (1978).
apparatus and is the device used at NIST. 13. IAEA, “Absorbed dose determination in external beam
radiotherapy,” An International Practice for Dosimetry
REFERENCES on Absorbed Dose to Water, IAEA Tech. Series No. 398
(2000).
1. D. W. O. Rogers and J. E. Cygler (ed), Clinical Dosimetry 14. L. V. Verhey, A. M. Koehler, J. C. McDonald, M. Goitein,
Measurements in Radiotherapy, Chapter 15: Primary I. Ma, R. J. Schneider, M. Wagner, “The determination
Standards of Air Kerma for 60Co and X-Rays and Absorbed of absorbed dose in a proton beam for purposes of
Dose in Photon and Electron Beams, M. McEwen (Medical charged-particle radiation therapy,” Radiation Research
Physics Publishing, Madison, WI, 2009). 79(1), 34–54 (1979).
2. J. Seuntjens and S. Duane, “Photon absorbed dose stan-
dards,” Metrologia 46(2), S39 (2009).
Chapter 4
Semiconductor Dosimeters
Giordano Biasi and Nicholas Hardcastle
University of Wollongong
Anatoly B. Rosenfeld
University of Wollongong
CONTENTS
4.1 Introduction 39
4.2 Diodes 40
4.2.1 Construction and Functioning 40
4.2.2 Dosimetry with Diodes 42
4.3 MOSFETs 44
4.3.2 Dosimetry with MOSFETs 47
4.4 Diamonds 48
4.4.2 Dosimetry with Diamonds 50
Notes 51
References 51
4.1 INTRODUCTION in real time, they can measure dose distributions that
Semiconductor dosimeters are a cornerstone of change with time.
dosimetry in modern radiation therapy. They can be At this time, the semiconductor dosimeter family
standalone devices, or fabricated into linear, 2D and includes silicon diodes, silicon metal-oxide-semicon-
quasi-3D arrays to provide high levels of flexibility in ductor field-effect transistors (MOSFETs), and natural
measurement geometry. Their radiation-sensitive vol- and synthetic diamonds. In brachytherapy and external-
ume is made of silicon, silicon dioxide, or diamond beam radiation therapy with electron and photon beams,
(Table 4.1); the sensitivity to ionizing radiation is high they are used for integral dosimetry in-phantom or in
and the sensitive volume can be small – down to a few vivo. The dose absorbed in the radiation-sensitive volume
micrometers across. Thus, semiconductor dosimeters over a time span is assessed by measuring radiation-gen-
can measure, with high spatial resolution, in small erated charge or current; alternatively, by quantifying
radiation fields. Further, because they can be read out the change of the device’s electrical characteristics, such
39
TABLE 4.1 Properties of Silicon, Natural Diamond, and Chemical Vapor Deposition
(CVD) Synthetic Diamond
Property at 300 K Silicon Diamond CVD Diamond
Atomic number 14 6 6
Density (g cm−3) 2.33 3.51 3.51
Band gap (eV) 1.12 5.47 5.47
Intrinsic resistivity (Ω cm) 2.3 × 105 >1011 ~1013–1014
Mean energy to produce e–h pairs (eV) 3.68 13 13
Electron drift mobility (cm2 V−1 s−1) 1450 200–2800 ~4500
Hole drift mobility (cm2 V−1 s−1) 480 ~2000 ~3800
as its conductivity. In proton therapy and heavy-ion the surfaces of the base is heavily doped; here, impurities
therapy (HIT), semiconductors dosimeters are niche are of the opposite type and their concentration is of two
players to verify the energy and range of primary par- orders of magnitude, or more, than that used for the base.
ticles, and can also be used to characterize the radiobio- On the n-side of the junction, electrons are majority and
logical efficiency (RBE) of a therapeutic beam.1 Further, holes are minority charge carriers; on the p-side, the other
they have been proposed for dosimetry in experimen- way around.
tal radiotherapy techniques such as synchrotron-based As soon as the p–n junction is created, electrons dif-
micro-beam radiation therapy (MRT). fuse from the n-side to the p-side; holes diffuse in the
The reader may complement the present chapter with opposite direction. Eventually, that diffusion ends up
recent reviews: on the application of semiconductors establishing a space-charge region, with a positive charge
dosimeters in brachytherapy by Carrara et al. [1] and in on n-side and a negative charge on the p-side; across the
particle therapy by Rosenfeld [2]; as well as older ones [3–5]. junction there is an electric field strong enough to pre-
vent further diffusion of majority carriers. The space-
4.2 DIODES charge region has a thickness of about 1 μm and is called
4.2.1 Construction and Functioning the depleted region; the electric field has a strength of
The life story of each silicon diode starts with a pure silicon about 103 V/cm and produces the built-in potential.
crystal – the base, or substrate – in which impurities are The result is a silicon diode based on a p–n junction
introduced. For instance, of phosphorous and aiming for with a thin radiation-sensitive volume – the depleted
concentrations in the range 1014–1016 cm−3, so that valence region. The region can be extended by applying a reverse
electrons are donated to the crystal lattice; that results in external bias; this extra step, which is necessary to use
an n-type base. Alternatively, of boron and aiming for con- the diode for spectroscopy of nuclear radiation, is not
centrations in the range 1015–1017 cm−3, so that holes (i.e., usually considered for dosimetry.
the absence of the covalent bond between atoms of silicon Take a diode with a p-type base2 (Figure 4.1). If the
and boron) are created in the lattice; that results in a p-type diode is exposed to ionizing radiation, ionization gen-
base. Then, to create a p–n junction (Figure 4.1), one of erates electron–hole pairs. If the absorbed energy is E,
about N = E/w pairs are expected, where w is the mean
energy required to produce a single electron–hole pair;
for silicon, w is 3.68 eV (Table 4.1). This idea can be
expressed by saying that the generation constant of elec-
tron–hole pairs in silicon is [6]:
g = 4.2 × 1013 cGy −1 cm −3 (4.1)
The excess minority charge carriers generated within

one diffusion length from the junction (Ln, for electrons)
are swept across the junction by the built-in potential.
FIGURE 4.1 Cross-sectional view of a silicon diode based on That results in a radiation-induced current, propor-
a p–n junction (schematics, not to scale). tional to the dose rate in silicon; its integral (charge) is
Semiconductor Dosimeters ◾ 41
proportional to the absorbed dose. Because only charges Ionizing radiation may come in bursts. For example,
generated within one diffusion length are collected, the a medical linear accelerator (linac) delivering radiother-
amount of charge Q collected, per cGy to silicon in a apy with photon beams produces pulses which have a
p–n junction of 1 cm2, is about: duration in the range 2–6 μs, with a frequency in the
range 180–400 Hz. The instantaneous dose rate – the
Q ∝ gLn (4.2) dose per pulse – is in the order of 5.6 × 10 −4 Gy/pulse.
That number translates to a dose rate of 140 Gy/s, if con-
The diffusion length is a function of the diffusion coef- sidering a pulse 4 μs long. At low dose rates, the G-R
ficient Dn and of the mean lifetime of minority carriers centers – assuming they have a constant concentra-
τn [7]: tion for successive pulses – are mostly empty; a frac-
tion of radiation-generated minority charge carriers is
Ln = Dnτ n (4.3) captured and recombines; the remaining charge is col-
lected at the electrodes. As the dose rate in each pulse is
Dn depends on the resistivity of the silicon substrate – a increased, more and more centers are filled. The fraction
function of the dopant concentration and of tempera- of minority charge carriers that recombine decreases;
ture; τn is a function of the structure of the generation- the fraction of charge collected increases. The dose-
recombination (G-R) centers for electron–hole pairs that rate dependence of a diode’s response, first discussed
exists in the crystal lattice. The general expression for by Rikner and Grusell [12], is thus determined by the
τn is based on the Read–Shockley recombination theory G-R centers. It can be controlled by adjusting either the
[8]. Ionizing radiation creates different centers as a func- mean lifetime – by priming the diode or by introduc-
tion of particle type and energy; each center introduces ing suitable G-R centers – or the resistivity of the silicon
its own energy level in the forbidden energy-band gap substrate [11, 13, 14].
and affects the mean lifetime as a function of injection Ionizing radiation may come from different direc-
level3 and temperature. tions. The geometry of a diode is planar and asymmetric
Because ionizing radiation produces G-R centers, the (Figure 4.1). If the degree to which a dosimeter perturbs
concentration of these centers increases with increas- radiation depends on the angle of incidence, a direc-
ing accumulate dose. The result is a degradation of the tionally dependent sensitivity is expected. The magni-
mean lifetime and, in turn, of the diode sensitivity – the tude of that dependence depends on the design of the
amount of charge collected per unit absorbed dose [9]. diode, compounded by asymmetries in its packaging.
That degradation depends on the history of the diode, It is possible to minimize the directional dependence
in terms of previous irradiations, and is non-linear of a diode – or an array of diodes – by investigating
with accumulated dose; also, it plateaus after a certain alternative, optimized packaging solutions, e.g., the
amount of accumulated dose. For that reason, priming removal or introduction of high-Z materials [15]; also,
has been used to stabilize diode sensitivity; a priming geometrical asymmetry can be compensated for by
dose can be larger than 10 kGy. An alternative approach mounting two diodes back-to-back [16]. More recently,
to stabilize diode sensitivity introduces suitable G-R an ‘edgeless’ diode was proposed; the conventional pla-
centers in the crystal lattice, for example using platinum nar design was replaced with a p–n junction that sur-
atoms; their concentration has to be high enough to rounds the base [17, 18]. The directionally independent
dominate, over radiation-induced centers, the degrada- sensitivity [19] was preserved with a recently developed
tion of the mean lifetime [10]. A more recent tactic to drop-in packaging into a flexible Kapton carrier; lack
stabilize diode sensitivity requires fixing the dimensions of high-Z materials prevents perturbations to incident
of the radiation-sensitive volume in two directions: lat- radiation.
erally, by using guard-rings, and in depth, by growing But what about perturbations caused by silicon itself?
an epitaxial silicon layer onto a highly conductive sub- After all, the aim is to use the diode to assess dose-to-
strate. The key is to have a radiation-sensitive volume of water; surely the atomic number of silicon, which is
dimensions shorter than the expected diffusion length higher than that of water (Z = 14 vs. Zeff ~ 7.42), requires
of minority charge carriers, accounting for its degrada- discussion. In clinical electron beams, the diode is a
tion with accumulated dose [11]. sensor that monitors the electrons; for those electrons,
the ratio water-to-silicon of the average mass-collision minority carriers. It is important to account for RTI
stopping-powers (Sel/ρ) is almost energy-independent. when the diode is used for in vivo dosimetry, e.g.,
In photon beams, the picture is more nuanced. In mega- placed on the patient’s surface for skin dosimetry.
voltage photon beams, if the diode is small with respect It may take up to 5 minutes to reach an equilibrium
to the range of the electrons (the secondary charged par- temperature, with the kinetics of the temperature of
ticles), as a first approximation it senses the fluence of the diode depending on its packaging. The internal
electrons that would exist in an unperturbed medium4; buildup in commercial diodes typically hinders those
the diode is, again, a sensor that monitors the elec- measurements anyway.
trons. In kilovoltage photon beams, however, the elec- To conclude, the diode sensitivity depends on accu-
trons originating from the interaction of the photons mulated dose (through radiation damage), dose rate,
inside the diode contribute a significant portion to the angle of incidence and energy of radiation, and temper-
absorbed dose; the diode becomes a sensor that moni- ature. Yet, at present, diode dosimeters are ubiquitous:
tors electrons and photons. At that point, it is necessary they are cost-effective, easy to use and reproducible; they
to consider that the ratio water-to-silicon of the average can be operated in passive or active mode (with or with-
mass energy-absorption coefficients (μel/ρ) varies with out external bias); it is possible to produce free-standing
energy; that is explained mainly in terms of a relative single diode or multidimensional arrays to capture, at
increase in cross section for photoelectric effect in the once, complex dose distributions; further, they can be
kilovoltage range [20]. read out in real time to measure dose distributions that
In large radiation fields produced by megavolt- change with time.
age photon beams, there is an important component
of scattered photons with energies in the kilovoltage 4.2.2 Dosimetry with Diodes
range. In that case, it may be desirable to reduce the Single diodes were first proposed as dosimeters in pho-
energy dependence by shielding the diode; a high- ton beams in 1963 [27]; two decades later, Grusell and
Z packaging filters the low-energy component of a Rikner pioneered discussions on their advantages and
photon beam [21]. That shielding is effective was also limitations [9, 21, 28, 29]. Diodes have been used in elec-
confirmed with Monte Carlo analysis [22]; a smaller tron beams to measure depth-dose and lateral-profile
shielding effect is also expected in unshielded diodes, distributions [30] in place of ionization chambers and
because of the presence of thin layers of materials not don’t require depth-dependent corrections [31]. Diodes
intended for shielding. The photon spectrum of mega- have also been used for in vivo dosimetry in electron and
voltage photon beams gets harder with decreasing field photon beams, typically on the skin [32, 33]. If used for
size [23]; in small fields the low-energy component can in vivo dosimetry, they require the amount of build-up
be neglected. Also, in megavoltage beams, the number material be minimized, regular calibration against an
of low-energy scattered photons increases with depth; ionization chamber and the application of correction
however, a relative over-response moving toward larger factors dependent on beam energy, temperature, inci-
depths may be offset by an under-response because of dent-beam angle, and field size.
dose-rate dependence [24]. At present, single diodes are mainly used for in-phantom
The relative temperature instability (RTI) – the integral dosimetry in megavoltage photon beams. Because
dependence on temperature of the diode sensitivity – they can be fabricated of small radiation-sensitive volumes
is a function of the structure of the G-R centers [25]; (in the order of 0.1 mm3), measurements have minimal
therefore, it also depends on the history of the diode volume-averaging and, by scanning in water tank, it is
in terms of previous irradiations. It can be described possible to achieve sub-millimeter resolutions. Other than
by [26]: lateral profiles, measurements of photon beam output fac-
tor, percentage depth-dose and tissue-maximum ratio or
1 dS d ln S 1 d ln(Dnτ n ) tissue-phantom ratio are also possible. Diodes are not used
= = (4.4)
S dT dT 2 dT to calibrate the output of a beam. For a bare diode (water
is the build-up material) dose-to-water DH2O can be calcu-
where S is the sensitivity, T the temperature, Dn the lated, using Bragg–Gray cavity theory, from the measured
diffusion coefficient, and τn the mean lifetime of dose-to-silicon DSi; it is necessary to consider the ratio
water-to-silicon of the mass-collision stopping powers aver- leading to the development of the Razor diode [46]. The
aged over the electron-fluence spectrum: Razor has superior stability, dose linearity and radiation
hardness [47]; similarly to its predecessor, it requires
S  field-size dependent corrections [48].
DH2O = DSi  el  P (4.5)
 ρ  H2O,Si Modern radiotherapy techniques deliver complex
intensity-modulated dose distributions. These are cre-
where P accounts for small perturbations, with respect to ated through summation of multiple small fields typi-
water, of the fluence of the secondary charged particles. cally in a dynamic manner, thus are impossible to be
In small megavoltage photon beams (side or diam- measured by scanning a single diode. Fortunately,
eter of less than 3 cm), dosimetry gets challenging [34]. diodes can be combined into planar arrays. Since 1D
Secondary electrons depositing dose in a small silicon arrays of single diodes were first investigated [49], they
volume are, to a large extent, originating from the mate- have gained popularity as stand-alone devices that could
rials surrounding it [35, 36]. The silicon diode and its be quickly positioned onto the treatment couch, or
packaging (extra-cameral components) perturb, with attached to the linac gantry for measurements in trans-
respect to water, the fluence of electrons; the pertur- mission mode. One of the first commercial 1D array was
bation is a function of the size of the radiation field. the Profiler (SunNuclear, Melbourne, FL, USA); it had
This translates into P in Equation 4.5 being field-size 46 p-type single diodes distributed along 22.5 cm spaced
dependent. The formalism required to correct the diode by 5 mm [49]. A further development added a second,
readings with a field-size dependent correction was orthogonal linear array [50]. The Profiler 2 (SunNuclear)
introduced by Alfonso et al. [37] and later codified in a had 83 and 57 single diodes, spaced by 4 mm, in the Y
Code of Practice dedicated to small-field dosimetry [38]. and X directions respectively [51].
Monte Carlo codes are effective in characteriz- In small fields, dosimetry requires a spatial resolution
ing diode response in small fields and in calculat- better than about 1 mm, a resolution that is impossible
ing field-size dependent corrections; see, for instance to achieve with arrays of single diodes. Instead, mono-
[39–42]. Corrections depend on diode design (radiation- lithic arrays – diodes are ion-implanted onto a com-
sensitive volume and extra-cameral components), linac mon substrate – can be fabricated in large areas with
treatment-head design and measurement conditions. radiation-sensitive volumes spaced by 1 mm or less.
Calculating corrections with Monte Carlo requires Prototypes have been proposed by several groups [52–
knowledge of dosimeter design, which may not be avail- 54]. For instance, the Dose Magnifying Glass, proposed
able. Designing a dosimeter maintaining a correction by the Centre for Medical Radiation Physics (University
factor close to unity is preferable, and was indeed shown of Wollongong, Australia), has 128 diodes with a spac-
to be possible by introducing appropriate modifications ing of 0.2 mm [55]. Other than in megavoltage photon
[43, 44]. beams, the Dose Magnifying Glass was also used for
Shielded diodes are preferable for dosimetry in large range verification in HIT [56].
fields. They should not be used for small fields; the While with 1D arrays it is easy to decrease the spac-
low-energy photon component is negligible and shielding between diodes, in the case of 2D arrays there is a
ing not only makes photon spectra very different from necessary trade-off between dosimeter area and spa-
unperturbed ones, but also perturbs electron spectra tial resolution, to be within the maximum number of
[22], which distorts the required corrections. Unshielded readout channels available. The Centre for Medical
diodes, instead, introduce smaller perturbation and Radiation Physics has investigated three different com-
have less directional dependence; provided field-size promises (Figure 4.2). The Magic Plate has 512 diodes
dependent corrections are applied, they measure output spaced by 2 mm, uniformly distributed onto a 5 × 5 cm2
factor and high spatial resolution lateral profiles identi- area [57]; the Duo has the same number of diodes spaced
cal to Monte Carlo calculations in water. by 0.2 mm, but along two perpendicular arrays [58];
One of the most discussed diodes dedicated to small- finally, to improve the description of 2D dose distribu-
field dosimetry was the stereotactic-field diode, or SFD tions, in the Octa diodes are arranged along 4 arrays at
diode (IBA Dosimetry, Germany) [45]. Short-term 45°, and are spaced by 0.3 mm [59, 60]. The Magic Plate,
stability and dose-rate dependence were suboptimal, the Duo and the Octa were used for dosimetry during
FIGURE 4.2 2D monolithic arrays: (a) the Magic Plate, (b) the Duo, and (c) the Octa [69].
adaptive multi-leaf collimator (MLC)-tracking, includ- the number of diodes in a 3D arrangement to improve
ing arc deliveries5 [62]; in transmission mode to verify spatial resolution is challenging; clever arrangements
dose maps at a depth in water [63]; for dosimetry in an were devised. However the more complex the configu-
environment resembling that of a magnetic resonance ration of the radiation-sensitive volumes, the more dif-
(MR)-guided linear accelerator6 [64]; to verify, in real ficult it becomes to evaluate the results – in particular if
time, the dosimetry in radiation fields produced with the response is directionally dependent.
a variable-aperture Iris™ collimator in a CyberKnife® At present, two quasi-3D diode arrays are commer-
system (Accuray Inc., Sunnyvale, CA, USA) [65]. Also, cially available (Figure 4.3): the Delta4 (ScandiDos,
the design of the Magic Plate, the Duo and the Octa was Uppsala, Sweden) and the ArcCHECK (SunNuclear).
improved to maintain a correction factor close to unity The Delta4 has 1069 cylindrical p-type diodes (diameter
in small-field dosimetry [66]; it was studied, in terms of 1 mm, thickness 0.05 mm) arranged onto two perpen-
uniformity of the diode sensitivity across the large-area dicular planes; diodes are spaced by 0.5 cm in the 6 ×
array [67], and in terms of dose-rate dependence and 6 cm2 central region and 1 cm elsewhere. The
directional dependence [68]. ArcCHECK has 1386 SunPoint® n-type diodes (0.8 ×
When using arrays, there are additional steps to con- 0.8 × 0.03 mm3) spaced by 1 cm on a HeliGrid™ inserted
sider to be able to relate readings to dose. For instance, into a doughnut-shaped phantom. These diode arrays
diodes sensitivity may be nonuniform. That may be a thus allow comparison of calculated doses from treat-
result of nonuniformities in the silicon crystal used to ment planning systems with those measured, typically
fabricate the diodes; also, possible variations involved still on a planar basis. For example the ArcCHECK
in the fabrication processes. A nonuniform sensitivity ‘unfolds’ the helical array to present the user with a
can be compensated for with an equalization procedure – planar dose map. The resultant measurements from
the dosimeter is irradiated with a flat field to calculate these devices can further be processed by interpolat-
equalization factors for each diode. ing dose between two detectors along a ray line in a
The introduction of planar arrays was a step forward given measurement interval, to result in quasi-3D dose
in dosimetry; however, they have an inherent limitation: distributions.
2D information collapses into 1D when radiation comes
from a direction parallel with their plane. A workaround 4.3 MOSFETs
is to keep the array always perpendicular to radiation at 4.3.1 Construction and Functioning
all times: the dosimeter can be attached to the linac gan- A MOSFET (Figure 4.4) is a device with three terminals.7
try or lodged into a phantom that rotates in synch with One of the terminals (the gate) controls the conductance
the gantry [70, 71]. As an alternative, arrays with quasi- of the channel between the other two terminals (the
3D geometries have also been also developed. Scaling up source and the drain).
FIGURE 4.3 (a) The Delta4 has diodes on two orthogonal planes. (Image courtesy of Sadagopan et al. [72].) (b) The ArcCHECK
has diodes on a HeliGrid™. (Image courtesy of Feygelman et al. [73].)
A MOSFET used for dosimetry is typically a p-MOS- the threshold voltage (Vth) is defined as the bias we must
FET (also referred to as p-MOS; p-channel MOSFET; apply on the gate to sustain the flow of that current [7].
radiation-sensitive FET, RadFET). It is fabricated start- Ionizing radiation generates electron–hole pairs in
ing from a pure silicon crystal – the substrate – into the oxide. The mean energy required to generate a pair is
which we introduce impurities to create an n-type base about 18 eV [74]. Electrons either recombine or are swept
(e.g., using phosphorous so that valence electrons are out of the oxide in a few picoseconds [75]. Holes have
donated to the crystal lattice). Two terminals (the source lower mobility [76]; if the device is irradiated under pos-
and the drain) are p+ type; the third (the gate) is placed itive bias, they disperse (stochastic hopping) toward the
on top of a silicon-dioxide layer (SiO2, the radiation- interface between oxide and substrate. Their transport
sensitive volume; hereafter, for simplicity: the oxide) is a function of temperature, bias and oxide thickness –
in contact with the substrate. The conductive channel and, to a lesser extent, of the processing technique used
is a p-channel formed by the minority charge carriers to grow the oxide and of the concentration of defects [77,
(holes). When the external bias on the gate (Vg) is zero, 78]. As a first approximation, the fraction of holes that
relative to the source, the conductance of the channel is escape recombination ( f E, fractional yield) is a function
very low; we can increase it by increasing the negative of the electric field in the oxide and of the initial density
bias. Let’s say we want a constant current, of our choice, of the pairs plasma generated; that density is a function
to flow through the channel (Ids, drain–source current); of the type and energy of the incident particle [77, 78].
Holes escaping recombination are trapped in long-
term sites – oxide-traps and interface-traps – close to
the interface between oxide and substrate. They inter-
fere with the conductance of the p-channel, causing an
increase in the threshold voltage required to produce any
given drain–source current. The change in the threshold
voltage (∆Vth, threshold-voltage shift) is proportional to
the dose absorbed in the oxide [77, 78]. This idea can be
expressed by writing the threshold-voltage shift as [79]:
Cross-sectional view of a p-channel MOSFET x ox Qox x qN

FIGURE 4.4 ∆Vth = − = − ox ox (4.6)
(schematics, not to scale). ε ox ε ox
where q = 1.602 × 10 −19 C is the electronic charge; xox number of available traps in the oxide; eventually, they
is the distance between the gate and the center of the reach saturation and the threshold voltage plateaus [7].
space-charge distribution Qox, i.e., it is the thickness of Also, the threshold voltage shift may increase up to a
the charge-collection region; εox = 3.365 × 10 −11 F/m is point where it is too high to be measured with a given
the electric permittivity of the oxide; Nox is the number reader.8 There are workarounds; for instance, boron
of trapped holes per unit area in the oxide. If approxi- implantation through the gate can be used to reduce the
mating xox to the oxide thickness tox, Equation 4.6 can initial threshold voltage and increase the device’s lifes-
be rewritten in terms of capacitance per unit area of the pan [81]; alternatively, MOSFETs can be annealed [82, 83].
def
oxide (Cox = εox/tox) [80]: Irradiation in a biased configuration has source, sub-
strate and drain all grounded; a voltage is applied on
qN ox the gate. The threshold voltage however is measured in
∆Vth ≈ − (4.7)
Cox a readout configuration: source and substrate are short-
circuited, gate and drain short-circuited, and a con-
Under the same approximation (xox ≈ tox), the number stant current is forced through the conductive channel.
of trapped holes per unit area in the oxide is expressed Readout may be affected by creep-up, 1/f noise and fading.
as [78, 79]: The creep-up effect refers to an instability of the thresh-
old voltage for consecutive queries as a result of charges
N ox ≈ g 0tox DAf E (4.8) being injected into the device by the readout electron-
ics; it can be prevented by keeping consecutive queries
where D is the absorbed dose; g0 = 8.1 × 1012 cGy−1 cm−3 ~60 s apart. The 1/f noise introduces instabilities to the
is the rate of generation of pairs in the oxide; A is the threshold voltage [84]; its importance depends on the
fraction of holes that escape recombination but end up device’s construction, such as processing technique used
trapped – it is a function of the electric field in the oxide, to grow the oxide, and operating characteristics, such as
of temperature, of the processing technique used to grow temperature and applied bias [85]; it also increases with
the oxide and of the concentration of defects [77, 78]. At increasing accumulated dose. Finally, fading describes a
this point, it is helpful to define a constant k as: long-term change in the value of the threshold voltage
after irradiation; the concentration of trapped charges,
def qg 0 mV
k= = 39 (4.9) in the oxide (∆Not) and at the interface with the sub-
εox cGy µ m 2
strate (∆Nit), varies in time depending on the device’s
construction, and on irradiation and post-irradiation
and express the MOSFET sensitivity S as:
bias [80, 86]. ∆Not tends to decrease monotonically after
def ∆Vth irradiation (thus reducing the initial threshold voltage
S= ≈ ktox
2 f
E (4.10) shift), while ∆Nit tends to exhibit an initial increase
D
(increasing the initial threshold voltage shift) followed
It is possible to optimize the MOSFET to operate in a by saturation [85]; in the case of positive bias, the net
specific dose range; for instance, by maximizing its effect is a negative shift in the threshold voltage. Fading
sensitivity. Equation 4.10 indicates two options. If the is expected to be most pronounced during the first few
requirement is known beforehand, the oxide can be hours after irradiation, but can be present after days [87].
grown of a different thickness; or, the bias can be varied MOSFETs are fabricated of silicon; the radiation-sen-
during operation if adapting an existing MOSFET to sitive volume is of silicon-dioxide. In low-energy photon
fit new conditions. There are limitations. Thick oxides beams (15–200 keV), MOSFET sensitivity normalized to
(1.0 μm and beyond) have poor reproducibility; they that in a 6 MV photon beam (effective energy 2 MeV) is
may be exposed to mechanical stress resulting in insta- larger by a factor of 3–4; sensitivity is maximized at pho-
bility of the threshold voltage. Similarly, a strong electri- ton energies of about 30–50 keV [88, 89]. In the mega-
cal field may also inflict mechanical stress. voltage photon range (4–25 MV) and in electron beams
Those are not the only trade-offs to consider. Any in the range 5–21 MeV, MOSFET sensitivity is almost
increase in sensitivity comes at the cost of a decrease constant [90–92]. The energy-dependence of MOSFET
in the device’s lifespan. This is because there is a finite sensitivity depends on the details of the manufacturing
process and packaging [89, 93, 94]; because of the rel- creep-up, 1/f noise and fading may affect the accuracy of
atively small size of the transistor, a relatively high the measurements.
fraction of secondary electrons depositing dose in the
oxide is generated in the surrounding packaging (dose 4.3.2 Dosimetry with MOSFETs
enhancement effect) [89]. MOSFETs dosimeters were first proposed in 1970 [103]
Also, MOSFET sensitivity is influenced by the angle and later demonstrated to monitor space-radiation on
of incidence of radiation. Similarly to diodes, this is satellites [104, 105]. At present, they have applications
explained by, and depends on, asymmetries in the geom- in brachytherapy [1]; radiation therapy with kilovolt-
etry and materials of the transistor and of its packaging. age [106] and megavoltage [107, 108] photon beams, and
Being device-dependent and measurement-condition with electron beams [109, 110]; also, to monitor radia-
dependent, values vary greatly in the literature; see, for tion in food-irradiation plants [111]. MOSFETs have
instance [90, 95] or, more recently [92, 96]. been proposed for use in intraoperative radiotherapy
Finally, because the mobility of the charge carriers (IORT) delivered with electron or low-energy photon
in the conductive channel fluctuates with temperature, beams [112, 113]. Despite the small dimensions of the
so does the MOSFET sensitivity. Of course, tempera- radiation-sensitive volume (the oxide), MOSFETs are
ture coefficients are device-dependent; however, they not recommended for small-field dosimetry in mega-
are also negligible because temperature effects are typi- voltage photon beams [38, 114, 115] because of the rela-
cally within the reproducibility of the MOSFET; they tively low measurement precision.
are neglected for clinical measurements [97] expect The use of MOSFETs in proton therapy is currently
for in vivo skin dosimetry. A typical time required for limited by their relatively large dependence on the linear-
MOSFETs to reach thermal equilibrium with a patient’s energy transfer (LET); the height of the Bragg Peak,
skin is 2 minutes [98]. We can minimize temperature normalized to the proximal plateau, is almost 40%
dependence by selecting a readout current correspond- lower estimated using a MOSFET than with an ioniza-
ing to a thermostable point on the Isd–Vg characteristic tion chamber [116]; however, it can be used to determine
[99]. Alternatively, we can use an internal p–n junc- the spatial accuracy of the Bragg peak location [117].
tion (drain or source onto substrate) for temperature MOSFETs can be used for dosimetry in x-ray micro-
measurements directly in the channel; or even a dual- beams, which are relevant to MRT, where beams are
MOSFET. In that case, the rationale is that MOSFETs separated center-to-center by about 200 μm [118–122].
irradiated under different gate bias have different radia- Most typically, today’s MOSFETs are used for in
tion-induced shifts of the threshold voltage, but similar vivo dosimetry in external beam radiotherapy. The
temperature coefficients. The temperature-independent MobileMOSFET system (Best Medical, Canada) consists
signal proportional to the absorbed dose will result from of MOSFET detectors, a reader module and Bluetooth
subtracting the shifts of the threshold voltage of the two wireless transceiver. Multiple detectors can be read out
MOSFETs [100]. at once, and different detectors are available for various
To summarize, MOSFETs dosimeters are robust, dose ranges. In vivo dosimetry, where the detector is
light-weight and cost-effective devices. They have placed on the patient surface, is typically used to verify
advantages over diodes: dose information can be stored treatment dose when commissioning new techniques,
permanently and read out, non-destructively, multiple treatment equipment or software, or to verify dosimetry
times. A power supply is strictly required only for the where there may be differences between machine perfor-
readout procedure and for active mode application. mance and treatment planning calculation. An example
Further, the radiation-sensitive volume (the oxide) has of the latter case is verification of junction doses when
dimensions an order of magnitude smaller than that a using matched jaw-defined fields.
diode and its sensitivity is adjustable by varying the bias MOSFETs have shown particular strengths in skin
during irradiation. MOSFET sensitivity is also inde- dosimetry. They are placed on the patient’s skin, either
pendent of dose rate up to 108 Gy/s [101, 102]. However, at the point of entry (entry-dose measurement) or exit
MOSFET dosimetry has limitations: it is key to account (exit-dose measurement) of incident radiation. Skin
for the energy-dependence, directional-dependence dosimetry requires a dosimeter with a reproducible
and temperature-dependence of the sensitivity. Also, effective-depth close to the 7 mg/cm2, i.e., a depth of
at the boundary between low- and high-density media

[132]. Brachytherapy applications of the MOSkin™ have
also been discussed extensively; e.g., see [133, 134].
A further application of MOSFETs to in vivo dosim-
etry uses a MOSFET implanted inside the planning
treatment volume (PTV). Commercial implantable wire-
less MOSFETs (DVS® dose verification system, Sicel
Technologies, Morrisville, NC) were investigated by several
groups [135–143]. Implanted MOSFETs can be imaged and
located with ultrasonography [140, 142], kilovoltage [138–
140, 142] or megavoltage [144] computerized tomography
FIGURE 4.5 Packaging of (a) a typical MOSFET dosimeter (CT) scans. Also, they can be used as fiducial markers in
and (b) a MOSkin™. Most manufacturers place the MOSFET image-guided radiotherapy (IGRT); a feasibility study on
on a Kapton film or on a thin printed-circuit-board carrier to their use for extracranial-target treatments delivered with
which the transistor is wire-bonded; a bubble of epoxy resin
the CyberKnife system was also reported [107].
is then placed on the top for protection. The MOSkin™ (c)
replaces that protective bubble with a thin polyamide film.
4.4 DIAMONDS
4.4.1 Construction and Functioning
0.07 mm in tissue, as recommended by the International
9
The crystal of a natural diamond has an energy gap of
Commission on Radiological Protection [123]. 5.47 eV and a resistivity in the range 1013–1016 Ω cm: it is
Initial investigations considered scientific-grade possible to apply an electric field across the crystal with-
MOSFETs which had their nickel casing removed to leave out generating a current.10 Also, when likened to silicon,
the bare chip exposed; the transistor was placed in a phan- the crystal is less prone to radiation damage because the
tom holder using wax to fill air cavities [124]. Another energy required for displacing atoms from their sites in
group [125] adapted for skin dosimetry a commercial the lattice is larger.
MOSFET, which had been previously characterized for In a crystal of density ρ and volume V, irradiated at a
in vivo dosimetry [95]; the MOSFET was protected by an dose rate dD/dt, charge carriers (electron–hole pairs) are
hemispheric layer of epoxy resin (Figure 4.5) and mea- generated at a rate (Figure 4.6):
sured dose at a water-equivalent depth between 0.6 and
1.8 mm, not complying with ICRP recommendations and dg ρV dD
= (4.11)
over-estimating skin dose [125]. dt w dt
The MOSkin™ (Figure 4.5) developed at the Centre
for Medical Radiation Physics (CMRP, University of
Wollongong, Australia) is a MOSFET for real-time in
vivo dosimetry. The transistor of the MOSkin™, which
has an oxide of thickness 0.55 μm, is mounted under-
neath a thin plastic layer with a uniform thickness of
0.07 mm [126]. The use of MOSkin™ for skin dosim-
etry has been reported by various authors; e.g., during
breast radiotherapy [126], intensity-modulated radiation
therapy (IMRT) for nasopharyngeal carcinoma patients
[127] and head & neck cancer patients [128]; for monitor-
FIGURE 4.6 Ionizing radiation generates electron–hole pairs
ing the eye-lens dose during cerebral angiography [129].
in the crystal of a diamond; the average energy spent to cre-
However, the package of the MOSkin™ can be adapted ate a pair is about 13 eV. An external bias inhibits recombi-
to different uses; for instance, to measure dose to the nation and promotes electrical conductivity; electrons and
rectal wall in external-beam radiotherapy [130, 131]; by holes drift toward the electrodes and are collected. Their drift
changing the build-up layer and selecting an appropri- mobility is higher in the crystal of a diamond than in silicon,
ate water-equivalent depth, to verify dose calculations so that collection time is shorter.
where w = 13 eV is the average energy spent to generate close to the negative electrode. The buildups contribute
a pair; the current I0 is then: an electric field opposed to that produced by the exter-
nal bias. For that reason, as accumulated dose increases,
dg at a constant dose rate, the effective field in the crystal
I0 = e (4.12)
dt decreases. A decreasing effective field also decreases
the charge-collection distance in Equation 4.16 and, in
where e is the electron charge. An external bias pro- turn, the gain in Equation 4.15. This effect is referred to
ducing an electric field E across the crystal inhibits the as polarization [146]; other than on the effective field,
recombination of charge carriers and promotes elec- the polarization depends on the flux of radiation [147]
trical conductivity σ – the crystal is then operated as and the quality of the crystal [148]; also, it becomes more
a resistor in which conductivity is proportional to the important if the crystal is damaged by radiation [149].
dose rate [145]. Charge carriers drift toward the elec- At any rate, eventually, traps are saturated: the polariza-
trodes with a velocity equal to the product of their drift tion plateaus and so does the current measured at the
mobility μ and the strength of the electric field: electrodes – albeit at a value lower than its initial value
at zero accumulated dose.
v = µE (4.13)
To minimize the polarization in dosimetry, the crys-
The current Iout collected at the electrodes of a crystal of tal can be primed, i.e., irradiated before its use. Priming
thickness L is: fills deep traps and stabilizes the crystal sensitivity. It
must be performed at the operational external bias.
µτ Also, the amount of priming required is determined by
I out = I 0 E (4.14)
L the quality of the crystal in terms of its impurities; that
quality can’t be predicted or controlled in the crystal of
where τ is the average lifetime of charge carriers. The a natural diamond. A typical priming is in the order of
crystal has a charge-collection efficiency that can be 5–10 Gy.
expressed using a gain factor G defined as the ratio of the Too many impurities cause the crystal to be insen-
current Iout measured at its electrodes to the current Io: sitive to radiation, because of a decrease in the recom-
bination time, and to be affected by polarization; but
def I out d impurities are necessary to make the current collected
G= = (4.15)
I0 L at the electrodes to increase linearly with dose rate. In
a parallel-plate biased structure, the current is propor-
where the charge-collection distance d, i.e., the average tional to the electrical conductivity σ in the crystal. The
length the charge carriers drift for, is: relationship between that conductivity and the genera-
def
tion rate of pairs is [150]:
d = µτ E (4.16)
∆
dg
σ ∝   (4.17)
Charge-collection efficiency is maximized with an  dt 
external bias that saturates the charge-collection dis-
tance; however, the bias should not be high enough to where Δ, a dose-rate index of linearity, depends on the
cause breakdown11 or leakage currents. quality of the crystal. Its value is 0.5 when the recom-
The crystal of a natural diamond contains impuri- bination of charge-carriers dominates over trapping;
ties such as atoms of nitrogen, boron, and aluminum for instance, in a pure crystal or when the dose rate is
in various concentrations; those impurities make the very high. It has values between 0.5 and 1.0 when trap-
crystal n- or p-type, depending on if electrons or holes ping dominates the lifetime of charge carriers; also, it
are donated to the lattice. Impurities with energy levels may be 1.0 or more for particular impurities and cross-
in the forbidden energy-gap may trap drifting charge sections for capture of charge-carriers [151, 152]. In the
carriers. Depending on the time-scale of the trapping literature, Δ has values between 0.49 and 1.16.12 There is
process, that may results in a buildup of negative charge controversy over if Δ varies with the bias, and with the
close to the positive electrode, and of positive charge energy and type of radiation [153].
The crystal of a natural diamond is an attractive 0.4 mm; it required an external bias of 100 V and a prim-
material for dosimetry because of its quasi-equivalence ing with up to 5 Gy. Crystals had different and uncon-
to human soft-tissue in terms of atomic number (Z = 6 trolled concentrations of impurities and defects; hence,
vs. Zeff ~ 7.4 for human soft-tissue). The ratio diamond- the amount of priming varied between specimens.
to-water of the mass collision-stopping-powers (Sel/ρ) Also, priming was required after any switching-off of
and that of the mass energy-absorption-coefficients the external bias. Once primed, however, crystals had a
(μel/ρ) are almost energy-independent for a wide range stable response, to within 0.1% for short-term stability,
of photon- and electron-energies. That the crystal is over 10 minutes, and to within 1% for long-term stabil-
energy-independent was confirmed with experiments in ity. Their sensitivity had no temperature dependence; its
megavoltage electron and photon beams [152, 154] and dose-rate dependence varied between 1.9% and 3.2% for
with Monte Carlo calculations [155] – however, the pack- dose rates between 0.9 Gy/min and 4.65 Gy/min [151,
aging is key because it may include high-Z components. 152, 154, 160, 161, 162].
Also, the crystal of a natural diamond has intrinsic sym- The crystal of a diamond however can also be arti-
metry; its sensitivity has a negligible directional depen- ficially grown. At present, the chemical vapor deposi-
dence; however, the packaging13 is again key [156, 157]. tion (CVD) and the high-pressure high-temperature
The crystal of a natural diamond is coveted for its (HPHT) techniques are widely used to grow synthetic
beauty, a property exploited by the jewelry industry; it is crystals under controlled and reproducible conditions;
also desired, for completely different reasons, in today’s for instance, groups that have reported on polycrystal-
dosimetry. Its sensitivity does not depend on angle of line diamonds grown using CVD include [163, 164];
incidence and energy of radiation; it can be cut of small those that have reported on synthetic single-crystal dia-
size, so that dosimetry with high spatial resolution is monds (SSCDs) grown using CVD or HPHT include
possible; it is radiation-hard, i.e., there is negligible [165, 166]. As growth techniques were improved and
radiation damage to the crystal lattice as a function perfected, synthetic crystals have become more widely
of accumulated dose. True, its sensitivity does depend available, reproducible, and affordable. Priming, still
on accumulated dose and dose rate (through polariza- required [156, 167], and the degree of dose-rate depen-
tion and dose-rate dependence), but those effects can be dence, vary between crystals produced by different
accounted for once characterized. However, the crystal – groups [165, 166, 168, 169] as a function of the concen-
that selected for dosimetry, at least, if not that for the tration of impurities.14
jewelry industry – is resented because it is exquisitely An SSCD in a Schottky-diode configuration (Figure 4.7)
expensive. The quest to overcome that limitation is was developed at Tor Vergata University (Rome, Italy).
revealed in the next section. An HPHT of area 4 × 4 mm2 is covered by a highly con-
ductive film of boron-doped CVD SSCD; a layer 22 μm-
4.4.2 Dosimetry with Diamonds thick of nominally intrinsic diamond is grown onto the
Natural diamonds were proposed as dosimeters in doped surface; onto that surface, a Schottky junction
1948 [158]; their commercialization and investigation is then mounted by thermal evaporation of an alumi-
in clinical photon and electron beams followed shortly num contact. The crystal is embedded in a waterproof
after [159]. Although diamonds were discussed as a reli-
able, small and near-tissue equivalent dosimeters, it was
apparent there were challenging in selecting optimal
ones from a large batch; those were rare and expensive.
Also, they were poorly reproducible, requiring bespoke
characterization and priming.
The PTW-60003 (PTW, Freiburg, Germany) is an
example of a widely discussed dosimeter based on a nat-
ural diamond crystal. Developed in cooperation with FIGURE 4.7 An SSCD in a Schottky-diode configuration.
the IPTP Institute (Riga, Latvia), it was commercially A film of CVD diamond, p-type doped, is deposited onto a
available in the 1990s. The radius of the crystal varied conductive HPHT substrate. A layer of intrinsic diamond is
from 1.0 mm to 2.2 mm and its thickness from 0.2 to grown onto the doped layer.
cladding made of poly(methyl methacrylate) (PMMA) That diamond has dimensions of 2.5 × 2.5 cm2, with a
and epoxy resin. Because of the built-in potential at the thickness of 300 µm; the front area is divided into 12 ×
interface between the metal contact and the layer of dia- 12 pixels spaced by 2 mm. Two Schottky barriers are cre-
mond, the SSCD works as a Schottky diode and can be ated at the interfaces between the diamond and metallic
operated without an external bias. The radiation-sensitive contacts; it is still necessary to improve the quality of the
volume, defined by the depleted region, is about 3.8 × contacts to ensure a uniform sensitivity [192].
10 −3 mm3, radius 1.1 mm and thickness 1 µm. The sensi-
tivity is stabilized by priming with a few Gy. That SSCD NOTES
has been made commercially available as the microDia-
1. In that case, devices are referred to as micro-dosimeters;
mond PTW-60019 (PTW, Freiburg, Germany); further
we will not cover micro-dosimetry applications.
details can be found in the literature [156, 171–173]. 2. For simplicity, the discussion is on p-type diodes; in
Since its introduction, the microDiamond has been the base, holes are majority and electrons are minority
extensively characterized: in photon and electron beams charge carriers. It is possible to swap around holes and
[167, 174–177]; for small-field dosimetry in megavoltage electrons so that the discussion is on n-type diodes.
3. Concentration of excess minority charge carriers with
photon beams [176, 178–183]; in x-ray beams produced
respect to that of equilibrium majority carriers.
with a synchrotron [184]; in a magnetic resonance (MR)- 4. Unless Bragg–Gray cavity theory breaks down; for
guided linear accelerator (linac) [185]; for applicability instance, in small-field dosimetry.
in a small-animal irradiation unit (SARRP) with x-ray 5. It was necessary to devise a framework for directionally
energies up to 220 kV. Those characterizations suggested dependent corrections; that methodology relies on real-
time information on incident-beam direction [61].
that the microDiamond, can be scanned in water tank
6. The magnetic field affects dose deposition and influ-
for high spatial resolution dosimetry; if used in the rec- ences the dosimeter sensitivity.
ommended dosimeter-beam orientation, has a sensitiv- 7. Never mind the fourth terminal (the substrate): it is
ity linear and stable with dose and dose rate, requiring short-circuited to either source or drain.
no correction factors; has a directionally dependent 8. In other words, readers have a maximum threshold
sensitivity if the beam is not parallel to the axis of the voltage read out.
dosimeter, particularly in the kV energy range; has a 9. Representing the depth of the radiation-sensitive basal-
layer cells.
low temperature dependence; requires a priming of 10 10. Compare that with the silicon crystal described in Section
Gy if used in the kV energy range. The microdiamond 3.2. In that case, to apply an electric field across the crystal
is particularly attractive for electron beam scanning without generating a current, we need to create a depleted
dosimetry; since the carbon-water stopping power ratio region, formed by a diode based on a p–n junction.
is nearly constant over typical clinical electron energy 11. In the order of 103 V µm −1.
ranges, direct dose measurement can be performed 12. For both natural and synthetic crystals.
13. Packaging includes encapsulation, electrodes, cables,
[174]. Further, it was proposed that the microDiamond is
etc.
suitable for in vivo dosimetry [156]; for reference dosim- 14. The size of the electrodes also impacts the dose-rate
etry in radiation fields down to 5 mm in length [186]; for dependence [170].
dosimetry in proton and heavy-ion beams – however, its
sensitivity has a dependence on the LET in heavy-ion REFERENCES
beams, for instance carbon and oxygen [187].
Different groups have proposed alternatives to the 1. M. Carrara, D. Cutajar, S. Alnaghy, et al., “Semiconductor
real-time quality assurance dosimetry in brachyther-
microDiamond. For instance, a group has developed apy,” Brachytherapy 17(1), 133–145 (2018). https://doi.
an SSCD, referred to as SCDD-Pro (single-crystal dia- org/10.1016/j.brachy.2017.08.013
mond dosimeter), for dosimetry in proton beams [188]. 2. A. B. Rosenfeld, “Novel detectors for silicon based
The SCDD-Pro is based on a high-quality free-standing microdosimetry, their concepts and applications,”
SSCD configuration: there is neither substrate nor dia- Nuclear Instruments and Methods in Physics Research,
Section A: Accelerators, Spectrometers, Detectors and
mond doping; a bias is required. The sensitivity of the
Associated Equipment 809 (February), 156–170 (2016).
SCDD-Pro exhibits no energy dependence and no dose- https://doi.org/10.1016/j.nima.2015.08.059
rate dependence [188]. Another group investigated a 2D 3. P. D. Bradley, A. B. Rosenfeld, M. Zaider, “Solid state
polycrystalline diamond based on CVD [189, 190, 191]. microdosimetry,” Nuclear Instruments and Methods
in Physics Research, Section B: Beam Interactions with 19. M. Petasecca, S. Alhujaili, A. H. Aldosari, et al.,
Materials and Atoms 184 (1–2), 135–157 (2001). https:// “Angular independent silicon detector for dosimetry
doi.org/10.1016/S0168-583X(01)00715-7 in external beam radiotherapy,” Medical Physics 42 (8),
4. A. B. Rosenfeld, “Electronic dosimetry in radiation ther- 4708–4718 (2015). https://doi.org/10.1118/1.4926778
apy,” Radiation Measurements 41 (December), S134–S153 20. P. Andreo, D. T. Burns, A. E. Nahum, et al., Fundamentals
(2007). https://doi.org/10.1016/j.radmeas.2007.01.005 of Ionizing Radiation Dosimetry (John Wiley & Sons,
5. A. B. Rosenfeld, “Advanced semiconductor dosimetry 2017).
in radiation therapy,” AIP Conference Proceedings 1345, 21. G. Rikner and E. Grusell, “Selective shielding of
48–74 (2011). https://doi.org/10.1063/1.3576159 a P-Si detector for quality independence,” Acta
6. G. C. Messenger and M. S. Ash, The Effects of Radiation Oncologica 24 (1), 65–69 (1985). https://doi.org/10.3109/
on Electronic Systems (Van Nostrand Reinhold 02841868509134367
Company Inc., New York, 1986). 22. H. Benmakhlouf and P. Andreo, “Spectral distribution
7. S. M. Sze and K. K. Ng, Physics of Semiconductor Devices, of particle fluence in small field detectors and its impli-
3rd ed. (John Wiley & Sons, Inc., Hoboken, NJ, 2006). cation on small field dosimetry,” Medical Physics 44 (2)
https://doi.org/10.1002/0470068329 (2017).
8. W. Shockley and W. T. Read, “Statistics of the recombi- 23. H. Benmakhlouf, J. Sempau, P. Andreo, “Output correc-
nations of holes and electrons,” Physical Review 87(5), tion factors for nine small field detectors in 6 MV radia-
835–842 (1952). https://doi.org/10.1103/PhysRev.87.835 tion therapy photon beams: a PENELOPE Monte Carlo
9. E. Grusell and G. Rikner, “Radiation damage induced study,” Medical Physics 41 (4), 041711 (2014). https://doi.
dose rate non-linearity in an n-type silicon detector,” org/10.1118/1.4868695
Acta Oncologica 23 (6), 465–469 (1984). https://doi. 24. M. Heydarian, P. W. Hoban, A. H. Beddoe, “A com-
org/10.3109/02841868409136050 parison of dosimetry techniques in stereotactic radio-
10. B. J. Baliga and E. Sun, “Comparison of gold, plati- surgery,” Physics in Medicine and Biology 41 (1), 93–110
num, and electron irradiation for controlling life- (1996). https://doi.org/10.1088/0031-9155/41/1/008
time in power rectifiers,” IEEE Transactions on 25. E. Grusell and G. Rikner, “Evaluation of tempera-
Electron Devices 24 (6), 685–688 (1977). https://doi. ture effects in P-type silicon detectors,” Physics in
org/10.1109/T-ED.1977.18803 Medicine and Biology 31 (5), 527–534 (1986). https://doi.
11. M. Bruzzi, “Novel silicon devices for radiation therapy org/10.1088/0031-9155/31/5/005
monitoring,” Nuclear Instruments and Methods in 26. A. S. Saini and T. C. Zhu, “Temperature dependence of
Physics Research, Section A: Accelerators, Spectrometers, commercially available diode detectors,” Medical Physics
Detectors and Associated Equipment 809 (February), 105– 29 (4), 622–630 (2002). https://doi.org/10.1118/1.1461842
112 (2016). https://doi.org/10.1016/j.nima.2015.10.072 27. A. R. Jones, “The application of some direct current
12. G. Rikner and E. Grusell, “Effects of radiation damage properties of silicon junction detectors to γ-ray dosim-
on the silicon lattice,” Physics in Medicine & Biology etry,” Physics in Medicine and Biology 8 (4), 307 (1963).
1261 (28), 1261–1267 (1983). https://doi.org/10.1088/0031-9155/8/4/307
13. J. Shi, W. E. Simon, T. C. Zhu, “Modeling the instan- 28. G. Rikner and E. Grusell, “General specifications for
taneous dose rate dependence of radiation diode silicon semiconductors for use in radiation dosimetry,”
detectors,” Medical Physics 30 (9) (2003). https://doi. Physics in Medicine and Biology 32 (9), 1109–1117 (1987).
org/10.1118/1.1602171 https://doi.org/10.1088/0031-9155/32/9/004
14. P. A. Jursinic, “Dependence of diode sensitivity on the 29. E. Grusell and G. Rikner, “Linearity with dose rate of
pulse rate of delivered radiation,” Medical Physics 40 (2) low resistivity P-type silicon semiconductor detectors,”
(2013). Physics in Medicine and Biology 38 (6), 785–792 (1993).
15. P. A. Jursinic, “Angular dependence of dose sensitivity https://doi.org/10.1088/0031-9155/38/6/011
of surface diodes,” Medical Physics 36 (6) (2009). 30. X. A. Li, C.-M. Ma, D. Salhani, “Measurement of per-
16. M. Westermark, J. Arndt, B. Nilsson, et al., “Comparative centage depth dose and lateral beam profile for kilo-
dosimetry in narrow high-energy photon beams,” voltage X-ray therapy beams,” Physics in Medicine
Physics in Medicine and Biology 45 (3), 685–702 (2000). and Biology 42 (12), 2561–2568 (1997). https://doi.
https://doi.org/10.1088/0031-9155/45/3/308 org/10.1088/0031-9155/42/12/019
17. M. J. Bosma, E. Heijne, J. Kalliopuska, et al., “Edgeless 31. R. Tailor, W. Hanson, G. Ibbott, “Response to ‘Comment
planar semiconductor sensors for a Medipix3-based on “Calculated absorbed-dose ratios, TG51/TG21, for
radiography detector,” Journal of Instrumentation 6 most widely used cylindrical and parallel-plate ion
(11), C11019–C11019 (2011). https://doi.org/10.1088/ chambers over a range of photon and electron energies”
1748-0221/6/11/C11019 ‘[Med. Phys. 30, 473–477 (2003)],” Medical Physics 30
18. X. Wu, J. Kalliopuska, S. Eränen, et al., “Recent advances (3), 478–480 (2003). https://doi.org/10.1118/1.1541252
in processing and characterization of edgeless detec- 32. D. Marre and G. Marinello, “Comparison of P-type
tors,” Journal of Instrumentation 7 (02), C02001–C02001 commercial electron diodes for in-vivo dosimetry,”
(2012). https://doi.org/10.1088/1748-0221/7/02/C02001 Medical Physics 31 (1), 50–56 (2004).
33. A. S. Saini and T. C. Zhu, “Energy dependence of com- 45. M. Tyler, P. Z.Y. Liu, K. W. Chan, et al., “Characterization
mercially available diode detectors for in-vivo dosim- of small-field stereotactic radiosurgery beams with
etry,” Medical Physics 34 (5), 1704–1711 (2007). https:// modern detectors,” Physics in Medicine and Biology
doi.org/10.1118/1.2719365 58 (21), 7595–7608 (2013). https://doi.org/10.1088/
34. P. Andreo, “The physics of small megavoltage pho- 0031-9155/58/21/7595
ton beam dosimetry,” Radiotherapy and Oncology 46. S. Derreumaux, C. Bassinet, C. Huet, et al., “SU-E-T-163:
126 (2), 205–213 (2018). https://doi.org/10.1016/j. Characterization of the response of active detectors and
radonc.2017.11.001 passive dosemeters used for dose measurement in small
35. H. K. Looe, D. Harder, B. Poppe, “Understanding photon beams,” Medical Physics 38 (6 Part 12), 3523–
the lateral dose response functions of high-reso- 3523 (2011). https://doi.org/10.1118/1.3612113
lution photon detectors by reverse Monte Carlo 47. G. Reggiori, P. Mancosu, N. Suchowerska, et al.,
and deconvolution analysis,” Physics in Medicine “Characterization of a new unshielded diode for small
& Biology 60 (16), 6585–6607 (2015). https://doi. field dosimetry under flattening filter free beams,”
org/10.1088/0031-9155/60/16/6585 Physica Medica 32 (2), 408–413 (2016). https://doi.
36. P. Andreo and H. Benmakhlouf, “Role of the density, org/10.1016/j.ejmp.2016.02.004
density effect and mean excitation energy in solid-state 48. P. Z. Y. Liu, G. Reggiori, F. Lobefalo, et al., “Small field
detectors for small photon fields,” Physics in Medicine correction factors for the IBA razor,” Physica Medica
and Biology 62 (4), 1518–1532 (2017). https://doi. 32 (8), 1025–1029 (2016). https://doi.org/10.1016/j.
org/10.1088/1361-6560/aa562e ejmp.2016.07.004
37. R. Alfonso, P. Andreo, R. Capote, et al., “A new for- 49. T. C. Zhu, L. Ding, C. R. Liu, et al., “Performance evalu-
malism for reference dosimetry of small and nonstan- ation of a diode array for enhanced dynamic wedge
dard fields,” Medical Physics 35 (11), 5179–5186 (2008). dosimetry,” Medical Physics 24 (7), 1173–1180 (1997).
https://doi.org/10.1118/1.3005481 https://doi.org/10.1118/1.598019
38. IAEA, Technical Reports Series No. 483 – Dosimetry of 50. G. A. Ezzell, J. W. Burmeister, N. Dogan, et al., “IMRT
Small Static Fields Used in External Beam Radiotherapy commissioning: multiple institution planning and
An International Code of Practice for Reference and dosimetry comparisons, a report from AAPM Task
Relative Dose Determination (2017). http://www-ns. Group 119,” Medical Physics 36 (11), 5359–5373 (2009).
iaea.org/standards/ https://doi.org/10.1118/1.3238104
39. P. Francescon, S. Cora, C. Cavedon, “Total scatter fac- 51. M. Ahmad, J. Deng, M. W. Lund, et al., “Clinical
tors of small beams: a multidetector and Monte Carlo implementation of enhanced dynamic wedges into
study,” Medical Physics 35 (2), 504–513 (2008). https:// the pinnacle treatment planning system: Monte
doi.org/10.1118/1.2828195 Carlo validation and patient-specific QA,” Physics in
40. G. Cranmer-Sargison, S. Weston, J. A. Evans, et al., Medicine and Biology 54 (2), 447–465 (2009). https://doi.
“Monte Carlo modelling of diode detectors for small org/10.1088/0031-9155/54/2/018
field MV photon dosimetry: detector model simplifica- 52. A. Bocci, M. A. Cortés-Giraldo, M. I. Gallardo, et al.,
tion and the sensitivity of correction factors to source “Silicon strip detector for a novel 2D dosimetric method
parameterization,” Physics in Medicine and Biology for radiotherapy treatment verification,” Nuclear
57 (16), 5141–5153 (2012). https://doi.org/10.1088/ Instruments and Methods in Physics Research, Section A:
0031-9155/57/16/5141 Accelerators, Spectrometers, Detectors and Associated
41. D. Czarnecki and K. Zink, “Monte Carlo calculated cor- Equipment 673, 98–106 (2012). https://doi.org/10.1016/j.
rection factors for diodes and ion chambers in small nima.2012.01.018
photon fields,” Physics in Medicine & Biology 58 (3) 53. E. Pappas, T. G. Maris, F. Zacharopoulou, et al.,
(2013). https://doi.org/10.1088/0031-9155/59/3/791 “Small SRS photon field profile dosimetry performed
42. C. Moignier, C. Huet, L. Makovicka, “Determination using a PinPoint air ion chamber, a diamond detec-
of the correction factors for detectors used with an 800 tor, a novel silicon-diode array (DOSI), and poly-
MU/Min CyberKnife system equipped with fixed colli- mer gel dosimetry. Analysis and intercomparison,”
mators and a study of detector response to small photon Medical Physics 35 (10), 4640–4648 (2008). https://doi.
beams using a Monte Carlo method,” Medical Physics org/10.1118/1.2977829
41 (7), 071702 (2014). https://doi.org/10.1118/1.4881098 54. D. Menichelli, M. Bruzzi, M. Bucciolini, et al., “Design
43. P. H. Charles, S. B. Crowe, T. Kairn, et al., “Monte Carlo- and development of a silicon-segmented detector for
based diode design for correction-less small field dosim- 2D dose measurements in radiotherapy,” Nuclear
etry,” Physics in Medicine & Biology 58 (13), 4501–4512 Instruments and Methods in Physics Research, Section A:
(2013). https://doi.org/10.1088/0031-9155/58/13/4501 Accelerators, Spectrometers, Detectors and Associated
44. T. S. A. Underwood, H. C. Winter, M. A. Hill, et al., Equipment 583 (1), 109–113 (2007). https://doi.
“Detector density and small field dosimetry: integral ver- org/10.1016/j.nima.2007.08.209
sus point dose measurement schemes,” Medical Physics 55. J. H. D. Wong, M. Carolan, M. L. F. Lerch, et al., “A
40 (8), 082102 (2013). https://doi.org/10.1118/1.4812687 silicon strip detector dose magnifying glass for IMRT
dosimetry,” Medical Physics 37 (2), 427–439 (2010). dosimetry,” IEEE Transactions on Nuclear Science 65
https://doi.org/10.1118/1.3264176 (9), 2640–2649 (2018). https://doi.org/10.1109/TNS.
56. E. Debrot, M. Newall, S. Guatelli, et al., “A silicon strip 2018.2860625
detector array for energy verification and quality assur- 68. G. Biasi, N. Hardcastle, M. Petasecca, et al., “On the
ance in heavy ion therapy,” Medical Physics 45 (2), 953– instantaneous dose rate and angular dependence of
962 (2018). https://doi.org/10.1002/mp.12736 monolithic silicon array detectors,” IEEE Transactions
57. M. Petasecca, M. K. Newall, J. T. Booth, et al., on Nuclear Science 66 (1), 519–527 (2019). https://doi.
“MagicPlate-512: A 2D silicon detector array for qual- org/10.1109/TNS.2018.2885017
ity assurance of stereotactic motion adaptive radiother- 69. N. Stansook, G. Biasi, K. Utitsarn, et al., “2D Monolithic
apy,” Medical Physics 42 (6), 2992–3004 (2015). https:// silicon-diode array detectors in megavoltage photon
doi.org/10.1118/1.4921126 beams: does the fabrication technology matter? A medi-
58. K. A. Shukaili, M. Petasecca, M. Newall, et al., “A 2D cal physicist’s perspective,” Australasian Physical &
silicon detector array for quality assurance in small Engineering Sciences in Medicine 42 (2), 443–451 (2019).
field dosimetry: DUO,” Medical Physics 44 (2), 628–636 https://doi.org/10.1007/s13246-019-00736-7
(2017). https://doi.org/10.1002/mp.12060 70. S. Stathakis, P. Myers, C. Esquivel, et al., “Characterization
59. G. Biasi, M. Petasecca, S. Guatelli, N. Hardcastle, of a novel 2D array dosimeter for patient-specific
et al., “A novel high-resolution 2D silicon array detec- quality assurance with volumetric arc therapy,”
tor for small field dosimetry with FFF photon beams,” Medical Physics 40 (7) (2013). https://doi.org/10.1118/
Physica Medica 45 (January), 117–126 (2018). https:// 1.4812415
doi.org/10.1016/j.ejmp.2017.12.010 71. C. K. McGarry, B. F. O’Connell, M. W. D. Grattan, et
60. G. Biasi, K. Al Shukaili, M. Petasecca, et al., “Today’s al., “Octavius 4D characterization for flattened and flat-
monolithic silicon array detector for small field tening filter free rotational deliveries,” Medical Physics
dosimetry: the Octa,” Journal of Physics: Conference 40 (9) (2013). https://doi.org/10.1118/1.4817482
Series 1154 (January), 012002 (2019). https://doi. 72. R. Sadagopan, J. A. Bencomo, R. L. Martin, et al.,
org/10.1088/1742-6596/1154/1/012002 “Characterization and clinical evaluation of a novel
61. N. Stansook, K. Utitsarn, M. Petasecca, et al., “Technical IMRT quality assurance system,” Journal of Applied
note: angular dependence of a 2D monolithic silicon Clinical Medical Physics 10 (2), 104–119 (2009). https://
diode array for small field dosimetry,” Medical Physics 44 doi.org/10.1120/jacmp.v10i2.2928
(8), 4313–4321 (2017). https://doi.org/10.1002/mp.12377 73. V. Feygelman, G. G. Zhang, C. Stevens, et al., “Evaluation
62. M. Duncan, M. K. Newall, V. Caillet, et al., “Real-time of a new VMAT QA device, or the ‘X’ and ‘O’ array geom-
high spatial resolution dose verification in stereotactic etries,” Journal of Applied Clinical Medical Physics 12 (2),
motion adaptive arc radiotherapy,” Journal of Applied 3346 (2011). https://doi.org/10.1120/jacmp.v12i2.3346
Clinical Medical Physics 19 (4), 173–184 (2018). https:// 74. G. A. Ausman and F. B. McLean, “Electron−hole pair
doi.org/10.1002/acm2.12364 creation energy in SiO2,” Applied Physics Letters 26 (4),
63. K. Utitsarn, G. Biasi, N. Stansook, et al., “Two- 173–175 (1975). https://doi.org/10.1063/1.88104
dimensional solid-state array detectors: a technique for 75. R. C. Hughes, “Charge-carrier transport phenomena
in vivo dose verification in a variable effective area,” in amorphous SiO: direct measurement of the drift
Journal of Applied Clinical Medical Physics 20 (11), 88– mobility and lifetimes,” Physical Review Letters 30 (26),
94 (2019). https://doi.org/10.1002/acm2.12744 1333–1336 (1973). https://doi.org/10.1103/PhysRevLett.
64. S. J. Alnaghy, T. Causer, M. Gargett, et al., “A feasibility 30.1333
study for high-resolution silicon array detector perfor- 76. R. C. Hughes, E. P. EerNisse, H. J. Stein, “Hole transport
mance in the magnetic field of a permanent magnet sys- in MOS oxides,” IEEE Transactions on Nuclear Science
tem,” Medical Physics 46 (9), 4224–4232 (2019). https:// 22 (6) (1975).
doi.org/10.1002/mp.13686 77. T. R. Oldham and F. B. McLean, “Total ionizing dose
65. G. Biasi, M. Petasecca, S. Guatelli, M. A. Ebert, et al., effects in MOS oxides and devices,” IEEE Transactions
“CyberKnife® fixed cone and IrisTM defined small radia- on Nuclear Science 50 III (3), 483–499 (2003). https://
tion fields: assessment with a high-resolution solid-state doi.org/10.1109/TNS.2003.812927
detector array,” Journal of Applied Clinical Medical 78. J. R. Schwank, M. R. Shaneyfelt, D. M. Fleetwood, et al.,
Physics 19 (5), 547–557 (2018). https://doi.org/10.1002/ “Radiation effects in MOS oxides,” IEEE Transactions
acm2.12414 on Nuclear Science 55 (4), 1833–1853 (2008). https://doi.
66. K. Utitsarn, Z. A. Alrowaili, N. Stansook, et al., “The org/10.1109/TNS.2008.2001040
effect of an air gap on a 2D monolithic silicon detec- 79. R. Freeman and A. Holmes-Siedle, “A simple model
tor for relative dosimetry,” Journal of Instrumentation for predicting radiation effects in MOS devices,” IEEE
14 (06), P06018–P06018 (2019). https://doi.org/10.1088/ Transactions on Nuclear Science 25 (6), 1216–1225
1748-0221/14/06/P06018 (1978). https://doi.org/10.1109/TNS.1978.4329516
67. G. Biasi, J. Davis, M. Petasecca, et al., “On monolithic 80. G. Ristić, S. Golubović, M. Pejović, “Sensitivity and fad-
silicon array detectors for small-field photon beam ing of PMOS dosimeters with thick gate oxide,” Sensors
and Actuators A: Physical 51 (2–3), 153–158 (1995). 93. A. B. Rosenfeld, M. G. Carolan, G. I. Kaplan, et al.,
https://doi.org/10.1016/0924-4247(95)01211-7 “MOSFET dosimeters: the role of encapsulation on
81. A. Jaksic, G. Ristic, M. Pejovic, A. Mohammadzadeh, dosimetric characteristics in mixed gamma-neutron
W. Lane, “Characterisation of radiation response of and megavoltage X-ray fields,” IEEE Transactions on
400 Nm implanted gate oxide RADFETs,” in: 2002 Nuclear Science 42 (6), 1870–1877 (1995). https://doi.
23rd International Conference on Microelectronics. org/10.1109/23.489229
Proceedings (Cat. No.02TH8595), vol. 2, pp. 727–730 94. B. Wang, C. H. Kim, X. G. Xu, “Monte Carlo model-
(2002). https://doi.org/10.1109/MIEL.2002.1003360 ing of a high-sensitivity MOSFET dosimeter for low-
82. S. Alshaikh, M. Carolan, M. Petasecca, et al., “Direct and medium-energy photon sources,” Medical Physics
and pulsed current annealing of P-MOSFET based 31 (5), 1003–1008 (2004). https://doi.org/10.1118/
dosimeter: the ‘MOSkin.’” Australasian Physical and 1.1688272
Engineering Sciences in Medicine 37 (2), 311–319 (2014). 95. P. Scalchi and P. Francescon, “Calibration of a MOSFET
https://doi.org/10.1007/s13246-014-0261-1 detection system for 6-MV in vivo dosimetry,”
83. G. W. Luo, Z. Y. Qi, X. W. Deng, et al., “Investigation of International Journal of Radiation Oncology Biology
a pulsed current annealing method in reusing MOSFET Physics 40 (4), 987–993 (1998). https://doi.org/10.1016/
dosimeters for in vivo IMRT dosimetry,” Medical Physics S0360-3016(97)00894-8
41 (5), 1–7 (2014). https://doi.org/10.1118/1.4871619 96. S. Qin, T. Chen, L. Wang, et al., “Angular dependence of
84. D. M. Fleetwood, T. L. Meisenheimer, J. H. Scofield, the MOSFET dosimeter and its impact on in vivo surface
“1/f noise and radiation effects in MOS devices,” IEEE dose measurement in breast cancer treatment,” Technology
Transactions on Electron Devices 41 (11), 1953–1964 in Cancer Research & Treatment 13 (4), 345–352 (2014).
(1994). https://doi.org/10.1109/16.333811 https://doi.org/10.7785/tcrt.2012.500382
85. C. Benson, R. A. Price, J. Silvie, et al., “Radiation- 97. J. Seco, B. Clasie, M. Partridge, “Review on the char-
induced statistical uncertainty in the threshold volt- acteristics of radiation detectors for dosimetry and
age measurement of MOSFET dosimeters,” Physics in imaging,” Physics in Medicine & Biology 59 (20),
Medicine and Biology 49 (14), 3145–3159 (2004). https:// R303–R347 (2014). https://doi.org/10.1088/0031-9155/
doi.org/10.1088/0031-9155/49/14/009 59/20/R303
86. A. Jaksic, G. Ristic, M. Pejovic, A. Mohammadzadeh, 98. T. Cheung, M. J. Butson, P. K. N. Yu, “Effects of tem-
C. Sudre, et al., “Gamma-ray irradiation and post- perature variation on MOSFET dosimetry,” Physics
irradiation responses of high dose range RADFETs,” in Medicine and Biology 49 (13), N191–N196 (2004).
IEEE Transactions on Nuclear Science 49 (3), 1356–1363 https://doi.org/10.1088/0031-9155/49/13/N02
(2002). https://doi.org/10.1109/TNS.2002.1039667 99. M. G. Buehler, B. R. Blaes, G. A. Soli, et al., “On-chip
87. M. W. Bower and D. E. Hintenlang, “The characterization p-MOSFET dosimetry (CMOS ICs),” IEEE Transactions
of a commercial MOSFET dosimeter system for use in on Nuclear Science 40 (6), 1442–1449 (1993). https://doi.
diagnostic X-ray,” Health Physics 75 (2), 197–204 (1998). org/10.1109/23.273520
https://doi.org/10.1097/00004032-199808000-00013 100. M. Soubra, J. Cygler, G. Mackay, “Evaluation of a
88. C. R. Edwards, S. Green, J. E. Palethorpe, et al., “The dual bias dual metal oxide-silicon semiconductor
response of a MOSFET, p-type semiconductor and field effect transistor detector as radiation dosimeter,”
LiF TLD to quasi-monoenergetic x-rays,” Physics in Medical Physics 21 (4), 567–572 (1994). https://doi.
Medicine and Biology 42 (12), 2383–2391 (1997). https:// org/10.1118/1.597314
doi.org/10.1088/0031-9155/42/12/006 101. D. M. Fleetwood, P. S. Winokur, J. R. Schwank,
89. T. Kron, L. Duggan, T. Smith, et al., “Dose response “Using laboratory X-ray and cobalt-60 irradia-
of various radiation detectors to synchrotron radiations to predict CMOS device response in strate-
tion,” Physics in Medicine and Biology 43 (11), 3235– gic and space environments,” IEEE Transactions on
3259 (1998). https://doi.org/10.1088/0031-9155/43/ Nuclear Science 35 (6), 1497–1505 (1988). https://doi.
11/006 org/10.1109/23.25487
90. R. Ramani, S. Russell, P. O’Brien, “Clinical dosimetry 102. J. R. Schwank, S. B. Roeske, D. E. Beutler, et al., “A dose
using MOSFETS,” International Journal of Radiation rate independent PMOS dosimeter for space applica-
Oncology Biology Physics 37 (4), 959–964 (1997). https:// tions,” IEEE Transaction of Nuclear Science 43 (6), 2671–
doi.org/10.1016/S0360-3016(96)00600-1 2678 (1996). https://doi.org/10.1109/23.556852
91. R. Ramaseshan, K. S. Kohli, T. J. Zhang, et al., 103. W. Poch and A. G. Holmes-Siedle, “The dosimeter:
“Performance characteristics of a MicroMOSFET as a new instrument to measure radiation dose,” RCA
an in vivo dosimeter in radiation therapy,” Physics in Engineers 16 (3), 56–59 (1970).
Medicine and Biology 49 (17), 4031–4048 (2004). https:// 104. A. Holmes-Siedle, “The space-charge dosimeter,”
doi.org/10.1088/0031-9155/49/17/014 Nuclear Instruments and Methods 121 (1), 169–179
92. P. H. Halvorsen, “Dosimetric evaluation of a new design (1974). https://doi.org/10.1016/0029-554X(74)90153-0
MOSFET in vivo dosimeter,” Medical Physics 32 (1), 105. L. Adams and A. Holmes-Siedle, “The development
110–117 (2005). https://doi.org/10.1118/1.1827771 of an MOS dosimetry unit for use in space,” IEEE
Transactions on Nuclear Science 25 (6), 1607–1612 117. H.-M. Lu, G. Mann, E. Cascio, “Investigation of an
(1978). https://doi.org/10.1109/TNS.1978.4329580 implantable dosimeter for single-point water equiv-
106. C. Ehringfeld, S. Schmid, K. Poljanc, et al., “Application alent path length verification in proton therapy,”
of commercial MOSFET detectors for in vivo dosimetry Medical Physics 37 (11), 5858–5866 (2010). https://doi.
in the therapeutic X-ray range from 80 kV to 250 kV,” org/10.1118/1.3504609
Physics in Medicine and Biology 50 (2), 289–303 (2005). 118. A. B. Rosenfeld, G. I. Kaplanl, T. Kron, et al., “MOSFET
https://doi.org/10.1088/0031-9155/50/2/008 dosimetry of an X-ray microbeam,” IEEE Transactions
107. P. Scalchi, R. Righetto, C. Cavedon, et al., “Direct tumor on Nuclear Science 46 (6) (1999).
in vivo dosimetry in highly-conformal radiotherapy: a 119. G. I. Kaplan, A. B. Rosenfeld, B. J. Allen, et al., “Improved
feasibility study of implantable MOSFETs for hypofrac- spatial resolution by MOSFET dosimetry of an X-ray
tionated extracranial treatments using the Cyberknife microbeam,” Medical Physics 27 (1), 239–244 (2000).
system,” Medical Physics 37 (4), 1413–1423 (2010). https://doi.org/10.1118/1.598866
https://doi.org/10.1118/1.3315370 120. A. B. Rosenfeld, M. L. F. Lerch, T. Kron, et al., “Feasibility
108. W. L. Jong, N. M. Ung, A. Vannyat, et al., “‘Edge-on’ study of online high-spatial-resolution MOSFET dosim-
MOSkin detector for stereotactic beam measurement etry in static and pulsed X-ray radiation fields,” IEEE
and verification,” Physica Medica 33, 127–135 (2017). Transactions on Nuclear Science 48 (6), 2061–2068 (2001).
https://doi.org/10.1016/j.ejmp.2016.12.020 121. A. B. Rosenfeld, E. A. Siegbahn, E. Brauer-Krish, et al.,
109. E. J. Bloemen-van Gurp, A. W. H Minken, B. J. Mijnheer, “Edge-on face-to-face MOSFET for synchrotron micro-
et al., “Clinical implementation of MOSFET detec- beam dosimetry: MC modeling,” IEEE Transactions on
tors for dosimetry in electron beams,” Radiotherapy Nuclear Science 52 (6), 2562–2569 (2005). https://doi.
and Oncology 80 (3), 288–295 (2006). https://doi. org/10.1109/TNS.2005.860704
org/10.1016/j.radonc.2006.07.002 122. E. A. Siegbahn, E. Bräuer-Krisch, A. Bravin, et al.,
110. W. L. Jong, N. M. Ung, A. H. L. Tiong, et al., “MOSFET dosimetry with high spatial resolution in
“Characterisation of a MOSFET-based detector for dose intense synchrotron-generated x-ray microbeams,”
measurement under megavoltage electron beam radio- Medical Physics 36 (4), 1128–1137 (2009). https://doi.
therapy,” Radiation Physics and Chemistry 144, 76–84 org/10.1118/1.3081934
(2018). https://doi.org/10.1016/j.radphyschem.2017.11.021 123. ICRP, The Biological Basis for Dose Limitation in the
111. A. Faigon, J. Lipovetzky, E. Redin, et al., “Extension of Skin (1992).
the measurement range of MOS dosimeters using radia- 124. M. J. Butson, A. Rozenfeld, J. N. Mathur, et al., “A new
tion induced charge neutralization,” IEEE Transactions radiotherapy surface dose detector: the MOSFET,”
on Nuclear Science 55 (4), 2141–2147 (2008). https://doi. Medical Physics 23 (5), 655–658 (1996). https://doi.
org/10.1109/TNS.2008.2000767 org/10.1118/1.597702
112. R. Consorti, A. Petrucci, F. Fortunato, et al., “In vivo 125. P. Scalchi, P. Francescon, P. Rajaguru, “Characterization
dosimetry with MOSFETs: dosimetric characterization of a new MOSFET detector configuration for in vivo
and first clinical results in intraoperative radiotherapy,” skin dosimetry,” Medical Physics 32 (6), 1571–1578
International Journal of Radiation Oncology Biology (2005). https://doi.org/10.1118/1.1924328
Physics 63 (3), 952–960 (2005). https://doi.org/10.1016/j. 126. I. S. Kwan, A. B. Rosenfeld, Z. Y. Qi, et al., “Skin
ijrobp.2005.02.049 dosimetry with new MOSFET detectors,” Radiation
113. M. Ciocca, V. Piazzi, R. Lazzari, et al., “Real-time in Measurements 43 (2–6), 929–932 (2008). https://doi.
vivo dosimetry using micro-MOSFET detectors dur- org/10.1016/j.radmeas.2007.12.052
ing intraoperative electron beam radiation therapy in 127. Z. Y. Qi, X. W. Deng, S. M. Huang, et al., “In vivo verifi-
early-stage breast cancer,” Radiotherapy and Oncology cation of superficial dose for head and neck treatments
78 (2), 213–216 (2006). https://doi.org/10.1016/j. using intensity-modulated techniques,” Medical Physics
radonc.2005.11.011 36 (1), 59–70 (2009). https://doi.org/10.1118/1.3030951
114. P. Francescon, S. Cora, C. Cavedon, et al., “Use of a new 128. A. Quinn, L. Holloway, D. Cutajar, et al., “Megavoltage
type of radiochromic film, a new parallel-plate micro- cone beam CT near surface dose measurements:
chamber, MOSFETs, and TLD 800 microcubes in the potential implications for breast radiotherapy,”
dosimetry of small beams,” Medical Physics 25 (4), 503– Medical Physics 38 (11), 6222–6227 (2011). https://doi.
511 (1998). https://doi.org/10.1118/1.598227 org/10.1118/1.3641867
115. O. A. Sauer and J. Wilbert, “Measurement of output fac- 129. M. J. Safari, J. H. D. Wong, K. A. A. Kadir, et al., “Real-
tors for small photon beams,” Medical Physics 34 (6), time eye lens dose monitoring during cerebral angiog-
1983–1988 (2007). https://doi.org/10.1118/1.2734383 raphy procedures,” European Radiology 26 (1), 79–86
116. R. Kohno, T. Nishio, T. Miyagishi, et al., “Experimental (2016). https://doi.org/10.1007/s00330-015-3818-9
evaluation of a MOSFET dosimeter for proton dose 130. N. Hardcastle, D. L. Cutajar, P. E. Metcalfe, et al., “In vivo
measurements,” Physics in Medicine and Biology real-time rectal wall dosimetry for prostate radiother-
51 (23), 6077–6086 (2006). https://doi.org/10.1088/ apy,” Physics in Medicine and Biology 55 (13), 3859–3871
0031-9155/51/23/009 (2010). https://doi.org/10.1088/0031-9155/55/13/019
131. S. J. Alnaghy, S. Deshpande, D. L. Cutajar, et al., “In vivo of radiation dose in tissue during cancer therapy,”
endorectal dosimetry of prostate tomotherapy using IEEE Sensors Journal 8 (1), 38–51 (2008). https://doi.
dual MOSkin detectors,” Journal of Applied Clinical org/10.1109/JSEN.2007.912542
Medical Physics 16 (3), 107–117 (2015). https://doi. 143. C. W. Scarantino, B. R. Prestidge, M. S. Anscher, et al.,
org/10.1120/jacmp.v16i3.5113 “The observed variance between predicted and mea-
132. E. A. Alhakeem, S. AlShaikh, A. B. Rosenfeld, et al., sured radiation dose in breast and prostate patients
“Comparative evaluation of modern dosimetry utilizing an in vivo dosimeter,” International Journal
techniques near low- and high-density heteroge- of Radiation Oncology Biology Physics 72 (2), 597–604
neities,” Journal of Applied Clinical Medical Physics (2008). https://doi.org/10.1016/j.ijrobp.2008.05.058
16 (5), 142–158 (2015). https://doi.org/10.1120/jacmp. 144. C. Esquivel, P. Rassiah, G. Beyer, et al., “SU-FF-T-203:
v16i5.5589 evaluation and performance characteristics of an
133. C. Tenconi, M. Carrara, M. Borroni, et al., “TRUS- implantable dosimeter with the TomoTherapy Hi-Art
probe integrated MOSkin detectors for rectal wall in System,” Medical Physics 34 (6), 2448 (2007). https://doi.
vivo dosimetry in HDR brachytherapy: in phantom org/10.1118/1.2760864
feasibility study,” Radiation Measurements 71, 379–383 145. E. A. Burgemeister, “Dosimetry with a diamond oper-
(2014). https://doi.org/10.1016/j.radmeas.2014.05.010 ating as a resistor,” Physics in Medicine and Biology
134. M. Carrara, C. Tenconi, G. Rossi, et al., “In vivo rectal 26 (2), 269–275 (1981). https://doi.org/10.1088/0031-
wall measurements during HDR prostate brachytherapy 9155/26/2/006
with MOSkin dosimeters integrated on a trans-rectal 146. S. F. Kozlov, R. Stuck, M. Hage-Ali, et al., “Preparation
US probe: comparison with planned and reconstructed and characteristics of natural diamond nuclear radiation
doses,” Radiotherapy and Oncology 118 (1), 148–153 detectors,” IEEE Transactions on Nuclear Science 22 (1),
(2016). https://doi.org/10.1016/j.radonc.2015.12.022 160–170 (1975). https://doi.org/10.1109/TNS.1975.4327634
135. C. W. Scarantino, D. M. Ruslander, C. J. Rini, et al., 147. W. Kada, N. Iwamoto, T. Satoh, et al., “Continuous
“An implantable radiation dosimeter for use in external observation of polarization effects in thin SC-CVD
beam radiation therapy,” Medical Physics 31 (9), 2658– diamond detector designed for heavy ion microbeam
2671 (2004). https://doi.org/10.1118/1.1778809 measurement,” Nuclear Instruments and Methods in
136. R. D. Black, C. W. Scarantino, G. G. Mann, et al., “An Physics Research Section B: Beam Interactions with
analysis of an implantable dosimeter system for exter- Materials and Atoms 331 (July), 113–116 (2014). https://
nal beam therapy,” International Journal of Radiation doi.org/10.1016/j.nimb.2013.11.040
Oncology Biology Physics 63 (1), 290–300 (2005). https:// 148. M. Pomorski, E. Berdermann, W. de Boer, et al., “Charge
doi.org/10.1016/j.ijrobp.2005.05.025 transport properties of single crystal CVD-diamond
137. T. M. Briere, A. S. Beddar, M. T. Gillin, “Evaluation of pre- particle detectors,” Diamond and Related Materials
calibrated implantable MOSFET radiation dosimeters 16 (4–7), 1066–1069 (2007). https://doi.org/10.1016/j.
for megavoltage photon beams,” Medical Physics 32 (11), diamond.2006.11.016
3346–3349 (2005). https://doi.org/10.1118/1.2065447 149. V. Grilj, N. Skukan, M. Jakšić, et al., “Irradiation of
138. A. S. Beddar, M. Salehpour, T. M. Briere, et al., thin diamond detectors and radiation hardness tests
“Preliminary evaluation of implantable MOSFET using MeV protons,” Nuclear Instruments and Methods
radiation dosimeters,” Physics in Medicine and Biology in Physics Research Section B: Beam Interactions with
50 (1), 141–149 (2005). https://doi.org/10.1088/0031- Materials and Atoms 306 (July), 191–194 (2013). https://
9155/50/1/011 doi.org/10.1016/j.nimb.2012.12.034
139. C. W. Scarantino, C. J. Rini, M. Aquino, et al., “Initial 150. J. F. Fowler, “Solid state electrical conductivity
clinical results of an in vivo dosimeter during exter- dosimeters,” in: Radiation Dosimetry: Volume II:
nal beam radiation therapy,” International Journal of Instrumentation, edited by E. Tochili, F. H. Attix, W. C.
Radiation Oncology Biology Physics 62 (2), 606–613 Roesch (Academic Press, 1966).
(2005). https://doi.org/10.1016/j.ijrobp.2004.09.041 151. P. W. Hoban, M. Heydarian, W. A. Beckham, et al.,
140. G. P. Beyer, C. W. Scarantino, B. R. Prestidge, et al., “Dose rate dependence of a PTW diamond detector
“Technical evaluation of radiation dose delivered in in the dosimetry of a 6 MV photon beam,” Physics in
prostate cancer patients as measured by an implantable Medicine and Biology 39 (8), 1219–1229 (1994). https://
MOSFET dosimeter,” International Journal of Radiation doi.org/10.1088/0031-9155/39/8/003
Oncology Biology Physics 69 (3), 925–935 (2007). https:// 152. W. U. Laub, T. W. Kaulich, F. Nüsslin, “A diamond detec-
doi.org/10.1016/j.ijrobp.2007.06.065 tor in the dosimetry of high-energy electron and photon
141. T. M. Briere, M. T. Gillin, A. S. Beddar, “Implantable beams,” Physics in Medicine and Biology 44 (9), 2183–
MOSFET detectors: evaluation of a new design,” 2192 (1999). https://doi.org/10.1088/0031-9155/44/9/306
Medical Physics 34 (12), 4585–4590 (2007). https://doi. 153. N. Ade and T. L. Nam, “The influence of defect lev-
org/10.1118/1.2799578 els on the dose rate dependence of synthetic diamond
142. G. P. Beyer, G. G. Mann, J. A. Pursley, et al., “An detectors of various types on exposures to high-
implantable MOSFET dosimeter for the measurement energy radiotherapy beams,” Radiation Physics and
Chemistry 108, 65–73 (2015). https://doi.org/10.1016/j. 166. D. Tromson, M. Rebisz-Pomorska, N. Tranchant, et al.,
radphyschem.2014.11.016 “Single crystal CVD diamond detector for high reso-
154. C. De Angelis, S. Onori, M. Pacilio, et al., “An inves- lution dose measurement for IMRT and novel radia-
tigation of the operating characteristics of two PTW tion therapy needs,” Diamond and Related Materials
diamond detectors in photon and electron beams,” 19 (7–9), 1012–1016 (2010). https://doi.org/10.1016/j.
Medical Physics 29 (2), 248–254 (2002). https://doi. diamond.2010.03.008
org/10.1118/1.1446101 167. W. U. Laub and R. Crilly, “Clinical radiation therapy
155. P. N. Mobit and G. A. Sandison, “An EGS4 Monte measurements with a new commercial synthetic single
Carlo examination of the response of a PTW-diamond crystal diamond detector,” Journal of Applied Clinical
radiation detector in megavoltage electron beams,” Medical Physics 15 (6), 92–102 (2014). https://doi.
Medical Physics 26 (5), 839–844 (1999). https://doi. org/10.1120/jacmp.v15i6.4890
org/10.1118/1.598593 168. G. T. Betzel, S. P. Lansley, F. Baluti, et al., “Clinical
156. I. Ciancaglioni, M. Marinelli, E. Milani, et al., investigations of a CVD diamond detector for radio-
“Dosimetric characterization of a synthetic single crys- therapy dosimetry,” Physica Medica 28 (2), 144–152
tal diamond detector in clinical radiation therapy small (2012). https://doi.org/10.1016/j.ejmp.2011.04.003
photon beams,” Medical Physics 39 (7 Part 1), 4493– 169. S. Spadaro, G. Conte, M. Pimpinella, et al., “Electrical
4501 (2012). https://doi.org/10.1118/1.4729739 and dosimetric characterization of a CVD diamond
157. S. Kampfer, N. Cho, S. E. Combs, et al., “Dosimetric detector with high sensitivity,” Radiation Measurements
characterization of a single crystal diamond detector 48 (January), 1–6 (2013). https://doi.org/10.1016/j.
in X-ray beams for preclinical research,” Zeitschrift Für radmeas.2012.11.017
Medizinische Physik 28 (4), 303–309 (2018). https://doi. 170. F. Marsolat, D. Tromson, N. Tranchant, et al., “Why
org/10.1016/j.zemedi.2018.05.002 diamond dimensions and electrode geometry are
158. K. G. Mckay, “Electron bombardment conductivity in crucial for small photon beam dosimetry,” Journal
diamond,” Physical Review 74 (11), 1606–1621 (1948). of Applied Physics 118 (23), 0–14 (2015). https://doi.
https://doi.org/10.1103/PhysRev.74.1606 org/10.1063/1.4937994
159. B. Planskoy, “Evaluation of diamond radiation doseme- 171. S. Almaviva, M. Marinelli, E. Milani, et al., “Synthetic
ters,” Physics in Medicine and Biology 25 (3), 519–532 single crystal diamond diodes for radiotherapy dosim-
(1980). https://doi.org/10.1088/0031-9155/25/3/011 etry,” Nuclear Instruments and Methods in Physics
160. M. Heydarian, P. W. Hoban, W. A. Beckham, et al., Research Section A: Accelerators, Spectrometers,
“Evaluation of a PTW diamond detector for electron Detectors and Associated Equipment 594 (2), 273–277
beam measurements,” Physics in Medicine and Biology (2008). https://doi.org/10.1016/j.nima.2008.06.028
38 (8), 1035–1042 (1993). https://doi.org/10.1088/ 172. S. Almaviva, I. Ciancaglioni, R. Consorti, et al.,
0031-9155/38/8/002 “Synthetic single crystal diamond dosimeters for
161. P. Bjork, T. Knoos, P. Nilsson, “Comparative dosimetry intensity modulated radiation therapy applications,”
of diode and diamond detectors in electron beams for Nuclear Instruments and Methods in Physics Research,
intraoperative radiation therapy,” Medical Physics 27 Section A: Accelerators, Spectrometers, Detectors and
(11), 2580–2588 (2000). https://doi.org/10.1118/1.1315317 Associated Equipment 608 (1), 191–194 (2009). https://
162. A. Fidanzio, L. Azario, R. Miceli, et al., “PTW-diamond doi.org/10.1016/j.nima.2009.07.004
detector: dose rate and particle type dependence,” 173. S. Almaviva, I. Ciancaglioni, R. Consorti, et al.,
Medical Physics 27 (11), 2589–2593 (2000). https://doi. “Synthetic single crystal diamond dosimeters for con-
org/10.1118/1.1318218 formal radiation therapy application,” Diamond and
163. A. Fidanzio, L. Azario, R. Kalish, et al., “A preliminary Related Materials 19 (2–3), 217–220 (2010). https://doi.
dosimetric characterization of chemical vapor depo- org/10.1016/j.diamond.2009.10.007
sition diamond detector prototypes in photon and 174. C. Di Venanzio, M. Marinelli, E. Milani, et al.,
electron radiotherapy beams,” Medical Physics 32 (2), “Characterization of a synthetic single crystal dia-
389–395 (2005). https://doi.org/10.1118/1.1851887 mond Schottky diode for radiotherapy electron beam
164. C. Descamps, D. Tromson, N. Tranchant, et al., “Clinical dosimetry,” Medical Physics 40 (2) (2013). https://doi.
studies of optimised single crystal and polycrystal- org/10.1118/1.4774360
line diamonds for radiotherapy dosimetry,” Radiation 175. J. E. Morales, S. B. Crowe, R. Hill, et al., “Dosimetry of
Measurements 43 (2–6), 933–938 (2008). https://doi. cone-defined stereotactic radiosurgery fields with a com-
org/10.1016/j.radmeas.2007.11.080 mercial synthetic diamond detector,” Medical Physics
165. N. Tranchant, D. Tromson, C. Descamps, et al., “High 41 (11), 111702 (2014). https://doi.org/10.1118/1.4895827
mobility single crystal diamond detectors for dosime- 176. A. Ralston, M. Tyler, P. Liu, et al., “Over-response of syn-
try: application to radiotherapy,” Diamond and Related thetic microdiamond detectors in small radiation fields,”
Materials 17 (7–10), 1297–1301 (2008). https://doi. Physics in Medicine and Biology 59 (19), 5873–5881
org/10.1016/j.diamond.2008.03.025 (2014). https://doi.org/10.1088/0031-9155/59/19/5873
177. J. M. Lárraga-Gutiérrez, P. Ballesteros-Zebadúa, M. synchrotron X-ray fields,” Medical Physics 43 (7), 4283–
Rodríguez-Ponce, et al., “Properties of a commer- 4293 (2016). https://doi.org/10.1118/1.4953833
cial PTW-60019 synthetic diamond detector for the 185. D. J. O’Brien, J. Dolan, S. Pencea, et al., “Relative dosim-
dosimetry of small radiotherapy beams,” Physics in etry with an MR-linac: response of ion chambers, dia-
Medicine and Biology 60 (2), 905–924 (2015). https://doi. mond, and diode detectors for off-axis, depth dose, and
org/10.1088/0031-9155/60/2/905 output factor measurements,” Medical Physics 45 (2),
178. G. Azangwe, P. Grochowska, D. Georg, et al., “Detector 884–897 (2018). https://doi.org/10.1002/mp.12699
to detector corrections: a comprehensive experimental 186. V. De Coste, P. Francescon, M. Marinelli, et al., “Is the
study of detector specific correction factors for beam PTW 60019 microdiamond a suitable candidate for
output measurements for small radiotherapy beams,” small field reference dosimetry?” Physics in Medicine
Medical Physics 41 (7), 072103 (2014). https://doi. and Biology 62 (17), 7036–7055 (2017). https://doi.
org/10.1118/1.4883795 org/10.1088/1361-6560/aa7e59
179. A. Chalkley and G. Heyes, “Evaluation of a synthetic 187. S. Rossomme, M. Marinelli, G. Verona-Rinati, et al.,
single-crystal diamond detector for relative dosimetry “Response of synthetic diamond detectors in proton,
measurements on a CyberKnifeTM,” The British Journal carbon, and oxygen ion beams,” Medical Physics 44 (10),
of Radiology 87 (1035), 20130768 (2014). https://doi. 5445–5449 (2017). https://doi.org/10.1002/mp.12473
org/10.1259/bjr.20130768 188. C. Moignier, D. Tromson, L. de Marzi, et al.,
180. P. Papaconstadopoulos, F. Tessier, J. Seuntjens, “On “Development of a synthetic single crystal diamond
the correction, perturbation and modification of dosimeter for dose measurement of clinical proton
small field detectors in relative dosimetry,” Physics in beams,” Physics in Medicine & Biology 62 (13), 5417–
Medicine & Biology 59 (19), 5937–5952 (2014). https:// 5439 (2017). https://doi.org/10.1088/1361-6560/aa70cf
doi.org/10.1088/0031-9155/59/19/5937 189. M. Zani, M. Scaringella, C. Talamonti, et al.,
181. P. Andreo, H. Palmans, M. Marteinsdóttir, et al., “On “Bidimensional polycrystalline CVD diamond detec-
the Monte Carlo simulation of small-field micro-diator for intensity modulated radiation therapy pre-
mond detectors for megavoltage photon dosimetry,” treatment verifications,” Journal of Instrumentation 10
Physics in Medicine and Biology 61 (1), L1–L10 (2015). (03), C03046–C03046 (2015). https://doi.org/10.1088/
https://doi.org/10.1088/0031-9155/61/1/L1 1748-0221/10/03/C03046
182. D. J. O’Brien, L. León-Vintró, B. McClean, “Small field 190. M. Scaringella, M. Zani, A. Baldi, et al., “First dose-map
detector correction factors kQclin, Qmsr (fclin, fmsr) measured with a polycrystalline diamond 2D dosim-
for silicon-diode and diamond detectors with circular 6 eter under an intensity modulated radiotherapy beam,”
MV fields derived using both empirical and numerical Nuclear Instruments and Methods in Physics Research
methods,” Medical Physics 43 (1), 411 (2016). https://doi. Section A: Accelerators, Spectrometers, Detectors and
org/10.1118/1.4938584 Associated Equipment 796 (8), 89–92 (2015). https://doi.
183. P. Francescon, W. Kilby, J. M. Noll, et al., “Monte Carlo org/10.1016/j.nima.2015.03.023
simulated corrections for beam commissioning mea- 191. A. Bartoli, I. Cupparo, A. Baldi, et al., “Dosimetric char-
surements with circular and MLC shaped fields on the acterization of a 2D polycrystalline CVD diamond detec-
CyberKnife M6 system: a study including diode, micro- tor,” Journal of Instrumentation 12 (03), C03052–C03052
chamber, point scintillator, and synthetic microdiamond (2017). https://doi.org/10.1088/1748-0221/12/03/C03052
detectors,” Physics in Medicine and Biology 62 (3), 1076– 192. A. Bartoli, M. Scaringella, A. Baldi, et al., “PO-0873: 2D
1095 (2017). https://doi.org/10.1088/1361-6560/aa5610 pixelated diamond detector for patient QA in advanced
184. J. Livingstone, A. W. Stevenson, D. J. Butler, et al., radiotherapy treatments,” Radiotherapy and Oncology
“Characterization of a synthetic single crystal dia- 127 (April), S459–S460 (2018). https://doi.org/10.1016/
mond detector for dosimetry in spatially fractionated S0167-8140(18)31183-6
Chapter 5
Film Dosimetry
Sina Mossahebi
University of Maryland School of Medicine
Baltimore, Maryland
Nazanin Hoshyar
New York-Presbyterian/Queens Department of Radiation Oncology
Flushing, New York
Rao Khan and Arash Darafsheh

St. Louis, Missouri
CONTENTS
5.1 Introduction 62
5.2 Radiographic Film 62
5.3 Radiochromic Film 64
5.3.1 Film Response to Radiation 65
5.4 Factors Affecting Radiochromic Film Response 67
5.4.1 Temperature and Humidity 67
5.4.2 Dose and Dose Rate 67
5.4.3 Beam Quality and Energy 67
5.4.4 Linear Energy Transfer (LET) 67
5.4.5 Batch 68
5.4.6 Post Irradiation Time and Temporal Response 68
5.4.7 Orientation 68
5.4.8 Storage 68
5.4.9 External Magnetic Fields 68
5.5 Other Considerations 69
5.5.1 Film Handling 69
5.5.2 Scanners 69
5.5.3 Calibration Procedures 69
5.5.4 Uncertainty Budget 70
5.6 Conclusion 71
References 71
61
5.1 INTRODUCTION For radiation dosimetry applications, different mod-

Film dosimeters provide two-dimensional (2D) mea- els of films have been developed. However, only a few
surement with high spatial resolution over a large area. models have survived the evolution; EBT3 and EBT-XD
There are in general two types of films used as dosim- models by GAFchromic™ (Ashland Inc., Bridgewater,
eters: radiographic and radiochromic. Radiographic NJ) which are currently widely used in external beam
films have been phased out of most of radiotherapy therapy applications. EBT-XD film holds the same
clinics in the United States and Canada; they have structure as EBT3 but has a thinner sensitive layer and
been replaced by 2D flat panel imagers and radiochro- slightly different chemical composition. In order to use
mic films. Radiochromic films are now widely used in films at higher doses, such as stereotactic radiosurgery
radiation therapy dosimetry and particularly in quality applications, EBT-XD films are preferable due to their
assurance (QA), especially for planar dose distribution higher dynamic range (up to 40 Gy).
comparisons. These films have been very useful for in Historically, MD-V3 film was designed to cover a dose
vivo measurements, small field dosimetry as well as dose range of 1–100 Gy, and HD-V2 film was later developed to
distribution evaluation of high dose-gradient regions cover higher dose range (10–1000 Gy). These models have
because of their high spatial resolution, water equiva- smaller thicknesses of sensitive layer compared to the EBT
lency, and ease of handling [1–6]. films [16]. In addition to the above models, other models
A number of clinical applications have benefitted have been developed to measure the low doses of diagnos-
from utilizing radiochromic films. Such applications tic radiology examinations at low x-ray energies. One such
include brachytherapy, intensity-modulated radiation model is the XR-QA2 film, which has a Bi2O3 radiation
therapy (IMRT) and volumetric arc therapy (VMAT) sensitive layer attached to the reflective polyester surface
commissioning and QA, CyberKnife, linac-based ste- layer and can measure doses as low as 1 mGy. Several inves-
reotactic radiosurgery (SRS) and stereotactic body tigators have reported on the use of radiochromic films
radiation therapy (SBRT), and charged particles therapy for dosimetry of low energy x-rays and electronic brachy-
dosimetry [7–18]. therapy systems such as percentage depth dose (PDD)
Since the inception of IMRT, the use of small fields measurements of Xoft 50 kVp source. These films require
for radiotherapy has significantly increased. There has calibration at the appropriate low energies due to variation
been a steady growth of both techniques and technology in response of the film compared to water [20–23].
such as CyberKnife and linac-based SRS/SBRT where This chapter provides a summary of film-based
small fields are unavoidable. Small field geometries dosimetry and its basic principles and applications in
suffer from the loss of lateral charged particle equilib- radiation dosimetry. Furthermore, the history of film,
rium, lower signal and steep dose gradients, which make origin and mechanism of response to radiation, factors
them difficult to perform accurate dose measurement. influencing the film response, applications, and prac-
Radiochromic films, because of their high spatial reso- tical examples are also discussed in more details for
lution and dynamic range, water-equivalent density, radiochromic films.
signal linearity, dose rate and energy independence
in megavoltage (MV) range, are highly desirable for 5.2 RADIOGRAPHIC FILM
small field dosimetry. The results of film dosimetry are In radiation therapy, films (as an inexpensive dosimeter)
reported to be in close agreement with various detectors provide a high-resolution 2D map of doses for photon
for small field sizes of 2 × 2 cm2. Several studies have and electron beams. Radiographic films have been used
investigated the small field sizes (< 2 × 2 cm2). Gonzalez- for over 100 years in radiology, radiation therapy, imag-
Lopez et al. have shown that for a field size of 1 × 1 cm2 or ing, and personnel dosimetry since the discovery of
smaller, due to the volume effect of the chambers, lower radioactivity and x-rays. They have become an integral
agreement (>15% difference) was observed in measuring part of megavoltage x-rays dosimetry for routine QA
output factors between EBT2 film and chamber mea- through verification of beam flatness and symmetry,
surement [19]. Morales et al. have observed a <3% dif- crossbeam profiles, depth doses in wider areas and also
ference between relative output factors measured with for complex techniques such as dynamic wedges, IMRT,
microdiamond and EBT3 films for cone sizes as small image-guided radiation therapy (IGRT), and small field
as 4 mm [2]. dosimetry like stereotactic radiosurgery.
Film Dosimetry ◾ 63
radiation in or around grain, make a sensitivity center

negatively charged. Mobile silver Ag+ rushes to the sen-
sitivity centers to neutralize and form an aggregate of
metallic silver and hence the latent image on the film.
Only a few metallic silvers primarily along the surface of
FIGURE 5.1 Films typically use silver bromide contents the grain (containing around ten billion silver bromide)
(30–40% by weight) with 1–2 μm grain diameter at 2 g/cm3 are enough to record the radiation signature.
density for x-rays, while higher AgBr contents (70–80% by Following the latent image formation, chemical pro-
weight), with grain diameter of about 0.3 μm and 3.3 g/cm3 cessing of the films is required to develop the film (as
density are used in nuclear emulsions. shown in Figure 5.2) while ensuring that the undevel-
oped silver is eliminated and not exposed due to the pro-
The use of radiographic films for quantitative dosim- cessing. The chemical processes convert all of the Ag+ in
etry is described in American Association of Physicists an exposed grain into atoms while all of the bromine
in Medicine (AAPM) Task Group (TG) 69 [24]. is washed away, leaving behind an opaque, microscopic
Figure 5.1 shows a schematic illustration of the radio- grain of silver.
graphic film. The radiographic film is composed of a It is obvious from the Gurney–Mott mechanism, that
transparent polyester base, coated on one or both sides the formation of opaque metallic silver can be traced
with a radiation-sensitive emulsion held in a water- back to the impingement of radiation. For dosimetry,
soluble gelatin matrix. The sensitivity of radiographic the amount and location of elemental silver can be cor-
film to ionizing radiation depends on the composition related to the absorbed dose by applying an appropriate
of the microscopic grains of the silver halide (mostly calibration. The opacity of the film is determined using
AgBr), grain size, density, and their microenvironment. optical density (OD), which is defined as:
Nuclear emulsions, which will not be discussed further
here, are used in cosmic-ray research and fast neutrons I
OD = − log10 (5.1)
personnel dosimetry. Io
Interaction with radiation, ionizes the microscopic
grains containing silver bromide; effectively converting where I o and I , respectively, are the intensities of the
ionic silver (Ag+) into silver (Ag) atoms. According to incident readout light and the transmitted light through
Gurney and Mott model of latent image formation [25], the film. The logarithm of transmitted light can be digi-
this happens in grains containing sensitive centers or tally obtained through the optical sensors. As the image
‘speck’. Electrons, which produced due to absorption of becomes darker, film becomes more opaque. This leads
FIGURE 5.2 Chemical processes involved in development of a radiographic film. For dose measurement, the consistency of
developing time, temperature, amount of agitation, and developer characteristics are important parameters.
the electron fluence and hence nonuniform dose dis-

tribution. Reciprocity law failure, also known as the
Schwarzschild effect [26], leads to nonlinear OD espe-
cially when exposed to low dose rates.
Despite their non-tissue equivalency (Zeff = 40–45),
energy dependence, cumbersome chemical process-
ing and development issues, radiographic films played
an important role in tumor localization, planning, and
dosimetry for both photon and electron beams up until
2010. With the recent introduction of radiochromic
films and other advanced dosimeters, the role of radio-
FIGURE 5.3 Plot of optical density of film with exposure
graphic films is diminishing in radiation therapy and
is called characteristics curve, sensitometric curve, or H&D radiology.
curve. An H&D curve has three main regions; the toe or fog
region defining minimal density threshold, the gradient or 5.3 RADIOCHROMIC FILM
latitude (straight-line portion) which is the exposure range Although the first reports of radiochromic films were
producing acceptable OD, and the shoulder which is a region published in the 1960s [27], the predecessors of modern
of the curve above the straight-line portion before reaching radiochromic film (GAFchromic™) with higher sensi-
the saturation or maximum density. tivity were developed in the 1980s. Unlike radiographic
films, radiochromic films do not need additional chem-
to increase of OD indicating higher integrated dose has icals for wet post-processing. Also, radiochromic films
been delivered to the film. OD is proportional to the are more water equivalent than radiographic films
electron fluence, and emulsion thickness and hence the due to their low effective atomic number of 6.0–6.8 vs.
absorbed dose. The relationship between absorbed dose 40–45 [1, 28].
and OD is called the sensitometric curve or H&D curve Various film models, such as EBT, EBT2, EBT3, and
(named after developers Hurter and Driffield) as shown EBT-XD, have been developed for clinical dosimetry
in Figure 5.3. applications over the last two decades. The protective
Note that the film properties, including fog, contrast, polyester layer differs between EBT2 and EBT3 such
and film speed are functions of developer temperature. that small silicon particles are integrated in the EBT3
For example, the OD of the film increases as the devel- polyester layer. The active layer in EBT3 is sandwiched
oper temperature is increased. between equal thickness layers of polyester with no
Energy dependency is one of the major disadvantages adhesive. The structure of both EBT-XD film and EBT3
of radiographic films. The photoelectric effect in silver film comprises of a sensitive layer of lithium penta-
bromide grains leads to absorption of x-ray 10–50 times cosa-10, 12-diynoate (LiPCDA) monomer with 25 and
more than that in tissue or air especially for x-ray ener- 28 μm thickness, respectively, sandwiched between
gies below 100 keV. This energy dependency needs to two symmetric matte-polyester layers each with
be accounted for when the ratio of the primary photons 125 μm thickness. Different chemical compositions and
to low-energy scattered photons varies, for example, in side-length ratio of the polymer crystal of EBT-3 and
the penumbra and under the collimator regions when EBT-XD creates differences in their active layers. The
measuring profiles of MV fields. The film over-response most recent film model is EBT-XD with smaller sensi-
can be compensated for by enclosing the film in a high- tive layer thickness which allows for extended dynamic
Z filter to reduce the contribution of low energy x-rays. range (up to 40 Gy for EBT-XD vs. 10 Gy for the EBT3).
AgBr films exhibit a significant dose rate effect that is Their higher dynamic range makes them ideal candidate
related to the grain composition and processing condi- for high dose applications such as SRS and SBRT. The
tions. Commercial x-ray films are individually packed response of both EBT3 and EBT-XD films is reported to
in a light tight envelope. Sometimes a small amount of be energy-independent for megavoltage photon beams,
air may get trapped inside the packing. The trapped air and electron beams. However, there is still dependency
can create a potentially non-uniform perturbation of on batches (or batch-to-batch variation) for these models
which should be addressed by individual batch calibra- As shown in Figure 5.4, the EBT-XD model shows a
tion [29, 30]. lower response compared to the EBT3 model due to the
When ionizing radiation interacts with the con- fact that the former has been designed to have a wider
stituent monomers of radiochromic film, polymers are dynamic range.
formed which leads to an observable film coloration, In EBT3 film, the primary absorption band reaches
which can be quantified by OD defined by Equation 5.1. to a saturation level earlier than the secondary bands.
A calibration curve allows for establishment of relation- Therefore, at lower dose level, the primary band (i.e.,
ship between the film response and the dose received by red channel) can provide higher sensitivity. However,
the film. Flatbed photograph scanners are a common at higher doses from 6 to 10 Gy, the green channel has
type of readout system used to analyze the films and higher sensitivity compared to the red channel [30].
often used clinically to measure the OD. A broad-band For absolute dosimetry, a calibration curve is required
light source and a charge-coupled device (CCD) array to convert OD into absorbed dose. The calibration curve
of detectors are important components of flatbed scan- is obtained by irradiating several film samples to vari-
ners which can provide either transmitted or reflected ous dose levels while staying within the dynamic range
intensity information using red, green, and blue color of the film. The net OD of these samples determined
channels. A mathematical convolution of the densitom- from a film scanner are fit to a polynomial function with
eter light source emission spectrum, the film absorption the following form [16]:
spectrum, and the spectral sensitivity of the densitom-
eter’s detector will result in the measured OD [31]. An D = a × Net OD + b × Net ODn (5.2)
alternate to flatbed scanner is a spectroscopy system
which can allow measurement of the absorbance over a where a, b, and n are the fitting parameters. Figure 5.5
broad spectral window or at a single wavelength [32, 33]. shows a series of EBT3 and EBT-XD films irradiated
The use of radiochromic film for quantitative dosimetry with 16 MeV electron beam. The corresponding calibra-
has been described in AAPM’s TG 55 [28]. tion curves are presented in Figure 5.6.
In conventional flatbed scanners, the digital image
5.3.1 Film Response to Radiation consists of three-color channels: red, green, and blue. In
The polymerization of the constituent monomers of the single channel dosimetry, a calibration curve is estab-
film occurs upon irradiation. In EBT film series, the lished for each color channel. Typically, the red channel
irradiated films have a blue appearance. There are two is most often used due to its higher sensitivity. However,
main absorption bands in the net absorbance spectrum at doses close to the upper dynamic range of the film,
of the films, a primary band around 636 nm and a sec- green channel can be used due to its higher sensitivity.
ondary band around 585 nm. Figure 5.4 shows the net Multichannel film dosimetry as described by Micke
absorbance (Net A) spectra of EBT3 and EBT-XD films et al. [34] varies the dose values until the correspond-
irradiated with 6 MV photon beams when measured ing scanned OD values for various color channels are
with a fiber coupled optical spectrometer. matched. The multichannel method balances the color
FIGURE 5.4 Net absorbance spectra of EBT3 and EBT-XD films irradiated at different dose levels using a 6 MV beam [29].
FIGURE 5.5 Series of film samples from EBT3 (left) and EBT-XD (right) irradiated with 16 MeV electron beam to obtain
calibration curve for the two EBT film models.
channel with the highest sensitivity and provides the studies have shown the benefit of multichannel dosim-
correction of a variety of errors associated with film etry in film-based patient specific QA [17, 18, 35, 36].
scan such as film non-uniformities and/or scanner It has been shown that the response of EBT3 and
related artifacts as well as covering the entire sensitivity EBT-XD films have minimal dose-rate and energy
range of the film by the spectral response of the scanner dependence on the megavoltage photon and electron
due to different sensitivity of each channel [34]. Multiple beams, therefore, a calibration obtained at one energy
FIGURE 5.6 Calibration curves using the red channel information for EBT3 (left) and EBT-XD (right) films corresponding to
images in the previous figure, irradiated by 16 MeV electron beam.
can be used for other energies [29, 30]. However, the film generations of the film [39]. The EBT films show a small
response still remains batch-dependent, which indicates dose-rate dependency (within 1%) when dose rate varies
that the calibration must be performed for each new between 16 and 520 cGy/min [39]. In a study of EBT3
film batch. film using a Novalis TX linear accelerator, it was found
that these films are dose rate independent to within 3%
5.4 FACTORS AFFECTING for a wide range of dose rates (50 to 1000 Gy/min) irre-
RADIOCHROMIC FILM RESPONSE spective of total dose delivered [40]. For flattening filter
There are several factors that affect the radiochromic free (FFF) beams, where dose rates of 1200 to 2400 cGy/
film response to radiation. Type of radiation, energy, min are often used, the dose-rate dependency of EBT2
dose, and dose rate can influence the film response to films in FFF beam profiles measurement was found to
varying degrees. Environmental factors such as temper- be negligible considering the uncertainty in scanner
ature, humidity, and UV light can also affect the OD of reproducibility and film stability [41]. Moreover, mul-
a radiochromic film. tiple studies have shown that the dose-rate dependency
of EBT3 and EBT-XD films were insignificant [29, 30].
5.4.1 Temperature and Humidity Additionally, radiochromic films may be useful for
The effect of temperature on OD of radiochromic film dosimetry at ultra-high dose rates (>40 Gy/s), such as
has been thoroughly investigated [37]. When the film FLASH radiation therapy. Overall 5% uncertainty in
temperature increases from 22 to 38°C, a linear change in radiochromic film dosimetry at these dose rates has been
absorbance causes the fluctuations in hydration of active reported in the literature for electron beams [42, 43].
film components, which affects the location of the absor-
bance peak and overall sensitivity of the film. It has been 5.4.3 Beam Quality and Energy
reported that the humidity introduces systematic dose Several studies evaluated the effect of beam quality on
errors for radiochromic film dosimetry [38]. A dose error the film; the dependence of calibration curve for mega-
of up to 15% can be introduced when the humidity varies voltage energies was negligible [17, 18, 29, 30]. However,
from 80% to 20%. Therefore, it is important to control the approximately 5% under-response in kV beams has
storage, humidity, and temperature of the film. been reported for EBT3 model compared to megavolt-
age beam irradiation [20].
5.4.2 Dose and Dose Rate
Film dosimetry relies on the calibration curve estab- 5.4.4 Linear Energy Transfer (LET)
lished during the calibration process with a reference In proton therapy beams, and other high LET beams, an
beam quality, dose-rate, and under specific conditions. under-response of about 10% in the Bragg peak has been
These factors can affect a dosimeter’s response. reported which makes absolute film dosimetry in pro-
The dose-rate dependency of radiochromic film has ton beams challenging since the film response depends
been evaluated in a number of studies across various on both dose and LET [44–46]. Figure 5.7 shows PDD
FIGURE 5.7 Percentage depth dose (PDD) measured using two different batches of EBT-XD film samples at discrete depths
compared to ionization chamber (solid lines) measurement. Film data were arbitrarily matched at 1.5 cm depth [47].
measurement using EBT-XD films compared with that 5.4.7 Orientation

using an ionization chamber. A matching point near the Considering the purpose of the measurement, dose
entrance at an arbitrary depth of 1.5 cm was selected. It can level to be determined, film orientation (coronal, sagit-
be seen that the film measurements underestimate the dose tal, transversal, etc.) should be carefully selected. After
by ∼1–5% and ∼7–8% in the spread-out Bragg peak (SOBP) cutting the film strips, it is important to label their ori-
region and at the distal end of the SOBP, respectively. The entation clearly to ensure a correct orientation is used
“quenching” effect is more pronounced in the distal region during film scanning. Film data are sensitive to film
of the Bragg peak, which has a high LET. orientation relative to the light polarization direction;
Correcting the quenching is not a trivial task and therefore, consistency must be followed during calibra-
requires a priori knowledge of the beam LET [48, 49]. tion, background scanning, and measurement steps [1].
One approach to correct for quenching is to determine
the value of the LET of the beam using Monte Carlo 5.4.8 Storage
simulations and establish a correction factor as a func- For optimal accuracy, it is recommended to store radio-
tion of depth. The goal of the correction factor is to cor- chromic films in a dark, temperature-controlled envi-
rect the dose determination in depth for proton beams. ronment (ideally in their original box) since they are
However, this assumes that the Monte Carlo simulation sensitive to excessive temperatures and UV-light expo-
and the determination of the LET are accurate and rep- sures. Pressure and movement during storage should be
resentative of the beam configuration used. avoided as it can cause rubbing and scratching of the film
pieces. Any visible damage or small scratches can cause
5.4.5 Batch significant change in the film response. Moisture, radia-
Several studies have shown a significant variation of tion, and magnetic environment should be avoided dur-
film response between different batches [29, 30] and a ing storage as it can affect the measurement accuracy and
minor variation (less than 1%) between the film sheets decrease the life of the film. In order to use films properly,
from the same batch for most types of radiochromic they should be kept in the original box in a temperature
films [1]. The batch dependence can be more than 10% controlled, dust-, smoke-, and moisture-free area, with-
[50], which suggests establishing a separate calibration out any radiation and magnetic field environment [1].
curve for each batch.
5.4.9 External Magnetic Fields
5.4.6 Post Irradiation Time and Temporal Response With the emergence of magnetic resonance image-
Several studies have investigated the temporal response guided radiation therapy (MRIgRT), that uses MRI for
of EBT films and shown a 6–9% increase in post- real time imaging of the anatomy during the radiation
irradiation OD for EBT films within 12 hours of irra- delivery, it is vital to realize to what extent the presence
diation in dose range of 1–5 Gy [1]. This increase in OD of a magnetic field would impact the response of dosim-
was higher (13–19%) for the older film models such as eters. The presence of an external magnetic field, typi-
MD-55-2, XR type T, and HS radiochromic films. cally 0.35–1.5 T required in the MRIgRT machines, does
For EBT3 and EBT-XD films, up to 15% increase in not directly affect the primary photon beams, however,
the OD (depending on the dose level and color channel) their secondary electrons are influenced by the mag-
can occur between the OD measured immediately after netic field through the Lorentz force [56, 57]. The mag-
the film exposure and that measured at 24 hours post- netic field will influence the trajectory of electrons and
irradiation. The increase in OD between 12 hours post- consequently the deposited dose in the medium which
irradiation and 24 hours post-irradiation can be as high can also to some extent impacts the response of rou-
as 5% [51]. It is recommended to wait at least 24 hours tinely used dosimeters.
post-irradiation before scanning the films to achieve The existing literature on radiochromic film dosim-
higher reproducibility in the measurement. If 24-hour etry in the context of dosimetry in the presence of an
post-irradiation wait time is not feasible, a correction external magnetic field is rather inconsistent (see Table 5.1).
function [52, 53] can be used to correct post-irradiation However, the common consensus is that below 1.5 T the
OD. Recent researches have shown that shorter wait influence of the magnetic field on the EBT3 and EBT-XD
time post irradiation may be used [54, 55]. film is within the measurement uncertainty.
TABLE 5.1 Summary of Recent Film Dosimetry Literature in the Context of MRIgRT
Concurrent Dose
Clinical B-field and Radiation type Batch Range
Study Film Model MRIgRT Irradiation Magnet B-field (T) (/Reference) Nos. (Gy) Result
Reyhan EBT2 No No MRI scanner 1.5 6 MV 1 0–8 ∼4% under-
et al. [58] (GE) measurement of dose.
Reynoso EBT2 Yes Yes MRIdian 0.35 Co (/Same
60 1 2–18 8.7%, 8%, and 4.3%
et. al. [59] (Viewray) system with reduction in the net
magnet off) optical density for red,
green, and blue
channels, respectively.
Roed EBT3 No Yes Electromagnet 1.5 60Co 1 2–8 <2% under-response,
et al. [60] but it was within the
measurement
uncertainty.
Delfs et al. EBT3 No Yes Electromagnet 0.35–1.42 6 MV 1 1–20 Up to 2.1% under-
[61] measurement of dose.
Barten EBT3 Yes Yes MRIdian 0.35 60Co (/Linac 1 1–8 Within 1% uncertainty of
et al. [62] (Viewray) for B = 0) the measurements. And
1.5% when real-time
MRI was performed.
Padilla-Cabal EBT3 No Yes Electromagnet 0.5–1 62.4–252.6 1 0.2–10 No significant change
et al. [63] MeV proton in film response.
Billas EBT3 No Yes Electromagnet 0.5–2 60Co 1 0.75–8 Within measurement
et al. [64] uncertainty for B ≤
1.5 T. But, 2.4%
over-response at B = 2 T.
Volotskova EBT2, EBT3, No No MRI scanner 1.5–3 6 MV & 6 MV 1 0–20 No significant change
et al. [65] EBT-XD (Siemens) FFF in film optical density.
Darafsheh EBT3, Yes Yes MRIdian 0.35 6 MV FFF 2 0–20 No significant change
et al. [66] EBT-XD (Viewray) (/Linac for in film response.
B = 0)
Source: Adapted from [66].
5.5 OTHER CONSIDERATIONS is dependent on the polarization of the incident light,

5.5.1 Film Handling and therefore, the orientation of the film. This should be
Radiochromic films are not overly sensitive to light like considered in the calibration films, as well for the test
their radiographic counterparts, however, they should be films actually used for dosimetry.
kept in a cool, dark, and dry place and never be exposed
to UV light, especially sunlight, as it will lead to an ion- 5.5.2 Scanners
izing effect on the monomers. The surface of the film The scanner should be kept turned on so that the reading
should be handled with latex gloves to prevent smudges electronics is warmed up. This is to stabilize the intensity
and fingerprints on the film’s surface. A sharp blade or of the light source. Also, at least some number (typically
a “guillotine” device should be used to cut the film in 10-20) of preview scans should be acquired to adequately
order to avoid unnecessary bending or deformation of warm up the light source. The user can experimentally
the film. Since there is a tendency for the separation find the number of sufficient warm up scans.
of the layers of the film, dosimetry should be avoided
around the edges of the film where cuts have been made. 5.5.3 Calibration Procedures
Thus, care should be taken to include a small buffer area For each batch of film, a calibration step is necessary
between the regions of interest and the perimeter of the to convert the response of the film into a dose distribu-
film to prevent unnecessary errors and uncertainties. tion, therefore, the model and lot number of the batch
Once the film is cut to the desired size, the corner for each calibration should be noted. Figure 5.8 demon-
of the film should be marked as the readout of the film strates the steps needed for calibration.
FIGURE 5.8 Radiochromic film calibration process.
The required tools and accessories for film calibra- can lead to larger uncertainties, and may require dose
tion are: the radiochromic films, a readout system which rescaling.
is typically a flatbed scanner, and an analysis software Any flatbed scanner, for example, Epson Expression
to extract the gray level information of the scanned 10000XL and Epson Perfection V750, can be used to
images. It is recommended to handle the films with read the films. Films should be positioned in the same
gloves and use a frame to place the films at the center orientation (i.e., portrait or landscape) on the scanner
of scanner. During the calibration process caution must close to the center of the scanner using some align-
be exercised to control the ambient temperature during ment frame device. The film is scanned in color mode,
irradiation, storage, and readout in a consistent fashion. 48-bit RGB (16 bits per channel), positive film, and at
Handle the films with care, avoid dust, fingerprints, or least 72 dpi resolution, without any default image cor-
over-bending. Store the films in a dry and dark environ- rections. Finally, an image processing software such as
ment; keep them away from light and avoid prolonged ImageJ (National Institutes of Health, USA) or a com-
exposure to UV light. mercial product can be used to readout the pixel values
About six to eight data points are needed for the of each image and a calibration fit can be generated by
calibration curve, covering the range of dose to be using Equation 5.2. The temperature, humidity, post-
determined. Ideally, at least two data points should irradiation time, and film orientation during irradiation
be higher than the desired maximum dose to be mea- and readout are factors that can affect the signal of the
sured. Based on the dynamic range of the film, several radiochromic film and should not change significantly
dose levels ranging from 0 cGy (the unexposed film) during calibration or test irradiation.
to approximately 110–130% of the maximum dose of Series of EBT3 and EBT-XD films irradiated with 16
interest are selected. Usually an 8″ × 10″ sheet of film MeV electron beam are shown in Figure 5.5 along with
is cut into strips of dimensions 1.25″ × 8″. Each strip their corresponding calibration curve in Figure 5.6.
should be marked to preserve and track its orienta-
tion and alignment (portrait or landscape). The ori- 5.5.4 Uncertainty Budget
entation directions for both irradiation and readout The uncertainty budget of radiochromic films itself
should be kept the same. Then, each strip is cut into depends on several factors such as the film model (EBT,
smaller pieces (>3 × 3 cm 2). In order to have accu- EBT2, EBT3, EBT-XD, etc.), application, dose range, scan-
rate calibration, each film should be scanned before ning (scanning parameters, orientation, stability), and
irradiation to remove the background and establish a calibration uncertainties. It should be noted that the total
baseline level. uncertainty in determining the dose depends not only on
Each of the film pieces is exposed to a reference the dosimetry system (radiochromic film) but also on the
field (e.g., 10 × 10 cm 2) of known absorbed dose. The known dose measurement uncertainties (i.e., measure-
time between irradiation and readout should be con- ment setup, measurement protocol, and output uncer-
sistent, typically waiting for the films to be stabilized tainties). Based on these uncertainties, one can determine
(at least 24 hours) prior to scanning them. Shorter error bars or confidence intervals for a particular film
waiting times, though, are not recommended but can measurement protocol.
be used by applying a correction factor or using a Several studies have evaluated the uncertainty of
separate calibration. Variable waiting-time windows radiochromic films used for photon, electron, and
TABLE 5.2 Summary of Selected Literature on Uncertainty Budget in Radiochromic Film Dosimetry
Study Film Model Radiation Type Dose Range (Gy) Results
van Battum et al. [67] EBT 6 MV photon beam 0–2.3 Gy The overall uncertainty in absolute EBT
film dose detection 1.8% using a single
film and 1.3% for two films.
Main factor: Intrinsic film inhomogeneity
Martišíková et al. [68] EBT 60Co 0.3–1 Gy Total uncertainty in the net OD
determination was estimated to be 1.6%
at 0.3 Gy and 0.8% at 1 Gy for 60Co.
This leads to a dose uncertainty of 1.6% at
0.3 Gy and 0.9% at 1 Gy.
Aland et al. [69] EBT2 6 MV photon beam 2 Gy Total uncertainty: Worst case (12.6%) and
Red and blue channels ideal case (2.6%)
EBT2 film can be used in routine quality
assurance testing for radiotherapy, in
situations where a dose uncertainty of up
to 2.8% is acceptable.
Palmer et al. [70] EBT3 Ir-192 and Co-60 HDR 0–20 Gy Film dosimetry fractional uncertainties
brachytherapy Calibration dose delivery on a Linac
Multichannel (0.5%), calibration fit (0.5%), film
position (0.9%), energy dependence
(1.5%), phantom scatter (0.5%), and film
scanning (0.2%)
Sorriaux et. al. [71] EBT3 Photon: 6 MV and 18 MV 0–10 Gy The total uncertainties on calibration
Electron: 6 MeV curve due to film reading and fitting were
Proton: 60 MeV and 100 MeV within 1.5% for photon and proton
Red, green, and blue channels beams. For electrons, the uncertainty was
within 2% for doses larger than 0.8 Gy.
León Marroquin et al. [72] EBT3 6 MV photon beam 0–120 Gy The overall uncertainty in the measured
Red, green, and blue channels dose for the red, green, and blue channels
was 3.2%, 4.9%, and 5.2%, respectively.
The relative orientation of the film, the
uniformity of response of the scanner,
and the response curves and fitting
procedure are the main factors in the
total uncertainties.
León-Marroquín et al. [29] EBT-XD Photon: 6 MV, 6 MV-FFF, 10 0.5–50 Gy The maximum uncertainty in the
MV-FFF, and 15 MV calculated dose, in ranges 5–40 Gy and
Electron: 6 and 20 MeV 8–50 Gy for red and green channels,
Red, green, and blue channels respectively, were ∼3.1% and ∼2.4%.
Net OD variations compared to 6 MV:
For dose ≤10 Gy: Red (<4%), Green
(<5.5%), Blue (10%).For dose >10 Gy:
Red (<3%), Green (<3.5%), Blue (5%)
proton beam dosimetry. For megavoltage photon beam protocol for the calibration and the use of radiochromic
using radiochromic film, overall uncertainties from 3% films can result in overall uncertainties within 2–5%.
to 5% have been reported. Table 5.2 provides a summary
of several studies on the evaluation of the uncertainties REFERENCES
in radiochromic film dosimetry systems. 1. I. J. Das, Radiochromic Film: Role and Applications in
Radiation Dosimetry (CRC Press, Boca Raton, FL, 2017).
5.6 CONCLUSION 2. J. E. Morales, M. Butson, S. B. Crowe, R. Hill, J. V. Trapp,
“An experimental extrapolation technique using the
Overall, film dosimeters offer the highest spatial reso- Gafchromic EBT3 film for relative output factor mea-
lution amongst currently available 2D dosimeters and surements in small x-ray fields,” Medical Physics 43(8),
are indispensable to small field dosimetry. A typical 4687–4692 (2016).
3. C. Fiandra, M. Fusella, F. R. Giglioli, A. R. Filippi, C. radiation therapy beams,” Proceedings of SPIE 11231,
Mantovani, U. Ricardi, R. Ragona, “Comparison of 112310I (2020).
Gafchromic EBT2 and EBT3 for patient-specific quality 16. S. Devic, N. Tomic, D. Lewis, “Reference radiochromic
assurance: Cranial stereotactic radiosurgery using volu- film dosimetry: Review of technical aspects,” Physica
metric modulated arc therapy with multiple noncopla- Medica 32, 541–556 (2016).
nar arcs,” Medical Physics 40(8), 082105 (2013). 17. T. Kairn, N. Hardcastle, J. Kenny, R. Meldrum, W. A.
4. C. Chen, S. Tang, D. Mah, M. Chan, “SU-E-T-286: Tomé, T. Aland, “EBT2 radiochromic film for quality
Dose verification of spot-scanning proton beam using assurance of complex IMRT treatments of the prostate:
GafChromic EBT3 film,” Medical Physics 42(6), 3399– Micro-collimated IMRT, RapidArc, and TomoTherapy,”
3399 (2015). Australasian College of Physical Scientists in Medicine
5. A. Chu, W. Feng, H. Lincoln, F. Su, R. Nath, “SU-E-T-118: 34(3), 333–343 (2011).
Small dynamic field dosimetry by Gfachromic film 18. L. Marrazzo, M. Zani, S. Pallotta, C. Arilli, M. Casati, A.
(EBT3) and 2D-array diode,” Medical Physics 40(6), Compagnucci, C. Talamonti, M. Bucciolini, “GafChromic®
230–231 (2013). EBT3 films for patient specific IMRT QA using a mul-
6. N. Corradini and S. Presilla, “SU-E-T-58: Submillimeter tichannel approach,” Physica Medica 31(8), 1035–1042
dose measurement with Gafchromic EBT2 and EBT3 (2015).
films,” Medical Physics 40(6), 216–216 (2013). 19. A. Gonzalez-Lopez, J. A. Vera-Sanchez, J. D. Lago-
7. A. L. Palmer, P. Di Pietro, S. Alobaidli, F. Issa, S. Doran, Martin, “Small fields measurements with radiochromic
D. Bradley, A. Nisbet, “Comparison of methods for the films,” Journal of Medical Physics 40(2), 61–67 (2015).
measurement of radiation dose distributions in high 20. J. E. Villarreal-Barajas and R. F. H. Khan, “Energy
dose rate (HDR) brachytherapy: Ge-doped optical fiber, response of EBT3 radiochromic films: Implications for
EBT3 Gafchromic film, and PRESAGER radiochromic dosimetry in kilovoltage range,” Journal of Applied clin-
plastic,” Medical Physics 40(6), 061707 (2013). ical Medical Physics 15(1), 331–338 (2014).
8. D. Cusumano, M. L. Fumagalli, M. Marchetti, L. Fariselli, 21. S. Devic, L. H. Liang, N. Tomic, H. Bekerat, M. Morcos,
E. De Martin, “Dosimetric verification of stereotactic M. Popovic, P. Watson, S. Aldelaijan, J. Seuntjens, “Dose
radiosurgery/stereotactic radiotherapy dose distribu- measurements nearby low energy electronic brachy-
tions using Gafchromic EBT3,” Medical Dosimetry 40, therapy sources using radiochromic film,” Physica
226–231 (2015). Medica 64, 40–44 (2019).
9. A. L. Palmer, A. Dimitriadis, A. Nisbet, C. H. Clark, 22. Y. Rong and J. S. Welsh, “Surface applicator calibration
“Evaluation of Gafchromic EBT-XD film, with compar- and commissioning of an electronic brachytherapy sys-
ison to EBT3 film, and application in high dose radio- tem for nonmelanoma skin cancer treatment,” Medical
therapy verification,” Physics in Medicine and Biology Physics 37(10), 5509–5517 (2010).
60(22), 8741–8752 (2015). 23. H. Bekerat, S. Devic, F. DeBlois, K. Singh, A. Sarfehnia,
10. I. Ali, S. Ahmad, S. Joel, J. F. Williamson, “Optimal J. Seuntjens, S. Shih, X. Yu, D. Lewis, “Improving
densitometry wavelengths that maximize radiochromic the energy response of external beam therapy (EBT)
film sensitivity while minimizing OD growth and tem- GafChromicTM dosimetry films at low energies (≤ 100
perature sensitivity artifacts,” Journal of X-Ray Science keV),” Medical Physics 41(2), 022101 (2014).
and Technology 17, 61–73 (2009). 24. S. Pai, I. J. Das, J. F. Dempsey, K. L. Lam, T. J. LoSasso,
11. I. Ali, J. F. Williamson, C. Costescu, J. F. Dempsey, A. J. Olch, J. R. Palta, L. E. Reinstein, D. Ritt, E. E.
“Dependence of radiochromic film response kinetics on Wilcox, “TG-69: Radiographic film for megavoltage
fractionated doses,” Applied Radiation and Isotopes 62, beam dosimetry,” Medical Physics 34(6), 2228–2258
609–617 (2005). (2007).
12. Y. Le, I. Ali, J. F. Dempsey, J. F. Williamson, “Prospects 25. R. W. Gurney and N. F. Mott, “The theory of photolysis
for quantitative two-dimensional radiochromic film of silver bromide and the photographic latent image,”
dosimetry for low dose-rate brachytherapy sources,” Proceedings of the Royal Society of London. Series A,
Medical Physics 33(12), 4622–4634 (2006). Mathematical and Physical Sciences 164(917), 151–167
13. J. I. Monroe, J. F. Dempsey, J. A. Dorton, S. Mutic, J. (1938).
B. Stubbs, J. Markman, J. F. Williamson, “Experimental 26. A. Djouguela, R. Kollhoff, A. Rubach, D. Harder, B.
validation of dose calculation algorithms for the Poppe, “The Schwarzschild effect of the dosimetry film
GliaSite™ RTS, a Novel 125I liquid-filled balloon brachy- Kodak EDR 2,” Physics in Medicine and Biology 50(21),
therapy applicator,” Medical Physics 28(1), 73–85 (2001). N317–N321 (2005).
14. Y. Zhu, A. S. Kirov, V. Mishra, A. S. Meigooni, J. F. 27. W. L. McLaughlin and L. Chalkley, “Measurement of
Williamson, “Quantitative evaluation of radiochromic radiation dose distributions with photochromic materi-
film response for two-dimensional dosimetry,” Medical als,” Radiology 84(1), 124–125 (1965).
Physics 24(2), 223–231 (1997). 28. A. Niroomand-Rad, C. R. Blackwell, B. M. Coursey, K.
15. A. Darafsheh, T. Zhao, R. Khan, “Optical spec- P. Gall, J. M. Galvin, W. L. McLaughlin, A. S. Meigooni,
tral analysis of radiochromic films irradiated with R. Nath, J. E. Rodgers, C. G. Soares, “Radiochromic
film dosimetry: Recommendations of AAPM Radiation 42. M. Jaccard, K. Petersson, T. Buchillier, J. F. Germond,
Therapy Committee Task Group 55,” Medical Physics M. T. Duran, M. C. Vozenin, J. Bourhis, F. Bochud,
25(11), 2093–2115 (1998). C. Bailat, “High dose-per-pulse electron beam dosim-
29. E. Y. León-Marroquín, D. J. Mulrow, A. Darafsheh, R. etry: Usability and dose-rate independence of EBT3
Khan, “Response characterization of EBT-XD radio- Gafchromic films,” Medical Physics 44(2), 725–735
chromic films in megavoltage photon and electron (2017).
beams,” Medical Physics 46(9), 4246–4256 (2019). 43. L. Karsch, E. Beyreuther, T. Burris-Mog, S. Kraft, C.
30. E. Y. León-Marroquín, D. J. Mulrow, R. Khan, A. Richter, K. Zeil, J. Pawelke, “Dose rate dependence for
Darafsheh, “Spectral analysis of the EBT3 radiochromic different dosimeters and detectors: TLD, OSL, EBT
films for clinical photon and electron beams,” Medical films, and diamond detectors,” Medical Physics 39(5),
Physics 46(2), 973–982 (2019). 2447–2455 (2012).
31. L. E. Reinstein, G. R. Gluckman, H. I. Amols, 44. M. Martišíková and O. Jäkel, “Gafchromic® EBT films
“Predicting optical densitometer response as a function for ion dosimetry,” Radiation Measurement 45(10),
of light source characteristics for radiochromic film 1268–1270 (2010).
dosimetry,” Medical Physics 24(12), 1935–1942 (1997). 45. M. Martišíková and O. Jäkel, “Dosimetric properties of
32. J. F. Dempsey, D. A. Low, S. Mutic, J. Markman, A. S. Gafchromic® EBT films in monoenergetic medical ion
Kirov, G. H. Nussbaum, J. F. Williamson, “Validation of a beams,” Physics in Medicine and Biology 55(13), 3741–
precision radiochromic film dosimetry system for quan- 3751 (2010).
titative two-dimensional imaging of acute exposure dose 46. L. Zhao and I. J. Das, “Gafchromic EBT film dosime-
distributions,” Medical Physics 27(10), 2462–2475 (2000). try in proton beams,” Physics in Medicine and Biology
33. A. Darafsheh, E. Y. León-Marroquín, D. Mulrow, M. 55(10), N291–N301 (2010).
Baradaran-Ghahfarokhi, T. Zhao, R. Khan, “On the 47. A. Darafsheh, T. Zhao, R. Khan, “Spectroscopic analy-
spectral characterization of radiochromic films irradi- sis of irradiated radiochromic EBT-XD films in proton
ated with clinical proton beams,” Physics in Medicine and photon beams,” Physics in Medicine and Biology 65,
and Biology 64(13), 135016 (2019). 205002 (2020).
34. A. Micke, D. F. Lewis, X. Yu, “Multichannel film dosim- 48. B. R. Smith, M. Pankuch, C. G. Hammer, L. A.
etry with nonuniformity correction,” Medical Physics DeWerd, W. S. Culberson, “LET response variability
38(5), 2523–2534 (2011). of GafchromicTM EBT3 film from a 60Co calibration in
35. D. Lewis, A. Micke, X. Yu, M. F. Chan, “An efficient clinical proton beam qualities,” Medical Physics 46(6),
protocol for radiochromic film dosimetry combining 2716–2728 (2019).
calibration and measurement in a single scan,” Medical 49. A. F. Resch, P. D. Heyes, H. Fuchs, N. Bassler, D. Georg,
Physics 39(10), 6339–6350 (2012). H. Palmans, “Dose- rather than fluence-averaged LET
36. T. Tessonnier, A. Dorenlot, N. Nomikossoff, “Patient should be used as a single-parameter descriptor of pro-
quality controls with Gafchromic EBT3 for tomother- ton beam quality for radiochromic film dosimetry,”
apy HI-ART 2: Analysis of the calibration’s condition to Medical Physics (2020).
achieve an absolute dosimetry,” Physica Medica 29, e45 50. S. Reinhardt, M. Hillbrand, J. J. Wilkens, W. Assmann,
(2013). “Comparison of Gafchromic EBT2 and EBT3 films for
37. A. Rink, D. F. Lewis, S. Varma, I. A. Vitkin, D. A. Jaffray, clinical photon and proton beams,” Medical Physics
“Temperature and hydration effects on absorbance 39(8), 5257–5262 (2012).
spectra and radiation sensitivity of a radiochromic 51. S. Momin, R. Khan, A. Darafsheh, “Growth kinetics of
medium,” Medical Physics 35(10), 4545–4555 (2008). the EBT3 and EBT-XD films response in radiotherapy
38. F. Girard, H. Bouchard, F. Lacroix, “Reference dosime- beams,” Proceedings of SPIE 11224, 112240J (2020).
try using radiochromic film,” Journal of Applied Clinical 52. C. Andrés, A. del Castillo, R. Tortosa, D. Alonso, R.
Medical Physics 13(6), 339–353 (2012). Barquero, “A comprehensive study of the Gafchromic
39. A. Rink, I. A. Vitkin, D. A. Jaffray, “Intra-irradiation EBT2 radiochromic film. A comparison with EBT,”
changes in the signal of polymer-based dosimeter Medical Physics 37(12), 6271–6278 (2010).
(GAFCHROMIC EBT) due to dose rate variations,” 53. L. Zhao, L. Coutinho, N. Cao, C.-W. Cheng, I. J. Das,
Physics in Medicine and Biology 52(22), N523–N529 “Temporal response of Gafchromic EBT2 radiochromic
(2007). film in proton beam irradiation,” IFMBE Proceedings
40. G. Twork and A. Sarfehnia, “SU-E-T-88: Evaluation of 39, 1164–1167 (2012).
the dose-rate dependency of GAFCHROMIC EBT3,” 54. M. F. Chan, Recent Advancements and Applications in
Medical Physics 40(6), 223–224 (2013). Dosimetry (Nova Science Publishers, Hauppauge, NY,
41. S. Oyewale, S. Ahmad, I. Ali, “SU-E-T-85: Dose rate 2018).
and energy dependence of EBT, EBT2, EDR2 films, and 55. M. F. Chan, D. Lewis, X. Yu, “Is it possible to pub-
Mapcheck2 diode arrays in beam profiles from a Varian lish a calibration function for radiochromic film?,”
TrueBeam system,” Medical Physics 39(6), 3722–3722 International Journal of Medical Physics, Clinical
(2012). Engineering and Radiation Oncology 3(1), 25–30 (2014).
56. B. W. Raaymakers, A. J. Raaijmakers, A. N. Kotte, D. 65. O. Volotskova, X. Fang, M. Keidar, H. Chandarana, I.

Jette, J. J. Lagendijk, “Integrating a MRI scanner with J. Das, “Microstructure changes in radiochromic films
a 6 MV radiotherapy accelerator: Dose deposition in due to magnetic field and radiation,” Medical Physics
a transverse magnetic field,” Physics in Medicine and 46(1), 293–301 (2019).
Biology 49(17), 4109–4118 (2004). 66. A. Darafsheh, Y. Hao, B. Maraghechi, J. Cammin, F.
57. C. Kirkby, T. Stanescu, S. Rathee, M. Carlone, B. Reynoso, R. Khan, “Influence of 0.35 T magnetic field on
Murray, B. G. Fallone, “Patient dosimetry for hybrid the spectral response of EBT3 and EBT-XD radiochro-
MRI-radiotherapy systems,” Medical Physics 35(3), mic films,” Medical Physics 47(9), 4543–4552 (2020).
1019–1027 (2008). 67. L. J. van Battum, D. Hoffmans, H. Piersma, S. Heukelom,
58. M. L. Reyhan, T. Chen, M. Zhang, “Characterization of “Accurate dosimetry with GafChromic EBT film of a
the effect of MRI on Gafchromic film dosimetry,” Journal 6 MV photon beam in water: What level is achievable?,”
of Applied Clinical Medical Physics 16(6), 325–332 (2015). Medical Physics 35(2), 704–716 (2008).
59. F. J. Reynoso, A. Curcuru, O. Green, S. Mutic, I. J. 68. M. Martišíková, B. Ackermann, O. Jäkel, “Analysis of
Das, L. Santanam, “Technical note: Magnetic field uncertainties in Gafchromic® EBT film dosimetry of
effects on Gafchromic-film response in MR-IGRT,” photon beams,” Physics in Medicine and Biology 53(24),
Medical Physics 43(12), 6552–6556 (2016). 7013–7027 (2008).
60. Y. Roed, H. Lee, L. Pinsky, G. Ibbott, “Characterizing the 69. T. Aland, T. Kairn, J. Kenny, “Evaluation of a Gafchromic
response of Gafchromic EBT3 film in a 1.5 T magnetic EBT2 film dosimetry system for radiotherapy qual-
field,” Radiotherapy and Oncology 123(Supplement 1), ity assurance,” Australasian Physical & Engineering
S403 (2017). Sciences in Medicine 34, 251–260 (2011).
61. B. Delfs, A. A. Schoenfeld, D. Poppinga, R.-P. Kapsch, P. 70. A. L. Palmer, C. Lee, A. J. Ratcliffe, D. Bradley, A.
Jiang, D. Harder, B. Poppe, H. K. Looe, “Magnetic fields Nisbet, “Design and implementation of a film dosim-
are causing small, but significant changes of the radio- etry audit tool for comparison of planned and delivered
chromic EBT3 film response to 6 MV photon,” Physics dose distributions in high dose rate (HDR) brachyther-
in Medicine and Biology 63, 035028 (2018). apy,” Physics in Medicine and Biology 58(19), 6623–6640
62. D. L. J. Barten, D. Hoffmans, M. A. Palacios, S. Heukelom, (2013).
L. J. van Battum, “Suitability of EBT3 GafChromic film 71. J. Sorriaux, A. Kacperek, S. Rossomme, J. A. Lee,
for quality assurance in MR-guided radiotherapy at 0.35 D. Bertrand, S. Vynckier, E. Sterpin, “Evaluation of
T with and without real-time MR imaging,” Physics in Gafchromic® EBT3 films characteristics in therapy pho-
Medicine and Biology 63(16), 165014 (2018). ton, electron and proton beams,” Physica Medica 29(6),
63. F. Padilla-Cabal, P. Kuess, D. Georg, H. Palmans, L. 599–606 (2013).
Fetty, H. Fuchs, “Characterization of EBT3 radiochro- 72. E. Y. León Marroquin, J. A. Herrera González, M. A.
mic films for dosimetry of proton beams in the presence Camacho López, J. E. Villarreal Barajas, O. A. García-
of magnetic fields,” Medical Physics 46(7), 3278–3284 Garduño, “Evaluation of the uncertainty in an EBT3
(2019). film dosimetry system utilizing net optical density,”
64. I. Billas, H. Bouchard, U. Oelfke, S. Duane, “The effect of Journal of Applied Clinical Medical Physics 17(5), 466–
magnetic field strength on the response of Gafchromic 481 (2016).
EBT-3 film,” Physics in Medicine and Biology 64(6),
06NT03 (2019).
Chapter 6
Thermoluminescence Dosimetry
Tomas Kron and Peta Lonski
CONTENTS
6.1 Introduction 75
6.1.1 What is TLD? 75
6.1.2 A Bit of History 76
6.1.3 Scope of the Chapter 76
6.1.4 Important Literature 77
6.2 Fundamental Principles Using LiF:Mg,Ti as Example 77
6.2.1 Basic Theory 77
6.2.2 Glow Curves 77
6.2.3 Dose Response 79
6.2.4 Other Influences on Dosimetry 80
6.3 TLD Materials 80
6.4 Equipment Required for TLD 82
6.4.1 TLD Handling 83
6.4.2 TLD Reader 83
6.4.3 TLD Evaluation, Quality Assurance, and Analysis 85
6.4.4 Annealing and Reuse 85
6.5 Applications in Radiotherapy 85
6.5.1 Radiation Protection 86
6.5.2 Phantom Measurements 87
6.5.3 In Vivo Dosimetry 88
6.5.4 Brachytherapy 89
6.5.5 Audits 90
6.6 Limitations of TLD 91
6.7 Conclusion 92
References 93
6.1 INTRODUCTION luminesce and the light intensity emitted is propor-

6.1.1 What is TLD? tional to the dose deposited in the material in the first
Thermoluminescence dosimetry or TLD for short is a place [1–3]. This is illustrated in Figure 6.1.
dosimetric technique with a lot of history and many This process will be discussed in more detail later
applications related to medicine. The name describes and we will explore the variety of materials that can be
the physical principles involved: after irradiation of used for TLD. TLD is a “passive” dosimetric technique:
TLD materials the material is heated which makes it there are no cables or readout mechanism connected to
75
FIGURE 6.1 Schematic illustration of the thermoluminescence process.
the detectors during irradiation. This allows placement (ISSDO) holding conferences every 3 years, which bring
of thermoluminescence (TL) dosimeters in virtually all together many protagonists of TLD. The 20th SSD con-
locations but prevents immediate “real time” readout. ference will be held in 2022 in Italy. Since 1980 the con-
The “free standing nature” of TL dosimeters allows tak- ferences publish proceeding in reputable journals [9],
ing measurements in many different scenarios of dose which has greatly helped to publicize the science and
levels that can vary by more than 10 orders of magni- applications of TLD.
tude. It is this versatility that makes TLD a useful tech-
nique that ensures its continued relevance in radiation 6.1.3 Scope of the Chapter
medicine. The present chapter aims to introduce TLD to scientists
working broadly in the medical field. This ranges from
6.1.2 A Bit of History radiation protection to imaging and radiation therapy.
TLD has a long history in medicine. Already in 1953 The objective is to link a basic understanding of the
Farrington Daniels, one of the pioneers of TLD, underlying science with good practice and useful appli-
reported the first “medical” application of TLD: “The cations of TLD.
crystals were swallowed by the patient (who had To this end the chapter commences with some
received an injection containing radioactive isotopes), theory before covering the instrumentation needed
recovered one or two days later, and the accumulated to setup a successful TLD service. As the authors are
dosage in roentgens was measured by matching ther- from a radiotherapy background the focus will be on
moluminescence intensity with that produced in crys- this; however, the concepts are easily translatable to
tals by a known roentgen dosage” [4]. radiation safety and medical imaging. We will also
John Cameron, who was the first professor for cover common applications of TLD with particular
Medical Physics in Madison, WI, was also a strong emphasis on phantom measurements, in vivo dosim-
promoter of TLD as a tool for medical dosimetry etry and audits. The chapter will conclude with a sum-
[5]. He established LiF:Mg,Ti (patented by Harshaw mary of limitations of TLD which will need to touch
Chemicals as TLD-100) as the original “standard” on competing techniques such as optically stimulated
material for TLD. This material and its properties were luminescence (OSL) dosimetry (Chapter 7) [10] and
then characterized extensively by Yigal Horowitz and radiochromic film (Chapter 5) [11]. All this is followed
his school [6–8]. by an attempt to predict the future of a technique,
Of particular importance for TLD has been the which is still widely used despite having been declared
International Solid State Dosimetry Organisation obsolete on many occasions.
Thermoluminescence Dosimetry ◾ 77
6.1.4 Important Literature The energy state of the trap exists between valence and
This chapter cannot provide a complete summary of TLD conduction band as illustrated in Figure 6.1. The rate at
in radiotherapy. In addition to the proceedings of the which electrons are released back into the conduction
Solid State Dosimetry conferences (e.g., [12–15]), which band is governed by Boltzmann statistics and depends
were mentioned above, the reader is referred to several upon the trap depth and temperature of the crystal.
books that provide more detail on thermoluminescence. Electrons can then recombine with the opposite charge
Of particular mention is the three-volume book carrier to return to valence band.
edited in 1984 by Horowitz [6, 16] and the text books of Three characteristics of this process are important for
McKeever et al. [17] and Oberhofer and Scharmann [2]. dosimetry:
More clinical applications are covered in review articles
by Kron [3, 18], Waligorski [19], and Olko [20] as well as 1. The number of trapped electrons is proportional to
a book chapter by Mobit and Kron [21]. the radiation dose received.
Finally, a recent report of the American Association of 2. The probability of releasing electrons back into
Physicists in Medicine (AAPM) task group 191, Clinical the conduction band depends on temperature. As
Use of Luminescent Dosimeters: TLDs and OSLDs, pro- such increasing the temperature of the TLD mate-
vides a lot of useful and practical information [22]. rial will increase the release of electrons. This can
6.2 FUNDAMENTAL PRINCIPLES mathematically be described as:
USING LiF:Mg,Ti AS EXAMPLE
6.2.1 Basic Theory dn  Eg 
− = n s exp  −  (6.1)
We are attempting a simplified description of the TLD dt  kT 
process with the objective to link features of TLD with
practical problems and approaches to use it for dosim- with n the number of electrons in the trap, t the
etry. For a more in depth coverage of TLD physics the time, s the frequency function associated with
reader can be referred to books by Horowitz [6, 23] and a particular trap, Eg the energy gap and kT the
McKeever et al. [17]. Boltzmann constant times the temperature in
Thermoluminescence is a physical process occurring Kelvin.
in insulating crystals such as salts that feature a large 3. In appropriately prepared crystals the recombina-
energy gap between valence and conduction band. This tion between electrons in conduction band and
is illustrated in Figure 6.1 and large in this context refers missing electrons in valence band can occur with
to “much larger than kT for room temperature” with k emission of visible light. In this case the intensity
being the Boltzmann constant (1.38064852 × 10 −23 m2 of the emitted light is proportional to the number
kg s−2 K−1) and T the temperature in Kelvin (K). kT for of trapped electrons and as such the initial dose
300 K or 27°C is approximately 0.025 eV and the typical received by the crystal.
band gap in TLD materials is a few eV. At room tem-
perature all charge carriers such as electrons and elec- 6.2.2 Glow Curves
tron vacancies (holes) are in the valence band. Radiation Figure 6.2 shows the emission of light in two TLD mate-
can free electrons (for the sake of simplicity “electron” rials as a function of the crystal temperature. Several
refers to any charge carrier) into the conduction band different traps with different abundance and differ-
from which they normally would recombine with their ent energy gaps are assumed resulting in several “glow
opposite charge to end up again in the valence band. peaks” and the curve shown is often referred to as the
The trick for TLD materials is that doping them with glow curve of a TLD material.
appropriate impurities creates lattice defects in the crys- In this context some of the complexities of TLD can
tal where electrons can be trapped. This is typically be demonstrated:
indicated by naming the crystal and adding the impu-
rities separated by a colon (e.g., LiF:Mg,Ti is a lithium • The number of traps of a certain type depends
fluoride crystal doped with magnesium and titanium, on the doping and thermal history of the crystal.
a common TLD material also referred to as TLD-100). Doping levels will determine the number of lattice
FIGURE 6.2 Schematic glow curves of LiF:Mg,Ti and LiF:Mg,Cu,P. (a) Glow curve of LiF:Mg,Ti; (b) The same glow curve after
a pre-read annealing at moderate temperature has been performed; (c) Glow curve for LiF:Mg,Cu,P.
imperfections and as such the number of traps. minimizing fading, the reduction of low energy
The thermal history will influence the type of peaks due to storage of the material at room
traps. Maintaining a high temperature for a long temperature.
time will allow the impurities to travel through the
• Evaluation of the TLD signal typically includes
crystal where they can start forming dimers, tri-
only light emitted at a temperature that avoids
mers and higher order combinations. In general,
impact of low energy peaks affected by fading.
the higher the order the deeper the trap. As such
An alternative is glow curve analysis where a fit-
the distribution of impurities and therefore energy
ting routine is used to fit the known glow peaks
levels of the traps depend on the thermal history.
to the acquired temperature/light intensity curve
• The exact description of the glow curve is not ana- [25, 26]. This glow curve analysis provides also an
lytical. However a simplified equation [24] can excellent quality assurance tool for the dosimet-
be used to describe the depletion of trapped elec- ric process and yields typically better accuracy.
trons as the temperature T is linearly increased, as However, it does require considerably more work
shown in Figure 6.2: to fine tune the process and usually relies on a
rather slow readout with slow ramp up of tempera-
I (T ) = I max exp(1 + W (T − Tmax ) − exp(W (T − Tmax ))) ture to resolve individual peaks.
(6.2)
6.2.3 Dose Response
with I the light intensity at temperature T in K
One of the main advantages of TLD is the wide dosimet-
and W = Egap /(kTmax
2 ) where E
gap is the depth of the
ric range which allows dose measurements over many
trap, k the Boltzmann constant and Tmax the tem-
orders of magnitude. This is illustrated in Figure 6.3
perature of the glow curve maximum.
which shows the signal of LiF:Mg,Ti TL material as a
• Heating the TL materials after irradiation to a function of the actual dose received.
moderate temperature will empty some of the As can be seen dose measurements are possible over at
lower electron traps thereby reducing the intensity least eight orders of magnitude from 0.1 mGy to several
of the low peaks during readout as illustrated in 100s of Gy. In some TLD materials such as LiF:Mg,Ti
Figure 6.2b. This “pre-read annealing” is useful in supralinearity is observed as shown in the Figure 6.3.
FIGURE 6.3 Dose response curve for LiF:Mg,Ti (solid line). Supralinearity can be seen for doses above approximately 1 Gy.
The large range of possible dose measurements from fractions of mGy to 100s of Gy can be appreciated.
At higher doses, there is an overresponse due to the fact Size and shape of the dosimeter may also affect angu-
that the signal shows a “dual hit” (or quadratic) compo- lar response. However, as materials are available in dif-
nent due to the microscopic distribution of dose deposi- ferent thicknesses and shapes this effect can be generally
tion. This is a complex process which depends on the minimized.
TLD material, the thermal preparation and the radia- Variation of TLD response with dose rate has not
tion quality [27–29]. been observed, which makes TLD suitable for some
In practice, either the supralinearity must be charac- novel very high dose rate (HDR) deliveries such as syn-
terized during commissioning of the system or standards chrotron irradiation [32] or FLASH, a novel approach to
used for comparison that are exposed to a dose similar to increase normal tissue sparing in radiotherapy through
the one in the measurement (compare also Section 6.4). ultra-high dose rates [33–35].
6.2.4 Other Influences on Dosimetry 6.3 TLD MATERIALS

Many other factors that affect TLD depend on the mate- One of the more important features of TLD is the wide
rial and the readout technique used. As such, some variety of materials that exhibit thermoluminescence.
influencing factors such as variation of response with Table 6.1 shows a summary of some common TLD
radiation quality will be discussed in Section 6.3. As TLD materials.
materials are usually denser than water, additional effects Many more materials exist and the table has been pro-
are related to the detector size and shape. In particular at vided more to illustrate certain features than to provide
low energy x-rays or other weakly penetrating radiation an exhaustive summary of materials. Another impor-
qualities self-attenuation can have an effect [30, 31]. tant material is Beryllium Oxide (BeO) [41] because of
TABLE 6.1 TLD Materials Commonly used in Medical Applications and their Characteristics
TLD Material (Harshaw LiF:Mg,Ti LiF:Mg,Cu,P Li2B4O7:Mn Al2O3:C CaSO4:Dy
Designation) (TLD 100) (TLD 100H) (TLD 800) (TLD 500) (TLD 900)
Physical density (g/cm3) 2.64 2.64 2.3 3.9 2.61
Effective atomic number 8.2 8.2 7.4 9.3 15.3
Sensitivity to 60Co radiation 1 About 30 0.3 40–60 About 15
(relative to LiF:Mg,Ti)
Energy response 30 1.7 1.25 0.9 3.5 12
keV/1.25 MeV
Temperature of main glow 195 210 200 180 220, 250
peak (°C)
Approximate wavelength of 400 380 600 420 480, 570
emitted light (nm)
Fading of main glow peak <10% per year Small 10% per month 5% per year stored 6% in 6 months
at 20°C in dark
Typical annealing 1 h: 400°C 1/4 h: 250°C 1/2 h: 300°C 1 h: 400°C 1/2 h: 400°C
procedure 20 h: 80°C 2 h: 100°C 20 h: 80°C
Useful dose range (Gy) 5 × 10−5 to 103 <10−6 to 100 10−4 to 104 <10−6 to 100 10−6 to 103
Available physical forms Powder, rods, crystals, Ribbons, powder Powder, Teflon- Chips, powder Ribbons, powder,
ribbons, Teflon chips based chips, cards, Teflon-based
ribbons chips
Toxicity High if ingested High if ingested High if ingested Low Low
Remarks Complex glow curve, Good signal to noise, Tissue equivalent Also OSL material; Complex glow
most common TLD, no supralinearity, forms also gem curve, high
available as 6Li and 7Li low-temperature stones sensitivity
annealing
Key references Cameron et al. [5] Moscovitch and Horowitz [6] and Mehta and Horowitz [6] and
and Horowitz [23] Horowitz, [8], Horowitz et al. [37] Sengupta [38] and Souza et al. [40]
Moscovitch [36], and Gaza et al. [39]
Duggan et al. [31]
Source: Adapted and updated from Kron (1994) [3].
its tissue equivalence and the fact that like Aluminum Unfortunately, not all of the variation of TL
Oxide it is also useful for OSL dosimetry which extends response with energy is determined by atomic
usability (and scientific interest). number, as can be seen in Figure 6.4b comparing
Considering Table 6.1 there are several important LiF:Mg,Ti and LiF:Mg,Cu,P [31, 42]. The difference
considerations that should be taken into account what- in response is in this case due to an anomalous
ever TL material is chosen for use: energy response of LiF:Mg,Cu,P described by Olko
et al. using microdosimetric analysis [28, 43, 44].
Even more complicated is the response of dif-
• The physical density provided here is the density
ferent detectors to particle irradiation with high
of a crystal of the material. In practice many TLD
linear energy transfer (LET) [27, 29, 37]. Similar to
crystals are ground to powder and then sintered
biological cell kill, supralinearity (= a linear qua-
into various forms such as rods, chips (often called
dratic component of dose response) plays a much
ribbons because the material is pressed into a “rib-
smaller role than for low LET radiation.
bon” and then cut in regular intervals) and pellets.
This will homogenize the sensitivity of the result- • In some circumstances it is advantageous to com-
ing detectors because it is difficult to grow crystals bine the TLD powder with a more robust matrix
with consistent doping. The resulting dosimeters such as Teflon [45, 46]. In this case also, the effec-
often have a different density in particular if bind- tive atomic number and physical density of the
ing agents are also used. detector will change and needs to be taken into
consideration. A particular interesting applica-
• In medical applications dosimetry aims to deter-
tion of the concept of mixing TL material with a
mine the dose as a quantity related to biological
matrix are “black TLDs.” For “black” TLDs, TLD
effect, be it detriment or tumor cell kill. This is
material (with or without Teflon) is mixed with
related to the term “tissue equivalent” which means
Carbon [47, 48]. Carbon does not affect the dose
that a detector has a dose response to different radi-
deposition in a significant way but it will reduce
ation qualities similar to relevant biological tissues.
the light that can be emitted from the detector
In practice this is often taken as water equivalent.
during readout (see next section). By adjusting
For photon dosimetry the atomic number of the
the carbon loading one can design detectors that
material will significantly affect its response to
report dose up to a depth of approximately 70 μm
radiation as different interaction processes (pho-
as is recommended for skin dose measurements.
toelectric effect, Compton interaction and pair
Figure 6.5 shows different TLD materials includ-
production) have vastly different dependencies on
ing “black TLDs.”
radiation quality for different atomic numbers.
Figure 6.4a shows the ratio of energy absorption • An interesting opportunity exists in mixed radia-
coefficient (broadly related to dose) for different tion fields. Given the fact that many TLD materials
biological tissues as a function of photon energy. contain Lithium or Boron, it is possible to modify
The significant increase in energy absorption the response to thermal neutrons by altering the
at lower energies is due to increasing importance isotope composition. The most common approach
of the photoelectric effect. The same phenomenon is to use pairs of dosimeters based on Li-6 and Li-7.
is observable for the TL detectors as illustrated Li-6 has a cross section for thermal neutrons that is
in Figure 6.4b. This means that to measure water several orders of magnitude larger than the one for
equivalent doses TLD materials with an effective Li-7. Results obtained using Li-7-based detectors
atomic number around 7.5 would be preferable provide information about the photon component
making LiF and BeO-based materials attractive. of the field, while Li-6-based detectors will exhibit
On the other hand it is also possible to assess neutron dose signal in addition to this. While this
dose to other mediums such as bone using TLD is in principle an attractive way to differentially
detectors that mimic the variation of response determine neutron and photon dose it does require
with photon energy of bone. Al2O3 is a reasonable the knowledge of the neutron spectrum which is
choice for this. usually not a given [22, 50, 51].
FIGURE 6.4 Variation of energy absorption coefficient with photon energy for different tissue types (a) and different TLD
materials (b).
• Table 6.1 also lists a selection of physical forms 6.4 EQUIPMENT REQUIRED FOR TLD
in which detectors are available. This is another Equipment and instrumentation used for TLD needs to
important advantage of TLD where the detec- ensure optimal use in terms of reproducibility, sensitiv-
tor can be designed to suit a particular applica- ity, linearity, lack of response variation with radiation
tion such as microcubes and pinworms [52–54] quality and incidence direction and of course time and
for small field and brachytherapy dosimetry, resource requirements. Not surprisingly, accurate tem-
respectively. In addition to this, the combination perature control is a key component in every step of
of detectors of different physical sizes allows to TLD. As discussed in Section 6.2, the response of TLD
inter- and extrapolate doses [55, 56] which has materials may be affected by thermal history, heating
been shown to be a useful approach in skin dose and cooling rates, fading, dirt or scratches, and other
measurements [57]. handling procedures. The success of a TLD system as a
FIGURE 6.5 Several different versions of LiF-based TL detec- FIGURE 6.6 TLD chips pre-packaged in a plastic sleeve using
tors (adapted from Kron, 1999 [49]). a heat sealer. The use of two detectors improves precision and
the plastic sleeve ensures easy handling and labeling (back of
whole depends upon careful control of each influencing the envelope), whilst not significantly attenuating radiation
factor. This section provides an overview of the equip- and increasing the size.
ment and processes required to establish a TLD system
for radiotherapy. the TLD. This method provides the flexibility to pack-
age multiple TLDs in various configurations and even
6.4.1 TLD Handling allows a series of TLDs to be sealed together to form an
Since dirt and scratches can affect the light output from array. An example of two TLD chips heat-sealed within
a TLD material, it is imperative that the dosimeter be a plastic sleeve is shown in Figure 6.6. Other TLD encap-
kept clean and free of damage. Tiny scratches on the sur- sulations will be discussed in the next section [51].
face of a TL dosimeter will not only reduce the amount Often in medical dosimetry it is necessary to take
of physical material available for evaluation, but also measurements at multiple locations. It is therefore desir-
reduce light emission for a given radiation exposure. able to group multiple TLDs together in “batches” of 50
Contaminants such as dirt and oils similarly reduce the or more. TLD disks, rods, and ribbons (chips) are stored
light output by blocking light at the surface of the mate- in dedicated high-temperature annealing trays which
rial. Vacuum tweezers can be used for TLD handling, are labeled to allow easy identification of individual
often being inexpensive and an effective means of avoid- TLDs. An example tray with spaces for up to 100 TLD
ing damage. chips is shown in Figure 6.7.
For patient in vivo dosimetry, TLDs must be prepared These trays should provide good thermal conduc-
in appropriate packaging suitable for the intended mea- tivity throughout the tray, ensuring all TLDs within a
surement. Often it is desirable to maintain the small batch are exposed to the same temperature/time profile.
physical size of the TLD to ensure patient comfort dur- This is particularly important during high-temperature
ing measurement and minimize any perturbation of annealing since any heterogeneity in thermal treatment
the radiation beam. Packaging procedures must ensure within a batch can lead to differences in dose response
the TLD material does not come into contact with the for individual TLDs, which may be unpredictable.
patient. This can be achieved by encasing TLDs within
plastic, which also allows the TLDs to be labeled so they 6.4.2 TLD Reader
can be correctly identified after treatment. Thin plastic The role of the TLD reader is to supply thermal energy
sleeves can be used to secure individual TLDs which can to the TL material and record subsequent light emis-
then be labeled. A heat-sealer may be used to secure the sions, usually captured by an inbuilt photomultiplier
edges of the plastic and prevent loss of TLDs, though tube. Commercially available readers vary in complex-
care should be taken to avoid heat sealing too close to ity, with more advanced models offering automated
benefit of increased throughput and can help to ensure

consistency in TLD readout. The insert in Figure 6.8
shows an example carousel used in automated readers
which can accommodate up to 50 TLD chips.
Different heating methods can be employed depend-
ing on the make of the TLD reader. The simplest
method of heating utilizes a hot planchet on which the
TLD materials rests [58]. Being in physical contact, this
method can lead to non-uniform heating of the mate-
rial. For low energy x-ray irradiation, one must also be
cautious to maintain the same orientation of the TLD
on a hot planchet, since any self-attenuation during irra-
diation can lead to changes in light emission, depending
on whether the TLD was placed on the planchet in the
FIGURE 6.7 High-temperature annealing tray with space for same orientation as it was during irradiation. An alter-
up to 100 TLD chips. Vacuum tweezers used for TLD han- native method involves passing a hot inert gas over the
dling are also shown. TLD as it is held on a vacuum needle. This method helps
to ensure uniform heating as well as uniform light emis-
readout of multiple chips. Simpler designs require the sion, and has the advantage that the TLD does not need
user to manually read individual TLDs. The benefit of to sit atop a potentially dirty surface, thereby reducing
simpler models, besides the reduction in initial cost, the risk of contamination.
is added diversity in the potential types of TLD which Automated readers can be controlled by PC-driven
can be used. Automated readers are often restricted to software, allowing the user to pre-program custom-
a particular form of TLD, requiring a carousel or stack- ized readout cycles with desired heating rates and
ing mechanism which is used to hold many TLDs or readout temperatures. A commercial automated reader
TLD cards during readout. Automated readers have the and control PC is shown in Figure 6.8. Programmable
FIGURE 6.8 Automatic programmable TLD reader (Harshaw 5500). The computer screen displays a glow curve with the
respective temperature profile indicated by a line. In the insert a carousel which can accommodate up to 50 TLD chips is
shown.
readers also allow for material-specific readout proce- TLD is read multiple times residual signal will still be
dures which can help to avoid accidental damage from present, which can affect subsequent measurements.
over-heating. The user can define areas of interest in The remaining signal depends upon the readout temper-
the TL glow curve used for evaluation. Often the area atures but in practice a high temperature anneal using
under the curve is used for evaluation since it is usu- a programmable oven is required to completely empty
ally more robust than the height of a peak. Evaluation remaining traps and ensure the TLD crystal is ready for
of higher energy peaks tends to be more robust than a new measurement. This is particularly important if a
the lower peaks, which can fade quicker after irradia- low dose is to be assessed after the same TL chip has
tion. In some cases it may be desirable to program a pre- received a relatively high dose.
read anneal cycle to eliminate lower, less stable peaks, Ensuring reproducible and uniform annealing also
thereby excluding their contribution to the final glow helps to control the thermal history of a TLD set, which
curve used for analysis. is a crucial factor to a TLD system as it relies on compar-
ison of TL signal from TLDs irradiated to an unknown
6.4.3 TLD Evaluation, Quality Assurance, dose to the others which have received a known one
and Analysis (typically referred to as standards). An example of a pro-
For clinical applications, timely analysis of multiple grammable oven used for TLD annealing is shown in
TLDs including standards is often required. A visual Figure 6.9.
representation of the glow curve should still be avail- TLD readers can also be used to anneal TLDs indi-
able for routine quality assurance and troubleshooting, vidually, however this process has some drawbacks.
since the shape of the curve can be useful for interpreta- First, only a single TLD can be annealed at a time, lead-
tion of TLD results. It is desirable to have the raw data ing to a prohibitively long time required to anneal a set.
exported to a personal computer for further analysis. Second, and perhaps most important, the anneal cycle of
A simple spreadsheet can be an effective means of ana- a TLD reader may not be as reproducible nor can higher
lyzing TLD results and templates can be created to aid temperatures be maintained. TLD readers also cannot
in maintaining records. For TLD readers which record achieve longer anneal times which may be required for
photomultiplier background signal and the signal from certain materials (compare Table 6.1).
a check light source, regular review of these metrics can
be useful as a constancy check for the performance of a 6.5 APPLICATIONS IN RADIOTHERAPY
TLD reader. Dose measurements are of great importance in radio-
When a system is initially commissioned glow curve therapy as very high and potentially lethal radiation
analysis can be a useful means of providing insight into doses are delivered. As a large part of this book deals
the performance of a TLD system and optimize tem- with a wide range of applications we cover only five
perature and time settings [24, 26]. In principle, equa- areas briefly where TLD has been or continues to be an
tions such as (6.2) are used to fit the experimentally important dosimetric technique.
determined glow curve. Given the finite time required Figure 6.10 attempts a summary of dosimetric needs
for thermal heating of the crystal to a given tempera- in radiotherapy. They range over at least 8 orders of
ture, glow curve de-convolution is best done using a magnitude in dose from μGy to several tens of Gy and
slow continuous temperature rise. Glow curve analysis it is beyond the scope of the present chapter to cover all
is also useful for defining regions of interest which are of them in detail. However, as discussed in the previous
to be used for subsequent TLD readings. Commercially sections and illustrated in Figure 6.3, TLD would be well
available software may be used, or if a glow curve can suited to cover this range and the related variation in
be exported from the TLD reader, this can be done dose rate.
manually. While the accuracy of dose measurements varies
between applications it is generally assumed that dosime-
6.4.4 Annealing and Reuse try in radiotherapy should yield dose with an uncertainty
During the readout process most (but not all) of the orig- better than 5% [59]. Quality assurance is an essential part
inal signal is lost as trapped electrons are released and of trying to achieve this and dose measurements and
interact with recombination centers. Even if the same independent audits play an important role.
FIGURE 6.9 High temperature programmable oven used for TLD annealing. A real-time temperature display is shown at the
bottom display.
Historically many different systems have been techniques that provide faster and multiple read out.
developed to introduce TLD into a radiotherapy clinic. However, for many applications such as finger rings [63]
Figure 6.11 shows a general flow of activities which and dosimeters to assess lens dose during diagnostic
would suit a system using TLD chips in batches as procedures [64], TLD is still a method of choice particu-
illustrated in the previous section. larly where the radiation quality is not well known. A
useful reference is the assessment of uncertainty in TL
6.5.1 Radiation Protection personnel monitoring provided by Veinot [65].
TLD has been for many years the most important dosi- In radiation oncology dose to staff has been decreas-
metric technique for personnel dosimetry [60, 61]. This ing significantly with the replacement of 60Co units with
has been now challenged by both OSL (Chapter 7) [10] linear accelerators and the introduction of remote after-
and radiophotoluminescence (RPL) [62] dosimetric loading units for brachytherapy. However, for regulatory
FIGURE 6.10 Application of dosimetry in radiotherapy as a function of dose.

FIGURE 6.11 Typical use cycle for a TLD system based on batches of TLD chips or rods. A batch consists here of some 50 TL
detectors which share an identical thermal history. This allows comparing unknown measured dose with known doses from
detectors in the same batch that have been exposed to a known dose with the same radiation quality as is measured. The rela-
tive response of the different detectors to each other is established using calibration runs in which all detectors in a set are
exposed to the same dose.
purposes and best practice in a high-dose environment, the geometry and radiological properties of a human
personnel dosimetry is usually required. [66–69]. This allows end-to-end (E2E) testing of proce-
As integrative dosimeters TLDs are also well suited dures from imaging to radiotherapy treatment delivery.
for environmental dosimetry. In a medical environ- Figure 6.12 shows an anthropomorphic phantom setup
ment, this can be used to verify the shielding of radio- for breast radiotherapy at a linear accelerator.
therapy or diagnostic facilities. TL dosimeters can be Anthropomorphic phantoms typically consist of many
easily packaged and placed in any location for a suitable slices as shown in Figure 6.12. Each slice is a few centime-
period of time. In addition to high sensitivity and small ters thick and includes a grid of pre-drilled holes where
volume they are largely independent of environmental detectors can be placed. Holes can be plugged with tissue
influences. The TLD measurement provides a real-life equivalent material if not in use for dosimeters and it is
indication of the dose at a certain location over a given possible to drill additional holes in areas of anatomical
period of time – it verifies the shielding calculation as interest as required for particular applications as shown
well as the assumed workload and the actual building in the insert of Figure 6.12. During the actual dose deliv-
and can be performed using a clinical TLD system. ery, TL dosimeters can be placed in selected locations
within the phantom and the dose delivery verified.
6.5.2 Phantom Measurements There are three major applications in radiotherapy for
TL dosimeters as passive dosimeters do not require these measurements:
cables or batteries during irradiation. This allows place-
ment in phantoms to mimic the treatment of patients. • Commissioning and testing of planning systems
While this can be done in principle for all types of
• Development of new treatment techniques
phantoms, the most important application is the use
of TLD in anthropomorphic phantoms representing • Quality assurance for individual patients
FIGURE 6.12 Anthropomorphic phantom setup for breast radiotherapy. TLDs are used on the surface of the phantom as well
as inserted into plugs typically provided in commercial phantoms in a 30 mm or 15 mm pattern. The insert to the figure shows
plugs that have been modified to accept TLD chips.
Commissioning of treatment planning systems in verifying the radiation dose but also in assessing the
is one of the most important tasks in setting up a workflow and providing training to staff. TLD measure-
radiotherapy department [70]. It requires beam data ments in anthropomorphic phantoms can provide the
acquisition, beam modeling, image manipulation, dosimetric verification.
contouring, as well as geometry and dose calculation Finally, the treatment of individual patients may
assessment. The final important step would be E2E require verification. While this is rather labor intensive
verification where a model representing a patient is and more commonly done using in vivo dosimetry as
imaged, planned and treated in exactly the same way described in the next section, there is scope to mock
a patient would be. Ideally this would include the cre- up uncommon treatments (e.g., total body irradiations,
ation of relevant structures and calculation of dose electron arcs) or unusual patients (e.g., bariatric, preg-
volume histograms. Anthropomorphic phantoms are nant, or pediatric patients) prior to treatment [73].
ideal for this purpose and TLDs are well suited for the
required measurements as it is not only important to 6.5.3 In Vivo Dosimetry
verify the high dose to the target but also the low dose TLD is one of the most commonly used techniques for
outside of the treated volume which is received by in vivo dosimetry [74]. Its typical applications are the
normal tissues. For these out-of-field measurements, assessment of the dose to critical organs and measure-
TLDs which can be close to tissue equivalence are the ments in difficult geometries. In clinical practice usu-
ideal dosimeters. ally LiF-based materials are used. This material has
A similar process would be required for the develop- relatively close tissue equivalence (Zeff = 8.3), and can
ment of new treatment techniques and audits which will be produced in the desired forms such as powder, rib-
be discussed in Section 6.5.5 [71, 72]. New treatment bons or rods with small dimensions of the order of few
techniques must be verified before put into clinical rou- mm. The typical accuracy reached in clinical practice of
tine. By their very nature new techniques are unfamil- about ±3% (2SD) is sufficient for the requirement of in
iar to staff and as such an E2E test is useful not only vivo dosimetry.
The major advantages of TLD for in vivo dosimetry In vivo dosimetry (Chapter 11) [75] can roughly be
are the small size of the detectors, their stand-alone separated into several different application groups [74]:
measurement characteristic and the fact that TL mate-
rials typically consist of one material only. Therefore, 1. Routine verification on every patient as is required
the reading of the TLD is generally independent of in some European countries. As these measure-
the angular distribution of the radiation. This is an ments are limited to surface dose assessment they
important feature for measurements in complicated are increasingly difficult to interpret as intensity
geometries where it may be difficult to estimate which modulated treatments become more common.
direction radiation may come from. TLD detectors are Other, real time dosimetric techniques such as
typically prepackaged prior to placing them on the diodes, MOSFETs (Chapter 4) [76], or electronic
patient’s skin in appropriate locations after the setup portal dosimetry (Chapter 8) [77] are more suitable
(compare Figure 6.6). This causes minimal interference for this and as such more commonly employed.
with patient setup and treatment.
2. Assessment of dose to critical structures, typically
In radiotherapy, skin dose can mean the dose to the
outside of the main radiation field. This is usually
basal cell layer around 0.07 mm deep. While TLDs are
done for documentation purposes with scrotal
typically too thick to give a meaningful result for such
and lens dose measurements being common. In
a shallow depth, TLD extrapolation or black TLDs can
the case of scrotal dose measurements these may
be used [48, 57]. This is relevant for skin toxicity such as
also be done in the presence of scrotal shielding
desquamation.
which make the prediction of the dose using plan-
On the other hand, the term “skin dose” is often used
ning systems virtually impossible. In some of these
to describe entrance dose more generally. In megavolt-
scenarios it is not possible to determine the dose
age radiotherapy, this includes the build-up effect and
directly. Figure 6.14 is an example for this where
the dosimeters need to be packaged to take this into
the lens dose is estimated from several detectors
consideration. Figure 6.13 shows a method to do this
placed around the lens [21]. This can be done for
using Perspex build-up domes [51]. Similar contain-
separate fields if individual fields are used or all
ers which make handling of the detectors easier and at
together in arc type deliveries.
the same time provide suitable buildup can now also be
made using additive manufacturing (3D printing). 3. Determination of dose where it is genuinely
For in vivo dosimetry, the major disadvantage of TLD unknown or very difficult to predict. The most com-
is the delayed readout. However, in fractionated radio- mon scenario for this is dose at distance from the
therapy it is feasible to obtain the results of measure- treated area. Dose calculation algorithms are lim-
ments after the first fraction of treatment and before the ited in their accuracy at distance from the field edge
next fraction is delivered to the patient [21]. as the dose depends on radiation leakage, internal
and external scatter something difficult to model. It
also yields a radiation spectrum which is difficult to
determine and as such the tissue equivalence and
high sensitivity of TL materials such as LiF:Mg,Cu,P
provides for highly suitable detectors [51].
6.5.4 Brachytherapy
TLD has some features which make it an ideal dosim-
eter for brachytherapy, the use of radioactive sources
implanted in the patient for treatment of malignancies
(Chapter 15) [78]. Brachytherapy implants are charac-
terized by radiation sources of various gamma-ray ener-
FIGURE 6.13 Perspex build-up dome to facilitate gies that are very close to tissues resulting in strong dose
entrance dose measurements in megavoltage external beam gradients. The tissue equivalence and good spatial reso-
radiotherapy. lution due to small physical size of TL dosimeters make
FIGURE 6.14 Suggestion for TLD use for in vivo dosimetry. (a) Lens dose measurement using TLDs under buildup that mim-
ics depth of the lens with closed eye lids. Three locations are used that can provide an estimate of lens dose in anterior (TLD2),
lateral (TLDs 1 and 3) and multifield/arc deliveries (judicious mix of all three); (b) Label for TLD placed inside a sealed satchel
for measurement in the oral cavity of a patient; (c) “Array” of TLDs wrapping around an irregular surface (neck) to investigate
early skin reaction.
them a promising tool for dosimetry in these difficult Shown in the figure is the dosimetry approach by the
circumstances. WHO/IAEA remote dosimetric audit using TLD cap-
As in external beam therapy, brachytherapy can sules filled with TLD powder [88–90].
benefit from in vivo dosimetry. For HDR brachyther- Audits are often categorized into three groups [91, 92]:
apy dose rates and fraction sizes are high enough to
allow these measurements to be done on skin [79, 80] • Level I dosimetric audits verify dose in reference
or in adjacent catheters [81, 82]. Of particular interest conditions. The setup of Figure 6.15 illustrates
are also doses to critical structures such as rectum such an approach where the dose is verified in a
or bladder which allow for relatively easy insertion of standardized simple scenario.
radiation detectors [83, 84]. However, as for other in
• Level II dosimetric audits are primarily aimed a
vivo dosimetry the delayed readout of the TL dosim-
treatment planning systems and verify how the
eters is not ideal for brachytherapy where real time
planning system handles typical dosimetric sce-
dosimetry would be preferential and other active
narios such as off-axis, simple inhomogeneities
detectors are more commonly used for in vivo dosim-
and different field sizes. Electron output verifica-
etry [80, 82, 85].
tion is somewhere in between levels I and II for
More recently TLDs have also been used to perform
audit purposes as a dose measurement at a point
dosimetric intercomparisons and audits for brachyther-
usually also requires an assessment of effective
apy [72, 86].
energy to determine dose response [93, 94].
6.5.5 Audits • Level III dosimetric audits finally are an E2E test
TLDs have been used for dosimetric audits for many of a typical (or challenging) treatment scenario,
years by many national and international organizations verifying all part of the radiotherapy chain from
[87, 88]. The simplest audit is the assessment of dose imaging for planning to dose optimization and
under reference conditions often referred to as level I calculation to verification and delivery. For this
audit. Figure 6.15 shows a very simple setup to perform type of dosimetric audit commonly anthropomor-
these measurements in water without the need for an phic phantoms as discussed above are used.
expensive scanning water phantom. This measurement
can easily be performed in virtually all radiotherapy TLDs are useful for all levels of audits in particular
centers in the world. when the audits are done remotely. In this case the light
FIGURE 6.15 Simple setup of capsules filled with LiF powder for remote auditing of radiotherapy beams at 5cm depth in
water.
weight, general robustness, and stand-alone nature of good service” this in itself can improve the quality of the
TLDs prove to be a significant advantage. As for person- TLD measurements.
nel dosimetry, OSL dosimeters have also become popu- Related to this is the need to meticulous and stan-
lar for level I dosimetry due to their faster and repeatable dardized work, which makes it often necessary to
readout [93]. For level III applications done remotely entrust TLD measurements to only few staff members.
TLDs still have the additional advantage of tissue equiv- However, once set up, a TLD service is very versatile
alence which is important when dose is assessed also and can become useful in areas such as radiation safety,
outside of the high-dose region. quality assurance and development of new procedures.
In practice, many audits combine measurements of Being busy also reduces the cost per measurement as the
point doses with TLDs with radiochromic film (Chapter 5) most significant investment is the TLD reader itself [95].
[11] which provides for an assessment of not only dose but Delayed readout is a limitation of TLD, which is
also dose distribution which has become essential in level shared amongst all “passive” dosimeters such as radio-
III audits of modern radiotherapy techniques. chromic film and most OSL detectors. This makes TLD
not suitable for measurements where real time results
6.6 LIMITATIONS OF TLD are required. In practice the turnaround of TLD mea-
TL dosimetry has a number of disadvantages that have surements is several hours which allows obtaining
both stimulated improvements in TL materials and results from one fraction in radiotherapy for the next.
techniques as well as the utilization of other technolo- Information loss after readout is a potential problem
gies. Some of them are contrasted to the respective for radiation monitoring of staff or the environment.
advantages in Table 6.2. The most important drawback While false-positive results are rare in TLD, a lost read-
in a medical environment is probably the fact that TLD ing is not replaceable and an unexpected reading has to
is comparatively labor intensive and time consuming. be taken at face value and acted upon. Pulsed OSL pro-
It typically requires several steps (annealing, calibra- vides a useful alternative to TL and has replaced TLD
tion, packaging, pre-read annealing, readout, signal in many personal monitoring systems. It is interesting
analysis), many of which need to be tightly controlled to note that most TLD materials also exhibit OSL prop-
to achieve acceptable reproducibility. Automated reader erties and the first dual evaluation systems have now
systems can help with this and as “a busy service is a become commercially available (https://www.lexsyg.
TABLE 6.2 Summary of Advantages and Disadvantages of TLD for Radiotherapy Applications
Advantages Disadvantages
Small physical size Small physical size (difficult handling, fragile)
Tissue equivalence facilitating out-of-field measurements No absolute dosimetry possible
Integrative, passive dose measurement Delayed readout
No cables required during measurement Concerns regarding reliability (black art)
High sensitivity – wide dosimetric range Concerns regarding accuracy (which can be overcome
through meticulous control of all aspects of the dosimeter
handling)
Reusable and cheap – multiple measurements can be made at the same time Expensive readout equipment required
Many forms and materials available (allows optimal choice for particular Many forms and materials available (confusing and difficult
purpose) to find appropriate technique for a particular application)
Mostly independent of environmental conditions such as temperature and No permanent dose record
pressure
Glow curve provides additional information which could be useful to Requires dedicated operator
improve accuracy or obtain information on radiation quality
Source: Adapted from Mobit and Kron (2006) [21].
com/tlosl-reader.html or https://www.nutech.dtu.dk/ more important. Tissue equivalence and high sen-

english/products-and-services/radiation-instruments/ sitivity are essential for this.
tl_osl_reader).
• The same features make TLD an important tool for
In this context, it also worth mentioning another
in vivo dosimetry. The stand-alone nature of the
related dosimetric technique: RPL [62, 96–98]. In RPL
measurement as well as the small size help as well.
ultraviolet light is used for readout of dosimeters. They
typically exhibit a small amount of fading and are resis- • Out-of-field and in vivo dosimetry often go
tant to most environmental influences. As such they are together. Of increasing relevance is the combined
predominantly used in personnel and accident dosim- dose from verification imaging and treatment
etry and their applications in radiotherapy are just delivery. As radiation quality is even more difficult
emerging [97, 99]. to predict in this case, TLD with its tissue equiva-
lence has a distinct advantage.
6.7 CONCLUSION • Remote dosimetric audits can continue to benefit
A wide range of dosimeters are available to medical from the relative robustness, long term stability
physicists to make meaningful assessments of dose in and stand-alone nature of TLDs. It can be expected
diagnostic and therapeutic procedures. It can be antici- though that many audits will use TLDs in combi-
pated that TLD will continue to feature as a relevant dose nation with other dosimeters.
measurement tool in the future as it has some unique
• The same features are relevant for environmental
characteristics, which will make it difficult to replace it –
dosimetry. While personnel dosimetry is unlikely
even if this means accepting the TLD disadvantages of
to revert back to wide spread use of TLD, assessing
workload and delayed readout.
new facilities through environmental monitoring
By comparing advantages and disadvantages listed in
will continue to be useful, particular in remote or
Table 6.2, it is possible to identify some areas in medi-
under-resourced locations.
cine where one can expect TLD to maintain a strong
presence or find even increasing use in radiotherapy • Due to reusability TLD is a relatively cheap tech-
dosimetry: nique. This helps with environmental dosimetry
and audits (if a TLD is lost in action its calibration
• Out-of-field dose is difficult to predict from treat- factor is probably the most costly loss).
ment planning but increasingly important as
patients live longer and very late effects such as The role of TLD to address other emerging problems in
secondary cancers or cardiac toxicity become radiotherapy such as hadron therapy [100], synchrotron
microbeams [101], and FLASH [35] is not clear yet. 13. Solid state dosimetry. Proceedings of the 14th
However, there are many great opportunities to perform International Conference on Solid State Dosimetry
(SSD14), June 27–July 2, 2004, New Haven, Connecticut,
meaningful research.
USA, Radiation Protection Dosimetry 119, 1–517 (2006).
Given this wide range of application for TLD one 14. Proceedings of the XII International Symposium on
can anticipate that the technique will continue to be an Solid State Dosimetry, September 5–9, 2011, Mexico
important dosimetric technique for medical applica- City, Mexico, Applied Radiation and Isotopes 71 (Suppl),
tions. While innovation is typically slow to be widely 1–101 (2012).
implemented in clinical practice, the rapid develop- 15. E. Goldfinch, S. W. S. Mckeever, A. Scharman, Solid
State Dosimetry – Proceedings of the 10th International
ment of new technology and TL materials will offer new Conference on Solid State Dosimetry, Held in Washington,
opportunities which will sooner or later benefit patients DC, July 13–17, 1992 (Nuclear Technology Publishing,
in medicine. Ashford, UK, 1993).
16. Y. S. Horowitz, “TL dose response,” in Thermoluminescence
REFERENCES and Thermoluminescent Dosimetry, edited by Y. S.
Horowitz (CRC Press, Boca Raton, 1984).
1. J. Cameron, N. Suntharalingam, G. Kenny, 17. S. W. S. McKeever, M. Moscovitch, P. Townsend,
Thermoluminescence Dosimetry (University of Wisconsin Thermoluminescence Dosimetry Materials: Properties
Press, Madison, 1968). and Uses (Nuclear Technology Publishing, Ashford,
2. M. Oberhofer and A. Scharmann, Applied UK, 1995).
Thermoluminescence Dosimetry (Adam Hilger, Bristol, 18. T. Kron, “Thermoluminescence dosimetry and its
1981). applications in medicine–Part 2: History and applica-
3. T. Kron, “Thermoluminescence dosimetry and its tions,” Australasian Physical and Engineering Sciences
applications in medicine–Part 1: Physics, materials in Medicine 18, 1–25 (1995).
and equipment,” Australasian Physical and Engineering 19. M. P. R. Waligorski, “What can solid state detectors
Sciences in Medicine 17, 175–199 (1994). do for clinical dosimetry in modern radiotherapy?,”
4. F. Daniels, C. Boyd, D. Saunders, “Thermoluminescence Radiation Protection Dosimetry 85, 361–366 (1999).
as a research tool,” Science 117, 343–349 (1953). 20. P. Olko, “Advantages and disadvantages of lumines-
5. J. Cameron, F. Daniels, N. Johnson, G. Kenny, cence dosimetry,” Radiation Measurements 45, 506–511
“Radiation dosimeter utilizing the thermoluminescence (2010).
of LiF,” Science 134, 233 (1961). 21. P. Mobit and T. Kron, “Applications of thermolumi-
6. Y. S. Horowitz, “General characteristics of TLD mate- nescence dosimeters in medicine,” in Microdosimetric
rials,” in Thermoluminescence and Thermoluminescent Response of Physical and Biological Systems to Low and
Dosimetry, edited by Y. S. Horowitz (CRC Press, Boca High LET Radiations, edited by Y. S. Horowitz (Elsevier,
Raton, 1984). Amsterdam, 2006).
7. Y. S. Horowitz, “The annealing characteristics of 22. S. F. Kry, P. Alvarez, J. E. Cygler, L. A. Dewerd, R. M.
LiF:Mg,Ti,” Radiation Protection Dosimetry 20, 219–230 Howell, S. Meeks, J. O’Daniel, C. Reft, G. Sawakuchi, E.
(1990). G. Yukihara, D. Mihailidis, “AAPM TG 191 clinical use
8. M. Moscovitch and Y. S. Horowitz, “Thermoluminescent of luminescent dosimeters: TLDs and OSLDs,” Medical
materials for medical applications: LiF: Mg, Ti and LiF: Physics 47, e19–e51 (2020).
Mg, Cu, P,” Radiation Measurements 41, S71–S77 (2006). 23. Y. S. Horowitz, “A unified and comprehensive theory of
9. G. Portal, Proceedings of the 6th International TL dose response of thermoluminescent systems applied
Conference on Solid State Dosimetry. Nuclear Instruments to LiF:Mg,Ti,” in Microdosimetric Response of Physical
and Methods, vol. 175 (1980). and Biological Systems to Low and High LET Radiations,
10. S. Kry and J. O’Daniel, “Optically stimulated lumi- edited by Y. S. Horowitz (Elsevier, Amsterdam, 2006).
nescent dosimeters in clinical practice,” in Radiation 24. A. Delgado, J. Gomezroz, J. Muniz, J. C. Portillo,
Therapy Dosimetry: A Practical Handbook, edited by A. “Application of glow curve analysis methods to improve
Darafsheh (CRC Press, Boca Raton, 2021), Chapter 7, TLD-100 dose reassessment performance,” Health
pp. 97–108. Physics 62, 228–234 (1992).
11. S. Mosahebi, N. Hooshyar, R. Khan, A. Darafsheh, 25. A. Bos, T. Pithers, J. Gomez Roz, A. Delgado, “An
“Film dosimetry,” in Radiation Therapy Dosimetry: intercomparison of glow curve analysis computer pro-
A Practical Handbook, edited by A. Darafsheh (CRC grams: I. Synthetic glow curves,” Radiation Protection
Press, Boca Raton, 2021), Chapter 5, pp. 61–74. Dosimetry 47, 473–477 (1993).
12. Proceedings of the 13th International Conference on 26. Y. S. Horowitz and D. Yossian, Computerised Glow Curve
Solid State Dosimetry (SSD2001), Athens, Greece, July Deconvolution: Application to Thermoluminescence
9–13, 2001, Radiation Protection Dosimetry 100, 1–552 Dosimetry (Nuclear Technology Publishing, Ashford,
(2002). UK, 1995).
27. P. Olko, P. Bilski, N. A. El-Faramawy, H. Y. Goksu, J. L. 40. J. Souza, L. Darosa, C. Mauricio, “On the thermo-
Kim, R. Kopec, M. P. Waligorski, “On the relationship luminescence glow curve of CaSO4:Dy,” Radiation
between dose-, energy- and LET-response of thermolu- Protection Dosimetry 47, 103–106 (1993).
minescent detectors,” Radiation Protection Dosimetry 41. A. Malthez, M. B. Freitas, E. M. Yoshimura, N. K.
119, 15–22 (2006). Umisedo, V. Button, “OSL and TL techniques combined
28. P. Olko and M. P. Waligorski, “Microdosimetric one in a beryllium oxide detector to evaluate simultaneously
hit detector model for calculation of dose and energy accumulated and single doses,” Applied Radiation and
response of some solid state detectors,” Radiation Isotopes 110, 155–159 (2016).
Protection Dosimetry 99, 381–382 (2002). 42. B. Obryk, C. Hranitzky, H. Stadtmann, M.
29. A. R. Lakshmanan and R. Bhatt, “Supralinearity and Budzanowski, P. Olko, “Energy response of different
LET dependence of thermoluminescence dosim- types of RADOS personal dosemeters with MTS-N
eters,” Physics in Medicine and Biology 26, 1157–1163 (LiF:Mg,Ti) and MCP-N (LiF:Mg,Cu,P) TL detectors,”
(1981). Radiation Protection Dosimetry 144, 211–214 (2011).
30. N. Nariyama, S. Tanaka, Y. Nakane, Y. Asano, H. 43. P. Olko, P. Bilski, E. Ryba, T. Niewiadomski,
Hirayama, S. Ban, H. Nakaskima, Y. Namito, “Responses “Microdosimetric interpretation of the anomalous photon
and glow curves of Li2B4O7:Cu, BeO and CaSO4:Tm energy response of ultra-sensitive LiF:Mg, Cu, P TL dose-
TLDs to 10–40 keV monoenergetic photons from syn- meters,” Radiation Protection Dosimetry 47, 31–35 (1993).
chrotron radiation,” Radiation Protection Dosimetry, 44. P. Olko, “Microdosimetric interpretation of photon
74, 155–161 (1997). energy response in TL systems,” in Microdosimetric
31. L. Duggan, C. Hood, H. Warren-Forward, M. Haque, Response of Physical and Biological Systems to Low and
T. Kron, “Variations in dose response with x-ray energy High LET Radiations, edited by Y. S. Horowitz (Elsevier,
of LiF:Mg,Cu,P thermoluminescence dosimeters: Amsterdam, 2006).
Implications for clinical dosimetry,” Physics in Medicine 45. G. Webb, “Thermal annealing effects in the thermo-
and Biology 49, 3831–3845 (2004). luminescence of lithium fluoride Teflon dosimeters,”
32. G. Grafstrom, B. A. Jonsson, S. E. Strand, P. Spanne, Health Physics 13, 814–816 (1967).
“Dosimetry at 28 keV synchrotron radiation for cell 46. A. Nash and F. Attix, “Use of LiF Teflon discs in lami-
irradiations,” Radiation Protection Dosimetry 108, 115– nated identification cards for personnel accident dosim-
122 (2004). etry,” Health Physics 21, 435–439 (1971).
33. P. G. Jorge, M. Jaccard, K. Petersson, M. Gondre, M. 47. S. Thomas and N. Palmer, “The use of carbon-loaded
T. Duran, L. Desorgher, J. F. Germond, P. Liger, M. C. thermoluminescent dosimeters for the measurement
Vozenin, J. Bourhis, F. Bochud, R. Moeckli, C. Bailat, of surface dose in megavoltage x-ray beams,” Medical
“Dosimetric and preparation procedures for irradiating Physics 16, 902–904 (1989).
biological models with pulsed electron beam at ultra- 48. P. Ostwald, T. Kron, C. Hamilton, J. Denham, “Clinical
high dose-rate,” Radiotherapy and Oncology 139, 34–39 use of carbon-loaded thermoluminescence dosimeters
(2019). for skin dose determination,” International Journal of
34. J. Bourhis, W. J. Sozzi, P. G. Jorge, O. Gaide, C. Bailat, F. Radiation Oncology, Biology, Physics 33, 943–950 (1995).
Duclos, D. Patin, M. Ozsahin, F. Bochud, J. F. Germond, 49. T. Kron, “Dose measuring tools,” in Modern Technology
R. Moeckli, M. C. Vozenin, “Treatment of a first patient of Radiation Oncology, edited by J. van Dyk (Medical
with FLASH-radiotherapy,” Radiotherapy and Oncology Physics Publishing, Wisconsin, 1999).
139, 18–22 (2019). 50. H. A. Nedaie, H. Darestani, N. Banaee, N. Shagholi, K.
35. M. C. Vozenin, M. Baumann, Coppes, R. P. & J. Mohammadi, A. Shahvar, E. Bayat, “Neutron dose mea-
Bourhis, “FLASH radiotherapy international work- surements of Varian and Elekta linacs by TLD600 and
shop,” Radiotherapy and Oncology 139, 1–3 (2019). TLD700 dosimeters and comparison with MCNP cal-
36. M. Moscovitch, New TLD materials and their merits: culations,” Journal of Medical Physics 39, 10–17 (2014).
LiF:Mg,Cu,P and Al2O3 (1991). 51. P. Lonski, S. Keehan, S. Siva, D. Pham, R. D. Franich, M.
37. Y. S. Horowitz, J. A. Kalef-Ezra, M. Moskovitch, H. L. Taylor, T. Kron, “Out-of-field in vivo dosimetry using
Pinto, “Further studies on the non-universality of TLD in SABR for primary kidney cancer involving
the TL-LET response in thermoluminescent LiF and mixed photon fields,” Physica Medica 37, 9–15 (2017).
Li2B4O7: The effect of high temperature TL,” Nuclear 52. C. Hood, L. Duggan, S. Bazley, J. Denham, M.
Instruments and Methods 172, 479–485 (1980). Budzanowski, T. Kron, “LiF:Mg,Cu,P ‘pin worms’:
38. S. K. Mehta and S. Sengupta, “Al2O3 phosphor for ther- Miniature detectors for brachytherapy dosimetry,”
moluminescence dosimetry,” Health Physics 31, 176–177 Radiation Protection Dosimetry 101, 407–410 (2002).
(1976). 53. P. Francescon, S. Cora, C. Cavedon, P. Salchi, S.
39. R. Gaza, E. G. Yukihara, S. W. McKeever, “The response Reccanello, F. Colombo, “Use of a new type of radio-
of thermally and optically stimulated luminescence chromic film, a new parallel-plate micro-chamber,
from Al2O3:C to high-energy heavy charged particles,” MOSFETs, and TLD 800 microcubes in the dosimetry
Radiation Measurements 38, 417–420 (2004). of small beams,” Medical Physics 25, 503–511 (1998).
54. E. Pantelis, A. Moutsatsos, K. Zourari, W. Kilby, C. Dahlgren, T. Lomax, G. Reitz, M. Durante, “Out-of-
Antypas, P. Papagiannis, P. Karaiskos, E. Georgiou, L. field dose studies with an anthropomorphic phantom:
Sakelliou, “On the implementation of a recently pro- Comparison of X-rays and particle therapy treatments,”
posed dosimetric formalism to a robotic radiosurgery Radiotherapy and Oncology 105, 133–138 (2012).
system,” Medical Physics 37, 2369–2379 (2010). 68. A. Molineu, D. S. Followill, P. A. Balter, W. F. Hanson,
55. T. Kron, A. Elliott, P. Metcalfe, “The penumbra of a M. T. Gillin, M. S. Huq, A. Eisbruch, G. S. Ibbott,
6MV x-ray beam as measured by thermoluminescence “Design and implementation of an anthropomorphic
dosimetry and evaluated using an inverse square root quality assurance phantom for intensity-modulated
function,” Medical Physics 20, 1429–1438 (1993). radiation therapy for the Radiation Therapy Oncology
56. T. Kron, A. Elliott, T. Wong, G. Showell, B. Clubb, P. Group,” International Journal of Radiation Oncology,
Metcalfe, “X-ray surface dose measurements using Biology, Physics 63, 577–583 (2005).
TLD-extrapolation,” Medical Physics 20, 703–711 69. M. P. Waligorski, R. Baranczyk, S. Hyodynmaa,
(1993). J. Eskola, J. Lesiak, B. Rozwadowska-Bogusz, A.
57. T. Kron, M. Butson, F. Hunt, J. Denham, “TLD extrapo- Kolodziejczyk, “A TL-based anthropomorphic bench-
lation for skin dose determination in vivo,” Radiotherapy mark for verifying 3-D dose distributions from external
and Oncology 41, 119–123 (1996). electron beams calculated by radiotherapy treatment
58. T. Kron, M. Butson, T. Wong, P. Metcalfe, “Readout of planning systems,” Radiation Protection Dosimetry 120,
thermoluminescence dosimetry chips using a contact 74–77 (2006).
planchet heater,” Australasian Physical and Engineering 70. IAEA, Commissioning and Quality Assurance of
Sciences in Medicine 16, 137–142 (1993). Computerized Planning Systems for Radiation
59. ICRU, “ICRU report 24: Determination of absorbed Treatment of Cancer, IAEA Technical report series N430
dose in a patient irradiated by X or gamma rays in (International Atomic Energy Agency, Vienna, 2004).
radiotherapy procedures,” ICRU Reports (International 71. E. Adolfsson, P. Wesolowska, J. Izewska, E. Lund, A. C.
Commission on Radiological Units and Measurements, Tedgren, “End-to-end audit: Comparison of TLD and
Bethesda, 1976). lithium formate EPR dosimetry,” Radiation Protection
60. P. Olko, L. Currivan, J. W. Van Dijk, M. A. Lopez, C. Dosimetry 186, 119–122 (2019).
Wernli, “Thermoluminescent detectors applied in 72. A. Haworth, L. Wilfert, D. Butler, M. A. Ebert, S.
individual monitoring of radiation workers in Europe Todd, J. Bucci, G. M. Duchesne, D. Joseph, T. Kron,
– A review based on the EURADOS questionnaire,” “Australasian brachytherapy audit: Results of the ‘end-
Radiation Protection Dosimetry 120, 298–302 (2006). to-end’ dosimetry pilot study,” Journal of Medical
61. C. Sneha, S. M. Pradhan, M. M. Adtani, “Study of mini- Imaging and Radiation Oncology 57, 490–498 (2013).
mum detection limit of TLD personnel monitoring 73. T. Kron, G. Donahoo, P. Lonski, G. Wheeler, “A tech-
system in India,” Radiation Protection Dosimetry 141, nique for total skin electron therapy (TSET) of an anes-
168–172 (2010). thetized pediatric patient,” Journal of Applied Clinical
62. N. Hocine, L. Donadille, C. Huet, C. Itie, I. Clairand, Medical Physics 19, 109–116 (2018).
“Personal monitor glass badge: Theoretical dosemeter 74. T. Kron, M. Schneider, A. Murray, H. Mameghan,
response calculated with the Monte Carlo transport “Clinical thermoluminescence dosimetry: How do
code MCNPX,” Radiation Protection Dosimetry 144, expectations and results compare?,” Radiotherapy and
231–233 (2011). Oncology 26, 151–161 (1993).
63. M. Ginjaume, S. Perez, X. Ortega, “Improvements in 75. D. Bollinger and A. Darafsheh, “Clinical considerations
extremity dose assessment for ionising radiation medi- and dosimeters for in vivo dosimetry,” in Radiation
cal applications,” Radiation Protection Dosimetry 125, Therapy Dosimetry: A Practical Handbook, edited by A.
28–32 (2007). Darafsheh (CRC Press, Boca Raton, 2021), Chapter 11,
64. N. Guberina, M. Forsting, A. Ringelstein, “Efficacy of pp. 151–171.
lens protection systems: Dependency on different cra- 76. G. Biasi, N. Hardcastle, A. Rosenfeld, “Scintillation
nial CT scans in the acute stroke setting,” Radiation fiber optic dosimetry,” in Radiation Therapy Dosimetry:
Protection Dosimetry 175, 279–283 (2017). A Practical Handbook, edited by A. Darafsheh (CRC
65. K. G. Veinot, “An estimate of the propagated uncer- Press, Boca Raton, 2021), Chapter 4, pp. 39–59.
tainty for a dosemeter algorithm used for personnel 77. B. Schott, T. Dvergsten, R. Caleb, B. Sun, “EPID-based
monitoring,” Radiation Protection Dosimetry 163, dosimetry,” in Radiation Therapy Dosimetry: A Practical
409–414 (2015). Handbook, edited by A. Darafsheh (CRC Press, Boca
66. J. A. Bencomo, C. Chu, V. M. Tello, S. H. Cho, G. S. Raton, 2021), Chapter 8, pp. 109–121.
Ibbott, “Anthropomorphic breast phantoms for qual- 78. C. L. Deufel, W. S. Culberson, M. J. Rivard, F.
ity assurance and dose verification,” Journal of Applied Mourtada, “Brachytherapy dosimetry,” in Radiation
Clinical Medical Physics 5, 36–49 (2004). Therapy Dosimetry: A Practical Handbook, edited by A.
67. C. La Tessa, T. Berger, R. Kaderka, D. Schardt, C. Darafsheh (CRC Press, Boca Raton, 2021), Chapter 15,
Korner, U. Ramm, J. Licher, N. Matsufuji, C. Vallhagen pp. 231–252.
79. F. Perera, F. Chisela, L. Stitt, J. Engel, V. Venkatesan, “TLD of high-energy radiotherapy photon beams in non-
skin dose measurements and acute and late effects after reference conditions,” Radiotherapy and Oncology 84,
lumpectomy and high-dose-rate brachytherapy only for 67–74 (2007).
early breast cancer,” International Journal of Radiation 91. T. Kron, C. Hamilton, M. Roff, J. Denham, “Dosimetric
Oncology, Biology, Physics 62, 1283–1290 (2005). intercomparison for two Australasian clinical trials
80. Z. Jamalludin, W. L. Jong, G. F. Ho, A. B. Rosenfeld, using an anthropomorphic phantom,” International
N. M. Ung, “In vivo dosimetry using MOSkin detector Journal of Radiation Oncology, Biology, Physics 52, 566–
during Cobalt-60 high-dose-rate (HDR) brachytherapy 579 (2002).
of skin cancer,” Australasian Physical and Engineering 92. T. Kron, A. Haworth, I. Williams, “Dosimetry for
Sciences in Medicine 42, 1099–1107 (2019). audit and clinical trials: Challenges and requirements,”
81. D. Adliene, K. Jakstas, B. G. Urbonavicius, “In vivo TLD Journal of Physics: Conference Series 444, 012014 (2013).
dose measurements in catheter-based high-dose-rate 93. P. Alvarez, S. F. Kry, F. Stingo, D. Followill, “TLD
brachytherapy,” Radiation Protection Dosimetry 165, and OSLD dosimetry systems for remote audits of
477–481 (2015). radiotherapy external beam calibration,” Radiation
82. J. Mason, A. Mamo, B. Al-Qaisieh, A. M. Henry, P. Measurements 106, 412–415 (2017).
Bownes, “Real-time in vivo dosimetry in high dose rate 94. A. S. Pradhan, “Influence on heating rate on the TL
prostate brachytherapy,” Radiotherapy and Oncology response of LiF TLD-700, LiF:Mg,Cu,P and Al2O3:C,”
120, 333–338 (2016). Radiation Protection Dosimetry 58, 205–209 (1995).
83. R. Das, W. Toye, T. Kron, S. Williams, G. Duchesne, 95. K. Kesteloot, A. Dutreix, E. Van Der Schueren, “A model
“Thermoluminescence dosimetry for in-vivo verifica- for calculating the costs of in vivo dosimetry and portal
tion of high dose rate brachytherapy for prostate can- imaging in radiotherapy departments,” Radiotherapy
cer,” Australasian Physical and Engineering Sciences in and Oncology 28, 108–117 (1993).
Medicine 30, 178–184 (2007). 96. Y. Garcier, G. Cordier, C. Pauron, J. Fazileabasse,
84. E. Jaselske, D. Adliene, V. Rudzianskas, B. G. “Intercomparison of passive dosimetry technol-
Urbonavicius, A. Inciura, “In vivo dose verification ogy at EDF facilities in France,” Radiation Protection
method in catheter based high dose rate brachyther- Dosimetry 124, 107–114 (2007).
apy,” Physica Medica 44, 1–10 (2017). 97. S. Hashimoto, Y. Nakajima, N. Kadoya, K. Abe, K.
85. G. Kertzscher, A. Rosenfeld, S. Beddar, K. Tanderup, Karasawa, “Energy dependence of a radiophotolumi-
J. E. Cygler, “In vivo dosimetry: Trends and prospects nescent glass dosimeter for HDR (192) Ir brachytherapy
for brachytherapy,” The British Journal of Radiology 87, source,” Medical Physics 46, 964–972 (2019).
20140206 (2014). 98. H. Vincke, I. Brunner, I., Floret, D. Forkel-Wirth, M.
86. A. Roue, J. L. Venselaar, I. H. Ferreira, A. Bridier, J. Van Fuerstner, S. Mayer, C. Theis, “Response of alanine and
Dam, “Development of a TLD mailed system for remote radio-photo-luminescence dosemeters to mixed high-
dosimetry audit for (192)Ir HDR and PDR sources,” energy radiation fields,” Radiation Protection Dosimetry
Radiotherapy and Oncology 83, 86–93 (2007). 125, 340–344 (2007).
87. D. S. Followill, D. R. Evans, C. Cherry, A. Molineu, G. 99. M. De Saint-Hubert, M. Majer, H. Hrsak, Z. Heinrich,
Fisher, W. F. Hanson, G. S. Ibbott, “Design, development, Z. Knezevic, S. Miljanic, P. Porwol, L. Stolarczyk, F.
and implementation of the radiological physics center’s Vanhavere, R. M. Harrison, “Out-of-field doses in chil-
pelvis and thorax anthropomorphic quality assurance dren treated for large arteriovenous malformations
phantoms,” Medical Physics 34, 2070–2076 (2007). using hypofractionated gamma knife radiosurgery
88. J. Izewska, P. Bera, S. Vatnitsky, “IAEA/WHO TLD and intensity-modulated radiation therapy,” Radiation
postal dose audit service and high precision measure- Protection Dosimetry 181, 100–110 (2018).
ments for radiotherapy level dosimetry. International 100. C. L. Tessa, T. Berger, R. Kaderka, D. Schardt,
Atomic Energy Agency/World Health Organization,” S. Burmeister, J. Labrenz, G. Reitz, M. Durante,
Radiation Protection Dosimetry 101, 387–392 (2002). “Characterization of the secondary neutron field pro-
89. J. Izewska, P. Andreo, S. Vatnitsky, K. R. Shortt, “The duced during treatment of an anthropomorphic phan-
IAEA/WHO TLD postal dose quality audits for radio- tom with x-rays, protons and carbon ions,” Physics in
therapy: A perspective of dosimetry practices at hos- Medicine and Biology 59, 2111–2125 (2014).
pitals in developing countries,” Radiotherapy and 101. R. P. Hugtenburg, A. E. Baker, S. Green, “X-ray synchro-
Oncology 69, 91–97 (2003). tron microdosimetry: Experimental benchmark of a
90. J. Izewska, D. Georg, P. Bera, D. Thwaites, M. Arib, M. general-purpose Monte Carlo code,” Applied Radiation
Saravi, K. Sergieva, K. Li, F. G. Yip, A. K. Mahant, W. and Isotopes 67, 433–435 (2009).
Bulski, “A methodology for TLD postal dosimetry audit
Chapter 7
Optically Stimulated Luminescence

Dosimeters in Clinical Practice
Stephen F. Kry
The University of Texas MD Anderson Cancer Center
Houston, Texas
Jennifer O’Daniel
Duke University School of Medicine
Durham, North Carolina
CONTENTS
7.1 Detector and Reader Theory 97
7.2 Dose Calculation 99
7.3 Precision 103
7.4 Reuse 103
7.5 Practical Applications 104
7.5.1 External Photon Beams 104
7.5.1.1 Surface Dosimetry 104
7.5.1.2 Small Fields 104
7.5.1.3 Out of Field 104
7.5.2 Electrons 105
7.5.3 Protons 105
7.5.4 Brachytherapy 105
7.5.5 Imaging 106
7.5.6 Quality Assurance 106
7.6 Summary 106
References 106
7.1 DETECTOR AND READER THEORY band). These electrons and holes will either recombine
Similar to thermoluminescent dosimetry, optically stim- or become captured in energy traps created by the added
ulated luminescence dosimetry relies upon a crystalline crystalline defects. Several types of these traps exist in
insulator with added impurities. When these crystals are the crystal that have different depths. Charges in shallow
exposed to ionizing radiation, the resultant ionizations energy traps typically escape spontaneously to recom-
raise a certain number of electrons to a higher energy bine within a few minutes. Charges in deeper energy
state (conduction band), leaving an equivalent number traps can remain caught for years. The number of ion-
of vacancies (holes) in the original energy state (valence izations and subsequent trapped electrons and/or holes
97
is approximately proportional to the amount of energy single reading of TLDs via heat exposure. Additionally,
incident on the detector. To extract signal from the opti- using heat requires fairly substantial ancillary equip-
cally stimulated luminescent dosimeter (OSLD), the ment (TLD reader, N2 or Ar gas, annealing equipment)
electrons and holes in deep energy traps are released by versus a simple desktop photo-multiplier tube setup.
application of light (although they can also be released At the time of writing this chapter, the Al2O3:C form
by heat as with thermoluminescent dosimeters; TLDs). remains the single commercially available OSLD.
The released electron can combine with a trapped hole, In practical OSLD, there are three types of traps: shal-
or vice versa, releasing its energy as luminescence. In low traps that are unstable at room temperature (and
reality the process is more complex, and competing make signal unstable for ∼10 minutes after irradiation),
defects may cause changes in OSLD signals based upon medium traps that are released when exposed visible light
factors such as the amount of radiation dose delivered (and are responsible for the dosimetric signal), and deep
and upon the detector’s exposure history [1, 2]. traps that are practically impossible to empty via light
An excellent overview of the long history of optically exposure (and remain filled permanently). The number
stimulated luminescence (OSL) is given by Yukihara and and relative proportion of each kind of trap affects the
McKeever [3]. There are reports in the literature on OSL sensitivity, fading, and reuse of OSLDs. The nature of the
phenomena throughout the 19th century. However, the simulation light (narrow or broad range of wavelengths,
materials studied held the trapped charges at shallow intensity, and duration) also affect how quickly OSLD
depths, requiring only a small amount of energy to be traps are emptied. A higher OSLD luminescence signal
released. Therefore their signals were unstable at room may be obtained by increasing the stimulation intensity
temperature, and unsuitable for radiation dosimetry. or duration, but this will also increase the rate of signal
The development of Al2O3:C in 1990 provided a sensitive loss. Most commercial readers utilize a 1 second stimula-
and stable luminescent dosimeter that reacted both to tion as a good balance between signal strength and signal
heat and to light in the visible spectrum [4]. As the first loss. Readers may also incorporate different light intensi-
OSLD material with sufficient signal stability at room ties to stimulate the detector depending on the dose level.
temperature, multiple academic research groups became For low doses, a higher intensity light is often used so that
involved in its development [5, 6]. The advantage of light more signal is generated, but more signal will be lost for
(as opposed to heat) is that it may be applied in a very subsequent readings.
precise and therefore reproducible fashion, in terms of At the time of writing this chapter and to the best of our
wavelength, intensity, and time. By using relatively short knowledge, there is only one commercially available OSLD
bursts of light exposure, only a small portion of the for clinical use: nanoDot™ by Landauer Inc. (Glenwood, IL).
stored energy is released with each reading. Therefore, As shown in Figure 7.1, the OSLD powder is shaped into
OSLDs may be read multiple times as compared to the a disk of approximately 4 mm diameter and contained
FIGURE 7.1 OSLD both within and without the plastic casing.
Optically Stimulated Luminescence Dosimeters in Clinical Practice ◾ 99
FIGURE 7.2 OSLD reader. (A) Picture of a reader, along with scanner and computer used for processing. (B) Schematic of
main OSLD reader components.
within a 1 × 1 cm2 light-tight plastic casing. During read- intensity and the PMT converts signal to a consistent
out, the OSLD disk is extruded from the plastic casing. number of counts. However, over the period of months
A current commercial OSLD reader, along with this cannot be assumed to be true as readers have been
ancillary equipment, is shown in Figure 7.2. The basic found to not be consistent in terms of how many counts
components of an OSLD reader include a light source are reported for readout of the same dose. An example
for stimulation, a photomultiplier tube (PMT) for detec- of this critical issue is shown in Figure 7.3 for a commer-
tion, and an amplifier to increase signal strength [3, 7–9]. cial OSLD reader. Over a period of 2 years, the figure
The light source could be a laser, light-emitting diode illustrates that the sensitivity of the reader changed by
(LED), or broadband bulb. Typically optical bandpass more than 20%. It is therefore important to monitor the
filters are used both in front of the light source, to select performance of the reader on a regular, periodic basis.
an appropriate stimulation wavelength as separate from
the OSL signal, and in front of the PMT to allow only 7.2 DOSE CALCULATION
the OSL signal to pass. For an example, with an Al2O3:C Dose (D) is measured with an OSLD based on the num-
OSLD the emission light is in the blue spectrum (∼420 ber of counts measured by the detector (M) and a cali-
nm), while the stimulation light is in the green spectrum bration coefficient (N) that relates counts to dose:
(∼525 nm) [10, 11]. The software associated with the
reader will either give the user a simple signal readout in D = M ⋅N (7.1)
counts accumulated or current integrated per unit inter-
val or will provide a way to calibrate the reader such that
the output signal will be converted into a dose reading.
The algorithms for such a conversion are discussed later
in this chapter. The stability of the reader operation is an
essential consideration for the precision of the readings.
Routine quality assurance to verify the consistency (or
correct for variations) of the stimulation light intensity,
duration of stimulation, and light sensitivity of the PMT
is essential. This may be done either through direct mea-
surements, using a calibration OSLD to verify reader
performance on a per-session basis, or by incorporating
a session-specific reader performance correction factor.
OSLD readers are relatively stable over short peri-
ods of time. That is, over the course of days or weeks, FIGURE 7.3 Daily assessment of commercial reader sensitiv-
the stimulation light source produces a consistent light ity (counts per dose) over a 2-year period.
Additional corrections are necessary to this simple given uncertainty). Alternately, a group of detectors with
relationship because the signal is not perfectly propor- similar sensitivities can be considered homogeneous –
tional to the dose deposited, but is also influenced by then the ks,i of each detector need not be tracked, but the
factors such as the amount of dose (linearity correc- uncertainty is larger because the detectors are of course
tion: kL), time since irradiation (fading correction, kF), not truly identical. Instead of a batch calibration, it is
energy (beam quality correction, kQ), and orientation of possible to determine and track each OSLD’s sensitivity
the dosimeter to the beam (angular correction, kθ). The individually, in which case each OSLD would not have
practical equation to determine dose is therefore given an associated ks,i, but each one would then have its own
by the following: calibration coefficient (N).
For the typical case, Mcorr is therefore given for J read-
Dw = Mcorr ⋅ N D ,w ⋅ kL ⋅ kF ⋅ kQ ⋅ kθ (7.2) ings of the ith detector as:
Each term is this equation is described below.

Mcorr is the corrected number of counts from the M corr

= ks ,i 
∑ (Mj
raw , j ⋅ kdj −1 ) 
− Mbkg  (7.3)
 J 
detector. This is the raw number of counts (Mraw) cor-  
rected for depletion of the signal associated with mul-
tiple readouts (kd) and any background signal (Mbkg). In Equation 7.3, ks,i can easily exceed 5% or more, and
Depletion (kd) is the correction applied to account for the is therefore critical to account for. In contrast, depletion
signal lost per reading, and is determined during detec- (kd) for a 1 second read is typically a 0.05% correction
tor commissioning by reading a dosimeter (or small (i.e., the detector loses 0.05% of the trapped signal dur-
sample of dosimeters) numerous times and determining ing a 1 second read out). While the user should know
the average loss per reading. The signal lost per reading and understand the depletion of their reader, this term
is typically 0.05% for the InLight reader operated in the can most often be neglected. Similarly, unless the doses
high-dose mode (i.e., with low stimulation intensity), being studied are very small or the background is high,
so kd is typically 1.0005 [10, 12, 13]. Background counts background signal is typically negligible and can also
(Mbkg) can be determined by keeping a dosimeter in the be neglected. Therefore, Equation 7.3 reduces for most
same environment as the experimental dosimeters, or applications to:
by reading the experimentals right before irradiation.
One additional factor is also typically incorporated into M corr = ks ,i ⋅ Mraw (7.4)
Mcorr: the relative sensitivity of each individual dosime-
ter. Because different OSLD will have different amounts ND,w is the calibration coefficient that relates signal
of actual active material and different trap distributions to dose. This is determined by irradiating OSLD to a
within the crystal, each OSLD will have a slightly dif- known dose and determining the associated counts.
ferent response. It is therefore common to determine These OSLD, used to determine the system calibration,
the typical sensitivity of the system, then apply an addi- are referred to as standards. The calibration coefficient is
tional correction factor for each OSLD relative to this determined from the signal readout with the standards
central/baseline sensitivity. This is often referred to as by simply rearranging Equation 7.1, where the subscript
using a batch calibration approach, and the relative (0) denotes that it is for the reference condition.
sensitivity of a given OSLD compared to the baseline
sensitivity of the system is given by ks,i (e.g., if the ith D0
N D ,w = (7.5)
detector’s sensitivity is lower than the baseline, ks,i cor- M 0,corr
rects its number of counts upwards). How ks,i is actu-
ally implemented in practice can vary. Each dosimeter Equation 7.5 defines the calibration coefficient under a
can be characterized and tracked with its own unique specific condition. As indicated by Equation 7.2, ND,w
ks,i value used during readout; this provides maximum will depend on beam energy, dose level, and the other
accuracy but requires additional effort (this is the model corrections that affect the proportionality of N and D.
for a “screened nanoDot™,” for example, where the ven- The user must therefore remember that ND,w is a func-
dor provides detectors with a known sensitivity within a tion of these parameters and therefore the conditions
with which the standards have been irradiated. ND,w because the user must manually switch to the nonlinear
can only be used if the experimentals (i.e., OSLD with calibration curve when appropriate.
unknown dose irradiated for an experiment) were irra- Regardless of how ND,w is determined, it is essential
diated under identical conditions, or if the differences to consider how often it needs to be evaluated. Because
are accounted for through the application of correction the components of the reader are not perfectly stable,
factors (i.e., as shown in Equation 7.2). if ND,w is determined at a given point in time it will not
While ND,w can be determined for a specific unique necessarily be appropriate at a later time. For maximum
condition (i.e., a specific dose, beam quality, etc.), it is precision, ND,w can be determined for every reading ses-
common to determine ND,w at several dose levels to create sion by irradiating standards for every reading session.
a calibration curve. Such a curve explicitly incorporates Particularly in the case of a calibration curve, it may not
dose level in the relationship between dose and counts. In be practical to directly establish the system response for
this case, the product (ND,w ∙ kL) is determined by the stan- every reading session; in such a case the user should at
dards instead of simply ND,w. An example of three differ- least verify the suitability of the calibration curve. This
ent calibration curves are shown in Figure 7.4, based on can be done by maintaining a “constancy dosimeter”:
the OSL MicroStar reader (Landauer, Inc., Glenwood, IL, a standard that is maintained for a long period of time
USA). The linear calibration curve is appropriate for low and read out before any session (correcting for fading
doses, <300 cGy. There are actually two linear calibration and depletion). The relative response of the system to the
curves for this system, one for the low dose mode and one constancy dosimeter indicates the relative performance
for the high dose mode. Higher levels of stimulation light of the system compared to historical values, and the
are required for dose levels below 10 cGy (“Linear Low calibration coefficient or calibration curve can be scaled
Dose” curve) while lower levels may be used for doses according to this.
between 10 and 300 cGy (“Linear High Dose” curve). The The calibration coefficient is defined with standards
MicroStar reader automatically switches to the appropri- for a specific dose level (unless a calibration curve is
ate level based on a short “pretest” exposure reading. For being used), at a specific time since irradiation, for a
doses above 300 cGy, the relationship between dose and specific beam quality, and with a specific orientation of
counts becomes nonlinear (“Non-Linear Dose” curve). the detector to the radiation field. If the experimentals
The calibration curve (based on a second order poly- are also irradiated under these conditions then ND,w can
nomial) is shown in contrast to a linear curve (fit up to be directly applied. Otherwise, the different irradiation
300 cGy). Particularly for high doses the different in the conditions of the experimentals must be incorporated
calibration curves is apparent. Care is required as well into the determination of dose. This is done through
FIGURE 7.4 Calibration curves (a) Linear curves for the low dose (<10 cGy) and high dose (10–300 cGy) scales (b) Non-linear
calibration curve for doses above 300 cGy. Dashed line depicts extrapolated calibration curve from the high dose linear region
(10–300 cGy), while solid line depicts polynomial calibration curve.
correction factors as described below. These correction correction can be neglected with an acceptable increase
factors do not refer to any absolute reference, but simply in the uncertainty budget. Fading can be determined by
to the reference defined by the standards. For example, irradiating a group of dosimeters at different times and
application of the fading correction does not account reading them out together, or by reading a sample of
for signal loss since irradiation, but rather accounts for OSLD repeatedly over an extended period of time (and
signal loss that is different than that of the standards. correcting for depletion).
If both the standards and experimentals are read out 10 kQ is a correction for beam quality and is necessary
days after irradiation, there is no fading correction nec- because OSLDs are not tissue/water equivalent and
essary because there is no difference in fading between respond differently to beams with different energies.
the experimentals and the calibration condition defined Al2O3:C has different mass attenuation coefficients and
by the standards. stopping powers than tissue or water which affects the
kL is a correction for non-linearity in the response of response of this dosimeter. Compared to the response
the detector to dose, and is necessary because OSLD do at 6 MV, OSLDs respond 2% higher in cobalt, but there
not respond uniformly as a function of dose. Instead, is typically less than 1% difference in response across
as the dose level increases (starting below 1 Gy and different high-energy photon beams. In electron beams,
increasing with increasing dose), the detector over- the response is around 2% lower for all energies as com-
responds. By 10 Gy, the detector over-responds by ∼15% pared to a 6 MV photon beam [9, 10, 15–18]. These val-
as compared to its response at low doses [14]. As such, ues describe the relative response of the detector under
this is an essential correction to be incorporated into the reference conditions for different nominal energies.
dosimeter response in order to ensure reasonable accu- However, beam quality also changes with parameters
racy. This correction factor can be determined during such as field size, depth, presence of wedges, etc., as
commissioning by irradiating a group of OSLD to dif- happens during measurements in non-reference loca-
ferent doses and determining the relationship between tions. Within the treatment field, these factors can make
signal per dose and dose. Then, for a given experimental, up to a 3% difference to the response of the dosimeter.
the signal can be corrected by the over/under-response Because the beam quality correction is very difficult to
at that dose as compared to the dose used to define ND,w determine for an arbitrary condition, measurements
by the standards. done under non-reference conditions are associated
Alternatively, it is common to establish a calibration with increased dosimetric uncertainty.
curve for OSLD detectors. In such a case, kL is implicitly While Al2O3:C shows relatively little variation with
incorporated into the calibration coefficient (as a func- energy within treatment fields of different energies, as the
tion of dose) of the system and need not be separately photon beam energy lowers below the megavoltage range it
determined. shows a dramatic over-response. Outside the treatment field
kF is a correction for loss of detector signal over time from a 6 MV beam, where there is an abundance of scat-
(fading) and is necessary because OSLD spontaneously tered radiation and a softer spectrum, OSLD over-respond
lose signal over time. During the first 10 minutes after by up to 30% (compared to an in-field calibration) [19]. In
irradiation this signal loss is dramatic (more than 40% kV applications such as CT dose measurements, OSLD
of signal is lost) as the shallow traps empty spontane- over-respond by as much as a factor of 3.5 compared to the
ously [10, 15]. It is not possible to achieve reasonable response in an MV beam. This over-response in kV ener-
accuracy if reading out during this time period, there- gies is not only sizeable, it also changes dramatically with
fore a minimum of 10 minutes must be waited between small changes in beam quality. For applications in CT, for
irradiation and readout of OSLD. After these shallow example, kVp, filtration, scan extent, size of the patient,
traps are emptied, the trapped signal in OSLD is highly and location of the measurement all impact the spectrum
stable, and fading is only ∼1% per month. If a long-term enough to induce measurable effects on the response of the
record is being maintained (or, for example, a constancy detector [20–23]. This is a major challenge for OSLD appli-
dosimeter is being used, see above on ND,w) then fad- cation in low energy beams and requires careful attention
ing differences between standards and experimentals and calibration.
may need to be quantified and accounted for. For many k θ is a correction for angular dependence and is nec-
applications the magnitude of fading is small and the essary because OSLD do not respond the same to en-face
irradiation as to edge-on. The disc-shape of the active (controlled) conditions, the detectors are irradiated
volume results in the detector under-responding when under reference conditions in the radiation field and have
irradiated edge-on as compared to en-face. A range of a consistent dosimeter orientation relative to the beam.
values have been reported for the magnitude of this This is the case for beam output audits performed by the
effect in a 6 MV beam, ranging from almost 0% [10] up Imaging and Radiation Oncology Core (IROC) [28, 29],
to 4% [24, 25]; robust and detailed measurements and the International Atomic Energy Agency (IAEA) [30],
Monte Carlo simulations found an angular dependence or the Australian Clinical Dosimetry Services (ACDS)
of 2% [26]. Practically, if the irradiations can be con- [13]. Under these optimal conditions, the uncertainty
trolled such that the standards and experimentals are may be as low as 1.6% or 3.5% (depending on if a high
irradiated in the same orientation, there is no difference precision or high efficiency calibration technique has
in response and no correction is necessary. In clinical been implemented). Most of the time in clinical prac-
situations where many beams impinge upon the detec- tice, measurements are made in “less controlled” condi-
tor it may be impossible to determine the average angle tions. If the dosimeter is used at an arbitrary location in
of incidence. In such a case, additional uncertainty must an arbitrary radiation field then the actual beam qual-
be included in the dose estimate. ity at the measurement location is unknown and will
introduce uncertainty. Similarly, the detector may no
7.3 PRECISION longer be irradiated under a controlled orientation rela-
There are three considerations relevant to the precision tive to the beam, introducing angular dependence into
achievable with OSLD. These same considerations apply the measurement that may be very hard to account for
to TLD as well, and for the same approach/under the (particularly if multiple beams are incident from a range
same considerations, very similar uncertainties exist for of different angles). It is also assumed that a less expe-
both OSLD and TLD. These considerations, and detailed rienced operator is reading the dosimeter. Under such
uncertainty budgets, are presented in the AAPM TG-191 “less controlled” conditions, a high precision approach
report [27]. to OSL dosimetry would be expected to have an uncer-
First is the manner in which the OSLDs are calibrated tainty around 3.9%, while a high efficiency approach to
and read. OSLDs can be handled in a range of manners; OSL dosimetry would be expected to have an uncer-
TG-191 defines a high precision method and a high effi- tainty around 5.2% (at the 1-sigma level).
ciency method. The high precision approach includes The third consideration is the clinical implementation
determining ND,w for every reading session, determin- of their use. Riegel et al. (2017) conducted a large clinical
ing ks,i for every detector, and characterizing and apply- study where therapists placed the dosimeter on patients
ing all correction factors to every reading. It is relatively to measure the dose in vivo from IMRT treatments [31].
time intensive (particularly during commissioning of Combining these clinical factors into a high efficiency
the system when all of the correction factors are deter- and less controlled environment led to a sizeable increase
mined), but provides the highest precision. In contrast, in the overall uncertainty. While the dosimeter had
the high efficiency approach uses a calibration curve to approximately a 5.2% uncertainty inherently, the added
determine ND,w∙ kL which is only spot checked and cor- uncertainty from positioning the dosimeters on patients
rected if necessary based on a constancy dosimeter. The to generate more than 11,000 measurements resulted in
relative sensitivity of each detector is provided by the an average (1-sigma) uncertainty of 10.3%.
manufacturer within a known window of uncertainty Achieving good precision with OSLD is feasible and
(e.g., “screened” nanoDot™). Fading and beam quality routinely done. However, there is a direct relationship
(across MV photons and electrons) are both neglected. between the effort committed, the control of the irra-
This approach is much faster to implement, but has a diation conditions, and the uncertainty in the final dose
higher associated uncertainty. At the 1-sigma level, the estimate.
best reasonable achievable precision with the high preci-
sion approach is 1.6%; with the high efficiency approach 7.4 REUSE
it is 3.5% (TG-191). OSLD can be reused by removing trapped electrons
The second consideration relevant to the uncertainty is and holes from the crystal lattice. This process is called
the control of the irradiation conditions. Under optimal bleaching and is comparable to annealing of TLD except
it is based on light. Many different light sources can and Phantom or in vivo measurements are typically per-
have been used to bleach Al2O3:C detectors, including formed by placing the OSLD on the surface. The effec-
fluorescent room lights, LEDs, and tungsten halogen tive point of measurement of a nanoDot™ OSLDs is
lamps [9, 10, 14, 15]. The time required to bleach the 0.8 mm below the surface of the plastic case [34]. This
dosimeters depends on the light source, but 12 hours places the OSLD in a high dose gradient region where
is generally ample time to reduce any residual signal to the uncertainty in the dose is high (and if the treat-
background levels. For any individual setup, the residual ment planning system calculation is being assessed,
signal after bleaching can be readily checked to ensure it the accuracy of the treatment planning system is also
has indeed reached background levels. limited). Therefore, to improve dosimetric accuracy
As described earlier in this chapter, OSLD have traps and agreement with the treatment planning system
of several different depths. Shallow traps spontaneously calculation, bolus or buildup caps can be placed over
empty shortly after irradiation, dosimetric traps are the OSLD to increase its depth of measurement beyond
highly stable at room temperature but are emptied dur- 5 mm. When using bolus, it is important that its lateral
ing the readout process, and deep traps that are so stable extent be at least 5 × 5 cm2 to generate lateral charged
that they are not emptied during readout. Similarly, particle equilibrium [31].
bleaching effectively empties dosimetric traps but does
not empty deep traps [32]. Therefore, after bleaching, the 7.5.1.1 Surface Dosimetry
OSLD does not return to its original state; rather, the The OSLD may also be used to verify surface/skin dose
deep traps remain filled. This is important because these from photon beams. These measurements are inherently
deep traps are competitors in the OSL process. The state less accurate. The OSLD measurement depth (0.8 mm) will
of occupancy of deep traps affects the sensitivity of the overestimate dose to superficial skin layers (0.07 mm, ICRP
detector by changing the relative trapping and recom- 26, 1977 [35]). Users should also expect worse agreement
bination efficiency. In practice, the accumulated dose with their treatment planning system calculations, given
in OSLDs affects the detector’s sensitivity as well as its the difficulty in accurately modeling the high dose gradi-
supralinearity (kL) [10–12, 33]. As an added complica- ent region near the surface.
tion, the changes in sensitivity are affected by bleach-
ing time, bleaching light spectrum, and accumulated 7.5.1.2 Small Fields
dose [14]. The complex and poorly understood results The diameter of the nanoDot™ disc is approximately
from the literature regarding changes in sensitivity 4 mm, making it suitable for measurements of fields
and supralinearity support using these dosimeters to a down to 2 cm × 2 cm. For smaller fields, use of
maximum accumulated dose of ∼10 Gy. Until an accu- nanoDot™ OSLD is not recommended both due to vol-
mulated dose of 10 Gy is reached, these dosimeters can ume-averaging effects due to its size as well as air gaps
be easily bleached and reused without changes to their present within the dosimeter [36]. When performing
behavior. measurements of small fields, be certain to align the
Reusing OSLD has the advantage of repeated use off-centered active area of the nanoDot™ OSLD with
of the detectors and therefore reduced cost. The big- the field’s central axis rather than centering the dosim-
gest challenge of reusing OSLD is that, because there is eter’s casing.
a finite 10 Gy lifespan, the accumulated dose for each
dosimeter must be tracked which may be logistically 7.5.1.3 Out of Field
challenging. Occasionally dose measurements will be needed out-
side the main field, such as to determine dose to a pace-
7.5 PRACTICAL APPLICATIONS maker. These measurements are complicated by a few
7.5.1 External Photon Beams issues. First, nanoDot™ OSLDs that have been calibrated
In the setting of external beam radiotherapy, OSLDs in-field will over-respond by 10–25% due to the lower
may be used to independently verify beam calibration, energy spectrum outside the field [19, 37]. This increases
in anthropomorphic phantoms either on the surface or the uncertainty of the dose measurement, although
at depth to verify dose calculations, or on patients to this can be offset by estimating appropriate correc-
measure in vivo doses. tion factors from the literature [19]. Second, outside the
treatment field the dose is elevated at the surface and fall-off region of the Bragg peak has substantial mea-
decreases down to dmax. Unless a surface dose measurement uncertainty due to the dramatic LET varia-
surement is desired, a bolus of ∼dmax thickness should tion, high dose gradient, and the fact that the particle
be placed over the dosimeter to avoid measuring this range becomes less than the detector thickness [43].
elevated surface dose [38–40]. Third, if the OSLD is OSLDs also have less non-linearity with increasing
out of the primary field but within the imaging field, doses when using protons as compared to x-rays, such
the OSLD may over-respond to the imaging dose (see that the same linearity correction or calibration curve
Section 7.5.5). This could result in a substantial over- established for photon or electron beams cannot be
estimation of out of field dose. Care should be employed used. Similar to electrons, surface dose measurements
in this scenario, for example by capturing the imaging are reasonably accurate due to shallower dose gradients
dose (e.g., through independent measurement or esti- than that of photon beams. Small field measurements
mation) separately from the therapeutic dose (e.g., by have been shown to be accurate down to 5 cm × 5 cm. At
placing another OSLD on the patient after the imaging the time of writing, only one report for a 2 cm × 2 cm field
is performed but before the treatment is delivered). size was available, which showed the need for a partial
irradiation correction of 2% [44]. Users should employ
7.5.2 Electrons caution when measuring proton fields smaller than 5 cm
The use of OSLD with electrons is similar to that with × 5 cm, and should not use OSLDs below 2 cm × 2 cm
photons. The energy dependence varies by less than 2% field sizes. The angular dependence for protons has been
due to small variations in collisional stopping power shown to be small (≤ 0.5%) within the spread-out-Bragg-
from 1 MeV to 20 MeV [10, 15]. In a high efficiency set- peak [24].
ting, it is acceptable to rely on reader calibrations from
a different modality or energy, but using a calibration 7.5.4 Brachytherapy
specific to the measurement’s modality and energy Given the small size of OSLDs, they are a good brachy-
improves the measurement accuracy. Note that combin- therapy dosimetry option. However, there are several
ing modalities (photon and electron) will increase the issues to be dealt with in order to optimize accuracy.
measurement uncertainty [41]. Surface dose measure- First, positioning accuracy is crucial, as it is for all
ments of electron beams have better agreement with dosimeters in a brachytherapy setting, due to the sharp
treatment planning dose calculations than in photon dose gradients (TG-138 [45]). Second, correction for the
beams due to the lower dose gradient. In particular total energy dependence at the lower brachytherapy energies
skin electron therapy has a very small build up region is more complex than with external beam therapy. The
due to additional scatter, and so OSLD measurements calibration could be done either at the brachytherapy
without bolus are appropriate. The angular depen- source energy, with increased uncertainty in absolute
dence of OSLD with low energy electrons has not been dose, or at an MV energy, with increased uncertainty
evaluated, which introduces additional uncertainty for in kQ. Both approaches are feasible [46, 47] (Casey et al.
oblique measurements. 2013). It is important to note that kQ can vary substan-
tially with source-to-detector distance due to beam
7.5.3 Protons hardening. Sharma and Jursinic [46] demonstrated a
Unlike with electrons, the behavior of OSLD changes change of ∼10% moving from 2 cm to 10 cm of solid
with protons. First, the greater linear energy transfer water with 192Ir. The kQ will also vary based on the source
(LET) of these particles leads to a higher ionization den- encapsulation. Casey et al. (2013) showed differences of
sity along the particle’s track. This creates a saturation 2.6% between Nucletron and Varian 192Ir sources. Third,
response in the OSLD resulting in decreased signal per the angular dependence has been shown to be ∼4% with
unit dose [31, 42]. This effect is approximately 4% for 192Ir [46]. Finally caution should be used when mea-
proton beams relative to photon beams, but is approxi- suring low energy betas because they may be partially
mately constant for different proton energies. This effect absorbed by the dosimeter (depending on the OSLD
is expected to be greater for carbon beams [15] and may thickness and beta energy), resulting in an underesti-
be dependent on the carbon energy and presence of sec- mation of dose. For additional discussion, the reader is
ondary particles. For any charged particle, the distal referred to the 2009 AAPM summer school [48].
7.5.5 Imaging standard doses) as well as the dosimeter’s functionality

In addition to therapeutic applications, OSLDs may (calibration curve or NDW + kL determination, determine
also be used to measure dose from imaging proce- magnitude of kF, kd, and kQ and decide whether/how to
dures. Some recent applications include diagnostic CT correct for them). A simple constancy check based on
[49], cone-beam CT [50, 51], and mammography [52]. an OSLD of a known dose should be done for every
With appropriate care taken to characterize the OSLD’s readout session. This OSLD may be reused for many ses-
response in a kV environment, accurate low dose mea- sions if the physicist accounts for depletion and fading.
surements may be performed. As discussed in the Additional QA should be performed on an annual basis
brachytherapy section, the OSLD may be calibrated for and compared to commissioning to determine if there is
low energy kV either via an MV source (with associ- any drift in the reader functionality.
ated large kQ) or at the low energy itself (with associated
uncertainties in absolute dose) [53]. Determining the 7.6 SUMMARY
appropriate energy spectrum for the associated imaging Similar to TLD, OSLD may be used to verify dose both
technique is also challenging, and introduces additional for equipment quality assurance measurements and
sources of uncertainty. Variations in kV from 80 kVp to for in vivo dosimetry. Their accuracy and precision is
140 kVp lead to changes in kQ of ±10–15% [20]. The pres- equivalent to that of TLD. OSLD have a clear advantage
ence or absence of a bow-tie filter, amount of material in in terms of having a faster and non-destructive readout
the beam path, and scan extent also affect kQ by 4 –10% process. If the dosimeter will be reused, bleaching of
[20, 50, 52, 54]. Due to the large variation in response OSLD is also a simpler process than annealing of TLD.
between kV and MV energies, a single OSLD cannot While light sensitivity is a conceptual disadvantage of
measure a cumulative imaging dose plus therapeutic OSLD, the light-tight casing of OSLD ensures this is
dose such as in the setting of image-guided radiation not a practical problem. Reuse of OSLD is more com-
therapy. Separate dosimeters with separate correction plicated than for TLD because of the finite lifespan of
factors would be required to appropriately measure each OSLD (10 Gy) before properties of the detector change.
component. In a clinical setting with the high efficiency Managing this requires tracking the dose history of
approach, the relatively small imaging contribution each detector.
to an in-field therapeutic OSLD measurement may be Ultimately, OSLD dosimeters provide point dose
ignored. If improved accuracy is desired, using two sets information and are broadly applicable to many clini-
of OSLD (one in place only during imaging, the other cal situations. The accuracy and precision of OSLD is
only during treatment) is necessary. affected by the calibration and handling practices of
Angular dependency is also a greater issue with the these detectors, with a direct trade-off between the
OSLD under-responding by 30–40% in radiography and accuracy and precision achieved and the effort involved
60% in mammography [52] for an edge-on versus en-face in characterizing the detectors, applying appropriate
irradiation. Interestingly, the angular dependency with correction factors, and performing appropriate quality
the rotating source in a CT scanner is a mere ∼5% in assurance testing. However, it is generally straightfor-
comparison [20, 54]. In mammography, self-attenuation ward to achieve good accuracy and precision with these
of the beam becomes a concern due to the very low detectors.
energy [55, 56]. Using an en-face position minimizes the
attenuation as much as possible. Finally, if a low dose REFERENCES
(≤ 5 mGy) is expected, subtracting the background sig- 1. R. Chen and S. W. S. McKeever, Theory of
nal will improve accuracy [20]. Thermoluminescence and Related Phenomena (World
Scientific, Singapore; River Edge, NJ, 1997).
7.5.6 Quality Assurance 2. R. Chen and V. Pagonis, Thermally and Optically
Quality assurance recommendations are discussed in Stimulated Luminescence: A Simulation Approach
detail in TG-191 [27]. In brief, the physicist needs to (Wiley, The Atrium, Southern Gate, Chichester, West
Sussex, UK; Hoboken, NJ, 2011).
verify the reader’s functionality (photodetector with 3. E. G. Yukihara and S. W. S. McKeever, Optically
dark current and consistent source reading, readout Stimulated Luminescence: Fundamentals and
precision, plus an end-to-end test with a range of known Applications (Wiley, Chichester, West Sussex, 2011).
4. M. S. Akselrod and V. S. Kortov, “Thermoluminescent 18. A. Viamonte, L. A. da Rosa, L. A. Buckley, A. Cherpak,

and exoemission properties of new high-sensitivity TLD J. E. Cygler, “Radiotherapy dosimetry using a commer-
α-Al2O3:C crystals,” Radiation Protection Dosimetry 33, cial OSL system,” Medical Physics 35, 1261–1266 (2008).
123–126 (1990). 19. S. B. Scarboro, D. S. Followill, et al., “Energy response of
5. B. G. Markey, L. E. Colyott, S. W. S. Mckeever, “Time- optically stimulated luminescent dosimeters for non-ref-
resolved optically stimulated luminescence from erence measurement locations in a 6 MV photon beam,”
α-Al2O3:C,” Radiation Measurements 24, 457–463 (1995). Physics in Medicine and Biology 57(9), 2505–2515 (2012).
6. S. D. Miller, Composite material dosimeters. U.S. Patent, 20. S. B. Scarboro, D. Cody, et al., “Characterization of
5,567,948, issued on 22 October 1996 (1996). the nanoDot OSLD dosimeter in CT,” Medical Physics
7. L. Bøtter-Jensen, S. W. S. McKeever, et al., Optically 42(4), 1797–1807 (2015).
Stimulated Luminescence Dosimetry (Elsevier, Amsterdam; 21. P. Mobit, E. Agyingi, G. Sandison, “Comparison of the
Boston; London, 2003). energy-response factor of LiF and Al2O3 in radiotherapy
8. S. W. McKeever and M. Moscovitch, “On the advantages beams,” Radiation Protection Dosimetry 119, 497–499
and disadvantages of optically stimulated lumines- (2006).
cence dosimetry and thermoluminescence dosimetry,” 22 C. Ruan, E. G. Yukihara, W. J. Clouse, P. B. Gasparian,
Radiation Protection Dosimetry 104(3), 263–270 (2003). S. Ahmad, “Determination of multislice computed
9. E. G. Yukihara, G. Mardirossian, et al., “Evaluation tomography dose index (CTDI) using optically stimu-
of Al2O3:C optically stimulated luminescence (OSL) lated luminescence technology,” Medical Physics 37,
dosimeters for passive dosimetry of high-energy photon 3560–3568 (2010).
and electron beams in radiotherapy,” Med Phys 35(1), 23. M. C. Aznar, J. Medin, B. Hemdal, A. Thilander Klang, L.
260–269 (2008). Botter-Jensen, S. Mattsson, “A Monte Carlo study of the
10. P. A. Jursinic, “Characterization of optically stimulated energy dependence of Al2O3:C crystals for real-time in
luminescent dosimeters, OSLDs, for clinical dosimetric vivo dosimetry in mammography,” Radiation Protection
measurements,” Medical Physics 34(12), 4594–4604 (2007). Dosimetry 114, 444–449 (2005).
11. E. G. Yukihara and S. W. McKeever, “Optically stimu- 24. J. R. Kerns, S. F. Kry, et al., “Angular dependence of the
lated luminescence (OSL) dosimetry in medicine,” nanoDot OSL dosimeter,” Medical Physics 38(7), 3955–
Physics in Medicine and Biology 53(20), R351–379 (2008). 3962 (2011).
12. I. Mrcela, T. Bokulic, J. Izewska, M. Budanec, A. Frobe, 25. D. W. Kim, W. K. Chung, D. O. Shin, M. Yoon, U. J.
Z. Kusic, “Optically stimulated luminescence in vivo Hwang, J. E. Rah, H. Jeong, S. Y. Lee, D. Shin, S. B.
dosimetry for radiotherapy: Physical characterization Lee, S. Y. Park, “Dose response of commercially avail-
and clinical measurements in (60)Co beams,” Physics in able optically stimulated luminescent detector, Al2O3:C
Medicine and Biology 56, 6065–6082 (2011). for megavoltage photons and electrons,” Radiation
13. L. Dunn, J. Lye, J. Kenny, J. Lehmann, I. Williams, T. Protection Dosimetry 149, 101–108 (2012).
Kron, “Commissioning of optically stimulated lumi- 26. J. Lehmann, L. Dunn, et al., “Angular dependence of
nescence dosimeters for use in radiotherapy,” Radiation the response of the nanoDot OSLD system for measure-
Measurements 51–52, 31–39 (2013). ments at depth in clinical megavoltage beams,” Medical
14. A. A. Omotayo, J. E. Cygler, G. O. Sawakuchi, “The Physics 41(6), 061712 (2014).
effect of different bleaching wavelengths on the sensi- 27. S. F. Kry, P. Alvarez, J. E. Cygler, L. A. DeWerd, R. M.
tivity of Al(2)O(3):C optically stimulated luminescence Howell, S. Meeks, J. O’Daniel, C. Reft, G. Sawakuchi,
detectors (OSLDs) exposed to 6 MV photon beams,” E. G. Yukihara, D. Mihailidis, “AAPM TG 191: Clinical
Medical Physics 39, 5457–5468 (2012). use of luminescent dosimeters: TLDs and OSLDs,”
15. C. S. Reft, “The energy dependence and dose response of Medical Physics 47(2), e19–e51 (2020).
a commercial optically stimulated luminescent detector 28. T. H. Kirby, W. F. Hanson, D. A. Johnston, “Uncertainty
for kilovoltage photon, megavoltage photon, and elec- analysis of absorbed dose calculations from thermolu-
tron, proton, and carbon beams,” Medical Physics 36(5), minescence dosimeters,” Medical Physics 19, 1427–1433
1690–1699 (2009). [Erratum: Medical Physics 39(9), (1992).
5788 (2012)]. 29. P. Alvarez, S. F. Kry, F. Stingo, D. Followill, “TLD
16. M. C. Aznar, C. E. Andersen, L. Botter-Jensen, S. A. and OSLD dosimetry systems for remote audits of
Back, S. Mattsson, F. Kjaer-Kristoffersen, J. Medin, radiotherapy external beam calibration,” Radiation
“Real-time optical-fibre luminescence dosimetry for Measurements 106, 412–415 (2017).
radiotherapy: Physical characteristics and applications 30. J. Izewska, P. Andreo, S. Vatnitsky, K. R. Shortt, “The
in photon beams,” Physics in Medicine and Biology 49, IAEA/WHO TLD postal dose quality audits for radio-
1655–1669 (2004). therapy: A perspective of dosimetry practices at hospi-
17. V. Schembri and B. J. M. Heijmen, “Optically stimulated tals in developing countries,” Radiotherapy & Oncology
luminescence (OSL) of carbon-doped aluminum oxide 69, 91–97 (2003).
(Al2O3:C) for film dosimetry in radiotherapy,” Medical 31. A. C. Riegel, Y. Chen, A. Kapur, L. Apicello, A. Kuruvilla,
Physics 34, 2113–2118 (2007). A. J. Rea, A. Jamshidi, L. Potters, “In vivo dosimetry
with optically stimulated luminescent dosimeters for sources: Report of AAPM Task Group No. 138 and
conformal and intensity-modulated radiation therapy: GEC-ESTRO,” Medical Physics 38(2), 782–801 (2011).
A 2-year multicenter cohort study,” Practical Radiation 46. R. Sharma and P. A. Jursinic, “In vivo measurements for
Oncology 7, E135–E144 (2017). high dose rate brachytherapy with optically stimulated
32. E. G. Yukihara, R. Gaza, et al., “Optically stimulated luminescent dosimeters,” Medical Physics 40(7), 071730
luminescence and thermoluminescence efficiencies (2013).
for high-energy heavy charged particle irradiation in 47. K. E. Casey, P. Alvarez, S. F. Kry, R. M. Howell, A.
Al2O3:C,” Radiation Measurements 38(1), 59–70 (2004). Lawyer, D. Followill, “Development and implementa-
33. P. A. Jursinic, “Changes in optically stimulated lumines- tion of a remote audit tool for high dose rate (HDR)
cent dosimeter (OSLD) dosimetric characteristics with Ir-192 brachytherapy using optically stimulated lumi-
accumulated dose,” Medical Physics 37, 132–140 (2010) nescence dosimetry,” Medical Physics 40, 112102 (2013).
34. A. H. Zhuang and A. J. Olch, “Validation of OSLD and a 48. J. E. Cygler and E. G. Yukihara (eds.), Optically stimu-
treatment planning system for surface dose determina- lated luminescence (OSL) dosimetry in radiotherapy.
tion in IMRT treatments,” Medical Physics 41(8), 081720 Clinical Dosimetry Measurements in Radiotherapy
(2014). (AAPM 2009 Summer School) (Medical Physics
35. International Commission on Radiological Protection Publishing, 2009).
(ICRP). Recommendations of the International 49. C. C. Brunner, S. H. Stern, et al., “CT head-scan dosim-
Commission on Radiological Protection. ICRP Publication etry in an anthropomorphic phantom and associated
26 (Pergamon Press, Oxford, UK, 1977). measurement of ACR accreditation-phantom imaging
36. P. H. Charles, S. B. Crowe, T. Kairn, J. Kenny, J. metrics under clinically representative scan condi-
Lehmann, J. Lye, L. Dunn, B. Hill, R. T. Knifght, C. tions,” Medical Physics 40(8), 081917 (2013).
M. Langton, J. V. Trapp, “The effect of very small air 50. G. X. Ding and A. W. Malcolm, “An optically stimu-
gaps on small field dosimetry,” Physics in Medicine and lated luminescence dosimeter for measuring patient
Biology 57, 6947–6960 (2012). exposure from imaging guidance procedures,” Physics
37. C. R. Edwards and P. J. Mountford, “Near surface pho- in Medicine and Biology 58(17), 5885–5897 (2013).
ton energy spectra outside a 6 MV field edge,” Physics in 51. T. Giaddui, Y. Cui, et al., “Comparative dose evaluations
Medicine and Biology 49(18), N293–N301 (2004). between XVI and OBI cone beam CT systems using
38. G. Starkschall, F. J. Stgeorge, et al., “Surface dose for Gafchromic XRQA2 film and nanoDot optical stimu-
megavoltage photon beams outside the treatment field,” lated luminescence dosimeters,” Medical Physics 40(6),
Medical Physics 10(6), 906–910 (1983). 062102 (2013).
39. S. F. Kry, U. Titt, et al., “A Monte Carlo model for cal- 52. R. M. Al-Senan and M. R. Hatab, “Characteristics of an
culating out-of-field dose from a Varian 6 MV beam,” OSLD in the diagnostic energy range,” Medical Physics
Medical Physics 33(11), 4405–4413 (2006). 38(7), 4396–4405 (2011).
40. S. F. Kry, U. Titt, et al., “A Monte Carlo model for out- 53. S. Scarboro, D. Cody, F. Stingo, P. Alvarez, D. Followill, L.
of-field dose calculation from high-energy photon ther- Court, D. Zhang, M. McNitt-Gray, S. Kry, “Calibration
apy.” Medical Physics 34(9), 3489–3499 (2007). strategies for the use of the nanoDot OLSD in CT appli-
41. M. J. Lawless, S. Junell, C. Hammer, L. A. DeWerd, cations,” The Journal of Applied Clinical Medical Physics
“Response of TLD-100 in mixed fields of photons and 20(1), 331–339 (2019).
electrons,” Medical Physics 40 102103 (2013). 54. L. Lavoie, M. Ghita, et al., “Characterization of a com-
42. G. O. Sawakuchi and E. G. Yukihara, “Analytical mod- mercially-available, optically-stimulated luminescent
eling of relative luminescence efficiency of Al2O3:C dosimetry system for use in computed tomography,”
optically stimulated luminescence detectors exposed Health Physics 101(3), 299–310 (2011).
to high-energy heavy charged particles,” Physics in 55. A. Konnai, N. Nariyama, et al., “Energy response of LiF
Medicine and Biology 57(2), 437–454 (2012). and Mg2SiO4 TLDs to 10–150 keV monoenergetic pho-
43. J. M. Edmund, C. E. Andersen, et al., “A track struc- tons,” Radiation Protection Dosimetry 115(1–4), 334–
ture model of optically stimulated luminescence from 336 (2005).
Al2O3:C irradiated with 10–60 MeV protons,” Nuclear 56. A. A. Nunn, S. D. Davis, et al., “LiF:Mg,Ti TLD response
Instruments & Methods in Physics Research Section as a function of photon energy for moderately filtered
B-Beam Interactions with Materials and Atoms 262(2), x-ray spectra in the range of 20–250 kVp relative to
261–275 (2007). 60Co,” Medical Physics 35(5), 1859–1869 (2008).
44. J. R. Kerns, S. F. Kry, et al., “Characteristics of optically 57. M. S. Akselrod and S. W. S. McKeever, “A radiation
stimulated luminescence dosimeters in the spread-out dosimetry method using pulsed optically stimulated
Bragg peak region of clinical proton beams,” Medical luminescence,” Radiation Protection Dosimetry 81(3),
Physics 39(4), 1854–1863 (2012). 167–175 (1999).
45. L. A. DeWerd, G. S. Ibbott, et al., “A dosimetric uncer-
tainty analysis for photon-emitting brachytherapy
Chapter 8
EPID-Based Dosimetry
Brayden Schott, Thomas Dvergsten, Raman Caleb,
and Baozhou Sun
Washington University in St. Louis
St. Louis, Missouri
CONTENTS
8.1 Introduction 109
8.2 Technology and Characteristics of Epid 109
8.3 Use of EPID for Linac QA 111
8.3.1 Imager QA 112
8.3.2 Machine QA 112
8.3.3 MLC Positioning 112
8.3.4 Jaw Positioning 113
8.3.5 Isocentricity 114
8.3.6 Variable Axis Angles 115
8.3.7 Output 115
8.3.8 Flatness/Symmetry 116
8.3.9 Wedge 116
8.4 Use of Epid for Patient QA 117
References 119
8.1 INTRODUCTION provide a review of EPIDs dosimetric characteristics

Electronic portal imaging devices (EPIDs) have been and application to both patient-specific and linac QA.
developed for verification of patient positioning prior
to or during the treatment delivery. Since the introduc- 8.2 TECHNOLOGY AND CHARACTERISTICS
tion of the amorphous silicon (aSi)-based EPIDs, almost OF EPID
every linac is equipped with an EPID panel. Due to its Beginning in the 1980s, several types of EPID panels
favorable dosimetric characteristics such as high resolu- emerged as alternatives to radiochromic film for mega-
tion, fast imaging acquisition, and digital format, EPIDs voltage imaging applications. In this chapter we will
have been used for radiation therapy dose measurements focus on modern aSi-based EPID technology, which
for more than two decades. It has broad dosimetry appli- is based on thin-film semiconductor technology. The
cations for both patient-specific intensity-modulated currently available commercial EPID panel consists of:
radiation therapy/volumetric-modulated arc therapy (1) a thin (typically 1 mm for Varian and Elekta EPID
(IMRT/VMAT) quality assurance (QA) including pre- panels) copper plate which acts as a buildup material
treatment and in vivo patient treatment verification and and absorbs scattered low energy radiation; (2) a scin-
linac machine QA. It can be used to verify almost every tillating phosphor screen made of Terbium-activated
aspect of machine performance. In this chapter we will Gadolinium Oxysulfide (Gd2O2S:Tb) to convert the
109
FIGURE 8.1 Schematic illustration of a flat panel imaging array. The external electronics controls the conductivity of the pixel
FETs and pre-amplify the signal.
incident radiation into optical photons, (3) an array of an acquisition and saves this sum as a single image. The
photodiodes and field-effect transistors (FETs) which raw EPID image can be processed by using a dark field
detect radiation both directly and indirectly; and (4) and a flood field calibration:
read out electronics to translate the charge readings
into an image. These parts of the EPID are enclosed in a I raw − I dark
I= (8.1)
protective cover. The buildup at the depth of the active I flood − I dark
imaging layer of the EPID is equivalent to 8 mm of water
[1]. Figure 8.1 shows the main components of Varian where Iraw, Idark, and Iflood are the raw, dark field, and flood
aS1000 EPID panel. field images, respectively. Both the dark and flood field
The image information acquired by the aS1000 is sent are acquired during EPID calibration. The dark field is
as an analog signal to the digitization unit where it is acquired by imaging without any linac output, giving
transformed into a digital signal [2]. The newest EPID the electronic noise level of the detector. The flood field
panel, the aS1200, is an improved version of aS1000 and is the image of an open field covering as much of the
incorporates a backscatter shield beneath the active EPID area as possible without irradiating the readout
area of the EPID [3]. The shielding reduces the amount electronics.
of backscatter contamination from the support arm Images given from the EPID is stored in DICOM
and provides a uniform backscatter component to all format, with 14, or 16 bit pixel values, where lower val-
images. In addition, the readout electronics have been ues correspond to higher intensities. These images have
redesigned and integrated to minimize the number of been corrected with dark field and flood field. To get
components. The thickness of the aSi-photodiode is dosimetric values, all pixel values are subtracted by the
reduced by 50%, which reduces image lag giving rise to maximum pixel value, and then multiplied by a calibra-
improved signal linearity. tion factor and the number of frames in the acquisition,
There are two acquisition modes of interest for dosi- to give so-called calibrated units (CU). The calibra-
metric applications: cine mode and integrated mode. tion factor is found through the definition of the CU; a
In cine mode, all images are recorded in a fixed time 10 × 10 cm2 open field with 100 MU at source-to-detec-
interval and formed in a series of movie images. In con- tor distance (SDD) of 100 cm should give a response of
trast, integrated mode creates a sum of all frames from 1 CU on the central axis (CAX).
EPID-Based Dosimetry ◾ 111
A Si-EPID demonstrates excellent dosimetric charac-

teristics including high resolution, fast readout, linearity,
reproducibility, high spatial resolution, and no dead time.
Varian’s aS1000 has the ability to acquire 10 frames/sec-
ond. The aS1200 can read with the maximum frame rate
up to 25 frames per second. The current aSi-EPID pan-
els provide a high spatial resolution compared with most
other available dosimeters, with the pixel size smaller
than (0.4 × 0.4 mm2). The most recent aS1200 utilizes an
active matrix of 43 × 43 cm2 consisting of 1280 × 1280
pixels, each one with size of 0.34 × 0.34 mm2.
Linearity of EPID response to dose and dose rate have
been well studied. EPID demonstrates an excellent lin-
earity of integrated dose and dose rate. Figure 8.2 shows
the linearity of EPID with dose rate and integrated dose
FIGURE 8.3 Central portal doses as a function of dose rate
[4]. The new aS1200 detector showed a significant dosi-
for 100 MU and 20 × 20 cm2 field size [6].
metric improvement when compared with the previous
aS1000 [5]. As shown in Figure 8.3, the aSi-1200 shows
an excellent linearity up to 2400 MU/min for 10 MV The aSi flat-panel EPIDs have been shown to have
flattening filter free (FFF) beam on Varian linac [6]. excellent image stability over time. The long-term
reproducibility of the Varian EPID system averaged
over all pixels has been shown to be <0.6% over a
3-year period for three EPIDs [7]. Louwe et al. found
variations of less than 1% (1 SD) for four aSi EPIDs
(Elekta iView-GT) over periods of up to 23 months;
this was improved by the application of a dynamic
dark field, i.e., a remeasurement of the dark field just
prior to image acquisition. With the dynamic dark
field correction, the reproducibility was improved to
within 0.5% (1 SD) [8].
8.3 USE OF EPID FOR LINAC QA

While the original purpose of EPID use was orientated
toward patient QA, the device’s utility has recently
been expanded to include machine QA. As noted
above, an EPID is an integrated imaging system that
acquires MV images with high mechanical and dosi-
metric precision. The seamless integration of these
FIGURE 8.2 Linearity of the EPID with dose rate and inte- imaging systems not only greatly reduces the setup and
grated dose. The dose rate to the EPID was varied by varying processing time that is required for routine film-based
the source to detector distance. Ion-chamber readings were or phantom-based QA tests, but also reduces the over-
recorded at each EPID position in a mini-phantom to deter- all cost of QA practices by minimizing the need for
mine the relative dose rate. The relative dose rate is expressed
third-party dosimetric instrumentation. Additionally,
in MU/min assuming a nominal output of 400 MU/min at
100 cm from the source. A measurement of integrated dose
EPID-based QA grants physicists the freedom to ana-
was obtained by multiplication of the image pixel values by lyze their acquired images by choosing from a vari-
the number of frames averaged, with and without correction ety of well-established commercial analysis software,
for dead time in image acquisition. These were also compared or by making use of verified in-house software tools.
to ion-chamber readings [4]. What follows is first a brief overview of common QA
practices in place to ensure device integrity, followed 8.3.2 Machine QA

by a discussion on EPID-based machine QA proce- With the advent of EPID-based machine QA, various
dures common in clinical use. AAPM Task Groups have been published to prescribe
protocol and tolerance suggestions that encompass
8.3.1 Imager QA both dosimetric and mechanical checks. AAPM
In order for an EPID to be a dependable tool for machine TG-142 [9] is the most comprehensive and will be the
QA, the functional integrity and stability of the imager focus of what follows. In terms of mechanical checks,
itself must first be ensured. Due to the numerous inte- standard protocols often include multileaf collima-
grated components that make up the EPID system (arm, tor (MLC) positioning, jaw positioning, isocentricity
panel, etc.) and the gravitational torque acting on the checks, and variable axis angles. In terms of dosimet-
device, the device and the gantry are susceptible to sag ric checks, output constancy, output factor, profiles,
effects, leading to geometric uncertainties. Not only is and wedge factor are often taken into account.
it critical to minimize sag for the sake of image-guided
radiation therapy (IGRT), but it is also necessary for the 8.3.3 MLC Positioning
sake of using the EPID as a machine QA tool. Several The leaves of an MLC precisely shape the beam during
AAPM Task Groups (TGs) including AAPM TG-142 [9] both static and dynamic fields. Following the intro-
and TG-179 [10] provide guidelines for imager QA tests duction of MLC in the mid-90s, several publications
that cover geometric accuracy, image quality, and dosi- offered suggestions to address MLC leaf QA. Most of
metric response for MV and kV planar imaging, as well the suggestions were normalized and summarized
as cone-beam computed tomography (CBCT) imag- in AAPM TG-50 [12], published in 1998. This Task
ing. For the remainder of this section, we will primarily Group, however, did not comprehensively address
focus on MV kV isocenter coincidence. procedures applicable to static field planning, and
To perform MV and kV isocenter analysis, physicists were not suitable for dynamic IMRT planning, which
usually make use of a phantom with some configura- was growing in popularity at the time. In response,
tion of BBs embedded within it. A set of 2D kV images several publications were released that began to out-
at four cardinal gantry angles can be acquired with the line dynamic leaf QA tests and procedures, and later,
EPID and kV imager located at a fixed source-to-imager suggestions for dynamic leaf QA were established by
distance (SID). A set of 2D MV images can be acquired AAPM TG-142. The task group suggests “strip test,”
at the same gantry angles as the 2D kV images. The devi- initially proposed by Chui et al., to be comprehen-
ation between the appropriate markers and electronic sive in that it is able to acquire leaf positional data
crosshairs (graticule) on each image represents the devi- as well as transmission information. Now commonly
ation between the imaging and treatment coordinates. referred to as the picket fence test, Chui et al.’s strip
In addition to simultaneously using MV and kV test [13] originally consisted of irradiating 1 mm wide
imagers to determine imager misalignments, some strips on film at known, regular intervals. Initially,
groups have proposed the ability to use a single imager the film was visually inspected for positional errors.
to determine its geometric integrity. For instance, Mao Visible leaf positional errors may include an abnor-
et al. determined it is possible to find the MV isocenter mally large gap (caused by either the leading leaf stop-
by first irradiating a phantom with embedded BBs at a ping too late, the trailing leaf stopping too early, or
nonzero gantry angle. One BB is embedded in the cen- a combination of the two), a narrow gap (caused by
ter of the phantom while the rest are dispersed strategi- either the leading leaf stopping too early, the trailing
cally along the phantom surface. The projections of the leaf stopping too late, or a combination of the two),
surface BBs are observable in the acquired portal image. or a deviated gap where both the leading and trailing
The location of ideal BB projections (during full struc- leaves either stop too late or too early. Going beyond
tural integrity) can be calculated. The real location of a simple visual inspection. Sastre-Padro et al. showed
the BB projections are extracted from the portal image it is possible to yield more quantitative results by
and optimized to determine the functional parameters digitizing the “strip-test” film and relating the rela-
of the treatment machine, which include the desired tive dose change observed at the leaf junctions to the
MV isocenter result [11]. position and width of each gap [14].
Relying on similar fitting techniques, several pub- et al. showed the feasibility of computing simple field
lications have highlighted the utility and increased size measurements using an EPID [19]. To acquire these
convenience of conducting MLC QA using an EPID measurements, the open field images were exported
[15]. It is possible to obtain sub-pixel accuracy through from the treatment machine and several signal profiles
various fitting techniques. One technique proposed were extracted from the image in both the inline and
by Mamalui-Hunter et al. made use of a modified crossline directions. The full-width at half-maximum
Lorentzian function whose parameters described the (FWHM) of each profile was converted from pixel val-
leaf positions and gap width pattern characteristic to a ues to cm, based upon the EPID’s pixel spacing and ver-
strip test image [16]. When repeating the acquisition and tical position. The EPID’s sensitivity to jaw positional
analysis process five times, the standard deviation of the shifts was tested by recording FWHM measurements
reproducibility of leaf deviation results was determined when increasing and decreasing the digital jaw positions
to be <0.1 mm [17]. These results give merit to the use in 0.1 cm increments. Sun et al. found that the measured
of EPID for MLC QA when using the Picket Fence test. field size reflected the digital jaw positional changes [19].
It was strongly established that the leaf positional data When computing the field size of an FFF beam, one
acquired from an EPID was acceptably comparable to should not simply follow the same method proposed
film measurements, and some even suggest the use of an above where the FWHM of each extracted profile is
EPID yields more consistent results than using film [17]. determined. As one may infer from the name, FFF
beams are not flat but forward-peaked. In this case,
8.3.4 Jaw Positioning the location of the 50% dose is not representative of the
In IMRT and SBRT fields, jaws are often used to spare field edges. As a work-around, Fogliata et al. proposed
healthy tissue by blocking radiation transmission a method that strategically re-normalizes the extracted
through MLC leaves, or by defining the extent of a treat- profiles so that the 50% dose location does align with the
ment field. Slight misalignments in the jaws position- field edge [20]. To accomplish this, one must find loca-
ing may result in significant tissue under or overdosing. tion in the dose profile that is similar for both flattened
Therefore, it is critical to ensure jaw positional accuracy. beams and FFF beams to normalize to the same dose
Light/Radiation congruence has long been the pre- level. Previously, Pönisch et al. suggested the inflection
ferred test for jaw positioning and beam steering. point of the profile to suffice [21]. Fogliata et al. pointed
During patient set-up, a therapist often makes use of the out, however, that the common step length (1 mm) used
light field for alignment. This light field is formed by a to locate the inflection point lacks the precision neces-
light source mounted in the gantry head which shines sary to accurately locate the point due to the high gradi-
through the X and Y jaws. The light that passes through ent region of the inflection point [20]. Such imprecision
the gantry head and onto the patient should be represen- can lead to 10% error in the dose level used for normal-
tative of how radiation will travel from the gantry head ization, ultimately potentially leading to a 40% error in
and onto the patient. Inconsistencies observed between central dose level after normalization. As an alterna-
the marked light field and radiation field edge may imply tive, Fogliata et al. suggest using what they refer to as
the machine needs to be steered or jaw misalignments the “shoulder point” to renormalize the signal profile.
are present. As an alternative to using film to conduct The “shoulder point” is located at the top of the sharp
the light/radiation congruence test, Prisciandaro et al. dose gradient defined by the field edge and right before
suggested an EPID-based method [18]. Instead of film, the dose of an FFF beam steadily increases toward the
Prisciandaro et al. utilized a phantom with a diamond CAX peak [20]. More quantitatively, the “shoulder point”
cut out in the middle. Each jaw was aligned to a vertex is the maximum of the signal profile’s third derivative
of a diamond using the light field. An open field image located inside of the profile’s inflection point. This loca-
was then acquired and the distance between the calcu- tion in the FFF profile is normalized to the same position
lated radiation field edge and a vertex is defined to be the in the flattened profile, and the relative dose at the FFF
light-radiation field in congruency [18]. CAX is used as the renormalization value. To simplify
An EPID also allows for the quick calculation of open the procedure, Fogliata et al. developed a function based
field size, which is often a required parameter for daily lin- on field size and EPID measuring depth to compute the
ear accelerator (linac) QA. As a part of linac daily QA, Sun renormalization value. Once the renormalization value
is determined, it is used to renormalize each FFF profile. that represents the center of each ray and to employ
The field size of the renormalized FFF profile can then a minimization routine to find the optimal intersec-
be computed as usual using the FWHM. tion point and “wobble” of the defined lines. To find
Beyond single open field jaw verification, the EPID the lines, many groups have first digitized the star shot
has shown utility in providing QA for more compre- film image and extracted multiple circular profiles
hensive jaw positioning. For instance, Clews et al. devel- about the estimated star shot center. The resulting pixel
oped an EPID-based technique that covers asymmetric profiles yield a series of peaks, which correspond to the
jaw QA [22]. When asymmetric jaws are employed, it is irradiated areas. To determine the exact radiation cen-
often the case that one set of jaws is positioned at the ter of each ray, several methods have been proposed
CAX while the other jaw forms the extent of the field. including subtracting the mean value of the profile
Following, a second field is acquired such that the jaw from the profile, finding the zero-crossing points, and
that was previously positioned at the CAX forms the finding their respective midpoints [23]. Depuydt et al.
extent of the field while the opposite jaw is now located suggested simply finding the middle of the FWHM of
at the CAX. If jaw misalignments are present during each ray to define their centers [24]. Once the intersect-
this configuration, a significantly high dose or low dose ing lines have been defined, the minimum circle that
line may form at the junction of the abutting CAX lines. intersects or is tangent to all the lines is found by some
Therefore, it is critical to minimize the over/under dose optimization routine that minimizes the distance to
caused by jaw misalignments. In their work, Clews et each line from some point. The position and size of this
al. acquired a set of jaw-abutted images and combined circle is defined as the variable axis isocenter size and
them. At the junction point, a dark or bright line may position [23, 24]. This method can be repeated for each
be visible due to underlap or overlap [22]. The ratio variable axis (gantry, collimator, and couch).
between the dose at the CAX and the dose at the field Perhaps a more robust test of isocentricity is the
center is computed. Clews et al.’s work showed a lin- Winston–Lutz test. This test was first introduced in 1988
ear relationship between the dose at the CAX and the by Lutz et al. for the purpose of stereotactic radiosur-
overlap/underlap size (in mm) [22]. Therefore, the ratio gery QA [25]. The original test consisted of fixing a steel
of the CAX dose to the field center dose was assumed ball at the location of treatment isocenter and attaching
to be a good indicator of asymmetrical jaw alignment. a film to the gantry head so that it sits perpendicularly to
Moreover, the EPID method proposed yielded compa- the beam path. The steel ball is positioned between the
rable results to similar film-based methods, and it was gantry head and the film. The film is irradiated at eight
suggested that the EPID-based method added precision distinct gantry and table angle configurations in an
while reducing time needed for test completion. open field orientation. The radiation field is larger than
the diameter of the ball. If the gantry positioning was
8.3.5 Isocentricity solid, the resulting “shadow” of the steel ball embed-
The motion of a linear accelerator is confined to rota- ded in the radiation field (visible on the film) should be
tion along three variable axes – gantry rotation, col- located in the exact center of the radiation field. Due to
limator rotation, and couch rotation. It is critical to misalignments (wobble), however, deviations occur and
test the rotation patterns along each variable axes and are decisive for determining the structural integrity of
to quantify how misalignments in each variable axes the linac. The resulting films were visually inspected for
contribute to the overall machine isocenter size and isocenter location discrepancies.
deviation. A common isocentricity test is known as the Moving beyond a simple visual inspection of the
Star Shot. Star Shot analysis is used for both 2D and irradiated films, several groups, including Low et al.
3D isocenter determination. Two-dimensional analy- [26] and Tsai [27], developed minimization algorithms
sis is accomplished by placing film perpendicular to in the mid-1990s to quantitatively describe image off-
a variable axis and irradiating a “strip” of the film at sets that allude to machine or setup misalignments. In
specified angles. The resulting film is made up of mul- more recent years, the effort required for the acquisition
tiple “strips,” or “rays,” that intersect near the center and analysis of the Winston–Lutz test has been greatly
of the film. To define the resulting isocenter size and reduced by making use of an EPID. Following a very
position, it is common to first define a straight line similar acquisition plan, Winkler et al. first introduced
a procedure of using an EPID for the Winston–Lutz test table until the crosshair shadow has rotated 90 degrees,
[28]. Like the original, images were acquired at various as displayed by the grid paper. This value is compared to
gantry, couch, and collimator angles, except instead of the digital angle readout to ensure the two are in agree-
irradiating films, the images were acquired digitized ment (within a set tolerance). Film is generally not used
by the EPID. The minimization routine employed by for these machine QA measurements.
Winkler et al. differs from the ones stated above in that As one may imagine, using a bubble level and grid
it divides up the data into variable axes categories. To paper for variable axis angle measurements can be
quantify table wobble, Winkler et al. asserted that the tedious and is susceptible to user subjectivity. Wilson
field center can be taken as a stationary point for each et al. utilized an EPID and rotation tracking to yield
table image and is not subject to change. The coordinates more quantifiable results [29]. To accomplish this, they
of the tungsten ball center, however, may change due to fabricated a phantom with five embedded BBs. The
table wobble. These displacements (between field center phantom was made to be fixed to the end of the table
and ball center) are plotted about an origin and the min- and extended out over machine isocenter. The EPID was
imum inscribed circle about the points is calculated and then positioned directly under the BBs in the phantom.
defined to be the table’s wobble radius and position. A When the table is positioned at 0 degrees, an image
similar approach is taken for collimator wobble, except was acquired and the BB locations in the images were
instead of defining the field to be a stationary reference detected and saved. For each additional angle to mea-
point, the tungsten ball center is now taken as the sta- sure, the table was rotated to that angle and an image
tionary point. Displacements between the variable field was acquired. Subsequently, the BB locations were
center and ball center are plotted and are also mini- detected and stored. The rotational transformation
mized using the minimum inscribed circle technique. matrix that describes the expected rotation based on
For gantry rotation, the tungsten ball center is defined three variables – the couch rotation center, the angle of
as stationary and the field center is expected to shift rotation, and an angle scaling factor – was optimized.
with misalignments. The displacement between the ball The resulting angle from the optimization was taken as
center and the field center on the image plane is “back- the physically measured angle. The angle scaling factor
projected,” forming a vector normal to the EPID. This and rotation center can be further analyzed for isocenter
is repeated for each image in which gantry positioned stability. Wilson et al. proposed the returned information
differs from the reference position, forming a collection is sufficient for isocenter QA and determined the results
of normal vectors (perpendicular to the x–y plane). If are comparable to Star Shot and Winston–Lutz analy-
the gantry rotates with no wobble, all lines would inter- ses [29]. Similar methods can be followed for collima-
sect at a single point. Since wobble does exist, however, tor rotation. And, with additional care taken to account
the gantry isocenter is determined by computing the for EPID tilt and projection geometries, Grelewicz et al.
smallest circle that is tangent to or intersects all lines. showed a similar method can be followed for measuring
This is accomplished through a gradient optimization gantry angle deviations as well as providing essential
technique which returns the gantry isocenter size and isocenter information [30].
location.
8.3.7 Output
8.3.6 Variable Axis Angles Apart from its utility in ensuring the structural
Apart from using an EPID as a direct replacement for integrity of a linac, an EPID has been shown to pro-
film measurements, the imager also enables physicists vide assurance of the dosimetric behavior of a linac.
to implement entirely new QA methods and procedures. However, the stability and output of an EPID must
For instance, Wilson et al. proposed a quantitative be verified against ion-chamber readings before it is
method to measure true table and collimator rotation used for regular dosimetric measurements. Budgell et
angle [29]. Measuring the physical angle of these vari- al. showed there is a linear relationship between mea-
able axes is generally completed by using a bubble level sured ion-chamber measurements and EPID measure-
and/or the use of physical grid paper. A physicist will ments normalized to 100 cGy for 100 MU. The MU
align grid paper under the crosshair shadows present ranged from 95 MU to 105 MU [31]. Budgell et al.
from the light field and will rotate the collimator or also monitored the output measurements of 100 MU
acquired using both an EPID and an ion chamber over symmetry present in an open field beam about the CAX.
the course of a year. From these measurements, it was AAPM TG-45 [33] defines symmetry as:
determined that almost 95% of the EPID measure-
ments were within 1.3% of the associated ion-chamber max(Dleft − Dright )
Symmetry = × 100 (8.3)
measurement, which is sufficient agreement for daily Dcenter
output monitoring [31].
Extending beyond simply determining agreement where Dleft and Dright are dose values at equidistant
among EPID and ion chamber for CAX output mea- points to the left and right of the CAX, and Dcenter is the
surements, it is also critical to ensure similar dosi- dose at the CAX. Budgell et al. notes that the measure-
metric properties are detectable for both methods of ment of flatness and symmetry does not take any addi-
measurement. One known parameter that affects an tional acquisition if the EPID is used for daily output
ion-chamber reading is field size. To understand how measurements by means of sampling the CAX dose of
field size affects EPID dosimetric measurements, Sabet an open field image [31]. The same open field images
et al. acquired relative EPID measurements for field can conveniently be used for flatness and symmetry
sizes ranging from 5 cm to 19 cm in length [32]. The ion- measurements.
chamber response was also recorded for each of these Budgell et al. monitored the flatness and symmetry
field sizes, and the ratio of EPID to ion-chamber mea- in terms of deviation from baseline results for 6 MV and
surements was computed. For the two buildup materials 8 MV over the course of a year [31]. During this period,
(Cu and solid water) and for both energies (6 MV and the EPID panel that was used for data acquisition was
18 MV), the difference between relative EPID measure- replaced and the gun and steering currents in the linac
ments and ion-chamber response was all within 1.5%. were periodically changed. These changes were all
Sabet et al. also evaluated the off-axis response for EPID reflected in the collected data and were approximately
and ion-chamber measurements by performing a one- sensitive to changes of half a percent. Such precision is
dimensional Gamma evaluation on the EPID and ion- adequate to catch abnormal beam behavior and to track
chamber measurements [32]. The results showed good profile trends over time, meaning an EPID is adequate
agreement for 6 MV and 18 MV, especially when copper for routine profile monitoring.
buildup was used. Thus, it was determined that an EPID
is an acceptable tool for continuous output measure- 8.3.9 Wedge
ment, especially when comparing output results to a set Physical wedges have long been used in radiation treat-
baseline. ment plans to modify the isodose distribution where
the isodose lines are oriented at some angle to the line
8.3.8 Flatness/Symmetry normal to the beam CAX. Using a physical wedge
As an extension of their investigation on the utility of comes with limitations. For instance, physical wedges
using an EPID for daily QA, Budgell et al. determined are manufactured with discrete field size dimensions.
that flatness and symmetry measurements were also Furthermore, physical wedges can be heavy, making it
possible using an EPID [31]. Both flatness and sym- difficult for a therapist to mount to the gantry head. Such
metry are routinely considered as a part of daily weight and positioning also presents a risk of injury to
machine QA. Flatness is simply the measurement of the patient. To obtain the same dosimetric effect, it is
dose uniformity across an axis (crossline or inline) of now much more common to use a combination of a
an open field image. AAPM TG-45 [33] defines flat- sweeping jaw motion and a varying dose rate such as is
ness as: used in Varian’s enhanced dynamic wedge (EDW) pro-
tocol. Using this method as opposed to a physical wedge
M −m allows for the incorporation of a wedge at any angle and
Flatness = × 100 (8.2)
M +m any field size and spares any risk of injury to the patient.
In order to ensure the accuracy of a wedge profile, it is
where M and m are the maximum and minimum signal common to use an ion-chamber array, a diode array,
values within profile width which is bound by the 80% or film to compare the profile to that in the treatment
of CAX dose locations. Similarly, symmetry resolves the planning system. Perhaps unsurprisingly, this method is
slowly being replaced with EPID-based methods, adding patient’s treatment as well as with (transit dosimetry) or
convenience and resolution while saving time and cost. without (non-transit dosimetry) an attenuating medium
Despite it having a linear dose response and a highly between the source and EPID. Further, the analysis can
reproducible CAX dose, using an EPID for wedge meth- be performed either at the level of the EPID or that of
ods presents some otherwise nonexistence problems. For the phantom/patient, if applicable. The resulting itera-
instance, scattering effects differ from an EPID to those tions of these measurement options allow centers to ver-
of solid water and backscatter from the imager arm is ify the deliverability of treatment plans in a way which
present. Moreover, the phosphor layer of the EPID yields suits their workflow while providing the desired treat-
an energy-dependent response. In order to quantify the ment information.
extent to which these limitations pose an issue to mea- EPID-level strategies generally involve evaluating the
suring a wedge factor, Greer et al. compared the EDW measured portal image with reference to a predicted
profiles obtained using an EPID to those obtained using image derived from the planned fluence [35, 36]. Either
ion-chamber measurements while monitoring the sta- the raw EPID response itself or the dose-to-water at
bility of an EPID and developed a correction method to the plane of the EPID may be predicted, with the latter
account for using an EPID for EDW measurements [34]. approach requiring conversion of the raw image into a
Greer et al. developed a quadratic function to cor- portal dose image. Although portal dose predictions may
rect for anomalies present in EPID measurements. The not represent a direct check of the method used by the
function was used for 30-degree and 60-degree wedge treatment planning system to generate the clinical plan’s
measurements for both 6 MV and 18 MV. The wedge dose distribution, appropriate calibrations do enable the
factors derived using an EPID were all within 0.6% of comparison of measured vs. predicted portal dose image
the ion-chamber reference factors. The largest deviation to be done on an absolute scale [35, 37]. Further, if the
in EPID wedge factors over the course of three weeks calculation of predicted portal dose is done by the treat-
was 0.9%. Furthermore, Greer monitored the ratio of ment planning system as opposed to an in-house devel-
the EPID dose to the ion-chamber dose over the course oped software, EPID-level dose prediction can serve as a
of over 200 days and found the standard deviation broader pre-treatment check of the treatment planning
among 6 MV measurements to be 0.4% and the stan- system. Measuring and predicting the EPID response
dard deviation among 18 MV measurements to be 0.3%, with a phantom in the beam path is a similar method of
which strongly supports the long-term reproducibility pre-treatment verification which more closely matches a
of EPID wedge measurements. To correct for the back- clinical environment [37]. Though both techniques can
scatter present due to the imager arm positioning, Greer assess the deliverability of plans, the communication
et al. suggests only taking EDW measurements in the between planning system and treatment machine, and
crossline direction. Greer et al. noted that measurement the general integrity of the planning system, neither can
acquired in the inline direction differ with EPID dis- identify errors unique to the treatment setting [36]. That
tance from the source due to beam broadening caused shortcoming is addressed when EPID-level transit mea-
by divergence. If an institution uses the Y Jaws for EDW surements are performed during an actual treatment,
measurements, Greer et al. simply suggest rotating the with the EPID image taken after the beam has passed
collimator to 90 degrees in order to minimize the effect through the patient [36].
of imager arm backscatter [34]. Even with the use of daily image guidance from
kV sources, it can be difficult to identify soft-tissue
8.4 USE OF EPID FOR PATIENT QA anatomy changes and visualize the impact of setup
The advantageous dosimetric properties of EPIDs have uncertainties or organ motion. With in vivo EPID-
inspired exploration into their utilization as patient- level dosimetry, these issues lead to an easily identifi-
specific QA tools, with a variety of clinical implementa- able discrepancy between the measured and predicted
tions reported. In addition to their ease of use and setup images which also enables an evaluation of the error’s
reproducibility, gantry-mounted EPIDs offer great flex- severity [36, 38]. However, the two-dimensional nature
ibility with respect to how such treatment verification of the analysis makes it difficult to precisely identify
is performed. EPID images can be taken before (pre- the source of any such discrepancy (doing so would
treatment verification) or during (in vivo verification) a likely require an additional simulation scan), and this
method’s focus on the EPID rather than the patient to derive the fluence of a clinical plan and use that as
inhibits an understanding of how any error impacts the input to a dose calculation engine [42–44]. The result-
dose distribution within the patient. Furthermore, per- ing phantom dose distribution, whether two- or three-
forming EPID measurements during treatment requires dimensional, can then be compared to that computed in
additional care to avoid collisions between the imager the treatment planning system.
and couch/patient [36]. Still, the method’s relative sim- Although phantom-level approaches expand upon the
plicity, the availability of commercial portal dose pre- benefits of pre-treatment EPID-level methods, they still
diction/analysis tools, and the possibility of combining share some of the same disadvantages. For example, even
multiple approaches make EPID-level verification an if differences between the measured and predicted three-
attractive means of patient-specific QA. dimensional dose distributions are observed in a phan-
Two significant limitations of EPID-level analyses not tom, the effect of these discrepancies on the patient would
resolved with an in vivo approach are their inherent lack not be immediately apparent, instead requiring further
of a third dimension and inability to directly validate investigation. This limitation, in addition to the increas-
the dose calculation method utilized by the treatment ing complexity of radiation therapy treatments, has made
planning system. The three-dimensional and tissue-like dose computations at the level of the patient more and
construction of many phantoms, meanwhile, allows the more valuable. While such reconstruction has previ-
use of phantom-level EPID approaches to build upon ously been performed using fluence measured with film,
the benefits of pre-treatment EPID-level verification EPID dosimetry is an alternative which allows for the
while addressing the above drawbacks. Generally, phan- possibility of easier, less time-consuming in vivo dosim-
tom-level methods involve utilizing EPID measure- etry as well as automated treatment verification [45, 46].
ments to reconstruct the (two- or three-dimensional) Consequently, there has been much exploration into dose
dose within a phantom and comparing the result with a reconstruction, both two- and three-dimensional, within
predicted distribution [39–44]. With a tissue-like phan- the patient using EPIDs. One such pre-treatment strategy
tom, the same dose calculation algorithm used to create is the use of EPID-measured fluence as input to a Monte
the treatment plan can be used to generate this predic- Carlo dose calculation with the patient’s planning CT
tion, enabling centers to use phantom-level methods scan. Comparing the resultant distribution with that
to verify the dose calculation model utilized for treat- of the treatment plan facilitates visualization of any
ment planning. Further, as long as the reconstruction differences and how they manifest within the patient’s
technique takes advantage of the three-dimensional anatomy. Further, as long as the reconstruction algo-
nature of a phantom, phantom-level approaches can rithm operates independently of the treatment plan-
provide better localization of dose discrepancies than ning system, this method can serve as a verification of
two-dimensional techniques. A common phantom-level the treatment plan’s dose computation model. Still, like
method utilizes transit EPID images measured with a other pre-treatment approaches, such in vivo measure-
phantom in the beam’s path to back-calculate the dose ments on their own can only reveal dose discrepancies
from the EPID to within the phantom [39, 40]. In this owing to more systematic errors, like issues with plan
approach, determining the beam transmission through deliverability or the treatment planning system. When
the phantom is required, as it enables an estimation of calculated with treatment-day CBCT scans instead of
the EPID dose contribution due to phantom scatter. the planning CT scan, however, this same general strat-
While this can be achieved with a second non-transit egy can also illuminate dosimetric errors from anat-
EPID image, modeling the transmission is a reported omy changes or setup uncertainties without the added
alternative which obviates the need for another mea- complications, such as scatter corrections and possible
surement [41]. It is important to note that the phantom collisions, of collecting transit portal images during
may or may not be physically present in the beam’s path treatment, thus equipping centers with a unique tool
during the measurement, as the categorization of ‘phan- for evaluating the necessity of plan adaptation.
tom-level’ does not require that dose be delivered to an The above method’s main limitation, which it shares
actual phantom, just that the phantom is the site of com- with other pre-treatment verification procedures, is
parison with the prediction. Indeed, one can also use that it is not based on a patient’s actual treatment.
EPID images from a pre-treatment, non-transit delivery Consequently, transit dosimetry, with EPID images
taken during a normal treatment session, has also been from the confirmation of leaf positions to three-dimen-
explored as a means of acquiring an in vivo dose distri- sional in vivo dosimetry [46, 51]. Furthermore, the abil-
bution [41, 46–50]. Because these transit measurements ity to use CBCT scans for dose reconstruction makes
are performed with the patient in the beam’s path, the EPID dosimetry an appealing tool for plan adaptation
reconstructed dose necessarily takes into consideration [52]. Ultimately, EPID dosimetry offers many possibili-
errors unique to the treatment setting, in addition to the ties for verification and continues to be applicable to the
more general concerns of plan deliverability, communi- evolving techniques in the field.
cation between treatment machine and planning system,
etc. The reconstruction itself can be performed using a REFERENCES
back-projection algorithm, though the information the 1. H. Gustafsson, P. Vial, Z. Kuncic, C. Baldock, P. B.
technique provides depends on whether it makes use of Greer, “EPID dosimetry: Effect of different layers of
the planning CT scan or a treatment-day CBCT scan. materials on absorbed dose response,” Medical Physics
For example, while a change in the patient’s anatomy 36, 5665–5674 (2009).
2. V. M. S. Inc, PortalVision aS1000: The State of the Art in
may very well cause a discrepancy between the planned
Electronic Portal Imaging (2016).
and measured dose distributions when using the plan- 3. B. W. King and P. B. Greer, “A method for removing
ning CT scan for reconstruction, identifying that change arm backscatter from EPID images,” Medical Physics
as the source of the difference could be difficult without 40, 071703 (2013).
additional imaging or the results of pre-treatment verifi- 4. P. B. Greer and C. C. Popescu, “Dosimetric properties of
an amorphous silicon electronic portal imaging device
cation. Further, if an anatomy change or setup error was
for verification of dynamic intensity modulated radia-
indeed the cause, the reconstructed distribution would tion therapy,” Medical Physics 30, 1618–1627 (2003).
not be fully valid. Thus, the use of the planning CT scan 5. V. Mhatre, S. Pilakkal, P. Chadha, K. Talpatra,
in reconstruction can still reveal mistakes from a wide “Dosimetric comparison of a-Si 1200 and a-Si 1000
range of sources, but it makes difficult a precise identi- electronic portal imager for intensity modulated radia-
fication of the source and an assessment of the dosimet- tion therapy (IMRT),” Journal of Nuclear Medicine &
Radiation Therapy 9, 2 (2018).
ric impact. Utilizing a treatment-day CBCT scan for the 6. Z. Xu, J. Kim, J. Han, A. T. Hsia, S. Ryu, “Dose rate
reconstruction, on the other hand, would more clearly response of Digital Megavolt Imager detector for flat-
show any change from the planning CT scan while pro- tening filter‐free beams,” Journal of Applied Clinical
viding a fully up-to-date dose distribution so that the Medical Physics 19, 141–147 (2018).
effect of any discrepancies with the treatment plan can 7. B. King, L. Clews, P. Greer, “Long-term two-dimen-
sional pixel stability of EPIDs used for regular linear
be readily evaluated [49, 50]. Moreover, while perform- accelerator quality assurance,” Australasian Physical &
ing patient-level dosimetry during (as opposed to before) Engineering Sciences in Medicine 34, 459–466 (2011).
treatment does present some complications, such as the 8. R. J. Louwe, L. N. McDermott, J. J. Sonke, R. Tielenburg,
need to avoid collisions between the couch/patient and M. Wendling, M. van Herk, B. J. Mijnheer, “The long‐
imager, it also has ramifications for patient safety. This term stability of amorphous silicon flat panel imag-
ing devices for dosimetry purposes: Stability of EPID
technique affords the opportunity to implement consis-
response,” Medical Physics 31, 2989–2995 (2004).
tent online treatment verification which, if a major dose 9. E. E. Klein, J. Hanley, J. Bayouth, F. F. Yin, W. Simon,
discrepancy is detected, could automatically halt treat- S. Dresser, C. Serago, F. Aguirre, L. Ma, B. Arjomandy,
ments [46]. Still, the lifetime of an imager used in that “Task Group 142 report: Quality assurance of medical
manner would be another concern, so such a procedure accelerators,” Medical Physics 36, 4197–4212 (2009).
10. J. P. Bissonnette, P. A. Balter, L. Dong, K. M. Langen,
may need to be limited to hypofractionated treatments.
D. M. Lovelock, M. Miften, D. J. Moseley, J. Pouliot,
Although EPIDs present many possibilities for patient- J. J. Sonke, S. Yoo, “Quality assurance for image‐guided
specific treatment verification, their utility would be radiation therapy utilizing CT‐based technologies:
reduced if measurement and analysis techniques were A report of the AAPM TG‐179,” Medical Physics 39,
unable to accommodate newer, more complex treatment 1946–1963 (2012).
methods. Instead, techniques such as VMAT and adap- 11. W. Mao, M. Speiser, P. Medin, L. Papiez, T. Solberg,
L. Xing, “Initial application of a geometric QA tool for
tive radiation therapy have been explored with EPID integrated MV and kV imaging systems on three image
dosimetry, demonstrating its continued relevance. For guided radiotherapy systems,” Medical Physics 38,
example, EPID verification of VMAT plans can vary 2335–2341 (2011).
12. A. Boyer, P. Biggs, J. Galvin, E. Klein, T. LoSasso, 26. D. A. Low, Z. Li, R. E. Drzymala, “Minimization of tar-
D. Low, K. Mah, C. Yu, “Basic applications of multileaf get positioning error in accelerator‐based radiosurgery,”
collimators. Report of the AAPM Radiation Therapy Medical Physics 22, 443–448 (1995).
Committee Task Group No. 50,” Medical Physics 72, 27. J.-S. Tsai, “Analyses of multi-irradiation film for sys-
16–40 (2001). tem alignments in stereotactic radiotherapy (SRT) and
13. C. S. Chui, S. Spirou, T. LoSasso, “Testing of dynamic radiosurgery (SRS),” Physics in Medicine & Biology 41,
multileaf collimation,” Medical Physics 23, 635–641 1597 (1996).
(1996). 28. P. Winkler, H. Bergmann, G. Stuecklschweiger, H. Guss,
14. M. Sastre-Padro, U. A. van der Heide, H. Welleweerd, “Introducing a system for automated control of rota-
“An accurate calibration method of the multileaf col- tion axes, collimator and laser adjustment for a medi-
limator valid for conformal and intensity modulated cal linear accelerator,” Physics in Medicine & Biology 48,
radiation treatments,” Physics in Medicine & Biology 49, 1123 (2003).
2631 (2004). 29. B. Wilson and E. Gete, “Machine‐specific quality assur-
15. P. Rowshanfarzad, M. Sabet, M. P. Barnes, D. J. ance procedure for stereotactic treatments with dynamic
O’Connor, P. B. Greer, “EPID‐based verification of the couch rotations,” Medical Physics 44, 6529–6537 (2017).
MLC performance for dynamic IMRT and VMAT,” 30. Z. Grelewicz, H. Kang, R. D. Wiersma, “An EPID based
Medical Physics 39, 6192–6207 (2012). method for performing high accuracy calibration
16. M. Mamalui‐Hunter, H. Li, D. A. Low, “MLC quality between an optical external marker tracking device
assurance using EPID: A fitting technique with subpixel and the LINAC reference frame,” Medical Physics 39,
precision,” Medical Physics 35, 2347–2355 (2008). 2771–2779 (2012).
17. S. J. Baker, G. J. Budgell, R. I. Mackay, “Use of an amor- 31. G. J. Budgell, R. Zhang, R. I. Mackay, “Daily monitoring
phous silicon electronic portal imaging device for of linear accelerator beam parameters using an amor-
multileaf collimator quality control and calibration,” phous silicon EPID,” Physics in Medicine & Biology 52,
Physics in Medicine & Biology 50, 1377 (2005). 1721 (2007).
18. J. Prisciandaro, M. Herman, J. Kruse, “Utilizing an 32. M. Sabet, P. Rowshanfarzad, F. W. Menk, P. B. Greer,
electronic portal imaging device to monitor light and “Transit dosimetry in dynamic IMRT with an a-Si
radiation field congruence,” Journal of Applied Clinical EPID,” Medical & Biological Engineering & Computing
Medical Physics 4, 315–320 (2003). 52, 579–588 (2014).
19. B. Sun, S. M. Goddu, S. Yaddanapudi, C. Noel, H. Li, 33. R. Nath, P. J. Biggs, F. J. Bova, C. C. Ling, J. A. Purdy,
B. Cai, J. Kavanaugh, S. Mutic, “Daily QA of linear J. van de Geijn, M. S. Weinhous, “AAPM code of prac-
accelerators using only EPID and OBI,” Medical Physics tice for radiotherapy accelerators: Report of AAPM
42, 5584–5594 (2015). Radiation Therapy Task Group No. 45,” Medical Physics
20. A. Fogliata, R. Garcia, T. Knöös, G. Nicolini, A. Clivio, 21, 1093–1121 (1994).
E. Vanetti, C. Khamphan, L. Cozzi, “Definition of 34. P. B. Greer and M. P. Barnes, “Investigation of an amor-
parameters for quality assurance of flattening filter phous silicon EPID for measurement and quality assur-
free (FFF) photon beams in radiation therapy,” Medical ance of enhanced dynamic wedge,” Physics in Medicine
Physics 39, 6455–6464 (2012). & Biology 52, 1075 (2007).
21. F. Pönisch, U. Titt, O. N. Vassiliev, S. F. Kry, R. Mohan, 35. A. van Esch, T. Depuydt, D. P. Huyskens, “The use of
“Properties of unflattened photon beams shaped by a an aSi-based EPID for routine absolute dosimetric
multileaf collimator,” Medical Physics 33, 1738–1746 pre-treatment verification of dynamic IMRT fields,”
(2006). Radiotherapy and Oncology 71, 223–234 (2004).
22. L. Clews and P. B. Greer, “An EPID based method for 36. S. L. Berry, C. Polvorosa, S. Cheng, I. Deutsch,
efficient and precise asymmetric jaw alignment quality K. C. Chao, C.-S. Wuu, “Initial clinical experience per-
assurance,” Medical Physics 36, 5488–5496 (2009). forming patient treatment verification with an electronic
23. A. Gonzalez, I. Castro, J. Martı ́nez, “A procedure to portal imaging device transit dosimeter,” International
determine the radiation isocenter size in a linear accel- Journal of Radiation Oncology Biology Physics 88,
erator,” Medical Physics 31, 1489–1493 (2004). 204–209 (2014).
24. T. Depuydt, R. Penne, D. Verellen, J. Hrbacek, S. Lang, 37. S. L. Berry, R. D. Sheu, C. S. Polvorosa, C. S. Wuu,
K. Leysen, I. Vandevondel, K. Poels, T. Reynders, “Implementation of EPID transit dosimetry based on a
T. Gevaert, “Computer-aided analysis of star shot films through‐air dosimetry algorithm,” Medical Physics 39,
for high-accuracy radiation therapy treatment units,” 87–98 (2012).
Physics in Medicine & Biology 57, 2997 (2012). 38. M. Kroonwijk, K. L. Pasma, S. Quint, P. C. Koper,
25. W. Lutz, K. R. Winston, N. Maleki, “A system for ste- A. G. Visser, B. J. Heijmen, “In vivo dosimetry for pros-
reotactic radiosurgery with a linear accelerator,” tate cancer patients using an electronic portal imaging
International Journal of Radiation Oncology Biology device (EPID); demonstration of internal organ motion,”
Physics 14, 373–381 (1988). Radiotherapy and Oncology 49, 125–132 (1998).
39. M. Wendling, R. J. Louwe, L. N. McDermott, J. J. 46. B. J. Mijnheer, P. González, I. Olaciregui-Ruiz, R. A.

Sonke, M. van Herk, B. J. Mijnheer, “Accurate two‐ Rozendaal, M. van Herk, A. Mans, “Overview of 3-year
dimensional IMRT verification using a back‐projec- experience with large-scale electronic portal imag-
tion EPID dosimetry method,” Medical Physics 33, ing device–based 3-dimensional transit dosimetry,”
259–273 (2006). Practical Radiation Oncology 5, e679–e687 (2015).
40. L. McDermott, M. Wendling, B. van Asselen, J. Stroom, 47. P. Francois, P. Boissard, L. Berger, A. Mazal, “In vivo
J. J. Sonke, M. van Herk, B. Mijnheer, “Clinical experi- dose verification from back projection of a transit dose
ence with EPID dosimetry for prostate IMRT pre‐treat- measurement on the central axis of photon beams,”
ment dose verification,” Medical Physics 33, 3921–3930 Physica Medica 27, 1–10 (2011).
(2006). 48. M. Wendling, L. N. McDermott, A. Mans, J.-J. Sonke,
41. R. Pecharromán‐Gallego, A. Mans, J. J. Sonke, J. C. M. van Herk, B. J. Mijnheer, “A simple backprojection
Stroom, Í. Olaciregui‐Ruiz, M. van Herk, B. J. Mijnheer, algorithm for 3D in vivo EPID dosimetry of IMRT
“Simplifying EPID dosimetry for IMRT treatment treatments,” Medical Physics 36, 3310–3321 (2009).
verification,” Medical Physics 38, 983–992 (2011). 49. L. N. McDermott, M. Wendling, J. Nijkamp, A. Mans,
42. B. Warkentin, S. Steciw, S. Rathee, B. Fallone, J.-J. Sonke, B. J. Mijnheer, M. van Herk, “3D in vivo dose
“Dosimetric IMRT verification with a flat‐panel EPID,” verification of entire hypo-fractionated IMRT treat-
Medical Physics 30, 3143–3155 (2003). ments using an EPID and cone-beam CT,” Radiotherapy
43. W. J. van Elmpt, S. M. Nijsten, R. F. Schiffeleers, A. and Oncology 86, 35–42 (2008).
L. Dekker, B. J. Mijnheer, P. Lambin, A.W. Minken, 50. W. van Elmpt, S. Nijsten, S. Petit, B. Mijnheer, P. Lambin, A.
“A Monte Carlo based three‐dimensional dose recon- Dekker, “3D In Vivo Dosimetry Using Megavoltage Cone-
struction method derived from portal dose images,” Beam CT and EPID Dosimetry,” International Journal of
Medical Physics 33, 2426–2434 (2006). Radiation Oncology Biology Physics 73, 1580–1587 (2009).
44. W. J. van Elmpt, S. M. Nijsten, A. L. Dekker, B. J. 51. M. Bakhtiari, L. Kumaraswamy, D. W. Bailey, S. de
Mijnheer, P. Lambin, “Treatment verification in the Boer, H. K. Malhotra, M. B. Podgorsak, “Using an EPID
presence of inhomogeneities using EPID‐based three‐ for patient-specific VMAT quality assurance,” Medical
dimensional dose reconstruction,” Medical Physics 34, Physics 38, 1366–1373 (2011).
2816–2826 (2007). 52. L. C. G. G. Persoon, A. G. T. M. Egelmeer, M. C. Öllers,
45. W. D. Renner, M. Sarfaraz, M. A. Earl, C. X. Yu, S. M. J. J. G. Nijsten, E. G. C. Troost, F. Verhaegen, “First
“A dose delivery verification method for conventional clinical results of adaptive radiotherapy based on 3D
and intensity‐modulated radiation therapy using mea- portal dosimetry for lung cancer patients with atelec-
sured field fluence distributions,” Medical Physics 30, tasis treated with volumetric-modulated arc therapy
2996–3005 (2003). (VMAT),” Acta Oncologica 52, 1484–1489 (2013).
Chapter 9
Scintillation Fiber Optic Dosimetry

Arash Darafsheh
St. Louis, Missouri
CONTENTS
9.2 Optical Fibers 124
9.2.1 Solid-Core Fibers 124
9.2.2 Hollow Waveguides 125
9.3 Scintillators 125
9.4 Detectors 127
9.4.1 Photomultiplier Tubes 127
9.4.2 Photodiodes 128
9.5 Cherenkov Radiation in Optical Fibers 128
9.6 Ionization Quenching and LET Dependencies 131
9.7 Characteristics of a Commercially Available System 132
9.8 Summary 133
References 133
9.1 INTRODUCTION the fiber) of the fiber. The optical signal is guided by the
Using scintillation properties of materials is one of optical fiber to a detector to measure the dose. However,
the oldest techniques for ionizing radiation detection the main problem with fiber optic dosimetry is that the
[1, 2]. In the context of radiotherapy dosimetry, scintil- signal received by the detector through the fiber is “con-
lation fiber optic dosimeters have drawn great attention taminated” with Cherenkov radiation, which may not be
due to their unique practical advantageous properties directly proportional to the dose. Cherenkov radiation
including the ability to perform in vivo, real-time, and has angular dependency and also depends on the length
intracavitary measurements with high spatial resolution of the fiber in the radiation field. It is a variable compo-
due to their small physical size and mechanical flex- nent that cannot be simply subtracted as a calibration fac-
ibility. These features make them ideal candidates for tor. Therefore, the total signal must be properly corrected
many applications in radiotherapy dosimetry, such as in for the contribution of Cherenkov radiation in order to
brachytherapy, intensity-modulated radiation therapy, accurately measure the absorbed dose in the scintillator.
superficial therapy, stereotactic radiosurgery, proton A significant issue related to scintillation dosimetry
therapy, and small-field dosimetry [3–19]. that occurs in proton therapy and other beams with
Working principle of fiber optic dosimeters is based high linear energy transfer (LET) is the non-proportion-
on radioluminescence properties of materials. A visible ality between the collected light from the scintillator
signal proportional to the absorbed dose is produced as and the proton dose [20–25]. At low stopping powers the
a result of the interaction of the ionizing radiation with scintillation signal is linear with respect to the energy
the sensitive portion (a scintillator attached to the tip of deposition, however, the scintillation signal “saturates”
123
at high stopping powers. This nonlinearity effect, mani- the core-cladding boundary and guided through the
fested as under-response of the optical signal to the core. Rays incident on the entrance face of the fiber with
radiation absorbed dose, is due mainly to the ionization angles greater than the acceptance angle (black ray in
quenching phenomenon resulting from non-radiative Figure 9.1) will impinge on the core-cladding boundary
de-excitations occurring at high density energy deposi- at angles smaller than θ c , will be only partially reflected
tion [26–29]. at each encounter with the core-cladding boundary and
In this chapter a brief overview of scintillation fiber will quickly leak out of the fiber.
optic dosimetry is given. Working principle of optical The numerical aperture (NA) is an important param-
fibers, characteristics of scintillators and detectors, and eter of optical fibers in quantifying their ability to collect
issues related to Cherenkov radiation generated in opti- light and radiate outgoing light. NA is related to the crit-
cal fibers are discussed. ical angle in the fiber that defines a cone of angles within
which all rays are guided in the fiber by total internal
9.2 OPTICAL FIBERS reflection, as shown in Figure 9.1. Mathematically,
In fiber optic dosimeter systems, optical fibers are NA = no sin θ a = nc2 − ncl2 , where no is the refractive
needed to collect and transport the radioluminescent index of the medium in which the fiber’s tip is inserted
signal from the scintillator to the detector. We briefly (no = 1 for air) and θ a is the half-angle of the fiber’s
describe basic principles and properties of solid-core acceptance cone, as illustrated in Figure 9.1. Practically,
optical fibers and hollow waveguides (HWGs) in this the NA is defined as the sine of the angle at which the
section. output optical power falls to 5% of the peak value. When
two fibers are coupled, it is important to match their NA
9.2.1 Solid-Core Fibers to optimize the coupling efficiency. Special couplers are
Typical materials of optical fibers are silicate-based designed to optimize the coupling between the fibers by
glasses and various plastics, such as polymethyl- matching their NAs. Coupling loss and reflection losses
methacrylate (PMMA) and polyethylene (PE) [30]. at both ends of the fiber can be minimized by antireflec-
Conventional optical fibers, fundamentally, consists of tion coating of the fiber ends.
two concentric dielectric cylinders: core and cladding. Depending of the modulation of the refractive index
Although the core of the fiber can have other shapes, of the core, fibers can have step-index or graded-index
such as square, most commonly fibers with circular profile. In step-index fibers, core has a constant refrac-
cross section are used. The core region is surrounded tive index nc with radius and total internal reflection
by the cladding and has slightly higher refractive index occurs at the core-cladding boundary. In graded-index
than the cladding (nc > ncl ). Usually, there are one or
more layers of protective coating outside the cladding,
such as a protective plastic buffer layer and a protective
jacket. Their main functions are to provide mechanical
support, protect the core and cladding against mechani-
cal damage and ambient conditions, such as moisture,
and shield against coupling of the ambient light into the
fiber. Their effects, however, on the optical properties of
the fiber are minimal.
Light is guided in the fiber by total internal reflec-
tion at the core-cladding boundary. The acceptance
angle of the fiber defines a cone of angles (acceptance
cone) within which rays are guided in the fiber. A ray of
light incident within the acceptance cone of the fiber on
the entrance surface of the core (red ray in Figure 9.1)
will impinge on the core-cladding boundary at an angle
greater than the critical angle, θ c = sin −1 (ncl / nc ), and is FIGURE 9.1 Schematic of the structure of an optical fiber
“trapped” inside the core, totally internally reflected at and its acceptance cone [31].
Scintillation Fiber Optic Dosimetry ◾ 125
fibers, the core index nc (r ) varies (gradually decreases)

with distance from the central axis of the fiber. In such
fibers, rays are gradually bent according to the index
profile of the core. Light is guided within all types of
fibers in the form of sets of guided electromagnetic
waves known as modes of the fiber. Only a certain
discrete number of modes can be guided by the fiber.
Fibers can be classified as single-mode or multimode. A
multimode fiber can sustain many hundreds of modes: FIGURE 9.2 Schematic of the axial cross section view of a
M ≈ (4a / λ 2 )(nc2 − ncl2 ), in which a is the core radius of HWG with silver/dielectric coating (left) and with silver-only
the fiber and λ is the wavelength of light. For optical coating (right).
power delivery as in fiber optic dosimeters, multimode
fibers are used which can be considered as “light pipes” for infrared light transmission, their transmission in
and ray picture of geometrical optics provides a good tool visible range is suboptimal. However, it has been shown
for their analysis. Rigorous analysis of the modal struc- that silver-only coated HWGs, i.e., without the dielectric
ture of light in the fiber optics is provided elsewhere [32]. silver iodide layer, have superior transmission in visible
The loss in an optical fiber is specified in terms of the range of the spectrum compared with the conventional
absorption coefficient, α, according to the following rela- HWGs with silver/dielectric coating that are optimized
tion: α = 10L log10 T1 = 10L log10 PPout
in
, where L is the length of for transmission of infrared light. Attenuation in vis-
the fiber, T is the transmission, Pin and Pout are the input ible spectral region ∼0.7 dB/m (∼85% transmission after
and output optical powers, respectively. Transmission 1 m) has been reported for such HWGs [34]. For visible
loss in optical fibers is extremely small for short-haul light transmission, solid-core fibers are preferred over
applications. For glass fibers the attenuation can be as HWGs.
low as ∼0.2 dB/km (>95% transmission after 1 km). For There are two types of sources for loss in HWGs:
plastic fibers, the attenuation is much higher, typically Propagation loss and bending loss. Propagation loss
∼200 dB/km (>95% transmission after 1 m). For practi- varies as 1/a3, where a is the bore radius. Bending loss
cal dosimetry applications, the fiber can be made long varies as 1/R, where R is the bending radius [33]. HWGs
enough to accommodate any configuration (including based on plastic and glass tubing have been fabricated
having the readout system in a remote place from the with bore diameters ∼0.1–2 mm. However, there is a
ionizing radiation field or in a separate room), with neg- tradeoff between mechanical flexibility, robustness, and
ligible loss of power in transmission. throughput. Since the power loss in a HWG is inversely
proportional to the cube of the bore radius, increas-
9.2.2 Hollow Waveguides ing the bore diameter reduces the transmission loss.
HWGs are generally used for infrared laser power deliv- However, there is a practical limit as the waveguide bore
ery due to their air core. The development of wave- size is increased, resulting in a rigid glass rod instead of
guides with hollow cores was a milestone in infrared a relatively flexible HWG. In practice, the largest hol-
light transmission, where conventional solid-core fibers low glass waveguide size that is typically used is ∼1 mm
dramatically suffer from optical power loss. HWGs inner diameter. HWGs made from plastic tubing are
structurally are composed of a glass or plastic capil- more flexible, however they have higher scattering losses
lary coated internally with a metal/dielectric layer to due to their inherently rougher inner surface compared
enhance the infrared transmission in the waveguide. to HWGs made from glass tubing.
The structure of a typical HWG is shown in Figure 9.2.
A metallic layer of silver (∼0.5–2 μm thick) is coated on 9.3 SCINTILLATORS
the inside of the silica tubing and then a dielectric layer Radioluminescence or scintillation properties of mate-
of silver iodide is formed over the metal film through an rials have attracted intense research effort for a myriad
iodization process in which some of the Ag is converted of applications in radiation detection and dosimetry,
to AgI [33]. The silica tubing has a polymer coating on radiotherapy quality assurance, molecular imaging,
its outside surface. Since HWGs are typically designed and so on. Luminescence is defined as spontaneous
emission of radiation from an electronically or vibra- Only a small fraction of the energy absorbed by the
tionally excited species not in thermal equilibrium with scintillator is converted into visible photons and the rest
its environment [35]. Various types of luminescence, is dissipated non-radiatively by phonons, vibrational
such as photo-, radio-, electro-, thermo-, chemo-, tribo-, modes within large structures. Light yield defined as
etc. are classified according to the mode of excitation number of photons per MeV of absorbed radiation is
[36]. Radioluminescence is an umbrella term for dif- one of the most important parameters of scintillators
ferent procedures (fluorescence, phosphorescence, and to quantify their scintillation efficiency. Anthracene has
delayed fluorescence) that lead to light generation as the highest light yield among organic scintillators and
a result of interaction of ionizing radiation with mat- often used as a reference for comparison. Its light yield is
ter. Fluorescence is emission between two states with ∼2 × 104 photons per MeV that corresponds to only ∼5%
equal multiplicity (i.e., singlet–singlet or triplet–triplet efficiency in the blue region of visible spectrum. The
transitions) whereas phosphorescence results from a light yield of typical plastic scintillators is ∼40–65% of
transition between two states with different spin mul- that of anthracene. NaI(Tl) is a typical inorganic scintil-
tiplicity (e.g., triplet–singlet emission after intersystem lator with light yield of ∼10%.
crossing). Fluorescence and phosphorescence processes Plastic scintillators are composed of a base mate-
have different time scales due to their allowed and for- rial, usually polyvinyltoluene (PVT), polystyrene, or
bidden transitions, typically 0.1–20 ns and 1 ms to 10 s, acrylic-naphthalene and one or more organic dyes.
respectively [32]. Although scintillation and radiolumi- When fast electrons or other charged particles pass
nescence are often used interchangeably, the former can through the scintillator, ∼3% of the energy deposited
be considered as a subcategory of the latter and usually is ultimately converted to scintillation photons in the
implies fast emission of light in response to radiation as range of 400–500 nm. A plastic scintillator is relatively
opposed to phosphorescence. transparent to its own light (attenuation length ∼4 m),
Scintillators are insulators with wide energy gap is very fast (<10 ns response time), is highly resistant
between their valence and conduction band. Based on to radiation damage (∼3% loss per 104 Gy) and can be
their chemical composition, the scintillators can be drawn into thin fibers or formed into sheets. The scin-
divided into two broad categories: organic and inor- tillation signal is independent of dose rate, total dose,
ganic. Main advantage of organic scintillators over inor- and angle of incidence. Plastic scintillator can be used
ganic scintillators for radiation therapy dosimetry is to form small, rugged, moderately sensitive detec-
that the former has almost water equivalent properties tors and with an atomic number that approximately
that helps avoid complicated dose calibration process matches that of water.
during scintillation dosimetry. Emission of observable scintillation is a complex two
Examples of organic scintillators include conjugated or three stage process [27]. A simplified description of
and aromatic compounds, such as aromatic hydrocar- this process is illustrated in Figure 9.3. The primary pro-
bons (naphthalene, anthracene, phenanthrene, pyrene, cess consists of excitation of PVT base molecules (often
perylene, porphyrins, phtalocyanins, etc.) and deriva- called solvent molecules), S, by secondary electrons.
tives, dyes (fluorescein, rhodamines, coumarins, and Only excitation of the weakly bound and delocalized
oxazines), polyenes, and diphenylpolyenes. These are phenyl π-electrons (vs. more tightly bound, inner-shell
employed as solutes in aromatic liquids and polymers σ-electrons), can result in fluorescence. Subsequently,
to form organic liquid and plastic scintillators, respec- an excited base molecule, S*, can dissipate its excitation
tively. Examples of inorganic scintillators are lanthanide energy by several mechanisms: self or internal quench-
ions (Eu3+, Tb3+), doped glasses (e.g., with ND, Mn, Ce, ing, (S* → S + heat), transfer of its excitation energy to
Cu), crystals (ZnS, CdS, ZnSe, CdSe, GaS, GaP, ruby: another solvent molecule (S* + S → S + S*), or emission
Al2O3/Cr3+), semiconductor nanocrystals, metal clusters, of a fluorescence photon (S* → S + hν). Since the UV
carbon nanotubes, and some fullerenes [36]. Inorganic photons produced by S* de-excitation in pure plastic
scintillators have higher photon interaction cross sec- base have a low yield and are strongly absorbed by plas-
tion and high light production efficiency. However, due tic and water-like media, pure plastic scintillator has
to their high atomic number they require complicated little practical value. To improve the quantum yield of
dose calibration process for scintillation dosimetry. the process and to shift the scintillation light to longer
metal-oxide semiconductor (CMOS) cameras can be

used as readout systems [39–42].
9.4.1 Photomultiplier Tubes

Conversion of scintillation photons into an electrical
signal and amplification of the signal are two functions
of a PMT. A PMT consists of an evacuated glass tube
with a very thin electrode (called a photocathode) at
its entrance, several electrodes (called dynodes) in the
interior, and an anode located at the end as the collec-
tor of electrons. Scintillation photons eject electrons
FIGURE 9.3 Simplified description of scintillation mecha- from the photocathode as a result of photoelectric effect.
nism for a ternary plastic scintillator that uses both primary Approximately one electron is released per five photons
and secondary wave-shifting dyes to produce visible photons
incident on the photocathode. The electrons released
in response to ionizing radiation. S* denote excited plastic
from the photocathode are accelerated toward the first
base molecules, F and F* denote ground state and excited
primary dye molecules, respectively, while Y and Y* denote dynode. Approximately five electrons are ejected as a
ground and excited states, respectively, of secondary dye result of impact of each electron with the dynode. These
molecules. secondary electrons are then accelerated toward the next
dynode (which is held at a more positive potential than
wavelengths, primary dyes, F, and secondary wave- the preceding one) which further increases their popu-
shifting dyes, Y, are added. The primary dye, F, typically lation upon incident with the dynode. This exponential
PBD, b-PBD (butyl-PBD), or p-terphenyl in ∼1% concen- increase in the population of electrons is repeated sev-
tration, absorbs energy from the excited PVT or polysty- eral times by letting the electrons pass through a series
rene base molecule via dipole–dipole non-radiative or of dynodes. Typical commercial phototubes may have
radiative transfer processes and itself enters an excited up to 15 dynodes. The electric field between dynodes is
state: S* + F → S + F*). De-excitation quickly occurs, established by applying a successively increasing posi-
(F* → F + hν p) accompanied by emission of 320–420 tive high voltage to each dynode. The voltage difference
nm “primary photons.” This primary process is local- between two successive dynodes is ∼100 V. The end result
ized, occurring within 1 mm of the original electron of this process is an output pulse having amplitude large
track. The primary photons travel through the plastic enough to be easily measured by the associated electron-
with a 0.1–3 mm mean-free path, until absorbed by the ics [37]. The gain of a PMT is strongly dependent on the
secondary dye molecules (typically bisMSB, t-PBD, or applied voltage, it can be higher than ∼106. The electron-
d-POPOP in ∼0.01% concentration). Energy absorption ics should be carefully selected since small changes in
by the secondary dye molecule (hν p + Y → Y* → hνs) the high voltage applied across the cathode and anode
is followed by reemission of 400–530 nm longer wave- can significantly change the output.
length photons, hνs, which constitute the final and easily PMTs are manufactured for working in ultraviolet,
observable optical signals. visible, and infrared wavelengths. Important param-
eters that should be considered in identifying a PMT
9.4 DETECTORS are their spectral response, quantum efficiency, sensitiv-
As mentioned in previous section, scintillation is not a ity, and dark current. These parameters are determined
very efficient process. However, highly efficient dosim- by the composition of the photocathode. Generally, the
eters can be developed by using photodetectors with long-wavelength cutoff is determined by the photocath-
high photon collection and efficiency [37]. Commonly ode, while the short-wavelength cutoff is determined
used photodetectors in scintillation dosimetry are pho- by the window material. The quantum efficiency of the
tomultiplier tubes (PMTs) and photodiodes [38]. When photon–electron conversion process in the photocath-
arrays of fibers or a scintillator sheet are used for 2D ode is low, typically ∼20–30% that implies ∼80–70%
measurements, multichannel PMTs, photodiode arrays, of the scintillation photons are wasted. For optimized
or charge-coupled device (CCD) and complementary results, the spectrum of the scintillator should match
the sensitivity of the photocathode. The dark current PIN system, respectively. Active photodetectors showed
consists mainly of the thermionic emission of electrons higher SNR at low dose rates and low energies. At high
from the photocathode and the first few dynodes, with dose rates, the SNR of passive photodetectors were on
much smaller contribution from cosmic rays. For exam- par with active photodetectors. The APD failed to read
ple a 5-cm-diameter photocathode may release ∼105 high dose rates because of the signal saturation. No tem-
electrons per second in the dark at room temperature. perature issues with the PMT were reported; the APD
Cooling of the photocathode greatly reduces the dark module, however, was affected by the heat. It was rec-
current, e.g., by a factor of ∼10 for temperature reduc- ommended to turn on the APD system ∼1 hour prior to
tion from 20°C to 0°C. However, caution must be exer- measurements to avoid background drifting [38].
cised to avoid condensation at the PMT window since
the moisture will reduce the amount of light incident 9.5 CHERENKOV RADIATION
on the photocathode. In addition, excessive cooling IN OPTICAL FIBERS
can cause voltage drop across the photocathode. Other Cherenkov radiation was discovered by Pavel Čerenkov,
ambient conditions influencing the performance of the for which he shared the 1958 Nobel Prize in Physics
PMT are changes in the humidity, presence of vibrations with Ilya Frank and Igor Tamm [43–45]. Cherenkov
and magnetic field. radiation has attracted a considerable amount of
recent research interest for its potential applications
9.4.2 Photodiodes in life sciences and engineering, such as in molecular
A photodiode detector is made of a semiconductor mate- imaging, particle detection, ionizing radiation quality
rial with appropriate doping optimized to work around assurance, and beam monitoring [46–64]. Cherenkov
the visible spectrum. The scintillation photons produce radiation is a visible light emitted from a dielectric
electron–hole pairs instead of just electrons as in a PMT medium when charged particles with velocities greater
photocathode. The quantum efficiency of a photodi- than the phase velocity of light in that medium, i.e.,
ode can be as high as ∼80% which makes it attractive v > c /n , pass through it. The passage of the charged par-
for use in low-level radiation detection. PMTs generally ticles induces dipole oscillations through polarization
have higher sensitivity than photodiodes. Photodiodes of the medium whose relaxation leads to emission of
produce more electrical noise than PMTs do but they light when v > c /n due to the constructive interference
are smaller and less expensive. Most photodiodes do of the emitted waves.
not amplify the signal. A special kind of photodiodes is Cherenkov radiation is a polarized, coherent, and
avalanche photodiode (APD) where electron–hole pairs directional emission; its direction is along the surface
produced are multiplied through avalanche process. The of a cone that makes the half-angle θ = cos −1 (nβ )−1 with
process is analogous to the electron multiplication in a the particle track, where n is the refractive index of the
PMT except that in this case there are no dynode-like medium and β = v /c is the ratio of the velocity of the
mechanical structures involved. The end result of the particle to that of light. To induce Cherenkov radiation,
electron–hole multiplication is the transformation of a a charged particle must satisfy the v > c /n condition; the
very low-level signal into a pulse having measureable minimum energy required is given as
amplitude [37]. The signal amplification in APD is not
as much as that in a PMT.  1 
Emin = m0c 2  − 1 (9.1)
A systematic evaluation of passive (PIN photodiode)  1− n −2

and active (APD and PMT) photodetectors performance
for plastic scintillation dosimetry based on the expected where m0 is the rest mass of the particle.
dose rate (<1 mGy/s, 1–10 mGy/s, >10 mGy/s) and beam The threshold electron energies to generate
energy (kV and MV energy range) is reported in [38]. Cherenkov radiation in water (n = 1.33), PMMA
The main issue associated with the passive photodetec- (n = 1.5), and pure silica (n = 1.55) are Emin,e = 264,
tors was their extremely low output signal in ∼pA range 174, and 158 keV, respectively. These energies are
that required a very sensitive electrometer to obtain reli- far below the energies of megavoltage beams used
able measurements. The APD and PMT generated ∼2 × in modern radiotherapy. The passage of these high
103 and ∼3 × 104 times more current compared to the energy primary and secondary electrons through the
FIGURE 9.4 (a) Minimum energy (Emin ) of an electron and proton required to generate Cherenkov radiation as a function of
the refractive index of the medium. (b) Maximum transferred energy (Wmax ) to an electron in a single collision with a proton
as a function of the proton’s energy.
fiber optic dosimeters generates Cherenkov light that The maximum transferred energy (Wmax ) to an elec-
is the dominant source of the unwanted background tron in a single collision with a proton as a function of the
signal in the output signal. proton’s energy, calculated from Equation 9.2 is plotted
The minimum energy of an electron and a pro- in Figure 9.4(b). It can be seen that Wmax = Emin,e in pure
ton to generate Cherenkov radiation is plotted in silica, PMMA, and water for 70, 77, and 114 MeV proton
Figure 9.4(a) as a function of the refractive index. It energy, respectively, indicating that for proton beams
can be seen that a proton requires much higher energy with energies greater than the above values Cherenkov
(∼1830 times higher) than an electron to generate radiation can occur from the secondary electrons liber-
Cherenkov radiation in a given medium. For example, ated as a results of columbic interaction of the protons.
the threshold energies of a proton to directly generate However, it has been shown that the amount of gener-
Cherenkov radiation in water, PMMA, and pure silica ated Cherenkov radiation from the secondary electrons
are Emin,p = 484, 320, and 290 MeV, respectively. It should liberated as a results of interaction of the proton beam
be mentioned that generation of Cherenkov radiation with the medium is not proportional to the absorbed
from high energy protons has been observed [65]; how- dose in the medium [10, 11].
ever, conventional clinical proton beams have energies The number of Cherenkov photons generated by a
below the threshold energy required for directly induc- charged particle with charge ze, where e is the elemen-
ing Cherenkov radiation. Therefore, the incident clini- tary charge, along path length dl in the wavelength
cal proton beams cannot directly generate Cherenkov region between λ1 and λ2 (λ1< λ2 ) is proportional to λ −2
radiation in water. Nevertheless, a proton can transfer and is given as
enough kinetic energy (Wmax > Emin,e) to the electrons in
the medium so that the electrons set in motion by the dN  1 1  1 
= 2πα z 2  −   1 − 2 2  (9.3)
protons can generate Cherenkov radiation. For a parti- dl  λ1 λ2   n β 
cle with rest mass M 0 and kinetic energy EK , maximum
energy transfer (Wmax ) to an electron in a single collision where α = e 2 / (4π 0 c ) ≈ 1/137 is the fine structure con-
is calculated from Equation 9.2 [66] stant. The amount of Cherenkov light contamination
recorded by the optical fiber dosimeter depends on the
2me c 2 β 2γ 2
Wmax = (9.2) angular configuration and spatial position and therefore
γ me
1 + 2M 0
+ ( )
me 2
M0 is not constant, so straightforward subtraction as a cali-
bration constant cannot be done. However, the spectral
where me and M 0 are the rest masses of an elec- characteristic of the Cherenkov radiation can be used to
tron and proton, respectively, and γ = E /( M0c 2 ) = decompose the output signal through rigorous spectros-
( EK + M 0c 2 )/( M 0c 2 ) and β = vc = 1 − γ −2 . copy. Specifically, Cherenkov radiation has a continuous
required for guided rays. Transmission of Cherenkov

radiation is, therefore, dependent on the angle between
the particle track and the fiber axis (see Figure 9.6), indi-
cating that the recorded raw signal is not directly related
to the absorbed dose by the scintillator.
The intensity of Cherenkov radiation can be ∼2 orders
of magnitude per mm of transport fiber lower than the
intensity of the scintillation signal per mm from the
scintillator piece [3]. However, the length of transport
fiber in radiation field is much longer than the scintil-
lation part (several cm vs. a few mm), which can lead
to a significant Cherenkov contamination in the total
signal recorded by the photodetector. In scintillation
fiber optic dosimetry, Cherenkov light is an unwanted
FIGURE 9.5 Theoretical spectrum of Cherenkov radiation
over visible spectrum with λ−3 intensity dependency on wave-
background signal and separating it from the total opti-
length. A typical spectrum of Cherenkov radiation collected cal signal is crucial; using raw signals without removing
by a silica optical fiber irradiated with a clinical electron that leads to a pronounced (up to ∼20%) discrepancy. It
beam is plotted for comparison. should be mentioned that in addition to the Cherenkov
radiation, the radioluminescence of the bare transport
spectrum spanning from near-ultraviolet to near-infra- fibers can contribute as an unwanted parasitic signal
red, restricted from both ends of the visible spectrum by in photon and electron dosimetry using scintillators
the absorption spectrum of the material in which it is [67]. However, their contribution is orders of magni-
generated, with light intensity decreasing proportional tude lower than the Cherenkov light and can be safely
to λ −3 as the wavelength increases. For comparison, a neglected [68].
typical experimental spectrum of Cherenkov radiation A comparison study of several transport fiber mate-
observed in an optical fiber irradiated with a clinical rials, including polystyrene, PMMA, and silica fibers
electron beam is presented in Figure 9.5, which shows a with low-OH content (“dry” silica) and high-OH con-
good agreement between experimental and theoretical tent (“wet” silica) irradiated with x-ray beam with
spectra. <150 kVp energy, showed that bare “dry” silica fibers
When optical fibers pass through ionizing radiation produce ∼4 orders of magnitude stronger radiolumines-
fields of high energy, such as in fiber optic dosimetry, cence than the bare PMMA fibers. The observed signal
Cherenkov radiation generated inside the fibers’ core is the radioluminescence in these fibers because the low
is guided through the fiber if the emitted ray hits the energy range selected for irradiation cannot generate
core-cladding boundary with angle greater than the Cherenkov radiation in the fibers [67]. The radiolumi-
critical angle to satisfy total internal reflection condition nescence in PMMA fibers reported to be ∼1–2 orders
FIGURE 9.6 Schematic illustration of Cherenkov radiation generated in the fiber when it is (a) guided through and (b) leaked
from the fiber. θ is the Cherenkov semi-cone angle and θi is the incident angle on the core-cladding boundary.
of magnitude weaker than that of polystyrene fibers. shown that HWGs based on silver-only coating have
PMMA fibers are the preferred choice of transport fiber higher transmission compared to their counterparts
material in fiber optic scintillation dosimetry due to with silver/dielectric coating and their performance in
their mechanical flexibility and lower economic cost scintillation fiber optic dosimetry has been investigated
(common to plastic fibers). recently [85–87].
Several methods have been proposed to minimize
the influence of Cherenkov radiation contamination 9.6 IONIZATION QUENCHING
in order to improve the accuracy of fiber optic dosim- AND LET DEPENDENCIES
etry [69–75]. These methods include the (i) Subtraction As mentioned earlier, a significant issue related to scin-
method based on using a parallel bare fiber identical tillation dosimetry that occurs in proton and other
to the one that is connected to the scintillator piece to high-LET fields is the non-proportionality between the
produce similar Cherenkov light that can be subtracted collected light from the scintillator and the proton dose.
from the total signal [76, 77]. However, this technique The light yield of a scintillating material depends not
is not reliable for radiation fields with high-dose gra- only on the energy of the ionizing particle, but also on
dient. (ii) Optical filtering where a long-wavelength- the stopping power of the particle. Birks’ semi empirical
emitting scintillator in conjunction with a long pass formula, Equation 9.4, is commonly used to determine
filter is used to selectively measure the signal in longer the quenching factor of an organic scintillator [26, 27]:
wavelengths of the spectrum where the intensity of the
Cherenkov radiation is weaker due to its λ −3 intensity dL S dE
= dr
(9.4)
profile [78]. However, this method is not very effec- dr 1 + kB dE
dr
tive since the filtered signal is still contaminated with
Cherenkov radiation due to the fact that the Cherenkov where r is the residual range of the particle, dL dr is the
radiation has a continuous spectrum. (iii) Temporal scintillation yield along the particle’s track, S is the abso-
separation that relies on different time scales associated lute scintillation factor, B dE
dr is the density of excitation
with Cherenkov emission (∼ps) and scintillation (∼ns or centers along the track, k is the quenching factor, and dE dr
longer) processes [79, 80]. In this methods, gated detec- is the stopping power.
tion is utilized to collect the signal, since the Cherenkov Ionization quenching in plastic scintillators can lead
emission stops immediately (∼ps) after the radiation to ∼10% under measurement of the dose near the Bragg
pulse, leaving the radioluminescence of the scintillator peak. In principle, quenching correction factors can be
to be detected. This method requires fast responding empirically determined [21–23]. Such factors, however,
electronics and works only with pulsed radiation fields. depend on the scintillator material and the energy spec-
(iv) Chromatic removal that requires two different opti- trum of the proton beam.
cal filters to measure the signal at two different spectral It was suggested that the visible emission from irradi-
regions; the dose is then calculated by using coefficients ated bare plastic fiber optics, i.e., without the additional
obtained from calibration [81–83]. (v) Rigorous spectral scintillator tip, can be used for proton therapy dosim-
separation based on acquiring the whole spectrum of etry [88, 89]. A remarkable distinction of this method
the transmitted signal and decomposing it into its con- compared to conventional fiber scintillators was that the
stituting components using a priori knowledge of the signals from the bare fibers were minimally affected by
spectral shape of the scintillation signal and Cherenkov the radiation quenching effect. The origin of this visible
radiation [73–75]. (vi) Using HWG with air core instead emission was initially attributed to Cherenkov radiation
of conventional solid-core fibers would reduce the dete- [88, 89]. However, Monte Carlo simulations did not pre-
riorating effect of Cherenkov radiation [84] due to the dict direct correlation between the absorbed dose and
fact that the production of Cherenkov light is minimal Cherenkov light production in a typical clinical pro-
in air since its refractive index is very close to 1; such ton beam [90–92]. The nature of the visible emission
HWGs, however, conventionally have optimal design responsible for proton dose measurement using PMMA
parameters for transmission of infrared wavelengths plastic and silica glass bare optical fibers was studied in
[33], whereas the scintillators of interest in fiber optic [10, 11] by performing rigorous luminescent spectro-
dosimeters emit primarily visible light. It has been scopic study validated with Monte Carlo simulations.
FIGURE 9.7 Radioluminescence spectra of (a) a bare PMMA fiber with peak emission at 410 nm and (b) a silica glass fiber
(with high-OH content) with emission peaks at 460 nm and 650 nm, placed at different depths in solid water phantom irradi-
ated with 100 MeV proton beam.
Monte Carlo simulation results, along with the observed with the depth in a phantom suggests a dependency on
experimental spectra, confirmed that Cherenkov radia- the LET of the beam and can be carefully examined for
tion cannot be responsible for the proton dose measure- LET sensing and high-LET beam monitoring. It should
ment ability of bare optical fibers [93–96]. be mentioned that in bare PMMA and bare silica fibers,
The luminescent spectroscopy of the visible light gen- the intensity of the radioluminescent signal is orders of
erated in proton-irradiated PMMA plastic bare fibers magnitude weaker than the signal from conventional
shows a continuous spectrum with a peak at 410 nm scintillators (e.g., BCF-12).
with spectral shape different from the spectrum of
Cherenkov radiation, see Figure 9.7(a) [10]. Comparing 9.7 CHARACTERISTICS OF A
the fiber’s spectrum with a typical spectrum of plastic COMMERCIALLY AVAILABLE SYSTEM
scintillator indicated that the nature of this visible light The first commercially available fiber optic dosim-
is the radioluminescence of the fiber material. A good etry system, Exradin W1 (Standard Imaging, Inc.,
agreement between depth dose measurement using Middleton, WI), was released in 2013. The fiber scintil-
ion chamber array and the fiber was observed. The fact lator is polystyrene with acrylonitrile butadiene styrene
that the acquired signal suffers minimal from quench- (ABS) plastic enclosure and polyimide stem. The sensi-
ing effect, which is promising in designing such devices tive scintillating piece has density of 1.05 g/cm3, 1.0 mm
for high-LET beam monitoring, may be qualitatively diameter, 3.0 mm length (0.0024 cm3 volume). Housing
explained by the lower density of excitation centers in of the tip has 2.8 mm diameter and 42 mm length. The
the bare fiber. In the case of the bare fiber, the density optical fiber used to transmit the scintillation signal
of the excitation centers along the track is very low, i.e., to the photodiode readout system is made of PMMA
B dE
dr 1, that leads to dr ∝ S dr indicating the minimal
dL dE with polystyrene jacket with 1.0 mm core diameter and
effect of ionization quenching effect in the bare fiber. 2.2 mm jacket diameter, and 3 m total length. The
In contrast to PMMA bare fiber, the emission spectrum manufacturer provides a procedure for dose calibration
of the silica glass fiber irradiated with a proton beam shows (obtain the “gain” coefficient) and Cherenkov radiation
two distinct peaks at 460 (blue) and 650 (red) nm, see correction. Several groups have studied its character-
Figure 9.7(b) [11], whose origin is connected to the silica istics, such as short-term repeatability, dose–response
point defects, namely oxygen-deficiency center (ODC) linearity, angular dependency, temperature dependen-
and non-bridging oxygen hole center (NBOHC) defects, cies, dose-rate dependencies, and long-term stability
respectively [97–100]. The intensity of the latter corre- [101–107].
lates with the absorbed dose, whereas that of the former The Exradin W1 was characterized under standard
does not directly follow the dose. The fact that the ratio conditions and short-term reproducibility of 0.3% and
of the signal peak intensities at 460 and 650 nm varies dose-rate independence within 0.5% was reported [101].
Consistency of the measure dose per MU was ∼0.1% 1D and 2D measurements, respectively. Cherenkov
(∼2%) when 50–100 MUs (5–100 MUs) were delivered; radiation generated in optical fibers is one of the main
however it was increased to ∼5% for extreme case of issues with photon and electron dosimetry and vari-
1 MU delivery, indicating that great consideration must ous solutions for dealing with it is available. However
be made when using such a system for very low dose the user must have a full understanding of the accu-
measurements. The calibration factor was noted to be racy of the Cherenkov radiation correction method
beam quality dependent and 1.5% ± 1.4% difference from used in their system, as well as other characteristics of
ion chamber measurement was reported. Variations in their scintillating fiber optic dosimetry system. For in
gain coefficient ∼3% was found when comparing dose- vivo dosimetry applications, the temperature depen-
to-water calibration on different days, indicating that dency should be evaluated in advance and properly
the calibration needs to be performed each time that the addressed. Cherenkov radiation generation by pro-
system is used for absolute dose measurement. Field size ton beams is of less importance, however, ionization
dependence was studied by measuring the output fac- quenching effect is the main issue with using scintil-
tors and 3.3% discrepancy with ion chamber measure- lators for proton therapy and other high LET beams.
ments was noted in very large field sizes, indicating that There is significant amount of work directed toward
the considered Cherenkov radiation correction method making scintillating fiber optic dosimeters for proton
is not very accurate for such large field sizes [101]. therapy, especially those with the ability to simultane-
In another study [102], maximum angular depen- ously measure LET and dose.
dency ∼0.34% and decrease in light output with tem-
perature ∼0.2–0.3% per °C was reported. The “effective
REFERENCES
point of measurement” was calculated by comparing the 1. G. F. Knoll, Radiation Detection and Measurement
tissue maximum ratio (TMR) curves measured with an (John Wiley & Sons, Inc., Hoboken, NJ, 2010).
ionization chamber and the scintillator system; its loca- 2. N. Tsoulfanidis and S. Landsberger, Measurement and
Detection of Radiation (CRC Press, Boca Raton, FL,
tion was found to be at 0.6 mm distance upstream with
2015).
respect to the central axis of the Exradin W1. The loss 3. L. Beaulieu and S. Beddar, “Review of plastic and liquid
of the signal per accumulated dose was 0.28% per kGy scintillation dosimetry for photon, electron, and pro-
(for 0–15 kGy) and 0.032% per kGy (for 15–127 kGy) ton therapy,” Physics in Medicine and Biology 61(20),
[102–104]. R305–R343 (2016).
4. S. Beddar and L. Beaulieu, Scintillation Dosimetry
For PDD measurement of photons and electrons, on
(CRC Press, Boca Raton, FL, 2016).
average 0.5% and 0.6% accuracy is reported comparing to 5. J. Lambert, D. R. McKenzie, S. Law, J. Elsey,
a CC04 (IBA Dosimetry GmbH, Germany) ion chamber, N. Suchowerska, “A plastic scintillation dosimeter for
respectively [108]. The PDD measurement in continuous high dose rate brachytherapy,” Physics in Medicine and
scanning mode using Exradin W1 was not available, and Biology 51(21), 5505–5516 (2006).
the measurement was done in step-and-shoot method. 6. B. Mijnheer, S. Beddar, J. Izewska, C. Reft, “In vivo
dosimetry in external beam radiotherapy,” Medical
Average precision ∼0.17% and daily reproducibility of Physics 40(7), 070903 (2013).
∼0.14% was reported for 2 Gy dose measurement. 7. F. Therriault-Proulx, T. M. Briere, F. Mourtada, S.
Aubin, S. Beddar, L. Beaulieu, “A phantom study of
9.8 SUMMARY an in vivo dosimetry system using plastic scintilla-
Scintillation fiber optic dosimeters have unique prac- tion detectors for real-time verification of 192Ir HDR
brachytherapy,” Medical Physics 38(5), 2542–2551
tically advantageous properties including the abil-
(2011).
ity to perform in vivo, real-time, and intracavitary 8. L. Wootton, R. Kudchadker, A. Lee, S. Beddar, “Real-time in
measurements with high spatial resolution due to vivo rectal wall dosimetry using plastic scintillation detec-
their small physical size and mechanical flexibility. tors for patients with prostate cancer,” Physics in Medicine
These features make them ideal candidates for many and Biology 59(3), 647–660 (2014).
applications in brachytherapy, intensity-modulated 9. I. Veronese, N. Chiodini, S. Cialdi, E. d’lppolito,
M. Fasoli, S. Gallo, S. La Torre, E. Mones, A. Vedda,
radiation therapy, superficial therapy, stereotactic G. Loi, “Real-time dosimetry with Yb-doped silica
radiosurgery, proton therapy, and small-field dosim- optical fibres,” Physics in Medicine and Biology 62,
etry. Individual and arrays of fibers can be used for 4218–4236 (2017).
10. A. Darafsheh, R. Taleei, A. Kassaee, J. C. Finlay, “The 24. A. Darafsheh, A. Soldner, H. Liu, A. Kassaee, T. Zhu,
visible signal responsible for proton therapy dosimetry J. Finlay, “Phosphor-based fiber optic probes for proton
using bare optical fibers is not Čerenkov radiation,” beam characterization,” Medical Physics 42(6), 3476
Medical Physics 43(11), 5973–5980 (2016). (2015).
11. A. Darafsheh, R. Taleei, A. Kassaee, J. C. Finlay, “Proton 25. A. Darafsheh, A. Kassaee, R. Taleei, D. Dolney, J.
therapy dosimetry using the scintillation of the silica Finlay, “Fiber optic microprobes with rare-earth-based
fibers,” Optics Letters 42(4), 847–850 (2017). phosphor tips for proton beam characterization,”
12. J. Lambert, Y. Yin, D. R. McKenzie, S. H. Law, Proceedings of SPIE 9700, 97000Q (2016).
A. Ralston, N. Suchowerska, “A prototype scintilla- 26. J. B. Birks, “Scintillations from organic crystals:
tion dosimeter customized for small and dynamic Specific fluorescence and relative response to different
megavoltage radiation fields,” Physics in Medicine and radiations,” Proceedings of the Physical Society. Section
Biology 55(4), 1115–1126 (2010). A 64(10), 874–877 (1951).
13. D. Létourneau, J. Pouliot, R. Roy, “Miniature scintillat- 27. J. B. Birks, The Theory and Practice of Scintillation
ing detector for small field radiation therapy,” Medical Counting (Pergamon Press, Oxford, 1964).
Physics 26(12), 2555–2561 (1999). 28. T. J. Gooding and H. G. Pugh, “The response of plas-
14. M. D. Belley, I. N. Stanton, M. Hadsell, R. Ger, B. W. tic scintillators to high-energy particles” Nuclear
Langloss, J. Lu, O. Zhou, S. X. Chang, M. J. Therien, Instruments and Methods 7(2), 189–192 (1960).
T. T. Yoshizumi, “Fiber-optic detector for real time 29. L. Torrisi, “Plastic scintillator investigations for relative
dosimetry of a micro-planar x-ray beam,” Medical dosimetry in proton-therapy,” Nuclear Instruments
Physics 42(4), 1966–1972 (2015). and Methods in Physics Research B 170, 523–530 (2000).
15. S. Safai, S. Lin, E. Pedroni, “Development of an inor- 30. Y. Koike, Fundamentals of Plastic Optical Fibers (Wiley-
ganic scintillating mixture for proton beam verifi- VCH, Weinheim, Germany, 2015).
cation dosimetry,” Physics in Medicine and Biology, 31. A. Darafsheh, Optical Super-resolution and Periodical
49(19), 4637–4655 (2004). Focusing Effects by Dielectric Microspheres (Ph.D. dis-
16. P. Woulfe, F. J. Sullivan, S. O’Keeffe, “Optical fibre sen- sertation, University of North Carolina at Charlotte,
sors: Their role in in vivo dosimetry for prostate cancer 2013).
radiotherapy,” Cancer Nanotechnology 7(1), 7 (2016). 32. B. E. A. Saleh and M. C. Teich, Fundamentals of
17. L. M. Moutinho, I. F. C. Castro, L. Peralta, M. C. Photonics (John Wiley & Sons, Hoboken, NJ, 2007).
Abreu, J. F. C. A. Veloso, “Brachytherapy dosimeter 33. J. A. Harrington, Infrared Fibers and Their Applications
with silicon photomultipliers,” Nuclear Instruments (SPIE Press, Bellingham, Washington, 2004).
and Methods in Physics Research Section A 787, 358– 34. J. E. Melzer and J. A. Harrington, “Investigation of sil-
360 (2015). ver-only and silver/TOPAS® coated hollow glass wave-
18. M. Eichmann and B. Thomann, “Air core detectors guides for visible and NIR laser delivery,” Proceedings
for Cerenkov-free scintillation dosimetry of brachy- of SPIE 9317, 93170H (2015).
therapy β-sources,” Medical Physics 44(9), 4900– 35. S. Braslavsky, “Glossary of terms used in photo-
4909 (2017). chemistry,” Pure and Applied Chemistry 79(3),
19. F. Therriault-Proulx, L. Beaulieu, S. Beddar, “Validation 293–465 (2007).
of plastic scintillation detectors for applications in 36. B. Valeur and M. N. Berberan‐Santos, Molecular
low-dose-rate brachytherapy,” Brachytherapy 16(4), Fluorescence: Principles and Applications (Wiley-VCH
903–909 (2017). Verlag & Co. KGaA, Weinheim, Germany, 2012).
20. A. Hitachi, T. Doke, A. Mozumder, “Luminescence 37. S. N. Ahmed, Physics and Engineering of Radiation
quenching in liquid argon under charged-particle Detection (Elsevier, Waltham, MA, 2015).
impact: Relative scintillation yield at different linear 38. J. Boivin, S. Beddar, M. Guillemette, L. Beaulieu,
energy transfers,” Physical Review B 46(18), 11463 (1992). “Systematic evaluation of photodetector performance
21. L. L. W. Wang, L. A. Perles, L. Archambault, N. Sahoo, for plastic scintillation dosimetry,” Medical Physics
D. Mirkovic, S. Beddar, “Determination of the quench- 42(11), 6211–6220 (2015).
ing correction factors for plastic scintillation detectors 39. M. Guillot, L. Beaulieu, L. Archambault, S. Beddar,
in therapeutic high-energy proton beams,” Physics in L. Gingras, “A new water-equivalent 2D plastic scintil-
Medicine and Biology 57(23), 7767–7781 (2012). lation detectors array for the dosimetry of megavoltage
22. D. Robertson, D. Mirkovic, N. Sahoo, S. Beddar, energy photon beams in radiation therapy,” Medical
“Quenching correction for volumetric scintillation Physics 38(12), 6763–6774 (2011).
dosimetry of proton beams,” Physics in Medicine and 40. J. Son, S. B. Lee, Y. Lim, S. Y. Park, K. Cho, M. Yoon,
Biology 58(2), 261–273 (2013). D. Shin, “Development of optical fiber based measure-
23. V. I. Tretyak, “Semi-empirical calculation of quenching ment system for the verification of entrance dose map
factors for ions in scintillators,” Astroparticle Physics in pencil beam scanning proton beam,” Sensors 18(1),
33, 40–53 (2010). 227 (2018).
41. A. M. Frelin, J. M. Fontbonne, G. Ban, J. Colin, 54. A. Ruggiero, J. P. Holland, J. S. Lewis, J. Grimm,
M. Labalme, A. Batalla, A. Vela, P. Boher, M. Braud, “Cerenkov luminescence imaging of medical isotopes,”
T. Leroux, “The DosiMap, a new 2D scintillating dosim- Journal of Nuclear Medicine 51(7), 1123–1130 (2010).
eter for IMRT quality assurance: Characterization 55. A. E. Spinelli, D. D’Ambrosio, L. Calderan, M. Marengo,
of two Cerenkov discrimination methods,” Medical A. Sbarbati, F. Boschi, “Cerenkov radiation allows in
Physics 35(5), 1651 (2008). vivo optical imaging of positron emitting radiotracers,”
42. H. Perera, J. F. Williamson, S. P. Monthofer, W. R. Physics in Medicine and Biology 55, 483–495 (2010).
Binns, J. Klarmann, G. L. Fuller, J. W. Wong, “Rapid 56. J. C. Finlay and A. Darafsheh, “Light sources,
two-dimensional dose measurement in brachyther- drugs, and dosimetry,” in Biomedical Optics in
apy using plastic scintillator sheet: Linearity, signal- Otorhinolaryngology: Head and Neck Surgery, edited by
to-noise ratio, and energy response characteristics,” B. Wong and J. Ilgner (Springer, New York, NY, 2016),
International Journal of Radiation Oncology Biology Chap. 19, pp. 311–336.
Physics 23(5), 1059–1069 (1992). 57. A. Darafsheh, S. Najmr, T. Paik, M. Tenuto, C. Murray,
43. P. A. Čerenkov, “Visible emission of clean liquids by J. C. Finlay, J. S. Friedberg, “Characterization of rare-
action of gamma radiation,” Doklady Akademii Nauk earth-doped nanophosphors for photodynamic ther-
SSSR 2, 451–454 (1934). apy excited by clinical ionizing radiation beams,”
44. I. Frank and I. Tamm, “Coherent visible radiation Proceedings of SPIE 9308, 930812 (2015).
of fast electrons passing through matter,” Comptes 58. G. Lucignani, “Cerenkov radioactive optical imag-
Rendus (Doklady) de l’Académie des Sciences de l’URSS ing: A promising new strategy,” European Journal of
14(3), 109–114 (1937). Nuclear Medicine and Molecular Imaging 38(3), 592–
45. J. V. Jelly, Čerenkov Radiation and Its Applications 595 (2011).
(Pergamon Press, London, 1958). 59. A. Darafsheh, T. Paik, M. Tenuto, S. Najmr, J. S.
46. J. L. Demers, S. C. Davis, R. Zhang, D. J. Gladstone, B. Friedberg, C. Murray, J. C. Finlay, “Optical character-
W. Pogue, “Čerenkov excited fluorescence tomography ization of novel terbium-doped nanophosphors excited
using external beam radiation,” Optics Letters 38(8), by clinical electron and photon beams for potential
1364–1366 (2013). use in molecular imaging or photodynamic therapy,”
47. R. Zhang, S. C. Davis, J. L. Demers, A. K. Glaser, D. Medical Physics 41(6), 436 (2014).
J. Gladstone, T. V. Esipova, S. A. Vinogradov, B. W. 60. B. Brichard, A. F. Fernandez, H. Ooms, F. Berghmans,
Pogue, “Oxygen tomography by Čerenkov-excited “Fibre-optic gamma-flux monitoring in a fission reac-
phosphorescence during external beam irradiation,” tor by means of Cerenkov radiation,” Measurement
Journal of Biomedical Optics 18(5), 50503 (2013). Science and Technology 18(10), 3257–3262 (2007).
48. Y. Helo, I. Rosenberg, D. D’Souza, L. Macdonald, 61. J. S. Cho, R. Taschereau, S. Olma, K. Liu, Y. C. Chen,
R. Speller, G. Royle, A. Gibson, “Imaging Cerenkov C. K. Shen, R. M. van Dam, A. F. Chatziioannou,
emission as a quality assurance tool in electron “Cerenkov radiation imaging as a method for quantita-
radiotherapy,” Physics in Medicine and Biology 59(8), tive measurements of beta particles in a microfluidic
1963–1678 (2014). chip,” Physics in Medicine and Biology 54(22), 6757–
49. K. Tanha, A. M. Pashazadeh, B. W. Pogue, “Review 6771 (2009).
of biomedical Čerenkov luminescence imaging appli- 62. S. Yamamoto, T. Toshito, S. Okumura, M. Komori,
cations,” Biomedical Optics Express 6(8), 3053–3065 “Luminescence imaging of water during proton-beam
(2015). irradiation for range estimation,” Medical Physics
50. P. Brůža, H. Lin, S. A. Vinogradov, L. A. Jarvis, D. J. 42(11), 6498–6506 (2015).
Gladstone, B. W. Pogue, “Light sheet luminescence 63. S. Yamamoto, M. Komori, T. Akagi, T. Yamashita,
imaging with Cherenkov excitation in thick scattering S. Koyama, Y. Morishita, E. Sekihara, T. Toshito,
media,” Optics Letters 41(13), 2986–2989 (2016). “Luminescence imaging of water during carbon-ion
51. R. Robertson, M. S. Germanos, C. Li, G. S. Mitchell, S. irradiation for range estimation,” Medical Physics
R. Cherry, M. D. Silva, “Optical imaging of Cerenkov 43(5), 2455–2463 (2016).
light generation from positron-emitting radiotracers,” 64. S. Yamamoto, S. Koyama, C. Yamauchi-Kawaura,
Physics in Medicine and Biology 54, N355–N365 (2009). M. Komori, “Luminescence imaging of biological
52. R. S. Dothager, R. J. Goiffon, E. Jackson, S. Harpstrite, subjects during X-ray irradiations lower energy than
D. Piwnica-Worms, “Cerenkov radiation energy trans- Cerenkov-light threshold,” Optical Review 24(3),
fer (CRET) imaging: A novel method for optical imag- 428–435 (2017).
ing of PET isotopes in biological systems,” PLoS One 65. R. L. Mather, “Čerenkov radiation from protons
5(10), e13300 (2010). and the measurement of proton velocity and kinetic
53. H. Liu, G. Ren, Z. Miao, X. Zhang, X. Tang, P. Han, S. S. energy,” Physical Review 84(2), 181–190 (1951).
Gambhir, Z. Cheng, “Molecular optical imaging with 66. Particle Data Group, “Review of particle physics,”
radioactive probes,” PLoS One 5(3), e9470 (2010). Chinese Physics C 38(9), 090001 (2014).
67. R. Nowotny, “Radioluminescence of some optical 80. B. L. Justus, P. Falkenstein, A. L. Huston, M. C. Plazas,
fibres,” Physics in Medicine and Biology 52(4), N67–N73 H. Ning, R. W. Miller, “Gated fiber-optic-coupled
(2007). detector for in vivo real-time radiation dosimetry,”
68. F. Therriault-Proulx, L. Beaulieu, L. Archambault, S. Applied Optics 43(8), 1663–1668 (2004).
Beddar, “On the nature of the light produced within 81. J. M. Fontbonne, G. Iltis, G. Ban, A. Battala, J. C.
PMMA optical light guides in scintillation fiber-optic Vernhes, J. Tillier, N. Bellaize, C. Le Brun, B. Tamain,
dosimetry,” Physics in Medicine and Biology 58(7), K. Mercier, J. C. Motin, “Scintillating fiber dosimeter
2073–2084 (2013). for radiation therapy accelerator,” IEEE Transactions
69. M. A. Clift, R. A. Sutton, D. V. Webb, “Dealing with on Nuclear Science 49(5), 2223–2227 (2002).
Cerenkov radiation generated in organic scintilla- 82. A. M. Frelin, J. M. Fontbonne, G. Ban, J. Colin, M.
tor dosimeters by bremsstrahlung beams,” Physics in Labalme, A. Batalla, A. Isambert, A. Vela, T. Leroux,
Medicine and Biology 45(5), 1165–1182 (2000). “Spectral discrimination of Čerenkov radiation in
70. P. Z. Liu, N. Suchowerska, J. Lambert, P. Abolfathi, scintillating dosimeters,” Medical Physics 32(9), 3000–
D. R. McKenzie, “Plastic scintillation dosimetry: 3006 (2005).
Comparison of three solutions for the Cerenkov chal- 83. M. Guillot, L. Gingras, L. Archambault, S. Beddar, L.
lenge,” Physics in Medicine and Biology 56(18), 5805– Beaulieu, “Spectral method for the correction of the
5821 (2011). Cerenkov light effect in plastic scintillation detectors:
71. L. Archambault, L. Beaulieu, S. A. Beddar, “Comment A comparison study of calibration procedures and
on ‘Plastic scintillation dosimetry: Comparison of validation in Cerenkov light-dominated situations,”
three solutions for the Cerenkov challenge’,” Physics in Medical Physics 38(4), 2140–2150 (2011).
Medicine and Biology 57(11), 3661–3665 (2012). 84. J. Lambert, Y. Yin, D. R. McKenzie, S. Law, N.
72. P. Z. Y. Liu, N. Suchowerska, J. Lambert, P. Abolfathi, Suchowerska, “Cerenkov-free scintillation dosimetry
D. R. McKenzie, “Reply to the comment on: ‘Plastic in external beam radiotherapy with an air core light
scintillation dosimetry: comparison of three solutions guide,” Physics in Medicine and Biology 53(11), 3071–
for the Cerenkov challenge’,” Physics in Medicine and 3080 (2008).
Biology 57(11), 3667–3673 (2012). 85. A. Darafsheh, J. E. Melzer, J. A. Harrington, A.
73. A. Darafsheh, R. Zhang, S. C. Kanick, B. W. Pogue, J. Kassaee, J. C. Finlay, “Radiotherapy fiber dosim-
C. Finlay, “Spectroscopic separation of Čerenkov radi- eter probes based on silver-only coated hollow glass
ation in high-resolution radiation fiber dosimeters,” waveguides,” Journal of Biomedical Optics 23(1),
Journal of Biomedical Optics 20(9), 095001 (2015). 015006 (2018).
74. A. Darafsheh, H. Liu, S. Najmr, M. Tenuto, J. S. 86. A. Darafsheh, H. Liu, J. E. Melzer, R. Taleei, J. A.
Friedberg, C. Murray, T. Zhu, J. C. Finlay, “Phosphor- Harrington, A. Kassaee, T. C. Zhu, J. C. Finlay, “Fiber
based fiber optic microprobes for ionizing beam radia- optic probes based on silver-only coated hollow glass
tion dosimetry,” Proceedings of SPIE 9317, 93170R waveguides for ionizing beam radiation dosimetry,”
(2015). Proceedings of SPIE 9702, 970210 (2016).
75. A. Darafsheh, R. Zhang, S. C. Kanick, B. W. Pogue, J. C. 87. A. Darafsheh, “Fiber optic dosimeters based on
Finlay, “Separation of Čerenkov radiation in irradiated silver-only coated hollow waveguides for radiation
optical fibers by optical spectroscopy,” Proceedings of therapy dosimetry,” Medical Physics 45(6), E684–
SPIE 9315, 93150Q (2015). E684 (2018).
76. S. Beddar, T. R. Mackie, F. H. Attix, “Cerenkov light 88. J. Son, M. Kim, D. Shin, U. Hwang, S. Lee, Y. Lim,
generated in optical fibres and other light pipes,” J. Park, S. Y. Park, K. Cho, D. Kim, K. W. Jang, M.
Physics in Medicine and Biology 37(4), 925–935 (1992). Yoon, “Development of a novel proton dosimetry
77. W. J. Yoo, S. H. Shin, D. Jeon, K. T. Han, S. Hong, system using an array of fiber-optic Cerenkov radia-
S. G. Kim, S. Cho, B. Lee, “Development of a fiber-optic tion sensors,” Radiotherapy and Oncology 117(3),
dosimeter based on modified direct measurement 501–504 (2015).
for real-time dosimetry during radiation diagnosis,” 89. K. W. Jang, W. J. Yoo, S. H. Shin, D. Shin, B. Lee, “Fiber-
Measurement Science and Technology 24(9), 094022 optic Cerenkov radiation sensor for proton therapy
(2013). dosimetry,” Optics Express 20(13), 13907 (2012).
78. S. F. de Boer, A. S. Beddar, J. A. Rawlinson, “Optical fil- 90. A. K. Glaser, R. Zhang, D. J. Gladstone, B. W. Pogue,
tering and spectral measurements of radiation-induced “Optical dosimetry of radiotherapy beams using
light in plastic scintillation dosimetry,” Physics in Cherenkov radiation: The relationship between light
Medicine and Biology 38, 945–958 (1993). emission and dose,” Physics in Medicine and Biology
79. M. A. Clift, P. N. Johnston, D. V. Webb, “A temporal 59(14), 3789–3811 (2014).
method of avoiding the Cerenkov radiation generated 91. A. Darafsheh, R. Taleei, A. Kassaee, J. C. Finlay, “The
in organic scintillator dosimeters by pulsed mega-volt- connection between Cherenkov light emission and
age electron and photon beams,” Physics in Medicine radiation absorbed dose in proton irradiated phan-
and Biology 47, 1421–1433 (2002). toms,” Medical Physics 43(6), 3418 (2016).
92. A. Darafsheh, R. Taleei, A. Kassaee, J. C. Finlay, “On plastic scintillator detector through comparison with
the nature of the background visible light observed in an in-house developed scintillator system,” Radiation
fiber optic dosimetry of proton beams,” Medical Physics Measurements 69, 50–56 (2014).
43(6), 3500 (2016). 102. P. Carrasco, N. Jornet, O. Jordi, M. Lizondo, A.
93. A. Darafsheh, R. Taleei, A. Kassaee, J. C. Finlay, “On the Latorre‐Musoll, T. Eudaldo, A. Ruiz, M. Ribas,
origin of the visible light responsible for proton dose “Characterization of the Exradin W1 scintillator for
measurement using plastic optical fibers,” Proceedings use in radiotherapy,” Medical Physics 42(1), 297–304
of SPIE 10056, 100560V (2017). (2015).
94. A. Darafsheh, R. Taleei, A. Kassaee, J. C. Finlay, “Proton 103. A. R. Beierholm, C. F. Behrens, C. E. Andersen,
therapy dosimetry by using silica glass optical fibers,” “Comment on “Characterization of the Exradin W1
Proceedings of SPIE 10058, 100580B (2017). scintillator for use in radiotherapy” [Med. Phys. 42,
95. A. Darafsheh, R. Taleei, A. Kassaee, J. C. Finlay, “Proton 297–304 (2015)],” Medical Physics 42(7), 4414–4416
therapy dosimetry by using the scintillation of glass (2015).
and plastic bare optical fibers,” Medical Physics 44(6), 104. P. Carrasco, N. Jornet, O. Jordi, M. Lizondo, A. Latorre‐
2858 (2017). Musoll, T. Eudaldo, A. Ruiz, M. Ribas, “Response to
96. A. Darafsheh, R. Zhang, A. Kassaee, J. Finlay, “Comment on ‘Characterization of the Exradin W1
“Characterization of the proton irradiation induced scintillator for use in radiotherapy’ [Med. Phys. 42,
luminescence of materials and application in radia- 297–304 (2015)],” Medical Physics 42(7), 4417–4418
tion oncology dosimetry,” Proceedings of SPIE 10478, (2015).
1047815 (2018). 105. A. Dimitriadis, I. Silvestre Patallo, I. Billas, S. Duane,
97. H. Zhang, Z. Yuan, J. Zhou, J. Dong, Y. Wei, Q. Lou, A. Nisbet, C. H. Clark, “Characterisation of a plastic
“Laser-induced fluorescence of fused silica irradiated scintillation detector to be used in a multicentre stereo-
by ArF excimer laser,” Journal of Applied Physics 110, tactic radiosurgery dosimetry audit,” Radiation Physics
013107 (2011). and Chemistry 140, 373–378 (2017).
98. L. Skuja, K. Kajihara, M. Hirano, H. Hosono, “Oxygen- 106. F. Alsanea, L. Wootton, N. Sahoo, R. Kudchadker, U.
excess-related point defects in glassy/amorphous Mahmood, S. Beddar, “Exradin W1 plastic scintillation
SiO2 and related materials,” Nuclear Instruments and detector for in vivo skin dosimetry in passive scattering
Methods in Physics Research Section B 286, 159–168 proton therapy,” Physica Medica 47, 58–63 (2018).
(2012). 107. C. Hoehr, C. Lindsay, J. Beaudry, C. Penner, V. Strgar,
99. L. Skuja, “Optically active oxygen-deficiency-related R. Lee, C. Duzenli, “Characterization of the exradin
centers in amorphous silicon dioxide,” Journal of Non- W1 plastic scintillation detector for small field appli-
Crystalline Solids 239(1–3), 16–48 (1998). cations in proton therapy,” Physics in Medicine and
100. T. Suzuki, L. Skuja, K. Kajihara, M. Hirano, T. Kamiya, Biology 63(9), 095016 (2018).
H. Hosono, “Electronic structure of oxygen dangling 108. L. Archambault and M. Rilling, “Basic quality assur-
bond in glassy SiO2: The role of hyperconjugation,” ance: Profiles and depth dose curves,” in Scintillation
Physical Review Letters 90(18), 186404 (2003). Dosimetry, edited by S. Beddar and L. Beaulieu (CRC
101. A. R. Beierholm, C. F. Behrens, C. E. Andersen, Press, Boca Raton, FL, 2016), Chap. 6, pp. 87–104.
“Dosimetric characterization of the Exradin W1
Chapter 10
Cherenkov and Scintillation

Imaging Dosimetry
Rachael L. Hachadorian*, Irwin I. Tendler *, and Brian W. Pogue
Dartmouth College,
CONTENTS
10.2 Cherenkov Light 140
10.3 Scintillation Light 141
10.3.1 1D–Point Measurement 142
10.3.2 2D–Surface and Planar Measurements 142
10.3.3 3D–Volumetric Measurements 142
10.4 Linac Pulsing and Time-Gated Real-Time Imaging 143
10.5 Strengths and Weaknesses of Cherenkov Imaging 145
10.5.1 Strengths 145
10.5.2 Weaknesses 146
10.6 Strengths and Weaknesses of Scintillation Imaging 146
10.6.1 Strengths 146
10.6.2 Weaknesses 147
10.7 Summary 147
Acknowledgments 147
References 148
10.1 INTRODUCTION on the deposited dose. With the appropriate time-gating

The use of radiotherapy in medicine has been prac- and sensitivity, this can provide the potential for video
ticed with the concept that the radiation beam and dose rate information to the clinical physics or radiotherapy
delivery could only be visualized by direct use of some teams. This potential is still developing as optimal camera
measurement probe or device, which samples the field. choices are refined and optical imaging geometries are
Imaging via film or external portal imaging devices is tested in clinical trials. The choice of the type of radio-
most common; however, these inherently sample the field luminescence signal to image will alter how the imaging
and not the dose delivery to the subject. The use of light is done, and what the application would be. In this chap-
emission from radiation can also provide a signal, which ter, the use of imaging to capture Cherenkov emission
can be used in real-time visualization of the beam in 2D or scintillation emission is discussed from fundamental
and 3D, and when planned carefully can directly report emission characteristics to how the signal can be used in
both quality audit and delivery verification applications.
Perhaps the leading discovery in recent years to make
* Equal Contribution Authors this type of imaging more attractive was that the use of
139
FIGURE 10.1 An illustration of the signal processing steps that go into capturing a cleaned up real-time signal from Cherenkov
or scintillation light imaging, starting with (1) time-synchronized detection, (2) followed by amplification in this case by a
microchannel plate photomultiplier tube (MCP-PMT), (3) followed by salt noise removal by median filtering in space and
time, and finally (4) background subtraction, to suppress any residual ambient light.
camera time-gating synchronized to the linear accel- range and transfers a substantial fraction of its energy
erator (linac) pulses allows for rejection of much of the to the secondary electrons within the host medium
ambient room lighting and capture of light just dur- via Compton scatter interactions. These liberated elec-
ing the linac pulse, as illustrated in Figure 10.1. This trons are scattered into velocities up to 0.75c, which in a
approach to synchronized imaging, when combined medium with index n > 1.33 makes them faster than the
with an amplified system, will allow for real-time video speed of light in water. The speed of light in the medium
rate capture and could be achieved in ambient room is cm = c/n, where c is the speed of light in a vacuum and
lighting [1]. This discovery has translated into a series n is the index [4].
of practical tools, which now allow for patient imaging As high energy electrons travel through a dielectric
of the radiotherapy beam delivery and accurate dose medium, they polarize and disrupt the local electric
quantitation [2]. The time synchronization of this and fields, causing a torque to the materials polarization,
the potential and limitations are outlined here. via the valence electrons. The electromagnetic inter-
Finally, the choice of using Cherenkov or scintillation action between the relativistic electron velocity and
luminescence to image translates into different applica- the dipoles of the medium leads to a traveling wave
tions because the quality of the signal intensity and lin- emission, as estimated from semiclassical electromag-
earity are different. Cherenkov imaging can provide the netic theory. An optical wavefront is emitted at an
signal directly from a patient’s tissue, inherently pro- angle of approximately 41 degrees from the direction
viding a map of the dose directly on the subject, albeit of the traveling electron in water, but this cone of solid
somewhat altered by the tissue optical properties of the angle varies proportionally with the velocity of the
patient [3]. Whereas scintillation imaging provides a particle and the index of the medium [1]. Cherenkov
brighter signal which is known to be highly linear with light exists over a wide spectrum of wavelengths from
dose. So, there are strengths and weaknesses associated ultraviolet (UV) through to near infrared (NIR), span-
with both Cherenkov and scintillation imaging, but ning the spectrum over an inverse square relationship
interestingly, the technology to image these two signals (I = c/ λ2) (Figure 10.2). However self-attenuation by
is substantially similar. This is a field that is still emerg- the medium leads to a loss of shorter UV photons,
ing, and each tool could turn into either a useful cali- and longer NIR photons, leaving detectable emission
bration tool or a time-saving clinical verification tool. largely in the UV-A to visible to short wave NIR wave-
The range of methods to image and the applications are lengths. By this inverse nature relationship, the shorter
outlined here. wavelengths (blue end) result in the highest intensi-
ties. The product is the well-recognized blue glow
10.2 CHERENKOV LIGHT commonly associated with it and can be seen visually
Cherenkov emission is an optical radiation by-product, within water in a nuclear reactor pool. It is important
produced when an electron travels faster than the phase to note though that because the light is broadband in
velocity of light in a dielectric medium. In the context nature, the spectrum exiting any medium is distorted
of radiation therapy, a linear accelerator delivers the by the absorption within that medium and the path of
treatment photon beam within the mega-voltage (MV) travel to leave it.
Cherenkov and Scintillation Imaging Dosimetry ◾ 141
FIGURE 10.2 (a) In radiotherapy, sufficient energy can be transferred to electrons within the tissue to liberate it via Compton
scatter, and cause it to travel faster than the velocity of light in water (v > cm). These high-speed electrons polarize the medium
transiently, and when the locally disrupted electric field in the medium begins to relax, an optical traveling wave of Cherenkov
light is emitted. (b) Cherenkov light is composed of a broad spectrum of wavelengths, from UV through visible to the infrared.
This emission spectrum occurs with an inverse square dependence upon wavelength, however within tissue blood absorbs the
shorter (blue) wavelengths of light, which substantially shifts the distribution of emitted wavelengths to red and NIR.
Within tissue, the blood in capillaries throughout the medium being the conduit for radiation interac-
the volume is a primary absorber of Cherenkov light. tion and electron liberation, and the fluorophore being
Hemoglobin preferentially absorbs out shorter wave- the electron/energy acceptor. A secondary fluorophore
lengths of the visible light spectrum, thus serving as a (a material that will absorb light at the lower end of the
high wavelength passing filter [5]. The smooth, inverse visible light spectrum) is sometimes added to shift the
square relationship is radically altered, with the result emission spectra to higher wavelengths [10]. Specifically,
that the red and NIR wavelengths will make it to the scintillators emit light when the excited organic fluoro-
surface of the tissue without being self-attenuated. In phore de-excites via a fluorescence, phosphorescence, or
some papers, patients reported having observed blue some alternate delayed fluorescence pathway [11].
light emitted throughout a stereotactic brain treatment. There exist two main categories of materials used for
This was most likely a product of Cherenkov in the vit- scintillation dosimetry: organic and inorganic. Popular
reous humor of the eye [6–8]. plastics (organic, usually near water-equivalent) used as
Because the Cherenkov light travels no more than base, bulk solvent, materials include polyvinyl toluene
several millimeters in vivo, there exists good potential (PVT) and polystyrene (PS); these materials are shaped
for accurate field verification in clinical imaging, for into volumes and fibers. A commercial example of a
robust mapping of the exact regions in which the radia- PVT-based scintillator is Saint Gobain BC-400, while
tion is delivered on the patient’s tissue. Saint Gobain BCF-12 is PS-based. These scintillators
are doped with proprietary organic luminescent agents
10.3 SCINTILLATION LIGHT termed fluors. Some examples of organic fluors used
Scintillation is a luminescence process – molecular in plastic scintillator dosimeters include anthracene
excitation caused by absorption of ionizing radiation and 3-hydrofluoride. Additionally, certain scintillating
results in emission of light during material electronic materials can be dissolved in solution, thus creating a
relaxation. Numerous studies have shown that under liquid volume which scintillates. The optical detection
appropriate conditions, scintillation signal can be of scintillation signal is enabled by the large photon out-
directly proportional to dose [9]. Plastic scintillators put yield associated with these materials. In the case of
used for dosimetry purposes are often composed of liquid scintillators, p-terphenyl and 2,5-diphenyloxazole
a bulk medium and organic fluorophores, where light (PPO) are common organic fluors, which can be incor-
production is enabled through fluorescence resonance porated into solution. Other dosimetry systems, such as
energy transfer between these two components, with the Saint Gobain PreLude 420 (cerium-doped lutetium)
of total skin electron therapy (TSET). Thin scintillator

discs attached to the skin surface of patients undergo-
ing TSET; by time-gating an intensified charge-coupled
device (ICCD) camera to linac pulses, light output from
these plastic discs was captured and related to surface
dose remotely and in a rapid manner [18, 19].
10.3.2 2D–Surface and Planar Measurements

Optical imaging of scintillators has been used to conduct
two-dimensional dosimetry in a variety of different treat-
ment regimes [20]. For example, it has been shown that
FIGURE 10.3 Emission spectra of various scintillators and
2D visualization of a proton beam is possible using only a
organic fluors: plastic (dark and light blue) and crystal (dark
commercial digital camera and solid plastic scintillator –
and light green) scintillators; solid (gray and black) and liquid
(orange and red) organic fluors. this can be useful for computing quenching correction
factors and conducting quality assurance beam testing,
see Figure 10.4B [21]. Detection of scintillator emission
and Polysin NaI(TI) (sodium iodide-doped thallium), has also been applied to the task of measuring surface
utilize inorganic scintillating crystals. dose during EBRT. For example, Jenkins et al. created
As part of developing an imaging-based scintilla- a flexible, thin, scintillating film using silicone and
tor dosimeter, selecting an appropriate scintillating Gd2O2S:Tb (GOS); the sheet was designed to be placed
material is highly dependent on the target application. over a patient’s skin surface during radiation therapy
Considering the wavelength of maximum emission is to track beam movement in real time [22]. Imaging of
important to ensure high detection efficiency, and here liquid scintillators has also been used for the purposes
Figure 10.3 shows the range of emission spectra, for of 2D dosimetry. Optical imaging of a light-shielded
example scintillators listed above. acrylic tank filled liquid scintillator via electron multi-
Furthermore, criteria such as scintillator response plying charge coupled device has been proven to accu-
time (temporal rise and decay of signal) and geometry rately (compared to treatment plans) measure depth
(dimensions and shape) are other examples of charac- dose lateral profiles and 2D dose distribution for thera-
teristics one would consider when designing a scintil- peutic photon beams [23].
lator dosimeter [12]. Examples of scintillator imaging
dosimetry systems (1D, 2D, and 3D) are listed below. 10.3.3 3D–Volumetric Measurements
Researchers have taken advantage of scintillator volumes
10.3.1 1D–Point Measurement for visualizing dose deposition in three dimensions. In
Scintillators have long been used for point dosimetry; the world of proton beam dosimetry, recent innovations
many of these scintillators follow a design first suggested have been made to enable to the construction of a real-
by Beddar et al. in 1992: a small (1 × 1 × 1 mm3)-sized time 3D system capable of tracking dynamic spot beam
scintillator is coupled to a fiber optic cable. Upon exci- delivery and potentially useful for future 3D dose map
tation, scintillator light emission is directed into the reconstruction. It should also be noted that this system is
cable and travels to a receiver, see Figure 10.4A. This capable of high-precision measurements in the low-dose
signal is converted from analog to digital and under- detection regime [24]. To visualize real-time dose depo-
goes amplification, filtering, and other processes [13, 14]. sition in 3D, researchers imaged a volume of scintillator
Modernized versions of this technology (incorporating material placed within the beam path. Utilizing a ple-
better Cherenkov elimination techniques and a more noptic camera, linac portal imaging device, and tomo-
robust design compared to earlier versions) have been graphic image reconstruction algorithms, a full, live, 3D
evaluated for use in proton, photon, and electron dosim- reconstruction of dose distribution was achieved [25].
etry (both for QA and in vivo dosimetry) [15-17]. Optical More recently, a plastic scintillator array (with elements
imaging of scintillator emission for point dose measure- extending in 3D) was imaged using a shielded comple-
ments has also recently been utilized within the context mentary metal-oxide-semiconductor (CMOS) camera,
FIGURE 10.4 Examples of 1D [13, 14], 2D [21], and 3D [26] setups used for optical imaging of scintillators. The 1D geometry
is for point sampling as used for point dosimetry by Beddar et al. [13, 14]. Planar measurements, B, have been used to profile
proton beams for maximum peak depth range analysis by Almurayshid et al. [21]. Finally, 3D imaging can be used with either
rotational or angular resolved imaging of the camera, combined with algorithms for tomographic recovery, as shown by a
number of groups in very different geometries [24–26].
where light was directed into the camera body using the high voltage, which amplifies the electron signal
a periscopic mirror and was presented for use in real- through the channels. This image intensifier compo-
time commissioning and verification of dynamic radio- nent is critical because it increases the time-resolution
therapy treatments, see Figure 10.4C [26]. Each of these capabilities of the optical shutter down to the picosec-
methods is a research study, and it is not obvious, which ond scale. Other types of time gates such as mechani-
methodology would gain long range adoption, but each cal shutters or electronic global shutters operate on the
has its strengths and weaknesses depending a lot on the millisecond timescale, and therefore are not sufficient
application need. for this fast gating. A unique requirement for these sci-
entific cameras is then to gate during the linac pulses,
10.4 LINAC PULSING AND TIME- but then to integrate these pulses on the camera sensor,
GATED REAL-TIME IMAGING which images the back end of the MCP-PMT, where a
While Cherenkov and scintillation imaging has been phosphor screen turns the electron image back to an
around for decades, much of the work published has optical image. Depending on the camera/sensor type
focused around traditional cameras [charge-coupled and software, the user may specify a number of param-
device (CCD) and CMOS] with frame rates near 1–100 Hz, eters, for example, the number of pulse accumulations
and long duty cycles [27, 28]. Pulsed or time-gated to be integrated together, prior to readout of the detec-
imaging has allowed Cherenkov and scintillation tor. Triggering of the camera to the linac pulsations is a
imaging with an acquisition system (camera) that is critical part of the functioning and allows the camera to
synchronized to the linear accelerator pulses. This can reject the majority of the ambient light. One Cherenkov
be achieved over an optimized time range where the or scintillation image therefore can consist of several
emitted light is emitted from the patient tissue or other pulses from the linac, as shown in Figure 10.5, followed
source and captured at video frame rates. Considering by one background exposure. This enables the user to
that a linear accelerator emits ionizing radiation in subtract the background light, in real time, thus iso-
short 3–5 microsecond pulses at repetition rates driven lating the Cherenkov light from the image [18]. When
by the klystron near 100–400 Hz, the camera can be attempting to capture Cherenkov or scintillation emis-
calibrated to acquire images during beam on via a fast sion, the goal must be to collect data in a time window
triggering mechanism. The fastest image gates tend to directly following a linac pulse, since both are rapid
be multichannel plate photomultiplier tubes (MCP- processes. By isolating the same time window over a
PMTs), often called image intensifiers. The image consistent trigger, the methodology adds reliability and
intensifier is turned on and off by electronic gating of robustness to the clinical acquisition process.
FIGURE 10.5 The linear accelerator is depicted on the top of the figure above, where the gantry rotates about a fixed isocenter.
The camera may be mounted to the ceiling for most consistent placement, but may also be mounted to a tripod for temporary
use. Below, the camera activity is illustrated in response to the linac pulses. Once a threshold voltage is reached, the intensi-
fier turns on and remains on over the duration of the intensifier gate. The Cherenkov + scintillation image is acquired and
both spatial and temporal median-filtered over several frames, followed by one background exposure. Before the background
exposure is collected, a short gate delay is employed.
When the camera is triggered using the pulsed radia- is advantageous for accurate and precise time-gated
tion delivery of the linac, this is known as an external imaging. For example, by setting the imaging system
trigger. The camera requires the signal from the linear to trigger of a rise in klystron voltage, one can cap-
accelerator (an external device) in order to initialize ture scintillator emission within nanoseconds after a
acquisition, as previously discussed. However, it is pos- linac pulse leaving a remaining time gap, which can
sible for the operator to have control over the time gate be used to collect background signal (image collected
over which Cherenkov is acquired. In contrast, this is when there is no active radiation field, e.g., in between
known as internal triggering, and employs an “always linac pulses) [29].
on” or continuous image acquisition mode. Simply put, Determining which camera hardware to utilize
the camera is turned on by the user; all optical output is in conducting imaging-based dosimetry is critical to
detected and recorded, then turned off by the user. This achieving accurate and precise measurements within
becomes useful for calibrating, focusing, for previewing a useful timeframe. Examples of camera types avail-
the field before the Cherenkov image is acquired, as well able, and commonly used in radiotherapy imaging, are
as for imaging sources of optical output that are not nec- CMOS, CCD, electron multiplying-intensified charge
essarily Cherenkov light. coupled device (EM-ICCD), and ICCD [30]. The cost of
Many scintillators in the field of imaging dosimetry these camera systems ranges in magnitude, thus, identi-
are selected to have the material property of fast rise fying and selecting the correct imaging criteria is impor-
and decay times (order of single nanoseconds); this tant. For example, one should consider the importance
FIGURE 10.6 (A–C) Cherenkov imaging shows the beam delivery in total skin electron therapy to a patient, with the image
sequence showing the total dose developed from top and bottom beams and them combined. (D–F) Scintillation imaging of
small local plastic scintillating discs show quantitative measures of dose, which can be captured with the same camera set up.
of low light sensitivity, desired frame rate, and back- No Extraneous Dose: Unlike diagnostic beams (such
ground light suppression when choosing equipment for as radiographs and CT scans), which deliver kV-range
Cherenkov imaging (Figure 10.6) [31]. radiation to large-scale body regions to gain informa-
tion, Cherenkov emission is a secondary radiation, which
10.5 STRENGTHS AND WEAKNESSES essentially yields “free information,” and comes with no
OF CHERENKOV IMAGING adverse side effects to the patient, namely dose [35].
10.5.1 Strengths Nuclear Medicine Applications: In recent research
Real-Time Field Verification: The immediate Cherenkov developments, it has been shown that luminescent
signal from linac pulses allows for image capture in real compounds can be strategically selected or manufac-
time. Therefore, the imaging is done while the patient is tured such that their excitation wavelengths exist in the
on the couch, without extra time needed [32]. Conebeam Cherenkov spectrum around some of the most abun-
CT or portal imaging provides different diagnostic dant Cherenkov emission wavelengths [36]. The phos-
information during the patient exam, the nature of the phorescence can then emitted over a longer time scale,
information gained with Cherenkov is more related to long enough to be imaged much like a radiotracer in
the deposited dose, instead of position and transmit- Cherenkov-Excited Luminescence Scanned Imaging
ted beam. Additionally, current software in use allows (CELSI) [37, 38]. The ability to trace a metabolic pathway
the team to watch the beam throughout the treatment, or any object therein is a common use for nuclear medi-
and quantitatively assess the treatment immediately cine modalities such as PET and SPECT, which require
postdelivery for patient shifts, day-to-day discrepancies additional modalities. If a patient were concurrently
between treatments, and therefore potentially capturing being treated using radiation, further development of
incidents or helping determine when adaptive planning CELSI would eliminate the need for the radiation asso-
is needed [33, 34]. ciate with an additional scan [39].
Real-Time in Vivo Dose Verification: A cumula- Strengths Weaknesses

tive Cherenkov image is the sum of all the fields of a Real-time signal from the actual Line-of-sight imaging only
given delivery. This image is what can be compared dose deposition (video rate)
to the treatment plan. As previously mentioned, Potential for in vivo tissue Surface-limited measurement
verification mapping (<10 mm deep)
research is being carried out to correct for the dis-
Intuitive visualization of procedure Limited to photons and
crepancies caused by tissue optical properties, beam for therapy team electrons
entrance and exit differences, and a number of other Molecular sensing potential, Limitations for exceptionally
parameters to establish linearity between dose and excited by Cherenkov in situ bright illumination rooms
Cherenkov light in nonhomogeneous media. Once
these parameters have been corrected for, real-time and 10.6 STRENGTHS AND WEAKNESSES
in vivo dose information is feasibly gained. Once OF SCINTILLATION IMAGING
finalized, verifying pixel-by-pixel dose immediately 10.6.1 Strengths
after a treatment fraction may become one of the Mature technology of scintillators: The field of scintillator
most promising aspects of a cumulative Cherenkov measurement has been advanced now with scintillators
image [40]. that have physical properties that match goal of con-
ducting quantitative optical dosimetry. Due to the fact
10.5.2 Weaknesses that large classes of scintillators have been developed
Surface-Limited Dosimetry: The optical wavelengths as “water equivalent” and can be produced in small
of Cherenkov emission limit its propagation inside in sizes, they can be designed to minimally disturb
the patient to a few millimeters, by the exponential the applied radiation field. Furthermore, the dose rate
attenuation with distance. This is its primary weak- independency and linear response to dose for most use-
ness because the lower energies have such a high prob- ful scintillators have been optimized over the years for
ability of being attenuated out by the time the photon potential dosimeter use. So, within the scope of imag-
reaches the surface. As a result, the most probable ing-based dosimetry systems, the high photon yield is
average sampling depth is <10 mm, depending upon provided by scintillator emission, and so this matches
the tissue type [41]. As a consequence, real-time field the concept of remote dosimetry with good signal-to-
verification (and eventually in vivo dose information) noise for camera based detection [44].
can only be used at subcutaneous depths. Therefore, Stability of signal and linearity with dose: Multiple
if the tumor or region in question is not at or close publications have shown that water-equivalent scin-
to the surface, information may only be gained indi- tillators are capable of accurately measuring dose in
rectly, from what is observable at the surface. Thus far, a small-field radiotherapy setting. The ability to pre-
the linearity established between absorbed dose and vent field perturbation and avoiding volume averag-
Cherenkov emission in a homogeneous and transpar- ing effects in converting signal to dose are just a few
ent medium implies that the imaged surface emission of the criteria that make optical detection of scintil-
reflects the deposited dose at the surface in homoge- lation effective for small-field dosimetry [45]. These
neous regions. characteristics have also enabled the optical imaging
Limited to MV Photon and Electron Therapy Beams: of scintillator emission to be used for dosimetry in
While photons and electrons constitute the major- brachytherapy treatments [46, 47]. Plastic scintil-
ity of radiotherapy treatments currently, it neglects lators have been shown to have strong resistance to
proton and heavy ion beams. Although the origin of radiation damage and, in certain cases, can effec-
Cherenkov comes from a relativistic charged particle, tively function independent of temperature (after
the observable effect for proton beams was substan- application of a linear correction factor) temperature
tially diminished by comparison to photon and elec- independence [15, 48-51].
tron beams. Darafsheh et al. demonstrated that the Insensitivity to optical properties: As compared to
visible signal responsible for proton therapy dosim- Cherenkov imaging, scintillation dosimetry is inde-
etry using bare optical fibers is not Cherenkov radia- pendent of tissue optical properties; converting light
tion, yet still the value of beam range finding in this signal to dose is significantly more straightforward
application is compelling [42, 43]. when tissue scattering and absorption interactions do
not need to be accounted for. As such, direct imaging Strengths Weaknesses

of scintillators instead of Cherenkov has some signifi- Linear with dose & dose rate Line-of-sight imaging
cant value [52]. Versatile to different geometries Cherenkov from fibers in the
field can be a confounding
10.6.2 Weaknesses signal
Water equivalent scintillators Slight angular dependence of
Complications of mixing Cherenkov & scintillation: available signal for remote imaging
Single channel detection for achieving high SNR has Independent of energy for certain Distance dependence for
been achieved with PIN photodiodes for fast perfor- scintillators remote imaging
mance at modest cost, while photomultiplier tubes are Independent of environmental
factors such as temperature
the optimal choice for high-gradient radiotherapy due
(following linear correction
to their superior signal sensitivity and larger area [53]. factor), humidity, and pressure
Multiple systems incorporating a small scintillator tip
attached to a fiber optic cable for signal transport have 10.7 SUMMARY
been described in the literature (see Chapter 9). The
The use and practical value of scintillation and/or
greatest issue these types of dosimetry systems face is
Cherenkov imaging are still being determined as
generation of Cherenkov within body of the fiber optic
research studies emerge. There is a range of techno-
cable; in this case, Cherenkov signal is a low-level con-
logical realizations of both, which are being evaluated
tamination to the scintillator output. Numerous solu-
both as devices and as methods for linac QA, pre-treat-
tions have been created to combat this issue; these
ment QA, and during treatment verification. Much
include, but are not limited to: basic optical filtering,
of the device development studies have been around
two-fiber subtraction, chromatic removal, and multi-
which different camera technologies to use, how to
spectral removal [11, 54]. Some groups have attempted
acquire, and to determine if the imaging needs to be
to solve the problem of Cherenkov contamination in
done in the setting of normal ambient room lighting.
systems, which use fibers in pulsed radiation environ-
The use of time-gated imaging is quite specialized,
ments using removal via temporal filtering; others have
and only a few systems allow for this, but it solves
focused on removing Cherenkov signal generated in liq-
the problem of being able to image in the presence of
uid scintillators [55, 56]. While this is a slight weakness,
ambient room lighting. The technological systems are
the methods to separate Cherenkov and scintillation
still evolving and commercial translation at this point
have been well developed at this point.
is still embryonic, but this should be changing in the
Direct View Limitations: In the case of in vivo dosim-
very near future.
etry, one of the greatest advantages of imaging scintil-
Linac or treatment plan QA is the most logical start-
lator dosimeters lies in the fact that no wires or signal
ing point for these technologies, and while the range of
transduction cables are necessary – scintillator output is
devices available for this are many, each has its strengths
detected remotely via camera. This fundamental method
and weaknesses. The strengths of Cherenkov or scintil-
of signal detection lends well to streamlining and auto-
lation sensing are the real-time nature of the data, and
mating dosimetry efforts. However, one of greatest
when imaging, the value of remote sensing is important.
drawbacks is that the dosimeter must be in direct view
Human measurement for treatment verification is
of the camera. As mentioned previously, many methods
now widely used for scintillation fiber dosimeters, but
avoid this issue by using a fiber optic; however, this elim-
beyond that, most efforts are in early-stage clinical trials.
inates the possibility of wireless signal detection and
results in Cherenkov generation within the body of the
fiber. Solution employing the use of multiple cameras, ACKNOWLEDGMENTS
mirrors, or unique image reconstruction techniques, The authors gratefully acknowledge collaboration in
and other technical methods has been described in the all discussions and imaging with other members of the
literature [57, 58]. It should also be noted that potential Radiation Oncology and Optics in Medicine research
angular and distance dependencies must be accounted groups at Dartmouth. Additionally, funding from this
for when dealing with wireless camera-based imaging work comes from NIH grant R01 EB023909 as well as
of scintillator emission. equipment and expertise from DoseOptics LLC.
REFERENCES 14. A. S. Beddar, T. R. Mackie, F. H. Attix, “Water-equivalent

plastic scintillation detectors for high-energy beam
1. A. K. Glaser, R. Zhang, S. C. Davis, D. J. Gladstone, B. dosimetry: II. Properties and measurements,” Physics in
W. Pogue, “Time-gated Cherenkov emission spectros- Medicine and Biology 37, 1901 (1992).
copy from linear accelerator irradiation of tissue phan- 15. P. Carrasco, N. Jornet, O. Jordi, M. Lizondo,
toms,” Optics Letters 37, 1193–1195 (2012). A. Latorre-Musoll, T. Eudaldo, M. Ruiz, M. Ribas,
2. C. Snyder, B. W. Pogue, M. Jermyn, I. Tendler, J. M. “Characterization of the exradin W1 scintillator for use
Andreozzi, “Algorithm development for intrafraction in radiotherapy: Characterization of exradin W1 for
radiotherapy beam edge verification from Cherenkov radiotherapy,” Medical Physics 42, 297–304 (2016).
imaging,” Journal of Medical Imaging 5, 015001 (2018). 16. C. Hoehr, C. Lindsay, J. Beaudry, C. Penner, V. Strgar,
3. R. Hachadorian, P. Bruza, M. Jermyn, A. Mazhar, D. R. Lee, C. Duzenli, “Characterization of the exradin W1
Cuccia, L. Jarvis, D. Gladstone, B. Pogue, “Correcting plastic scintillation detector for small field applications
Cherenkov light attenuation in tissue using spatial fre- in proton therapy,” Physics in Medicine and Biology 63,
quency domain imaging for quantitative surface dosim- 095016 (2018).
etry during whole breast radiation therapy,” Journal of 17. M. Rahman, P. Brůža, K. M. Langen, D. J. Gladstone,
Biomedical Optics 24(7), 071609 (2018). X. Cao, B. W. Pogue, and R. Zhang, “Characterization
4. A. K. Glaser, R. Zhang, D. J. Gladstone, B. W. Pogue, of a new scintillation imaging system for proton pencil
“Optical dosimetry of radiotherapy beams using beam dose rate measurements,” Physics in Medicine &
Cherenkov radiation: The relationship between light Biology 65(16), 165014 (2020).
emission and dose,” Physics in Medicine and Biology 59, 18. P. Bruza, S. Gollub, J. Andreozzi, I. Tendler,
3789 (2014). B. Williams, L. Jarvis, D. Gladstone, B.W. Pogue, “Time-
5. W. G. Zijlstra, A. Buursma, W. P. M. der Roest, gated scintillator imaging for real-time optical sur-
“Absorption spectra of human fetal and adult oxyhe- face dosimetry in total skin electron therapy,” Physics
moglobin, de-oxyhemoglobin, carboxyhemoglobin, in Medicine and Biology 63, 095009 (2018). https://doi.
and methemoglobin,” Clinical Chemistry 37, 1633–1638 org/10.1088/1361-6560/aaba19
(1991). 19. I. Tendler, P. Brůža, J. Andreozzi, M. Jermyn, B.
6. F. Newman, M. Asadi-Zeydabadis, V. Durairaj, Williams, L. Jarvis, B. Pogue, and D. Gladstone, “Rapid
M. Ding, K. Stuhr, B. Kavanagh, “Visual sensations dur- multisite remote surface dosimetry for total skin elec-
ing megavoltage radiotherapy to the orbit attributable to tron therapy: scintillator target imaging,” International
Cherenkov radiation,” Medical Physics 35, 77–80 (2008). Journal of Radiation Oncology, Biology, Physics 103(3),
7. G. G. Fazio, J. V. Jelley, W. N. Charman, “Generation 767–774(2019).
of Cherenkov light flashes by cosmic radiation within 20. D. A. Alexander, R. Zhang, P. Brůža, B. W. Pogue, and
the eyes of the Apollo astronauts,” Nature 228, 260–264 D. J. Gladstone, “Scintillation imaging as a high‐resolu-
(1970). tion, remote, versatile 2D detection system for MR‐linac
8. D. T. Blumenthal, B. W. Corn, N. Shtraus, “Flashes of quality assurance,” Medical Physics 47(9) 3861–3869
light-radiation therapy to the brain,” Radiotherapy & (2020).
Oncology 116, 331–333 (2015). 21. M. Almurayshid, Y. Helo, A. Kacperek, J. Griffiths,
9. A. S. Beddar, “Plastic scintillation dosimetry and its J. Hebden, A. Gibson, “Quality assurance in proton
application to radiotherapy,” Radiation Measurements beam therapy using a plastic scintillator and a com-
41, S124–S133 (2006). mercially available digital camera,” Journal of Applied
10. S. F. de Boer, A. S. Beddar, J. A. Rawlinson, “Optical fil- Clinical Medical Physics 18, 210–219 (2017).
tering and spectral measurements of radiation-induced 22. C. H. Jenkins, D. J. Naczynski, S.-J. S. Yu, L. Xing,
light in plastic scintillation dosimetry,” Physics in “Monitoring external beam radiotherapy using real-time
Medicine and Biology 38, 945 (1993). beam visualization,” Medical Physics 42, 5–13 (2015).
11. L. Beaulieu, S. Beddar, “Review of plastic and liquid 23. F. Pönisch, L. Archambault, T. Briere, N. Sahoo,
scintillation dosimetry for photon, electron, and pro- R. Mohan, S. Beddar, M. Gillin, “Liquid scintillator for
ton therapy,” Physics in Medicine and Biology 61, R305– 2D dosimetry for high-energy photon beams,” Medical
R343 (2016). Physics 36, 1478–1485 (2009).
12. I. Tendler, Quantitative Scintillation Imaging for 24. C. D. Darne, F. Alsanea, D. G. Robertson, N. Sahoo,
Dose Verification and Quality Assurance Testing in S. Beddar, “Performance characterization of a 3D liquid
Radiotherapy (Ph.D. dissertation, Dartmouth College, scintillation detector for discrete spot scanning proton
2020). beam systems,” Physics in Medicine and Biology 62,
13. A. S. Beddar, T. R. Mackie, F. H. Attix, “Water- 5652–5667 (2017).
equivalent plastic scintillation detectors for high-energy 25. M. Goulet, M. Rilling, L. Gingras, S. Beddar, L. Beauliue,
beam dosimetry: I. Physical characteristics and theoret- L. Archambault, “Novel, full 3D scintillation dosimetry
ical considerations,” Physics in Medicine and Biology 37, using a static plenoptic camera,” Medical Physics 41,
1883 (1992). 082101 (2014).
26. M. W. Jennings, T. P. Rutten, D. J. Ottaway, “Evaluation of 38. X. Cao, S. R. Allu, S. Jiang, M. Jia, J. R. Gunn, C. Yao,
the signal quality of an inexpensive CMOS camera towards E. P. LaRochelle et al. “Tissue pO2 distributions in
imaging a high-resolution plastic scintillation detector xenograft tumors dynamically imaged by Cherenkov-
array,” Radiation Measurements 104, 22–31 (2017). excited phosphorescence during fractionated radiation
27. B. Robinson, “A new type of Cherenkov imaging detec- therapy,” Nature Communications 11, 573 (2020).
tor,” Physica Scripta 23, 716 (1981). 39. P. Brůža, H. Lin, S. Vinogradov, L. Jarvis, D. Gladstone,
28. A. L. Wintenberg, C. G. Moscone, J. P. Jones, G. R. B. W. Pogue, “Light sheet luminescence imaging with
Young, “A CMOS integrating amplifier for the PHENIX Cherenkov excitation in thick scattering media,” Optics
ring imaging Cherenkov detector,” IEEE Transactions Letters 41, 2986–2989 (2016).
on Nuclear Science 45, 758–763 (1998). 40. Hachadorian, R. L., P. Bruza, M. Jermyn, D. J. Gladstone,
29. M. Ashraf, M. Ramish, P. Bruza, V. Krishnaswamy, D. B. W. Pogue, and L. A. Jarvis, “Imaging radiation dose in
J. Gladstone, and B. W. Pogue, “Time‐gating to medi- breast radiotherapy by x-ray CT calibration of Cherenkov
cal linear accelerator pulses: Stray radiation detector,” light,” Nature Communications 11 (2020) 2298.
Medical physics 46(2), 1044–1048 (2019). 41. B. W. Pogue, R. Zhang, H. Lin, J. Gunn, A. Glaser,
30. J. M. Andreozzi, R. Zhang, A. Glaser, L. Jarvis, B.W. Pogue, J. Andreozzi, S. Jiang, D. J. Gladstone, L. A. Jarvis,
D. Gladston, “Camera selection for real-time in vivo radia- “Molecular Imaging through centimeters of tissue:
tion treatment verification systems using Cherenkov High resolution imaging with Cerenkov excitation,” in
imaging,” Medical Physics 42, 994–1004 (2015). Frontiers in Optics 2015 (2015), paper FW5E.1 (Optical
31. D. A. Alexander, P. Bruza, J. C. M. Farwell, V. Society of America, 2015). https://doi.org/10.1364/
Krishnaswamy, R. Zhang, D. J. Gladstone, and B. FIO.2015.FW5E.1
W. Pogue, “Detective quantum efficiency of intensi- 42. A. Darafsheh, R. Taleei, A. Kassaee, J. C. Finlay, “The
fied CMOS cameras for Cherenkov imaging in radio- visible signal responsible for proton therapy dosimetry
therapy,” Physics in Medicine and Biology 65(22), using bare optical fibers is not Čerenkov radiation,”
225013(2020). Medical Physics 43, 5973 (2016).
32. I. Tendler, A. Hartford, M. Jermyn, E. LaRochelle, X. 43. A. Darafsheh, R. Taleei, A. Kassaee, J. C. Finlay, “Proton
Cao, V. Borza, D. Alexander et al. “Experimentally therapy dosimetry using the scintillation of the silica
observed cherenkov light generation in the eye during fibers,” Optics Letters 42(4), 847-850 (2017).
radiation therapy,” International Journal of Radiation 44. L. Beaulieu, M. Goulet, L. Archambault, S. Beddar,
Oncology, Biology, Physics 106(2), 422–429 (2020). “Current status of scintillation dosimetry for megavolt-
33. M. Jermyn, L. Jarvis, S. Gollub, J. Andreozzi, T. Miao, age beams,” Journal of Physics: Conference Series 444,
P. Bruza, D. Gladstone, B. Pogue, “Cherenkov video 012013 (2013).
imaging during breast radiation therapy verifies sta- 45. S. Beddar and L. Beaulieu (eds.), Scintillation Dosimetry
ble beam shapes across treatment days,” International (CRC Press, Boca Raton, 2016).
Journal of Radiation Oncology, Biology, Physics 99, 46. G. Kertzscher and S. Beddar, “OC-0172: Development
S230 (2017). of an inorganic scintillation detector system for in vivo
34. L. A. Jarvis, R. A. Hachadorian, M. Jermyn, P. Bruza, dosimetry for brachytherapy,” Radiotherapy & Oncology
D. A. Alexander, I. I. Tendler, B. B. Williams et al. 127, S89–S90 (2018).
“Initial Clinical Experience of Cherenkov Imaging in 47. F. Therriault-Proulx, L. Beaulieu, S. Beddar, “Validation
EBRT Identifies Opportunities to Improve Treatment of plastic scintillation detectors for applications
Delivery,” International Journal of Radiation Oncology, in low-dose-rate brachytherapy,” Brachytherapy 16,
Biology, Physics (https://doi.org/10.1016/j.ijrobp.2020. 903–909 (2017).
11.013). 48. A. S. Beddar, “Water equivalent plastic scintillation
35. R. Zhang, D. Gladstone, L. Jarvis, R. Strawbridge, detectors in radiation therapy,” Radiation Protection
J. Hoopes, O. Friedman, A. Glaser, B.W. Pogue, “Real- Dosimetry 120, 1–6 (2006).
time in vivo Cherenkoscopy imaging during external 49. L. Wootton and S. Beddar, “Temperature dependence of
beam radiation therapy,” Journal of Biomedical Optics BCF plastic scintillation detectors,” Physics in Medicine
18, 110504 (2013). and Biology 58, 2955–2967 (2013).
36. X. Cao, S. R. Allu, S. Jiang, J. R. Gunn, C. Yao, J. Xin, P. 50. F. Alsanea, L. Wootton, N. Sahoo, R. Kudchadker,
Bruza et al. “High-Resolution pO2 Imaging Improves U. Mahmood, S. Beddar, “Exradin W1 plastic scintil-
Quantification of the Hypoxic Fraction in Tumors lation detector for in vivo skin dosimetry in passive
During Radiation therapy,” International Journal scattering proton therapy,” Physica Medica, European
of Radiation Oncology, Biology, Physics (https://doi. Journal of Medical Physics 47, 58–63 (2018).
org/10.1016/j.ijrobp.2020.09.046). 51. I. I. Tendler, P. Bruza, M. Jermyn, X. Cao, B. B.
37. R. Zhang, A. D’souza, J. Gunn, T. Esipova, Williams, L. A. Jarvis, B. W. Pogue, D. J. Gladstone,
S. Vinogradov, A. Glaser, L. Jarvis, D. Gladstone, “Characterization of a non-contact imaging scintillator-
B. W. Pogue, “Cherenkov-excited luminescence scanned based dosimetry system for total skin electron therapy,”
imaging,” Optics Letters 40, 827–830 (2015). Physics in Medicine and Biology 64(12), 125025 (2019).
52. I. I. Tendler, P. Bruza, M. Jermyn, A. Fleury, B. B. International Journal Devoted to the Applications of
Williams, L. A. Jarvis, B. W. Pogue, D. J. Gladstone, Physics to Medicine and Biology: Official Journal of the
“Improvements to an optical scintillator imaging-based Italian Association of Biomedical Physics (AIFB) 42,
tissue dosimetry system,” Journal of Biomedical Optics 185–188 (2017).
24(7), 075001 (2019). 56. J. Caravaca, F. B. Descamps, B. J. Land, M. Yeh,
53. J. Boivin, S. Beddar, M. Guillemette, L. Beaulieu, G. D. Orebi Gann, “Cherenkov and scintillation light
“Systematic evaluation of photodetector performance separation in organic liquid scintillators,” European
for plastic scintillation dosimetry,” Medical Physics 42, Physical Journal C 77, 811 (2017).
6211–6220 (2015). 57. A. S. Beddar, L. Archambault, D. Robertson, Large-
54. A. Darafsheh, J. E. Melzer, J. A. Harrington, A. Kassaee, volume scintillator detector for rapid real-time 3-D
J. C. Finlay, “Radiotherapy fiber dosimeter probes based dose imaging of advanced radiation therapy modalities
on silver-only coated hollow glass waveguides,” Journal (2017).
of Biomedical Optics 23(1), 015006 (2018). 58. J. Dalmasson, G. Gratta, A. Jamil, S. Kravitz, M. Malek,
55. J. Archer, L. Madden, E. Li, M. Carolan, M. Petasecca, K. Wells, J. Bentley, S. Steven, J. Su, “Distributed imag-
P. Metcalfe, A. Rosenfeld, “Temporally separating ing for liquid scintillation detectors,” Physical Review D
Cherenkov radiation in a scintillator probe exposed 97 (2018).
to a pulsed X-ray beam,” Physica Medica: PM: An
Chapter 11
Clinical Considerations and

Dosimeters for In Vivo Dosimetry
Douglas Bollinger
Philadelphia, Pennsylvania
Arash Darafsheh
St. Louis, Missouri
CONTENTS
11.2 Online Review Dosimeters 152
11.2.1 Diodes 152
11.2.2 Metal Oxide Semiconductor Field Effect Transistors (MOSFETs) 154
11.2.3 Electronic Portal Imaging Devices (EPIDs) 156
11.2.4 Scintillation Fiber Optic Dosimeters 158
11.2.5 Ion Chambers 159
11.3 Offline Review Dosimeters 160
11.3.1 Film Dosimeters 160
11.3.2 Radiophotoluminescent Dosimeters (RPLDs) 162
11.3.3 OSLDs and TLDs 163
11.4 Summary 164
References 166
11.1 INTRODUCTION machine maintenance to pre-treatment patient-specific

Quality assurance (QA), as defined by American measurements. The form of QA we are concerned with
Association of Physicists in Medicine (AAPM)’s Task in this chapter is in vivo dosimetry. In vivo dosimetry
Group (TG)-100 [1], “confirms the desired level of qual- is a powerful tool to verify dose delivery accuracy. By
ity by demonstrating that the quality goals for a task or placing a radiation detector at a point of interest, we
parameter are met.” In practice, this means confirma- can validate that the expected dose is delivered to the
tion that the various aspects of the radiation therapy patient (to within some tolerance). However, in order to
system are able to perform at the level required to cor- get an accurate estimate of dose, it is critical to under-
rectly predict and deliver a given dose distribution to stand the abilities as well as the limitations of differ-
a stated target. QA comes in many forms from routine ent dosimeters. Not all dosimeters are appropriate in
151
all clinical situations and different dosimeters will give If the detector is to be placed perpendicular to the
the operator different information. For example, the incoming beam(s), however, then a hemispherical diode
readings from some dosimeters are often represented may be appropriate. Diodes also differ in the material
as point measurements while others lend themselves and thickness of the build-up cap. Ideally, there would
to two-dimensional (2D) planar data. An alternative be a different build-up cap for every clinical energy cor-
method of classifying dosimeters is by the availability of responding to a different depth of dmax. This, however, is
the readings. Some dosimeters provide readings that are not practical in many cases and indeed, having a large
instantaneously available (online review) while others variety of build-up caps can increase the complexity and
produce data that need to be analyzed following treat- potential for error with in vivo dosimetry. A common
ment in order to extract a dose reading (offline review). compromise is to have one set of build-up caps for low
This chapter begins by discussing dosimeters that can energy photons (≤12 MV), one for high energy photons
be used for online review, while the second section is (>12 MV), and one for the electron beams.
devoted to those designed for offline review. Where pos- Diodes must be calibrated before their use for in
sible, a wide variety of clinical examples of the uses of vivo measurement. In addition, as diodes are known to
these dosimeters are included. degrade with radiation damage, this calibration should
be periodically checked and, if necessary, repeated.
Diodes may be used for both photon and electron
11.2 ONLINE REVIEW DOSIMETERS dosimetry. For electron beams, diodes have the addi-
11.2.1 Diodes tional advantage of low energy dependence due to the
Diodes have long been one of the most popular dosime- fact that the water to silicon stopping power ratio is rela-
ters for in vivo measurements due to their instantaneous tively independent of electron energy [9].
read-out ability, small physical size, and high sensitiv- There are many examples available in the literature
ity. However, their response to radiation suffers from a for the use of diodes for entrance and exit dose mea-
variety of dependencies, including temperature, angle surements [10–12]. One such example of diodes in clini-
of incidence of radiation, dose rate, source-to-surface cal practice is provided by Alecu et al., who reported on
distance (SSD), dose per pulse, integrated dose, radia- their use to record the exit and entrance dose for more
tion type, and energy [2–5]. As such, it is important than 300 pelvic treatments [13]. They calibrated their
to evaluate the significance of each of these factors for diodes against ion chamber readings under reference
any application to minimize them or correct for them conditions in order to use them for dose reporting to
as needed [6]. In addition to the diode itself, an elec- patients. They did not correct for changes in response
trometer is required to read out the diode measurement. caused by difference in the SSD and field size between
Historically, diodes have required a cable connection to the measurements and the calibration condition.
an electrometer but some newer models are now able to However, they reported that the accuracy of their results
operate wirelessly. This greatly facilitates their ease of could be increased if they apply the appropriate correc-
use and opens the way for further applications. In 2005, tion factors. Under this system, they found that 95% of
the AAPM released the TG-62 to provide guidance on measured readings on their diodes agreed to within 5%
the use of diodes for in vivo dosimetry. While there have of the predicted values. Similar results were reported
been a number of innovations in diode dosimetry since for thoracic measurements. While this study was for 3D
then, TG-62 remains an excellent resource for under- plans, diodes may also be used for dose verification of
standing the fundamentals of diode dosimetry [7]. intensity-modulated radiation therapy (IMRT) treat-
There are a large variety of commercially available ments [14]. In this context, in order to predict the dose
diodes for in vivo dosimetry. Two common classes of to the diode with high accuracy, it is important to use a
diodes are cylindrical and hemispherical; the appropri- fine dose grid (e.g., 1 mm) in planning due to the rela-
ate geometry will depend on its desired clinical use [8]. tively small size (∼0.01 mm3) of the detection volume in
Cylindrical diodes have the advantage of not having diodes.
intrinsic directional dependence perpendicular to the While diodes can be used in a large number of clinical
detector axis and so are appropriate when there are likely scenarios, extra caution must be exercised for measure-
to be a variety of beam angles impinging on the detector. ments of small fields and at short SSDs. An over-response
Clinical Considerations and Dosimeters for In Vivo Dosimetry ◾ 153
of 5–10% has been reported in such scenarios in diode the treatment field increasing the impact of leakage on
measurements [15]. In addition, diodes have a known diode readings. This effect is compounded by the longer
temperature dependence, which can affect their read- treatment times common for TBI. This leakage should
ing when placed on a patient, the magnitude of which be evaluated and, if found to be significant, an alterna-
depends on the diode design. One simple solution to cir- tive setup or equipment should be considered. TBI diode
cumnavigate this problem is to perform the diode cali- dosimetry is made even more complex by the common
bration at a higher temperature. Welsh et al. and Saini use of a tray on the gantry. Both this factor and the large
et al. have published lists of temperature dependen- field sizes need to be corrected for if high accuracy is
cies for a few commercially available diodes (0%/°C desired and the diode is calibrated under standard clini-
to 0.5%/°C) [16, 17]. Accounting for this temperature cal conditions. As mentioned previously, an alternative
dependence is important if one wishes to use diodes for approach is simply to calibrate the diode under the con-
skin surface measurement. It is also worth noting that dition that mimics the exact TBI treatment geometry.
this temperature dependence can vary as a function When placing diodes for TBI in vivo dosimetry, unless
of beam energy. Another confounding issue for skin deliberately accounting for the angular dependence,
measurements is that the effective point of measure- it is important to orient them as perpendicular to the
ment for typical diodes is just under 1 mm, while the incoming beam as possible. The exact locations will vary
International Commission on Radiological Protection depending on the patient orientation and the anatomi-
(ICRP) definition of skin dose is at a depth of 70 µm [18]. cal regions we wish to evaluate the dose. If the diodes
Recently, diodes designed specifically for skin mea- are used to estimate the dose at the patient’s midpoint,
surements with an effective depth of ∼70 µm have been they should be left in the same location on the patient for
developed [19]; however, their use is not yet widespread. all of the delivered fields. Depending on the size of the
Due to the photon energy dependence of most diodes, build-up cap or bolus thickness, we may need to apply
diodes are not commonly used as in vivo dosimeters for a correction factor to the recorded entrance and/or exit
kilovoltage beams. Work by Saw et al., however, has dose before they can be used to calculate the midpoint
demonstrated that for some diodes, the energy depen- dose. Popular points of measurement include the head,
dence between 70 kVp and 100 kVp can be negligible neck, chest, and abdomen. If a lung block is used, its
as long as the diode is calibrated at a similar low energy effect can also be evaluated with diodes measurements.
[20]. This cannot be assumed for any diode and the user Like TBI, total skin electron therapy (TSET) treat-
should determine this dependency for themselves if they ment typically involves large field sizes at extended SSDs
wish to use a diode in the kilovoltage settings. and as such the diode readings must either be corrected
Diodes are commonly applied in total body irradia- to account for that or the calibration must be performed
tions (TBI). Given the low margin for error, the real- under the TSET conditions. In addition, given the rela-
time response of diodes can usefully inform clinical tively low depth of electron penetration in tissue, unin-
decisions during treatment. There is also no need for tentionally shielding of a portion of the skin that we
large distance (e.g., extended SSD) corrections provided wish to treat, with the diode is a real concern. As such, it
that the diodes are initially calibrated at the correspond- is important to use minimal buildup on the diodes and
ing treatment distance. However, if they are calibrated correct the readings accordingly. In addition, as TSET
at the isocenter for an extended distance technique, a is often delivered with patient setup alternating every
correction should be applied. In addition, diodes used other fraction, to get a full picture of the delivered dose
for TBI dosimetry are typically placed on the patient’s over the entire course, it is necessary to get readings for
skin for extended period of time, which is warmer than at least two subsequent fractions. When doing this, it is
the phantoms (solid water, etc.) used for calibration. As critical that the location of the diodes is recorded and
such, it is important to factor in the diode’s temperature reproduced every fraction. As with TBI, the leakage cur-
dependence for TBI dosimetry, either by calibrating at a rent in TSET can be significant and should be properly
realistic temperature that will be used for treatment or evaluated before the diode is used.
by applying a known temperature correction factor. In As mentioned before, diodes have a known depen-
addition, due to the large field sizes involved in TBI, a dence on energy for photon beams. This can make their
greater length of the diode cable is likely to be present in use to measure out-of-field dose (which has a higher
proportion of lower energy scattered photons) challeng- position can shift. As such, an in vivo reading by a diode
ing. With that said, data presented by Chan et al. show can be a useful QA check. With that said, the position-
that it is possible to calculate a correction factor as a ing uncertainty of the diode within the rectum does lead
function of distance from the treatment field [21]. This to a lower level of precision; one study found a range of
can be useful for measuring dose to pacemakers or any diode readings for GYN brachytherapy nearly 40% dif-
other areas of concern, such as the genitals. ferent from those predicted by the treatment planning
Volumetric modulated arc therapy (VMAT) can be system (TPS) [29]. The authors were unable to defini-
challenging for diodes due to their directional depen- tively determine whether this discrepancy was due to
dence. Hui et al. conducted a study on the use of diodes positional uncertainty surrounding the diode place-
in the context of VMAT and while they found a trans- ment or errors in dose delivery; although for an error of
verse directional dependence for a cylindrical diode can this magnitude the latter seems unlikely. Nevertheless,
be <1%, they were only able to establish a diode accu- the use of diodes for in vivo dosimetry in brachyther-
racy of 10% for dose to a small target [22]. This must also apy remains a viable second check provided the user is
be considered for units in which a rotation delivery is aware of its potential pitfalls and the accuracy that can
typical of most or all of its plans, such as TomoTherapy® be ascribed to it. For example, it could be used to catch
units and HalcyonTM systems. An additional variable gross errors in treatment delivery as demonstrated by
one must consider when performing in vivo dosimetry Alecu et al., who set a threshold of ±20% for their rectal
on a TomoTherapy® unit is the pitch of the unit itself. In diode readings [30].
the aforementioned study, Hui et al. also reported that as Another confounding factor effecting diodes use in
pitch increases, diode measurement accuracy decreases. brachytherapy is their demonstrated energy dependence
Recent studies have demonstrated that with the [31]. In the realms of high dose rate (HDR) brachyther-
proper calibration, diodes may be used for in vivo apy, Ir-192 has an average energy of 380 keV, significantly
dosimetry for proton therapy [23, 24]. This introduces below traditional megavoltage energies used in external
a whole host of interesting new applications. One sug- beam treatments. On the other hand, diodes benefit
gested use has been the placement of a diode dosimeter from having a high sensitivity which is useful in HDR
in the esophagus of an anesthetized pediatric craniospi- where the dose falls off dramatically as the distance from
nal irradiation (CSI) patient to verify distal range of the the source increases. One potential use for diodes in this
proton beam [25]. Another interesting application would context is not to measure the dose itself, but instead to
be placement of a diode in a rectal balloon as proposed assess source position accuracy. For example, one group
by Gottschalk et al. [26]. One should be careful with the has looked at Monte Carlo simulations of a diode array
placement of diodes in the context of proton therapy embedded in the treatment couch to verify the source
as they will change the range of any proton beam that position in vivo with a reported accuracy of 1 mm [32].
passes through them, thus impacting the final dosim-
etry of the plan. In addition, as with photons, the more a 11.2.2 Metal Oxide Semiconductor Field
diode is used in a proton beam, the more it will degrade. Effect Transistors (MOSFETs)
However, this degradation will be more dramatic per Basic working principle and design of MOSFET-based
unit dose with protons than it is with photons [27]. dosimeters have been described by multiple authors (e.g.,
The use of diodes placed in the patient’s rectum for in see Chapter 4) [33–36]. Like a diode, a MOSFET-based
vivo brachytherapy dosimetry was first reported a few dosimeter has a relatively small size and can be used to
decades ago [28]. Rectal dose is of particular interest in measure dose in real time. Unlike a diode, MOSFETs store
both gynecology (GYN) and prostate brachytherapy. In dose information, meaning that they can be read multiple
contrast to traditional external beam three-dimensional times. However, they have a more limited lifetime com-
conformal radiation therapy (3D CRT), a very high level pared to the diodes; most MOSFET-based dosimeters
of spatial dose delivery accuracy is critical in brachy- should not be used once they have recorded more than
therapy. However, compared to conventional exter- ∼100 Gy (the actual threshold for a given device depends on
nal beam radiation, there is a far longer time interval the manufacturer and model) [37]. Anyone commission-
between imaging and treatment. In that time, a patient’s ing a MOSFET in vivo dosimetry program should be aware
anatomy can change dramatically and the applicator of this, lest much of the useful “lifespan” of a MOSFET
be used in the commissioning measurements. Even at this system is that it has an electromagnetic positioning
lower dose levels, as integrated dose to the MOSFET sensor that can be used to verify the spatial coordinates
increases, the response of the MOSFET also increases. of the MOSFET as it records dose. This is particularly
One study found that for 200 cGy fractions delivered useful in treatments with high dose gradients, such as
consecutively, the variation in response between one in brachytherapy or stereotactic body radiation therapy
fraction and the next could be higher than 6% [38]. (SBRT)/stereotactic radiosurgery (SRS). In particular,
Most MOSFETs come pre-calibrated from the manu- the CyberKnife® system makes use of circular fields
facturer. While many clinics will accept the manufac- ranging from 5 mm to 60 mm diameter. In such sys-
turer calibration during the lifetime of the MOSFET, tems, an in vivo dosimeter that is both physically small
a higher accuracy can be obtained if the MOSFET is (the dimensions of the sensitive volume of RADPOS are
recalibrated prior to every use [39]. Depending upon 0.2 × 0.2 × 5 × 10 −4 mm3) [49] and whose position can
the measurement conditions, most MOSFETs can be be accurately determined is essential. MOSFETs have
expected to be accurate to within ±5% [40]. been evaluated on CyberKnife® for both breast and lung
MOSFETs can exhibit an angular dependence treatments and found to be accurate to within 4% [50].
(though typically less than diodes), which varies in mag- With that said, CyberKnife® treatments typically involve
nitude based on the manufacturer and measurement multiple KV images taken throughout treatment and, as
conditions of the device [41]. Despite this, they have mentioned earlier, this may necessitate a dose correc-
found wide use beyond conventional 3D CRT and IMRT tion. In addition, for small field sizes, a correction fac-
and are frequently utilized in the context of VMAT and tor may be necessary for the accurate use of a MOSFET
TomoTherapy® applications [42]. MOSFETs also have a dosimeter.
temperature dependency of the response; however, this is MOSFETs are also used in the in vivo intraoperative
usually minimized through dual biasing (two MOSFETs radiation therapy (IORT) setting [51]. This often involves
on a single chip are measured at different levels of bias) physically placing the MOSFETs in the patient’s body.
[43]. As well as application in conventional external In these cases, it is important to consider proper ster-
beam radiation therapy (EBRT), MOSFETs have found ilization and it will be necessary to place the MOSFETs
uses in many of the same applications as diodes rang- in a small sterile container (such as a small plastic bag)
ing from TBI measurements to brachytherapy. Also before placing them in the patient. The user should
like diodes, if MOSFETs are intended for entrance or be aware that the choice of the container may have an
exit dosimetry, a build-up cap should be used. Such a impact on the MOSFET reading, particularly as IORT is
build-up cap can increase the angular dependence of the often carried out using electrons or low energy photons
MOSFET [44] so proper attention must be paid to the [52]. Numerous studies have demonstrated the feasibil-
obliquity of the incoming beam(s). ity for this use of MOSFETs [53, 54]. In the IORT setting,
It is important that the user be aware that MOSFET it is important to accurately characterize the angular
dosimeters have a strong energy dependence between the dependence of the MOSFET system beforehand, as it
KV and MV energy ranges [45]; they will over-respond is typically challenging to know the exact angle of the
to KV photons. If many KV images are to be taken of a MOSFET in relation to the radiation source.
patient with the MOSFET in place, then a dose correc- As mentioned previously, MOSFETs can also be
tion may be required. Despite this, MOSFETs have been used in the context of TBI. As MOSFETs are typically
used in the KV setting, and many groups have reported attached to wires, they can be trickier to handle than
their potential to measure imaging dose, such as that other dosimeters in this setting. However, their small
from a CBCT [46–48]. Such attempts have typically sizes combined with their instant readout make them a
relied on cross-calibrating the MOSFETs with an ion popular choice for TBI measurements in many clinics.
chamber to get suitable dose readings. Reported uncer- However, as MOSFETs typically come pre-calibrated for
tainties in the referenced publications range from 2.25% more common clinical conditions (lower SSD, smaller
to 20% depending on the exact application. field size, etc.), they should be recalibrated if they are
MOSFETs can also be purchased in the form of an intended for TBI dosimetry [55].
array. The RADPOS system (Best Medical Canada) is Implantable MOSFET devices can be used for in vivo
one such example. What is particularly interesting about dosimetry of patients. These devices usually consist of
one or two MOSFETs attached to an antenna, all con- dramatic changes and advances in both the technologies
tained within a small capsule that can be implanted in play and their clinical uses, which include, but are not
in the patient’s breast, prostate, or other sites of clini- limited to, patient alignment, patient-specific IMRT QA,
cal interest [56]. When implanted in the prostate, this and, of course, in vivo dosimetry [66]. In 2001, AAPM’s
device not only serves to record the dose but can also TG-58 [67] was published, which relates specifically to
be used as a fiducial marker, which helps to minimize EPIDs and their use in the clinic. While this is a use-
the positional uncertainty of the dosimeter placement. ful resource for understanding the basic principles upon
Within the context of brachytherapy, MOSFETs may which EPIDs operate, the modern systems differ signifi-
be used to verify skin dose in partial breast treatments cantly from what is described. In addition, many clin-
[57]. Whenever accurate positioning of the MOSFETs ics have become significantly more reliant on EPIDs as
is critical, as is the case in brachytherapy due to steep part of their workflow since the release of TG-58. A per-
dose gradients, it is wise to have a method to identify fect example of this is the new HalcyonTM unit (Varian
their location on the patient in the TPS beyond rough Medical Systems, Palo Alto, CA). This modality has a
approximation. A simple solution for patients receiving built-in EPID panel that automatically records the exit
a CT is to mark the location where the dosimeter will fluence from all clinical patient plans, whether or not it
be placed with a BB marker. Then, when the patient is is needed. EPIDs can be utilized for in vivo dosimetry, in
treated, replace the BB with a MOSFET. If a considerable both an online and offline fashion.
time interval is anticipated before treatment can begin, Unlike the other dosimeters mentioned thus far, the
then one can mark the patient’s skin in the location of EPID is primarily concerned with “exit dosimetry.” That
the BB and use that for MOSFET positioning instead. is to say the dosimetry of particles that have already
In addition, as mentioned earlier, MOSFETs have an traveled through the patient. This somewhat limits their
energy dependence. While the average energy of Ir-192 application and it is challenging to envision a scenario
used in HDR treatments is high enough not to warrant a where they would be useful for brachytherapy or skin
significant correction for MOSFET response, MOSFETs dose measurements. Even with these limitations, there
will over-respond to a number of common LDR sources, are a variety of “pros” to EPID-based dosimetry. These
such as I-125 or Pd-103. With that being said, their small include the convenience of having the device actually
size makes them attractive for measuring quantities attached to the accelerator, lack of angular dependence
such as urethral dose in a prostate seed implant. When as the device rotates with the gantry, and a real-time
correctly calibrated, such a use has been demonstrated and high-resolution 2D measurement. On the other
to be feasible by placing the MOSFETs in a urinary cath- hand, they have a limited field of view (FOV), although
eter [58]. on some modern EPIDs, their detection area can be as
MOSFET dosimeters can also be used in the context large as ∼40 × 40 cm2. This raises the question of where
of proton therapy [59, 60]. However, they have a strong to place the detector in relation to the patient. If a large
dependence on the linear energy transfer (LET) of the FOV is desired, the EPID should be as close to the
protons [61]. Available methods to account for this range patient as possible. However, this brings with it a risk
from simple correction factors to more complex meth- of collision, either to the couch or the patient. As such,
ods relying on Monte Carlo simulations [62]. MOSFETs adequate room should be left to ensure patient safety as
can be expected to be accurate to within 5% in the con- well as avoiding mechanical damage.
text of proton therapy when an appropriate correction When using EPIDs for in vivo dosimetry, it is impor-
method is applied. Without those corrections, errors as tant to first decide if they are intended for 2D or 3D dose
great as 26% have been reported [63]. When MOSFETS verification. While 2D dose comparison is inherently
are used to measure the water equivalent path length, and logistically simpler, it is possible to reconstruct a
they can be accurate to within 1 mm [64]. 3D dose based on EPID images. A number of different
groups have implemented methods to achieve this [68,
11.2.3 Electronic Portal Imaging Devices (EPIDs) 69]. For example, one option is back-projecting the mea-
Electronic portal imaging devices (EPIDs) have been in sured fluence and then applying an iterative convolution
use in radiation therapy departments since the 1980s technique to determine the delivered dose [70]. One of
(see Chapter 8) [65]. The intervening decades have seen the many advantages of exit dosimetry such as this is
that it offers the ability to catch not only errors in deliv- measurements [74]. Using a combination of phantom
ery but any anatomical changes along the beam path. and in vivo patient measurement for 75 prostate plans,
This, however, can also be a confounding factor. If the they were able to demonstrate that similar results to
user only wishes to evaluate delivery accuracy, indepen- IMRT QA could be obtained by averaging the in vivo
dent of setup uncertainty, then positional corrections dosimetry results of a patient over at least three fractions.
need to be applied. Alternatively, looser tolerances for They calculated this could reduce the time required for
in vivo measurements can be implemented. To illus- patient specification by a factor of ∼3. If EPIDs are to be
trate this, in a study on the use of EPID in in vivo breast used in this way, it is important to average the results
dosimetry, when setup variations were not accounted over more than one fraction. Any single fraction might
for researchers found that only 72% of plans delivered be subject to random errors and so can give erroneous
resulted in a match between the predicted and measured results when assessing the delivery accuracy of the plan
dose with an agreement of <5% [71]. When setup varia- over the entire course of treatment. With that stated, if
tions were incorporated into the results, however, 93% of there is a systematic error in delivery, the more fractions
plans had a predicted dose agreement <5%. Exit dosime- that are averaged to obtain results, the fewer fractions
try is also advantageous in that the user does not need to will be left to correct said error.
be concerned with perturbation to the dose distribution Many of the early works in in vivo EPID dosim-
caused by the placement of the dosimeter in the beam etry looked at correlating the EPID signal with some
path. known measurement of dose, such as an ion chamber
Unlike most of the other dosimeters discussed in in a cylindrical phantom, for either a variety of setup
this chapter (with the notable exception of film), EPID’s conditions that could be interpolated or for patient-
typical use in in vivo dosimetry results in a full array of specific plans [75]. Obviously, performing a relative
measurements to be analyzed instead of more discrete EPID calibration for every patient requires signifi-
individual points, potentially yielding more informa- cant clinical resources. Recently, there has been an
tion. As such, the EPID is not typically used for indi- increased effort to extract dose from EPID readings
vidual point dose comparisons in the context of in vivo using more advanced techniques. A full mathemati-
dosimetry. When looking at a 2D image instead of a cal description of these techniques is beyond the
point, a single dose threshold criterion may no longer scope of this chapter. To give one example, Mans et
be the most appropriate tool for evaluation. Some pop- al. published a case study where through the use of
ular evaluation metrics include looking at the distance EPID-based in vivo dosimetry, their group was able
to agreement (DTA) or using gamma analysis, although to catch a plan transfer error for a rectum patient
looking at an absolute dose difference map may also be [76]. Using the data from their EPID, they predicted
helpful. A full description of gamma analysis is beyond an 11.6% under-dose to their PTV unless the problem
the scope of this chapter but readers who are unfamil- was rectified. This highlights the benefits of the EPID
iar with the concept are referred to the work of Low et for evaluating dose to internal target anatomy (using
al. [72]. When using gamma analysis, it is important an aforementioned back-projection approach), which
to choose a criterion based on the treatment scheme. is more challenging or often impossible with other
For example, SBRT would warrant a stricter tolerance online dosimeters.
than a more fractionated IMRT regime. In addition, Recent years have also seen the emergence of com-
when viewing an array, it is likely that not all measured mercially available software packages for dose recon-
points will be of equal significance. For example, per- struction, such as EPIgray (DOSIsoft) or iViewDose
cent dose deviations in portions of the field where the (Elekta). These software packages not only facilitate
planned dose is <10% of the max dose are less likely the in vivo process, but work for a variety of treatment
to be clinically significant then percent dose deviations options including 3D CRT, IMRT, and VMAT. When
within the target volume. One solution is to implement EPID dosimetry is intended for a variety of treatment
a dose threshold, below which dose deviations are not techniques, it may be a good practice to set different
evaluated [73]. in vivo tolerances for different techniques and sites, as
McDermott et al. have looked into the feasibility of average agreement values have been reported to differ
replacing patient-specific IMRT QA with in vivo EPID between them [77].
As mentioned earlier, EPIDs are not often a practical should give consideration to using fiducials with known
choice for brachytherapy applications that is not to say spatial positions in relation to the scintillator. This is
that no investigation has been pursued down this avenue. particularly the case with SBRT, one of the features of
While getting an accurate dose measurement for an in which is typically a very steep dose falloff. If attention
vivo brachytherapy source can be challenging, a simpler is paid to careful positioning of the dosimeter, however,
application of using the EPID to verify source position there is no reason that scintillators cannot be used in
during treatment has already been implemented. Smith this context too, as has been demonstrated by the work
et al. have demonstrated this, using a flat-paneled detec- of Cantley et al. [81] This group also used scintillators on
tor embedded underneath a treatment couch to confirm the surface of rectal balloons for in vivo measurements,
that their source was traveling to the correct position however, within the context of 5 fraction SBRT. Here,
during multiple HDR prostate treatment deliveries [78]. they found agreement within 6% between the predicted
It is possible to envisage such an application applied to TPS dose and the measured dose.
other brachytherapy sites. The use of scintillators has also been investigated for
out-of-field dose measurements [82, 83]. Traditionally,
11.2.4 Scintillation Fiber Optic Dosimeters these measurements are a challenge for many dosim-
Of all of the dosimeters mentioned in this chapter eters due to a variety of factors including the relatively
thus far, scintillators have perhaps experienced the low signal recorded (typically <5% of the maximum
least use in the context of routine in vivo dosimetry. field dose), the angle of incidence compared to the pri-
Despite this, they possess a number of advantageous mary beam, and the difference in energy spectrum from
traits, not least of which is their water equivalence the main beam [84]. For EBRT, this change in energy
(see Chapter 9). Also, they have small physical size spectrum not only affects the photon component of the
and mechanical flexibility, do not require a correction measurement, but there will also be an increase in the
for angle, energy, and dose rate, and have the benefit relevant proportion of contaminant electrons. Once
of MR compatibility. With the emergence and grow- again, the water equivalence of plastic scintillators helps
ing popularity of real-time MR-guided EBRT this last to overcome some of these challenges. Out-of-field dose
point could be particularly important [79]. is particularly relevant to pregnant patients or patients
In an in vivo study, Wooten et al. measured the rectal with an implanted cardiac device, such as a pacemaker.
wall dose during prostate IMRT by placing fiber optic Work has been done using scintillators to estimate the
plastic scintillators on a rectal balloon [80]. Rectal bal- reduction in out-of-field dose to such devices using lead
loons are typically used as immobilization devices dur- shielding.
ing prostate radiation therapy. They also afford a unique Bourgouin et al. [82, 83] looked at using scintil-
opportunity to measure the dose to the rectal wall with lators to measure out-of-field dose for both photon
the affixation of a dosimeter, such as a scintillator, to the (6 MV and 23 MV) and electron beams (6 MeV and
surface of the balloon. The researchers took bi-weekly 18 MeV). Placing the detectors 3 cm away from the
readings of five patients over the course of radiation field edge, they collected percentage depth dose (PDD)
therapy resulting in 142 scintillator dose measurements. curves with both a scintillator and an ion chamber
They found excellent agreement (0.4%) between these for the purpose of comparison. They also collected a
measurements and the predicted dose from the TPS. An series of readings with both detectors placed at dif-
important takeaway from this study, other than the fea- ferent out-of-field distances. When compared to the
sibility of using scintillators in this setting, is that the TPS, both the readings of the ion chamber and the
scintillators themselves were attached to three fiducial scintillator showed significant variation (up to 12%
markers, thus allowing their localization on CT. As for the peripheral dose measurements and up to 80%
mentioned earlier, one of the advantages of scintilla- in the PDD measurements at shallower depths than
tors is their water equivalence; however, due to this very the d max); however, they agreed more closely with
equivalence, they are difficult to detect on CT, which each other. It is important to note that the TPS also
derives its contrast from variations in electron density. struggles with accuracy for out-of-field measure-
As such, when using scintillators for in vivo dosimetry, if ments and so this is not necessarily an indicator of
the placement accuracy is of great importance, the user poor dosimeter performance. The ion chamber and
the scintillator showed good agreement for all mea- per °C has been reported [92]. The loss of the signal per
surements with the exception of the results recorded accumulated dose in scintillators ∼0.28% per kGy (for
for the 6 MeV beam. Here, the authors attributed the 0–15 kGy) and 0.032% per kGy (for 15–127 kGy) has
discrepancy to the relatively large collection diameter also been reported.
of the ion chamber. The other important takeaway
from this study is that the scintillator used was able to 11.2.5 Ion Chambers
accurately measure doses as low as ∼0.5 cGy for both Ion chambers, while are more typically used for refer-
the photon and electron beam. While further valida- ence dosimetry and machine calibration, have also
tion is advised, these results do indicate that plastic found a use in in vivo programs. Certainly, they are use-
scintillators may have a use in in vivo dosimetry for ful for cross-calibration of a number of other dosimeters,
out-of-field measurements. such as diodes. With that stated, their requirement for a
Scintillators can also be used for in vivo brachyther- high voltage bias introduces unique safety concerns that
apy measurements due to their mechanical flexibility are not present for the other dosimeters discussed here.
and relatively small size. In these cases, it is especially In addition, this means that the measurement volume
important to consider the potential temperature must be connected with cables to an electrometer which
dependence of the scintillator, which can vary hugely can make the setup more unwieldy. All of this makes
depending on the type of scintillator [85]. The user is the ion chamber somewhat unusual as a choice for in
advised to consult with the manufacturer, or ideally vivo dosimetry. Furthermore, unsealed ion chambers
perform their own measurements, to asses this depen- need to be carefully corrected for changes in tempera-
dency for their own scintillators. There is considerable ture and pressure. Their ability to measure dose directly,
debate about the significance of the background signal however, means that in certain situations, they may be
created by Cherenkov radiation; however, many groups an appropriate dosimeter.
have recommended either for correcting it or avoiding In using ion chambers for in vivo dosimetry, one
it using specially designed scintillators [86, 87]. If using needs to be aware of the size of the chamber. If being
scintillators for in vivo dosimetry in brachytherapy, it used to measure a region with a high dose gradient, as
is also important to be aware that they can be some- is often the case in IMRT and VMAT, an ion chamber
what mechanically delicate and may degrade quickly can suffer from a volume averaging effect if its volume is
if bent too much or placed under too much strain. too large in relation to the dose gradient. Furthermore,
One example of their use would be to place them in the relatively large physical size of ion chambers makes
unused catheters for patients undergoing interstitial them unsuitable for intracavitary dosimetry applica-
therapy [88]. tions, for example, internal measurements in brachy-
Cherenkov radiation generated in optical fibers is therapy, where a dosimeter with a small physical size is
one of the main issues with scintillation fiber optic needed.
dosimetry of photon and electron fields. Cherenkov Despite all of these drawbacks, there are a number of
radiation generation by proton beams is of less impor- published works on in vivo dosimetry with ion cham-
tance; however, ionization quenching effect is the main bers [93, 94]. For example, Piermattei et al. looked at
issue with proton therapy and other high LET beams using a micro ion chamber to directly measure dose
dosimetry using scintillators. Several methods have to 30 patients treated to the pelvis. The ion chamber
been proposed to minimize the influence of Cherenkov was placed behind each patient on the beam’s central
radiation contamination in order to improve the accu- axis by affixing it to the EPID on the machine. In this
racy of fiber optic dosimetry [89–91]. However, the user way, they were able to relate the recorded signal to the
must have a full understanding of the accuracy of the dose predicted by the TPS for simple 3DCRT plans at
Cherenkov radiation correction method used in their a variety of beam angles. Their reported accuracy was
system, as well as other characteristics of their scintil- within 4% for 95% of measured points. It is important
lating fiber optic dosimetry system. For in vivo dosim- to note that this study did not place the ion chamber
etry applications, the temperature dependency should directly on the patient’s surface, thus circumnavi-
be evaluated in advance and properly addressed as a gating the potentially hazardous voltage dilemma. If
decrease in light output with temperature ∼0.2–0.3% using ion chambers in vivo, this is a prudent strategy.
It is also worth mentioning that at the time this study the MU calculations is correct nor that the patient is
was published, their EPID was not able to accurately positioned correctly.
measure the dose. In most modern clinical settings,
an EPID could easily replace an ion chamber in this
context. 11.3 OFFLINE REVIEW DOSIMETERS
Entrance dosimetry is another application of ion 11.3.1 Film Dosimeters
chambers that avoids affixing them to or within the Unlike the other dosimeters discussed up to this point,
patient. Poppe et al. looked at this in the context of film dosimeter is used in an offline setting when per-
IMRT. They used a prototype wire ion chamber posi- forming in vivo dosimetry. Both radiographic and
tioned in the linac gantry head, below the multi-leaf radiochromic films have been used for in vivo dosim-
collimators (MLCs). Transmission-type ion chambers etry [95–97]. A full description of their construction as
were used to minimize disturbances to the beamline well as the physics behind their interactions is described
caused by the presence of the ion chamber. This tech- elsewhere (see Chapter 5) [98, 99]. Interested readers are
nique of dosimetry cannot account for changes in the also referred to AAPM’s TG-55 [100] and TG-69 [101]
skin surface or patient position, however, it does offer for more information about radiochromic and radio-
an easy way to validate and isolate the accuracy of graphic films, respectively. Since radiographic films are
delivery in terms of radiation leaving the gantry head rarely used nowadays, the focus of this section is on the
and MLC movement. Wires were placed along the gan- application of radiochromic films. As far as their practi-
try head such that each wire position corresponded to cal use is concerned, the user should be aware that the
each MLC pair. During routine patient-specific QA, a radiographic film requires development in a dedicated
baseline reading for each of the wires could be estab- film processor unit before it can be read. Radiochromic
lished, and then the measured reading for each wire for film is “self-developing” and does not share this require-
each treatment could be compared with this baseline ment. However, the reading on the radiochromic film
value. Using this technique, the investigators claimed (e.g., optical density of the film) is known to change
they were able to verify MLC leaf position accuracy based on how much time has elapsed since the radia-
during delivery to within 1 mm. If placing an ion tion exposure. For example, it has been shown that the
chamber in the beam head, it is not only important response of GafchromicTM EBT-3 and EBT-XD films
that it does not disturb the beamline but it also must can change ∼8 and ∼16% during the first 24 hours post-
either be translucent or positioned in such a way as to irradiation [102]. However, a stable response is reached
not block the light field if that is used for patient setup. after ∼24 hours in both film models. A common recom-
In addition, it must be robustly attached to prevent it mendation is to wait at least 24–48 hours before reading.
falling during treatment or setup and causing injury to What is most critical is that the time between delivery
the patient and/or the therapists. and measurements is the same as that used for the cali-
Ion chambers can also find a use in TBI in vivo bration. As such, for in vivo dosimetry necessitating
dosimetry. Due to the large field sizes typically immediate and quantitative results, film is not a suit-
employed, it is often possible to place an ion chamber able choice for most scenarios. Radiochromic films have
in a solid water box or phantom within the treatment the advantage of being nearly tissue equivalent and so
field at the same time as the patient. For example, if the their response is largely independent of energy in the
patient is positioned sitting up with their knees bent, megavoltage range [103, 104]. There are also a number
the ion chamber within the phantom can be placed of other benefits to film dosimetry. We will not go into
under their knees. When used in such a way, if the all of them here but chief among them is the high spa-
proper calibrations and corrections (such as tempera- tial resolution of film (which is most often limited by the
ture and pressure) are applied, the ion chamber can choice in scanner settings as opposed to the film itself)
be used to ensure that the expected dose is delivered and the ability to perform 2D dose measurement using
with each field and MU settings. The downside of this a single film sheet as opposed to a single point dose (1D)
approach is that we are not measuring the dose to the measurement offered by most other dosimeters.
patient themselves and, as such, this technique cannot In vivo film dosimetry has also found a use in proton
be used to verify that the patient thickness used for therapy. There is a large variation in the reported energy
dependence of radiochromic film with some groups dependency and concluded that film as small as 5 ×
reporting variations on the order of 10% [105] while 5 mm2 could be used for in vivo dosimetry [111]. When
other have claimed it to be minimal [106]. The energy cutting film, a guillotine paper cutter should be used
dependence of film also depends on what portion of if available.
the proton depth dose curve it is being used to measure One example of internal use of film is provided by
and under-doses as high as 20% have been reported in Chuang et al., who used small squares of radiochro-
the Bragg peak region [107]. As such, while film may be mic film placed inside the patients’ mouths to verify
used for relative dosimetry in a proton beam, care and the delivered dose for a head and neck IMRT plan
careful consideration should be applied before using it [112]. Another example is the use of film in the breast
for any kind of absolute dosimetry [108]. Despite this, IORT setting. Commonly in IORT, a protective disk
the use of film for absolute dose measurements has been is placed in the surgical cavity prior to the radiation
investigated with one group claiming dose accuracy to delivery. This disk serves to protect the healthy tissue
within 5% [109]. If using film for relative measurements, beyond the cavity that the operator does not wish to
it need not be calibrated in a proton beam and may be irradiate. Radiochromic film may be affixed to that
calibrated using photons on a linac if that is logistically disk to not only verify the dose delivery accuracy, but
simpler. Obviously if being used for absolute dosimetry, also that the disk is functioning as expected. Interested
the film should be calibrated using protons. When scan- readers are referred to a paper by Severgnini et al. for
ning the film, LET corrections can also be applied to a more detailed description [113]. IORT is sometimes
mitigate the energy dependence [110]. performed using low energy photons and radiochro-
A potential use for film within the context of proton mic film is also an appropriate dosimeter in this sce-
therapy is for range verification. If the film is placed nario. However, the user should be cautioned that
along the proton beam distal to the patient, it can be some GafchromicTM film models (e.g. EBT2, EBT3) can
used to record the distal fall-off of the beam. Using this display large energy dependencies in the kilovoltage
information, the range of the beam can be calculated range. The magnitude of this effect depends on the film
and compared to the predicted range from the TPS. itself so the user should be aware of both the manu-
Zheng et al. demonstrated the feasibility of this tech- facturer and model of the film they are using to see
nique for in vivo measurements for a variety of proton if reference to the energy dependence of their film is
patient sites, including lung [106]. They reported that the available in the current literature. GafchromicTM film
film was accurately able to predict the beam range to has also found a limited use within the confines of
within 1 mm. A maximum deviation of 3 mm between brachytherapy. For example, it can be wrapped around
the planned and delivered range was reported in their the surface of a vaginal cylinder and used to verify
work for the lung treatment cases. dose at the cylinder surface during treatment. Such
Film dosimetry can be used for electrons, 3DCRT, a technique was utilized by Pai et al., who reported a
IMRT, and VMAT plans. It can also be used both exter- dose delivery accuracy within 10% [114].
nally and internally to the patient; however, due to its While using films for internal in vivo dosimetry,
larger size, film is not typically applied to as diverse a such as the one suggested above, is interesting, in the
range of internal applications. Smaller dosimeters like majority of in vivo film dosimetry cases, the films are
MOSFETs and OSLDs tend to be popular choices for placed externally to the patients. When used to mea-
this. When internal use is intended, the film should sure the skin dose, the film is typically placed on the
be wrapped in some type of plastic to protect the film. patient’s surface with no buildup. However, if the aim
Since radiographic film typically comes in sealed film of in vivo dosimetry is to measure the entrance dose
sleeves, radiochromic film is the more suitable choice at a certain depth, then appropriate amount of bolus
for internal dosimetry as it can be cut into smaller material should be placed over the film. When film is
pieces. It is important to exercise caution when cutting used to measure the skin dose, it is typically placed in
radiochromic films as the cutting itself can damage the build-up region of the beam. As such, the effective
the film and reduce the accuracy of the results around point of measurement may not be at the correct depth
the region of the cut affected by strain artifact. With to give us the clinically relevant skin dose measurement
that stated, Moylan et al. investigated the film size we want. As such, a correction factor should be applied
to get an accurate dose estimate. This correction factor IMRT conditions, however, they found that RPLDs were
is typically on the order of 15–16% [115]; however, it can able to accurately measure dose within a PTV region to
vary depending on the manufacturer. It is important as within 1.2% (compared to an ion chamber reading) and,
well to avoid stressing the film too much by bending it. as such, a correction factor may not be required unless
Ashburn et al. found a change in film response of >5% treating a particularly small target, such as with SBRT.
when it was significantly bent [116]. A common use of Another useful feature regarding RPLDs in the context
film is verification of junction matching, such as that of in vivo measurements is the relative lack of angu-
between the spinal and cranial fields on a CSI case. In lar dependence for angles <45° on the device, which
such cases, the film is typically placed in a block beneath increases the simplicity of setup on the patient [120].
the patient’s head to verify that we are neither under not Use of RPLDs for in vivo TBI measurements has been
over-irradiating at the match line. reported by Rah et al. [121] Compared to other popu-
Film can also be used in the context of TSET in vivo lar dosimeters for TBI, such as diodes, RPLDs do not
dosimetry. In this context, the film is typically cut into have large angular, SSD, or energy dependence in the
small squares, wrapped in plastic or paper sleeves, MV range, which simplifies the measurement process.
and placed on the patient at the desired measurement Rah et al. used RPLDs along with TLDs and MOSFETs
points. Due to the small size (∼2 × 2 cm2) of the film on three TBI patients and estimated an uncertainty in
pieces needed for this application, a single sheet of the RPLD readings of <3%. They also found that the
GafchromicTM film can provide enough samples for readings of the RPLDs agreed with those of the TLDs
multiple patients. This raises the possibility of having a and MOSFETs to within ∼6.5%, which they attributed
sheet-specific (instead of batch-specific) calibration for largely to differences in position of the dosimeters on the
all films used, if a high level of accuracy is desired. As patient’s surface. When using RPLDs in the context of
alluded to previously, TSET is often delivered by plac- TBI, it is important to remember to place adequate bolus
ing the patient in six different positions with three posi- over the dosimeter to allow for dose buildup and reduce
tions used per day. As such, the radiation cycles are on the effect of electron contamination, particularly if the
a two-day basis and to get an accurate estimate of the RPLD is placed behind a lung block. Another impor-
dose for the entire course of treatment, film will need tant takeaway from this study is that Rah et al. found the
to be placed in the same anatomical locations on two RPLDs to have a linear response with dose in the range
consecutive treatment days. of 0.5 Gy–10 Gy, although other reports go further and
claim a linearity all the way up to 500 Gy [122].
11.3.2 Radiophotoluminescent Dosimeters (RPLDs) As mentioned previously, the response of RPLDs used
While not as ubiquitous as some of the other dosimeters in the megavoltage range is largely energy-independent.
mentioned in this chapter, RPLDs have been used in However, when they are used for in vivo brachytherapy
radiation therapy for decades, particularly in the con- measurements, a correction is required if the calibration
text of personnel monitoring [117]. Similar to TLDs and energy is not the same as the source energy [123]. RPLDs
OSLDs, RPLDs have a relatively small physical size (e.g., are known to over-respond to low energy photons as
the GD-301 model is 1.5 × 8.5 mm2). RPLDs are also demonstrated by Hashimoto et al., where they found a
largely energy-independent in the megavoltage range 13% increase in dose response at a 7 cm distance from
(provided a suitable energy compensating filter is used) an Ir-192 source compared to a 3 cm distance. They also
and can be read multiple times [118]. However, they provide a list of correction factors for detector response
cannot be used in an online setting. They may also be and absorbed dose energy dependence as a function of
annealed (typically by heating them to a temperature distance for RPLDs calibrated in 4 MV beam to be used
of ∼400°C for 1 hour) to allow for repeated use. RPLDs in conjunction with Ir-192.
have been investigated for use in the context of IMRT Despite the energy dependence of RPLDs for brachy-
by Hashimoto et al. They found that for small field sizes therapy, they have been used in this application [124].
(<3 × 3 cm2), RPLDs suffer from both a density pertur- Takayuki et al. investigated using RPLDs to moni-
bation and volume averaging effect. As such, a correc- tor dose to 61 head and neck intestinal brachytherapy
tion factor may be necessary if the user wishes to use patients. They used a variety of techniques to implant
RPLDs for in vivo dosimetry of small fields [119]. Under and localize the RPLDs, including sealing with adhesive
and suturing to the mouth, sandwiching them between the imaging dose, OSLDs, and TLDs are both known
two plastic plates in a template, and placing them within to over-respond [134] and significant correction factors
a silicone impression of the mandible. Small radiopaque must be applied if they are to be used in that setting.
markers were also attached to the RPLDs to accurately Also arising from this effect, if measuring out-of-field
assess their location for dose comparison purposes. photon dose for a megavoltage beam, they will also over-
Unlike TLDs, RPLD readers are able to distinguish respond to the lower energy scattered photons. As such,
the signal caused by surface contamination from that careful consideration to appropriate corrections needs
caused by dose [125]. This is particularly relevant in the to be made if a high accuracy is desired in that scenario.
context of head and neck treatment verifications due to It should be noted that TLDs tend to exhibit less energy
the presence of saliva. It also reduces (but by no means dependence than OSLDs as they are typically made of
eliminates) the need for careful handling and the avoid- a more tissue-equivalent material (e.g., LiF vs. Al2O3:C)
ance of fingerprints. [135]. This is also relevant to HDR brachytherapy dosim-
The largest in vivo RPLD study carried out at the time etry, where the radiation has already lower energy spec-
of this writing was also performed by Takayuki et al., trum which would further decrease with depth resulting
and consisted of over 1000 measurements taken dur- in the potential need for a depth-dependent correction
ing pelvis brachytherapy irradiations [126]. This study factor to maintain accuracy [136]. Despite this, the sig-
spanned multiple treatment sites including the prostate nificance of this energy softening has been debated and
as well as gynecological malignancies. What is particu- the argument can be made that for many applications,
larly interesting is that the RPLDs were placed using a this correction is negligible compared to far larger fac-
variety of different techniques to measure dose in the tors of experimental uncertainty. When OSLDs and
urethra, rectum, and vagina. For vaginal readings, the TLDs are used for entrance or exit dose measurement,
RPLDs were placed in a small slot within a vaginal cylin- bolus or a build-up cap should be used to allow for elec-
der being used for treatment. For the urethral readings, tronic equilibrium. OSLDs may also be used to measure
a “train” of RPLDs was made by placing 10 end-to-end surface dose with no additional buildup [137].
into a single Teflon tube which, in turn, was placed in While external applications are the most common,
the patient’s urinary catheter. The rectum dose readings OSLDs and TLDs have also been used internally. For
were achieved using a similar technique, however, had example, one interesting work by Garner et al. looked
sutures placed at the ends and middle of the train to into using TLDs to measure cardiac and esophageal dose
hold it in place during treatment. This range of applica- to a canine receiving radiation on a CyberKnife® system
tions in a single study demonstrates the versatility of the [138]. In this case, the intent was assessing the accuracy
RPLD as an in vivo dosimeter. It is worth noting that the of the CyberKnife® when delivering dose to targets sig-
authors only found an accuracy to within 20% for the nificantly affected by cardiac motion. Traditional sur-
measurements of the rectum, which they attributed to face dosimeters such as diodes were inappropriate due
organ motion rather than any uncertainties originating to the large variety of entrance angles, and their inabil-
in the detector. ity to account for cardiac motion. Due to their small
size, Garner et al. were able to surgically implant TLD
11.3.3 OSLDs and TLDs crystals on to the surface of the heart itself where they
The physics of OSLDs and TLDs has been described in remained during treatment. The TLDs were visible dur-
Chapter 7 and Chapter 6 of this book, respectively. Both ing simulation of the animals, allowing for an accurate
OSLDs and TLDs have a long history of successful imple- assessment of their spatial position for dose prediction.
mentation for in vivo dosimetry [127–130]. They have a Despite their small size, the high dose gradients pres-
number of advantages including their small size, ability ent in CyberKnife® treatment necessitated averaging the
for reuse, and relatively small energy dependence in the expected dose across the entire volume of the TLD. This
megavoltage range [131]. However, caution must be exer- was achieved by contouring the TLDs on the simulation
cised in their use since most OLSDs and TLDs exhibit scan, and recording the average dose in the region of the
supralinearity beyond 10–20 Gy [132]. For some mod- contour. It is important to note that this technique can
els, the threshold is even lower than 10Gy [133]. In addi- be applied to other in vivo dosimeters that are scanned
tion, for lower energy applications, such as measuring with the patient for an accurate estimate of dose. Due to
its invasive nature, this particular form of in vivo dosim- each patient, other than the time commitment, is the
etry is unlikely to be popular with human subjects (the uncertainty in dose distribution. In vivo dosimetry with
canine was euthanized in order to retrieve the TLDs). OSLDs in these cases can be a valuable tool to verify the
However, the use of TLDs in this context is testament accuracy of the treatment delivery. There are a few ways
to their versatility and highlights a number of consid- to go about predicting the dose to the OSLDs in such
erations that need to be made for many more common instances. Perhaps the simplest is to scan the patient
clinical scenarios. with both the custom applicator and OSLDs in place.
Their small size also means that OSLDs and TLDs Alternatively, if the OSLDs cannot be visualized with
are popular choice for in vivo brachytherapy dosimetry. this technique, the applicator can be scanned again with-
A number of studies have looked at using them to mea- out the patient, but instead with BBs placed at each loca-
sure dose within the context of prostate HDR [139–141]. tion where an OSLD will be present during treatment.
One such example is the work by Anagnostopoulos Assuming the applicator is rigid, this scan can then be
et al., who used 50 TLDs to record the dose to either the registered to the patient scan and a dose estimate to the
urethra or posterior rectal wall during 18 fractions of OSLDs can be calculated. This alternative approach also
HDR prostate brachytherapy. The TLDs were placed in a avoids exposing the OSLDs to imaging dose.
chain within one of the needles that had been implanted A full history of the clinical implementation of TLDs
for treatment or, if such needle were not available, within and OSLDs is beyond the scope of this chapter, and
an extra needle placed explicitly for the purposes of in indeed, this book; however, it would be remiss to con-
vivo dosimetry. There are advantages and disadvantages clude this brief summary without mentioning the appli-
associated to both techniques. Implanting an extra needle cation of OSLDs for both TBI and TSET [144, 145]. Due
for dosimetry results in increased trauma to the patient. to their relative ease of use, OSLDs and TLDs are popular
However, using a needle that will also be used for deliv- dosimeters for verifying dose to TBI and TSET patients.
ery required an interruption in the treatment before the In the case of TBI, they can be placed at various anatomi-
source gets to the needle in question. Using this method, cal points of interest during delivery and are often used
Anagnostopoulos et al. were able to establish a dose deliv- with bolus to get an estimate of entrance and/or exit dose.
ery accuracy to within a mean of 7% for all treatments. Bolus is rarely used in TSET cases due to the sparing
Similar to the work by Garner et al., they also compared effect it would have on the tissue beneath it and instead
their measured dose to the dose predicted to the cathe- OSLDs and TLDs (often covered by a plastic wrap) are
ter that contained the TLDs on the simulation CT scan. placed directly on the patient’s surface. Most OSLDs and
As the catheter itself was visible, and TLDs ran down its TLDs also have the advantage of low angular dependence
length, no additional markers to indicate the location of which is particularly useful in TBI and TSET where
the TLDs were necessary. It is of interest to note that in dosimeters are often placed in locations with an oblique
this study, the TLDs were calibrated in a 6 MV photon angle of incidence to the beam. The user should be aware
beam and no correction for the lower energy of Ir-192 was that the design of the dosimeter and its casing will affect
made. Brezovich et al. also used TLDs to measure urethral the angular dependence so angular independence should
dose during HDR prostate brachytherapy procedures not be freely assumed without evidence for the model in
[142]. However, instead of using needles, they placed the question. If the OSLDs or TLDs used for TBI and TSET
TLD directly in the patient’s urethra via a Foley catheter. measurements are calibrated under standard clinical ref-
Looking at three patients, they found the mean urethral erence conditions, it may be necessary to apply a correc-
dose to be within 4% of what was predicted for each case. tion factor to their readings [146]. This correction factor
The use of OSLDs within brachytherapy is not lim- will depend on the specific dosimeters as well as the clini-
ited to the prostate treatment. Another example of their cal setup being used for TBI and TSET.
application is in verification of the dose distribution of
custom-made applicators [143]. While applicators spe- 11.4 SUMMARY
cializing in skin surface treatments do exist, many clin- As this chapter shows, there are a wide variety of
ics will create their own using a variety of techniques options available for in vivo dosimetry. It is up to the
and materials, including thermoplastics and wax bolus. clinical physicist to figure out how to make best use of
The downside of creating such a unique applicator for the resources available to them and, if those resources
are insufficient, make recommendations to the hospital the kV energy range, then its energy dependence needs
administration as to what devices are needed. There is no to be carefully established and/or well documented in the
single dosimeter that will work in all scenarios but simi- literature. When applied correctly, an in vivo dosimetry
larly, there are few, if any, departments that will need all program can be an incredibly valuable QA tool to help
of the dosimeters discussed here. The needs of a depart- verify the safety and efficacy of dose delivery. However,
ment will also strongly depend on how many special- if such a program is applied with little thought or under-
ized procedures they are doing (e.g., TBI, TSET, IORT). standing given to the limitations of the dosimeters used,
Each dosimeter has its advantages and disadvantages, then any results obtained may be meaningless, or worse,
and it is important to be aware of those before beginning lead to incorrect clinical judgments and treatments. Pros
a new in vivo dosimetry program. For example, if the and cons relevant to in vivo dosimetry for each of the
dosimeter is going to be used to measure treatments in discussed dosimeters is provided in Table 11.1.
TABLE 11.1 Advantages and Disadvantages of Dosimeters for In Vivo Dosimetry Applications
Dosimeter Pros for In Vivo Dosimetry Cons for In Vivo Dosimetry
Diodes • Small • Temperature dependence
• Highly sensitive • Angular dependence
• Can be reused • Dose rate and Dose per pulse dependence
• Immediate results • Energy dependence
• Large body of supporting literature • Most models require wires and an electrometer
• Sensitivity changes with accumulated dose
MOSFET • Small • Angular dependence
• Immediate results • Dose per pulse dependence
• Can be reused • Energy dependence between KV and MV ranges
• Stores dose information permanently • Sensitivity changes with accumulated dose
• Large body of supporting literature
• Minimal temperature dependence with a
dual bias
EPIDs • Immediate results • Cannot be placed within or on the patient
• Can be reused • Not practical for brachytherapy
• Doesn’t require additional equipment • Not practical for entrance dose measurements
• 2D planar dosimetry • Not practical for TBI or TSET
• Large body of supporting literature
Scintillators • Small • Less widely used than most other dosimeters
• Immediate results • Temperature dependence
• Can be reused • Mechanically delicate
• Water equivalent • Susceptible to “noise” caused by Cherenkov radiation
• Minimal angular or dose rate dependence
• MR compatible
Ion Chambers • Immediate results • Bulky
• Can be reused • Requires a high voltage bias
• Can directly measure dose • Requires wires and an electrometer
• Temperature and pressure dependence
• Not practical for brachytherapy
Film Dosimeters • 2D planar dosimetry • Not appropriate for online review
• Gafchromic film is tissue equivalent • Reading changes dramatically based on time since
• Large body of supporting literature exposure
• Requires careful handling
• Requires use of an appropriately calibrated scanner
RPLDs • Small • Not appropriate for online review
• Can be reused • Less widely used than most other dosimeters
• Minimal angular dependence • Requires a reader
OSLDs and TLDs • Small • Not appropriate for online review
• Can be reused • TLDs require careful handling
• Large body of supporting literature • Requires a reader
• TLDs are close to tissue equivalent
• OSLDs are easy to use
A physicist should begin the process of dosimeter 10. T. Loncol, et al., “Entrance and exit dose measure-
selection by first asking themselves what is the ulti- ments with semiconductors and thermoluminescent
dosimeters: A comparison of methods and in vivo
mate goal of the program. Is it to catch gross errors or
results,” Radiotherapy and Oncology 41(2), 179–187
to observe trends? Is absolute dosimetry needed or is a (1996).
relative measurement sufficient? Is a single point mea- 11. D. I. Thwaites, et al., “Experience with in vivo diode
surement good enough or is a 2D dose plane needed? dosimetry for verifying radiotherapy dose delivery:
Do in vivo readings need to be able to identify a prob- Practical implementation of cost-effective approaches,”
lem with delivery immediately or will there be time to International Atomic Energy Agency (IAEA) 245–246
(2002).
process and analyze the results offline? Without hav- 12. S. Heukelom, et al., “Comparison of entrance and exit
ing a concrete goal in mind from the outset, it is easy dose measurements using ionization chambers and
to get confused an overwhelmed by the sheer volume silicon diodes,” Physics in Medicine and Biology 36(1),
of available choices out there. Hopefully, the review 47–59 (1991).
provided by this chapter helps to clarify the roles 13. R. Alecu, et al., “A method to improve the effectiveness
of diode in vivo dosimetry,” Medical Physics 25(5),
that different dosimeters play in different treatment 746–749 (1998).
scenarios. 14. P. D. Higgins, et al., “In vivo diode dosimetry for
routine quality assurance in IMRT,” Medical Physics
30(12), 3118–3123 (2003).
REFERENCES 15. J. G. Wierzbicki and D. S. Waid, “Large discrepan-
1. M. S. Huq, et al., “The report of Task Group 100 of the cies between calculated and diode readings for small
AAPM: Application of risk analysis methods to radia- field sizes and small SSDs of 15 MV photon beams,”
tion therapy quality management,” Medical Physics Medical Physics 25(2), 245–246 (1998).
43(7), 4209–4262 (2016). 16. K. T. Welsh and L. E. Reinstein (2001). “The thermal
2. R. Alecu, et al., “Guidelines on the implementation characteristics of different diodes on in vivo patient
of diode in vivo dosimetry programs for photon and dosimetry,” Medical Physics 28(5), 844–849.
electron external beam therapy,” Medical Dosimetry 17. A. S. Saini and T. C. Zhu (2002). “Temperature depen-
24(1), 5–12 (1999). dence of commercially available diode detectors,”
3. F. Ghahramani, et al., “Dependency of semiconductor Medical Physics 29(4), 622–630.
dosimeter responses, used in MDR/LDR brachyther- 18. ICRP, “The 2007 Recommendations of the International
apy, on factors which are important in clinical condi- Commission on Radiological Protection,” ICRP
tions,” Reports of Practical Oncology & Radiotherapy Publication 103. Ann. ICRP 37 (2–4) (2007).
13(1), 29–33 (2008). 19. N. Vicoroski, et al., “Development of a silicon diode
4. P. A. Jursinic, “Angular dependence of dose sensitivity detector for skin dosimetry in radiotherapy,” Medical
of surface diodes,” Medical Physics 36(6 Part 1), 2165– Physics 44(10), 5402–5412 (2017).
2171 (2009). 20. C. B. Saw, et al., “Energy dependence of a new solid
5. S. F. Kry, et al., “AAPM TG 158: Measurement and state diode for low energy photon beam dosimetry,”
calculation of doses outside the treated volume from Medical Dosimetry 23(2), 95–97 (1998).
external-beam radiation therapy,” Medical Physics 21. M. Chan, et al., “Estimating dose to ICD outside
44(10), e391–e429 (2017). the treatment field using skin QED diode,” Medical
6. K. González-Pérez, et al., “Calibration and correc- Physics 33(6), 2218 (2006).
tion factors of a set of semiconductor diodes for its 22. S. Hui, et al., “SU-FF-T-150: In vivo diode dosimetry
use in external radiation therapy,” AIP Conference for helical tomotherapy dose verification,” Medical
Proceedings 1747(1), 040001 (2016). Physics 32(6), 1984–1984 (2005).
7. E. Yorke et al., Diode In Vivo Dosimetry for Patients 23. A. Toltz, et al., “Time-resolved diode dosimetry cali-
Receiving External Beam Radiation Therapy. Report of bration through Monte Carlo modeling for in vivo
Task Group 62 of the Radiation Therapy Committee passive scattered proton therapy range verification,”
(Medical Physics Publishing, Madison, 2005). Journal of Applied Clinical Medical Physics 18(6), 200–
8. X. R. Zhu, “Entrance dose measurements for in-vivo 205 (2017).
diode dosimetry: Comparison of correction factors 24. U. Sowa, et al., “Dosimetric characteristics of active
for two types of commercial silicon diode detectors,” solid state detectors in a 60 MeV proton radiotherapy
Journal of Applied Clinical Medical Physics 1(3), beam,” Nukleonika 57, 491–495 (2012).
100–107 (2000). 25. A. Toltz, et al., “SU-E-T-524: In-vivo diode dosimetry
9. L. L. W. Wang and D. W. O. Rogers, “Monte Carlo proton therapy range verification validation study
study of Si diode response in electron beams,” Medical for pediatric CSI,” Medical Physics 42(6), 3455–3456
Physics 34(5), 1734–1742 (2007). (2015).
26. B. Gottschalk, et al., “Water equivalent path length 43. T. Cheung, et al., “Effects of temperature variation on
measurement in proton radiotherapy using time MOSFET dosimetry,” Physics in Medicine and Biology
resolved diode dosimetry,” Medical Physics 38(4), 49(13), N191–N196 (2004).
2282–2288 (2011). 44. S. Qin, et al., “Angular dependence of the MOSFET
27. J. R. Srour, et al., “Review of displacement damage dosimeter and its impact on in vivo surface dose mea-
effects in silicon devices,” IEEE Transactions on surement in breast cancer treatment,” Technology in
Nuclear Science 50(3), 653–670 (2003). Cancer Research & Treatment 13(4), 345–352 (2014).
28. K. Tanderup, et al., “In vivo dosimetry in brachyther- 45. T. Cheung, et al., “Energy dependence corrections to
apy,” Medical Physics 40(7), 070902 (2013). MOSFET dosimetric sensitivity,” Australasian Physics
29. M. Allahverdi, et al., “Evaluation of treatment plan- & Engineering Sciences in Medicine 32(1), 16–20 (2009).
ning system of brachytherapy according to dose to the 46. P. Ravindran, et al., “In vivo dose measurement dur-
rectum delivered,” Radiation Protection Dosimetry ing IGRT with KV cone beam CT,” Medical Physics 40,
150(3), 312–315 (2011). 188 (2013).
30. R. Alecu and M. Alecu, “In-vivo rectal dose measure- 47. M. Amin, et al., “SU-FF-T-387: Performance of a por-
ments with diodes to avoid misadministrations during table MOSFET dosimetry system for in-vivo dose
intracavitary high dose rate brachytherapy for carci- measurements in orthovoltage treatments and kilo-
noma of the cervix,” Medical Physics 26(5), 768–770 voltage cone beam CT,” Medical Physics 36(6 Part 15),
(1999). 2611–2611 (2009).
31. A. S. Saini and T. C. Zhu, “Energy dependence of com- 48. G. X. Ding and C. W. Coffey, “Dosimetric evaluation
mercially available diode detectors for in-vivo dosim- of the OneDose™ MOSFET for measuring kilovoltage
etry,” Medical Physics 34(5), 1704–1711 (2007). imaging dose from image-guided radiotherapy proce-
32. J. Poder, et al., “HDR brachytherapy in vivo source dures,” Medical Physics 37(9), 4880–4885 (2010).
position verification using a 2D diode array: A Monte 49. N. Ploquin, et al., “Use of novel fibre-coupled radiolu-
Carlo study,” Journal of Applied Clinical Medical minescence and RADPOS dosimetry systems for total
Physics 19(4), 163–172 (2018). scatter factor measurements in small fields,” Physics in
33. A. S. Kumar, et al., “Characteristics of mobile MOSFET Medicine and Biology 60, 1–14 (2014).
dosimetry system for megavoltage photon beams,” 50. R. Marants, et al., “Evaluation of the 4D RADPOS
Journal of medical physics 39(3), 142–149 (2014). dosimetry system for dose and position quality assur-
34. J. H. Koivisto, et al., “Characterization of MOSFET ance of CyberKnife,” Medical Physics 45(9), 4030–
dosimeters for low-dose measurements in maxillofa- 4044 (2018).
cial anthropomorphic phantoms,” Journal of Applied 51. M. Ciocca, et al., “Real-time in vivo dosimetry using
Clinical Medical Physics 16(4), 266–278 (2015). micro-MOSFET detectors during intraoperative elec-
35. D. J. Gladstone and L. M. Chin, “Automated data col- tron beam radiation therapy in early-stage breast
lection and analysis system for MOSFET radiation cancer,” Radiotherapy and Oncology 78(2), 213–216
detectors,” Medical Physics 18(3), 542–548 (1991). (2006).
36. A. Holmes-Siedle, “The space-charge dosimeter: 52. F. W. Hensley, “Present state and issues in IORT phys-
General principles of a new method of radiation ics,” Radiation Oncology 12(1), 37 (2017).
detection,” Nuclear Instruments and Methods 121(1), 53. A. Petoukhova, et al., “In vivo dosimetry with
169–179 (1974). MOSFETs and GAFCHROMIC films during elec-
37. D. Shah, Radiation Effects on MOSFETs (2016). tron IORT for accelerated partial breast irradiation,”
38. J. A. Tanyi, et al., “MOSFET sensitivity dependence Physica Medica 44, 26–33 (2017).
on integrated dose from high-energy photon beams,” 54. R. Consorti, et al., “In vivo dosimetry with MOSFETs:
Medical Physics 35(1), 39–47 (2008). Dosimetric characterization and first clinical results
39. G. Toncheva, et al., “SU-GG-I-67: Variation of MOSFET in intraoperative radiotherapy,” International Journal
calibration factor as a function of dosimeter age,” of Radiation Oncology, Biology, Physics 63(3), 952–960
Medical Physics 37(6 Part 3), 3116–3117 (2010). (2005).
40. P. H. Halvorsen, “Dosimetric evaluation of a new 55. T. M. Briere, et al., “Patient dosimetry for total body
design MOSFET in vivo dosimeter,” Medical Physics irradiation using single-use MOSFET detectors,”
32(1), 110–117 (2005). Journal of Applied Clinical Medical Physics 9(4),
41. J. Koivisto, et al., “Characterization of MOSFET 200–205 (2008).
dosimeter angular dependence in three rotational 56. T. M. Briere, et al., “Implantable MOSFET detectors:
axes measured free-in-air and in soft-tissue equiva- Evaluation of a new design,” Medical Physics 34(12),
lent material,” Journal of Radiation Research 54(5), 4585–4590 (2007).
943–949 (2013). 57. S. Kim, et al., “In vivo dosimetry of skin dose during
42. R. A. Kinhikar, et al., In vivo dosimetry using MOSFET HDR breast brachytherapy using balloon catheter,”
and TLD for Tomotherapy (Springer Berlin Heidelberg, International Journal of Radiation Oncology, Biology,
Berlin, Heidelberg, 2009). Physics 72(1), S524–S525 (2008).
58. E. J. Bloemen-van Gurp, et al., “In vivo dosimetry 71. A. Fidanzio, et al., “Breast in vivo dosimetry by EPID,”
using a linear MOSFET-array dosimeter to determine Journal of Applied Clinical Medical Physics 11(4), 249–
the urethra dose in 125I permanent prostate implants,” 262 (2010).
International Journal of Radiation Oncology, Biology, 72. D. A. Low, et al., “A technique for the quantitative
Physics 73(1), 314–321 (2009). evaluation of dose distributions,” Medical Physics
59. C.-W. Cheng, et al., “Dosimetric characteristics of a 25(5), 656–661 (1998).
single use MOSFET dosimeter for in vivo dosimetry 73. M. Miften, et al., “Tolerance limits and methodolo-
in proton therapy,” Medical Physics 37(8), 4266–4273 gies for IMRT measurement-based verification QA:
(2010). Recommendations of AAPM Task Group No. 218,”
60. G. A. Pablo Cirrone, et al., “Preliminary investiga- Medical Physics 45(4), e53–e83 (2018).
tion on the use of the MOSFET dosimeter in proton 74. L. N. McDermott, et al., “Replacing pretreatment
beams,” Physica Medica 22(1), 29–32 (2006). verification with in vivo EPID dosimetry for prostate
61. A. B. Rosenfeld, et al., “A new silicon detector for IMRT,” International Journal of Radiation Oncology,
microdosimetry applications in proton therapy,” IEEE Biology, Physics 67(5), 1568–1577 (2007).
Transactions on Nuclear Science 47(4), 1386–1394 75. K. Slosarek, et al., “EPID in vivo dosimetry in
(2000). RapidArc technique,” Reports of Practical Oncology
62. R. Kohno, et al., “In vivo proton dosimetry using and Radiotherapy: Journal of Great Poland Cancer
a MOSFET detector in an anthropomorphic Center in Poznan and Polish Society of Radiation
phantom with tissue inhomogeneity,” Journal of Oncology 15(1), 8–14 (2010).
Applied Clinical Medical Physics 13(2), 3699–3699 76. A. Mans, et al., “Catching errors with in vivo EPID dosim-
(2012). etry,” Medical Physics 37(6 Part 2), 2638–2644 (2010).
63. R. Kohno, et al., “Proton dose distribution measure- 77. S. Celi, et al., “EPID based in vivo dosimetry system:
ments using a MOSFET detector with a simple dose- Clinical experience and results,” Journal of Applied
weighted correction method for LET effects,” Journal Clinical Medical Physics 17(3), 262–276 (2016).
of Applied Clinical Medical Physics 12(2), 326–337 78. R. L. Smith, et al., “OC-0177: Clinical implementation
(2011). of in vivo source position verification in high dose rate
64. H.-M. Lu, et al., “Investigation of an implantable prostate brachytherapy,” Radiotherapy and Oncology
dosimeter for single-point water equivalent path 115, S87–S88 (2015).
length verification in proton therapy,” Medical Physics 79. M. van Herk, et al., “Magnetic resonance imag-
37(11), 5858–5866 (2010). ing-guided radiation therapy: A short strengths,
65. L. E. Antonuk, “Topical review: Electronic portal weaknesses, opportunities, and threats analysis,”
imaging devices: A review and historical perspec- International Journal of Radiation Oncology, Biology,
tive of contemporary technologies and research,” Physics 101(5), 1057–1060 (2018).
Physics in Medicine and Biology 47, R31–R65 80. L. Wootton, et al., “Real-time in vivo rectal wall
(2002). dosimetry using plastic scintillation detectors for
66. W. van Elmpt, et al., “A literature review of patients with prostate cancer,” Physics in Medicine and
electronic portal imaging for radiotherapy dosim- Biology 59(3), 647–660 (2014).
etry,” Radiotherapy and Oncology 88(3), 289–309 81. J. L. Cantley, et al., “Real-time in vivo dosimetry for
(2008). SBRT prostate treatment using plastic scintillating
67. M. G. Herman, et al., “Clinical use of electronic por- dosimetry embedded in a rectal balloon: A case study,”
tal imaging: Report of AAPM Radiation Therapy Journal of Applied Clinical Medical Physics 17(6), 305–
Committee Task Group 58,” Medical Physics 28(5), 311 (2016).
712–737 (2001). 82. A. Bourgouin, et al., “Estimating and reducing dose
68. M. Wendling, et al., “A simple backprojection algo- received by cardiac devices for patients undergoing
rithm for 3D in vivo EPID dosimetry of IMRT treat- radiotherapy,” Journal of Applied Clinical Medical
ments,” Medical Physics 36(7), 3310–3321 (2009). Physics 16(6), 411–422 (2015).
69. B. Mijnheer, et al., “TU-C-BRE-11: 3D EPID-based 83. A. Bourgouin, et al., “Technical Note: Out-of-field
in vivo dosimetry: A major step forward towards dose measurement at near surface with plastic scin-
optimal quality and safety in radiation oncology tillator detector,” Journal of Applied Clinical Medical
practice,” Medical Physics 41(6 Part 26), 457–457 Physics 17(5), 542–547 (2016).
(2014). 84. A. Skrobala, et al., “Low dose out-of-field radiother-
70. E. Van Uytven, et al., “Validation of a method for in apy, part 2: Calculating the mean photon energy
vivo 3D dose reconstruction for IMRT and VMAT values for the out-of-field photon energy spec-
treatments using on-treatment EPID images and a trum from scattered radiation using Monte Carlo
model-based forward-calculation algorithm,” Medical methods,” Cancer/Radiothérapie 21(5), 352–357
Physics 42(12), 6945–6954 (2015). (2017).
85. F. Therriault-Proulx, et al., “A method to correct for 100. A. Niroomand-Rad, et al., “Radiochromic film dosim-
temperature dependence and measure simultane- etry: Recommendations of AAPM Radiation Therapy
ously dose and temperature using a plastic scintilla- Committee Task Group 55,” Medical Physics 25(11),
tion detector,” Physics in Medicine and Biology 60(20), 2093–2115 (1998).
7927–7939 (2015). 101. S. Pai, et al., “TG-69: Radiographic film for megavolt-
86. C. E. Andersen, et al., “Characterization of a fiber- age beam dosimetry,” Medical Physics 34(6 Part 1),
coupled luminescence dosimetry system for online in 2228–2258 (2007).
vivo dose verification during brachytherapy,” Medical 102. A. Darafsheh, T. Zhao, R. Khan, “Spectroscopic analy-
Physics 36(3), 708–718 (2009). sis of irradiated radiochromic EBT-XD films in proton
87. J. Lambert, et al., “A plastic scintillation dosimeter for and photon beams,” Physics in Medicine and Biology
high dose rate brachytherapy,” Physics in Medicine and 65, 205002 (2020).
Biology 51(21), 5505–5516 (2006). 103. E. Y. León-Marroquín, et al., “Spectral analysis of
88. C. E. Andersen, et al., “Time-resolved in vivo lumines- the EBT3 radiochromic films for clinical photon
cence dosimetry for online error detection in pulsed and electron beams,” Medical Physics 46(2), 973–982
dose-rate brachytherapy,” Medical Physics 36(11), (2019).
5033–5043 (2009). 104. E. Y. León-Marroquín, et al., “Response characteriza-
89. A. Darafsheh, “Scintillation fiber optic dosimetry,” in tion of EBT-XD radiochromic films in megavoltage
Radiation Therapy Dosimetry: A Practical Handbook, photon and electron beams,” Medical Physics 46(9),
edited by A. Darafsheh (CRC Press, Boca Raton, 2021), 4246–4256 (2019).
Chapter 9, pp. 123–137. 105. L. Zhao and I. J. Das, “Gafchromic EBT film dosim-
90. A. Darafsheh, et al., “Radiotherapy fiber dosime- etry in proton beams,” Physics in Medicine and Biology
ter probes based on silver-only coated hollow glass 55(10), N291–N301 (2010).
waveguides,” Journal of Biomedical Optics 23(1), 1–7 106. Y. Zheng, “SU-E-T-449: In-vivo dosimetry and range
(2018). verification for proton therapy,” Medical Physics 40(6
91. A. Darafsheh, et al., “Spectroscopic separation of Part 17), 308–308 (2013).
Čerenkov radiation in high-resolution radiation fiber 107. S. Reinhardt, et al., “Comparison of Gafchromic EBT2
dosimeters,” Journal of Biomedical Optics 20(9), 1–7 and EBT3 films for clinical photon and proton beams,”
(2015). Medical Physics 39(8), 5257–5262 (2012).
92. P. Carrasco, et al., “Characterization of the Exradin 108. A. Darafsheh, et al., “On the spectral characterization
W1 scintillator for use in radiotherapy,” Medical of radiochromic films irradiated with clinical proton
Physics 42(1), 297–304 (2015). beams,” Physics in Medicine & Biology 64(13), 135016
93. A. Piermattei, et al., “In-vivo portal dosimetry by an (2019).
ionization chamber,” Physica Medica 21(4), 143–152 109. V. Devaraju and R. Slopsema, “SU-E-T-120:
(2005). Characterization of EBT2 film for absolute dosimetry
94. B. Poppe, et al., “DAVID – A translucent multi- in proton beams,” Medical Physics 38(6 Part 12), 3513–
wire transmission ionization chamber for in vivo 3513 (2011).
verification of IMRT and conformal irradiation 110. D. Kirby, et al., “LET dependence of GafChromic films
techniques,” Physics in Medicine and Biology 51, and an ion chamber in low-energy proton dosimetry,”
1237–1248 (2006). Physics in Medicine and Biology 55(2), 417–433 (2009).
95. G. K. Svensson, “Quality assurance in external beam 111. R. Moylan, et al., “Dosimetric accuracy of
radiation therapy,” RadioGraphics 9(1), 169–182 Gafchromic EBT2 and EBT3 film for in vivo dosim-
(1989). etry,” Australasian Physical & Engineering Sciences in
96. H.-W. Liu, et al., “Role of in vivo dosimetry with radio- Medicine 36(3), 331–337 (2013).
chromic films for dose verification during cutane- 112. K. Chuang, et al., “SU-FF-T-237: Gafchromic film in
ous radiation therapy,” Radiation Oncology 10(1), 12 vivo dosimetry for oral radiotherapy,” Medical Physics
(2015). 34 (2007).
97. M. Williams and P. Metcalfe, “Radiochromic film 113. M. Severgnini, et al., “In vivo dosimetry and shielding
dosimetry and its applications in radiotherapy,” AIP disk alignment verification by EBT3 GAFCHROMIC
Conference Proceedings 1345(1), 75–99 (2011). film in breast IOERT treatment,” Journal of Applied
98. S. Mossahebi, N. Hoshyar, R. Khan, and A. Clinical Medical Physics 16(1), 112–120 (2015).
Darafsheh, “Film dosimetry,” in Radiation Therapy 114. S. Pai, et al., “The use of improved radiochromic film
Dosimetry: A Practical Handbook, edited by A. for in vivo quality assurance of high dose rate brachy-
Darafsheh (CRC Press, Boca Raton, 2021), Chapter 5, therapy,” Medical Physics 25(7), 1217–1221 (1998).
pp. 61–74. 115. S. Devic, et al., “Accurate skin dose measurements
99. S. Devic, “Radiochromic film dosimetry: Past, present, using radiochromic film in clinical applications,”
and future,” Physica Medica 27(3), 122–134 (2011). Medical Physics 33(4), 1116–1124 (2006).
116. J. Ashburn, “SU-E-T-160: Response of EBT2 multicenter cohort study,” International Journal of
Gafchromic film to possible stresses associated with Radiation Oncology, Biology, Physics 96(2), E605–
in vivo measures,” Medical Physics 39(6 Part 12), E606 (2016).
3740–3740 (2012). 130. N. D. MacDougall, et al., “In vivo dosimetry in UK
117. N. Navab Moghadam, et al., “Response of TLD and external beam radiotherapy: Current and future
RPL personal dosimeters in a national inter-compari- usage,” The British Journal of Radiology 90(1072),
son test program,” International Journal of Radiation 20160915–20160915 (2017).
Research (IJRR) 14, 73 (2016). 131. R. Ponmalar, et al., “Dosimetric characterization of
118. A.-L. Manninen, et al., “The applicability of radio- optically stimulated luminescence dosimeter with
photoluminescence dosemeter (RPLD) for measuring therapeutic photon beams for use in clinical radio-
medical radiation (MR) doses,” Radiation Protection therapy measurements,” Journal of Cancer Research
Dosimetry 151(1), 1–9 (2012). and Therapeutics 13(2), 304–312 (2017).
119. S. Hashimoto, et al., “Field-size correction factors of a 132. P. A. Jursinic, “Characterization of optically stimu-
radiophotoluminescent glass dosimeter for small-field lated luminescent dosimeters, OSLDs, for clinical
and intensity-modulated radiation therapy beams,” dosimetric measurements,” Medical Physics 34(12),
Medical Physics 45(1), 382–390 (2018). 4594–4604 (2007).
120. E. H. Silva, et al., “ENERGY and angular depen- 133. Y. R. Ponmalar, et al., “Response of nanodot optically
dence of radiophotoluminescent glass dosemeters for stimulated luminescence dosimeters to therapeutic
eye lens dosimetry,” Radiation Protection Dosimetry electron beams,” Journal of Medical Physics 42(1), 42–
170(1-4), 208–212 (2016). 47 (2017).
121. J.-E. Rah, et al., “Clinical application of glass dosime- 134. S. B. Scarboro, et al., “Characterization of the nanoDot
ter for in vivo dose measurements of total body irradi- OSLD dosimeter in CT,” Medical Physics 42(4), 1797–
ation treatment technique,” Radiation Measurements 1807 (2015).
46(1), 40–45 (2011). 135. S. B. Scarboro and S. F. Kry, “Characterization of
122. D. Y. C. Huang and S.-M. Hsu, “Radio-photo energy response of Al(2)O(3):C optically stimulated
luminescence glass dosimeter (RPLGD),” in Advances luminescent dosimeters (OSLDs) using cavity theory,”
in Cancer Therapy (IntechOpen, 2011). Radiation Protection Dosimetry 153(1), 23–31 (2013).
123. S. Hashimoto, et al., “Energy dependence of a radio- 136. A. S. Meigooni, et al., “Influence of the variation of
photoluminescent glass dosimeter for HDR 192Ir energy spectra with depth in the dosimetry of192Ir
brachytherapy source,” Medical Physics 46(2), 964– using LiF TLD,” Physics in Medicine and Biology
972 (2019). 33(10), 1159–1169 (1988).
124. T. Nose, et al., “In vivo dosimetry of high-dose-rate 137. A. H. Zhuang, and A. J. Olch, “Validation of OSLD
brachytherapy: Study on 61 head-and-neck cancer and a treatment planning system for surface dose
patients using radiophotoluminescence glass dosim- determination in IMRT treatments,” Medical Physics
eter,” International Journal of Radiation Oncology, 41(8 Part 1), 081720 (2014).
Biology, Physics 61(3), 945–953 (2005). 138. E. A. Gardner, et al., “In vivo dose measurement using
125. M. Tsuda, “A few remarks on photoluminescence TLDs and MOSFET dosimeters for cardiac radio-
dosimetry with high energy X-rays,” Igaku Butsuri surgery,” Journal of Applied Clinical Medical Physics
20(3), 131–139 (2000). 13(3), 190–203 (2012).
126. T., Nose, et al., “In vivo dosimetry of high-dose- 139. G. Anagnostopoulos, et al., “In vivo thermolumines-
rate interstitial brachytherapy in the pelvic region: cence dosimetry dose verification of transperineal
Use of a radiophotoluminescence glass dosimeter 192Ir high-dose-rate brachytherapy using CT-based
for measurement of 1004 points in 66 patients planning for the treatment of prostate cancer,”
with pelvic malignancy,” International Journal International Journal of Radiation Oncology, Biology,
of Radiation Oncology, Biology, Physics 70(2), Physics 57(4), 1183–1191 (2003).
626–633 (2008). 140. R. Das, et al., “Thermoluminescence dosimetry for in-
127. J. A. Nieto, “Thermoluminescence dosimetry (TLD) vivo verification of high dose rate brachytherapy for
and its application in medical physics,” AIP Conference prostate cancer,” Australasian Physical and Engineering
Proceedings 724(1), 20–27 (2004). Sciences in Medicine 30(3), 178–184 (2007).
128. F. H. Yusof, et al., “On the use of optically stimulated 141. W. Toye, et al., “An in vivo investigative protocol for
luminescent dosimeter for surface dose measurement HDR prostate brachytherapy using urethral and rectal
during radiotherapy,” PLoS One 10(6), e0128544– thermoluminescence dosimetry,” Radiotherapy and
e0128544 (2015). Oncology 91(2), 243–248 (2009).
129. A. C. Riegel, et al., “In vivo dosimetry with optically 142. I. A. Brezovich, et al., “In vivo urethral dose measure-
stimulated luminescent dosimeters for conformal ments: A method to verify high dose rate prostate treat-
and intensity modulated radiation therapy: A 2-year ments,” Medical Physics 27(10), 2297–2301 (2000).
143. E. Ekwelundu, et al., “Custom-designed mouthpiece 145. T. Kron, et al., “A technique for total skin electron
for HDR brachytherapy of embryonal rhabdomyo- therapy (TSET) of an anesthetized pediatric patient,”
sarcoma of the soft palate,” Journal of Contemporary Journal of Applied Clinical Medical Physics 19(6), 109–
brachytherapy 6(3), 300–303 (2014). 116 (2018).
144. C. Holloway, et al., “Total body irradiation (TBI) opti- 146. P. Alvarez, et al., “TLD and OSLD dosimetry systems
cally stimulated luminescence in vivo dosimetry,” for remote audits of radiotherapy external beam calibra-
Medical Physics 40, 223 (2013). tion,” Radiation Measurements 106, 412–415 (2017).
Chapter 12
Dosimeters and Devices for IMRT QA

Nesrin Dogan and Matthew T. Studenski
University of Miami
Miami, Florida
Perry B. Johnson
University of Florida Health Proton Therapy Institute
Jacksonville, Florida
CONTENTS
12.2 Patient-Specific QA 174
12.2.1 Rationale 174
12.2.2 Quantitative Evaluation Tools 175
12.2.3 Dosimeters and Devices 175
12.2.3.1 Point Measurements 175
12.2.3.2 Measurement Using Film 176
12.2.3.3 Detector Arrays 177
12.2.3.4 EPID-Based Analysis 179
12.2.3.5 Log File Analysis 179
12.2.3.6 Transmission Detectors 179
12.2.3.7 3D Dosimeters 180
12.3 Periodic QA 180
12.3.2 Acceptance and Commissioning 180
12.3.2.1 Tests 180
12.3.2.2 Equipment 181
12.3.3 Daily QA 181
12.3.3.1 Tests 181
12.3.4 Monthly QA 181
12.3.4.1 Tests 181
12.3.5 Annual QA 182
12.3.5.1 Tests 182
12.4 Summary 182
References 182
173
12.1 INTRODUCTION updated recommendations for machine-specific QA,

Intensity-modulated radiation therapy (IMRT) and which include more stringent tolerances for IMRT.
volumetric modulated arc therapy (VMAT) are sophis- AAPM TG-119 report [12] also provides patient-specific
ticated treatment techniques that are widely available IMRT QA guidelines, including evaluation criteria for
in routine clinical practice [1–5]. In IMRT, treatments QA. The NCS Reports 22 and 23 [20, 21] present a com-
are delivered by a series of multiple small beams or prehensive review of many QA methods and tests pub-
beamlets using either static or dynamic multi-leaf col- lished in the literature and recommended an extensive
limator (MLC)-based techniques. With VMAT, both set of QA techniques and tools (both measurement and
treatment planning and delivery are especially complex non-measurement) for IMRT and VMAT. Recently pub-
as patient treatment is delivered while the gantry rotates lished AAPM TG-218 presents a comprehensive analysis
around the patient, the MLC is dynamically shaping the of performance of tolerance limits and methodologies
beam and the dose rate and gantry rotation speed are such as dose difference/distance-to-agreement (DTA)
constantly changing. As compared to conventional 3D and gamma index metrics and provides specific guide-
conformal radiation therapy (3D CRT), delivery of such lines on tolerance and action limits for patient-specific
complex treatments imposes new demands and require- IMRT QA [22]. The report suggests a universal tolerance
ments for machine and patient-specific quality assur- limit of >95% gamma passing rate with 3%/2 mm DTA
ance (QA) and verification methods [6–7]. and 10% dose threshold for evaluation of IMRT QA.
Currently, there are variety of commercially avail- With that knowledge in mind, it is essential to choose
able dosimeters and measurement devices specifically suitable dosimetry devices and tools for accurately test-
designed for IMRT QA, including treatment planning as ing the limitations of different components of an IMRT
well as machine and patient-specific QA purposes. The system such as treatment planning, delivery systems,
selection of a specific dosimeter system depends on the and accurate comparisons of calculated and measured
type of measurement and the endpoint of the preferred IMRT dose distributions. Appropriate choice of dosime-
QA process [8–10]. ters and phantoms is expected to significantly reduce the
Many reports and papers on the use of QA devices magnitude and frequency of errors associated with each
and techniques have been published since the initial component in the IMRT QA program. Patient-specific
clinical utilization of IMRT. Most of the IMRT tech- QA is one of the most important aspects of an IMRT QA
niques published in the early 2000s relied on the use of program since it assures appropriate treatment plans are
ionization chambers and radiographic film dosimetry created, transferred, and delivered correctly for a spe-
for QA of IMRT fields [11–13]. Use of such traditional cific patient. In this chapter, we provide a comprehen-
methods is time-consuming and limits the analysis of sive review of the dosimeters and devices needed for
dose to a 2D plane. However, over the last decade, IMRT proper patient-specific QA, including a description of
techniques and tools have rapidly evolved and a variety each device, the acceptance testing and commissioning
of new IMRT QA devices such as two-dimensional (2D) process, and the performance of periodic QA tests.
array detectors, 3D detectors, electronic portal imag-
ing devices (EPIDs), radiochromic films, and 3D gel 12.2 PATIENT-SPECIFIC QA
dosimetry have become available [14–18]. AAPM Task 12.2.1 Rationale
Group (TG) Report 120 on dosimetry tools and tech- Patient-specific QA, also known as delivery QA, IMRT
niques for IMRT [8] provides a comprehensive review QA, or measurement-based verification QA, defines a
of available dosimeters and phantoms for IMRT com- category of tests related to the comparison between the
missioning and verification of IMRT plans. This report dose planned for a specific patient and the dose mea-
includes how ionization chambers, films, EPIDs, array sured when that plan is delivered to a QA device. The
detectors, and a variety of phantoms (geometric and comparison can be made at a single point, but more
anthropomorphic) should be used, including limita- often than not is made between spatial dose distribu-
tions and appropriate use of each dosimeter and quan- tions. Patient-specific QA has been a prevalent aspect
titative comparison techniques for dose distribution of radiation oncology since the introduction of IMRT
evaluation. AAPM TG-142 report [19] indicates the in the 1990s [23–25]. At the time, there arose the need
need for dosimetry and measurement tools and provides for the validation of complex dose distributions whose
Dosimeters and Devices for IMRT QA ◾ 175
quality could no longer be guaranteed through conven- 12.2.2 Quantitative Evaluation Tools
tional machine QA and measurement-based, indepen- Dose difference, DTA, and gamma evaluation are the
dent dose calculations, that is, hand calculations. In most commonly used tools for the quantitative com-
terms of hardware, it also became necessary to assess parison of dose distributions (e.g., measured versus
the deliverability of plans given the mechanical limita- calculated) for patient-specific QA although there are
tions of treatment delivery systems that may influence a number of QA tools for patient-specific IMRT QA
the physical manifestation of the calculated fluence reported in the literature [33–35]. The DTA was first
pattern, for example, leaf speed, leaf positioning, and used by van Dyk et al. for QA of treatment planning
dose rate. With regard to patient safety, patient-specific systems [36]. The gamma evaluation, which combines
QA was also identified as a critical barrier protecting the dose difference and DTA criterion for each point of
against failures such as plan corruption, missing data, interest in a single criterion, was first developed by Low
or poor beam modeling [26]. These factors helped lead et al. for evaluation of 2D dose distributions [34]. The
to the establishment of patient-specific QA as a rou- tolerance and acceptance levels of gamma evaluation
tine aspect of the pre-treatment review process. This may be influenced by many factors, including IMRT
practice was later codified in several society-based rec- QA equipment and delivery technique. The DTA is
ommendations [22, 27–30], and subsequently added by influenced by the spatial resolution of the dose mea-
many payers as a requirement for billing codes related suring device and the dose calculation grid size. The
to IMRT planning. accurate determination of an appropriate dose toler-
In the present, the validation of complex dose distri- ance level for gamma evaluation can only be obtained
butions has been addressed, to an extent, through the by carrying out an error analysis of the IMRT QA
proliferation of software systems capable of providing device used for the measurement, treatment planning
independent dose calculations for IMRT plans. This has system, and the choice of 2D versus 3D gamma eval-
opened a debate as to whether or not measurement-based uation. A criterion of at least 3%/3 mm DTA with a
validation is warranted in lieu of computational alterna- 10% dose threshold with minimum pass rate of 90%
tives [31, 32]. A key argument in favor of measurement is generally considered a good agreement between cal-
is that with the adoption of advanced modalities such as culated and measured dose using absolute dosimetry
SBRT, VMAT, and the forthcoming 4π radiotherapy, the [20, 21, 34]. A 3 mm or less dose calculation grid size
physical demands on treatment delivery systems have should be used to achieve 3 mm or better DTA [12].
increased. In order to produce a fluence pattern that Recently published AAPM TG-218 report provides a
matches the one calculated for an advanced modality, comprehensive review of all IMRT QA tools, including
modern treatment delivery systems must operate within practical considerations using gamma evaluation crite-
tighter tolerances while accounting for an increase in ria [22]. The report recommended universal tolerance
the degrees of freedom made available to the treatment limits of ≥95% (and 90%) gamma passing rate with
planning system. As such, the deliverability of IMRT 3%/2 mm DTA and 10% dose threshold and local tol-
plans remains a key issue that cannot be addressed by erance limits for IMRT QA using gamma evaluation
software alone. Another aspect of patient-specific QA, method.
which justifies measurement, is the way in which it
validates a variety of factors through a single test, for 12.2.3 Dosimeters and Devices
example, beam model, dose calculation, plan transfer, 12.2.3.1 Point Measurements
and deliverability. It could be argued that many of these Patient-specific QA in its simplest form compares the
factors should be addressed through proper IMRT com- dose predicted at a single location to the dose delivered
missioning, yet, a patient-specific measurement goes to that location when measured using a point detector,
further by extending these checks to the unique circum- that is, a zero-dimensional measurement. The point
stances that exists for a given patient and a given plan. detector must be capable of providing absolute dose
It is the comprehensive nature of the method that helps through either a calibration factor traceable to a pri-
define its value and maintains its status as one of the mary standard or by cross-calibration with a traceable
most commonly practiced QA measures in the radiation device. The size of the detector must be such that par-
oncology clinic. tial volume effects are minimized. This depends on the
active volume of the detector, the dose gradient in the immediate readout and offer an improved spatial resolu-
area surrounding the measurement location, and the tion. MOSFETS, however, show a decreased linearity for
size and shape of the individual IMRT fluence patterns. low doses (<50 cGy) and can over-respond to phantom
Other desirable traits include common features such as scatter. Previous systems utilizing MOSFETs for patient-
energy independence, high sensitivity, limited direc- specific QA implemented patterns of detectors within
tional dependency, linearity with dose, tissue equiva- water equivalent plastic phantoms, both cylindrical and
lence, dose rate independence, and low cost. slab-based [38]. In this way, they were able to provide a
In the patient-specific QA setting, vented ion cham- sample of a 2D or 3D dose distribution. While TLDs,
bers meet most of needs described above and have com- OLSDs, and MOSFETs are commonly used for in vivo
monly been used for this purpose. As noted, the volume dosimetry and end-to-end testing, their use as a routine
of the ion chamber should be small enough to limit par- tool for patient-specific QA is limited due to better alter-
tial volume effects for the given application. An active natives as described in the following sections.
volume of 0.125 cm3 or less has been previously recom-
mended as a practical cutoff [8, 12]. Consideration must 12.2.3.2 Measurement Using Film
also be given to the size and composition of the mate- Historically, the transition from point dose to the mea-
rial within which the ion chamber is placed, that is, the surement of spatial dose distributions came through the
physical phantom. The application of cavity theory for combination of ion chambers and film. While a prop-
gas filled detectors is most directly applied when charged erly calibrated film is capable of providing absolute dose,
particle equilibrium (CPE – or transient CPE) exists. the accuracy and reproducibility of the measurement
As such, the ideal measurement conditions include a are dependent upon the measurement conditions, film
physical phantom with a homogenous elemental com- processing, and analysis technique [39]. The challenges
position and density, and a measurement location that are such that in the patient-specific QA setting, film is
avoids regions where CPE may be lost. Examples of the often used to verify the relative dose distribution while a
latter include the initial build-up region and any area point measurement made with a calibrated ion chamber
immediately above or below an imbedded heterogeneity. is used to verify absolute dose. In order to acquire a rela-
The physical phantom should closely match the virtual tive dose distribution, a film must be exposed, processed
phantom that was used to create the verification plan. (depending on film type), scanned, digitized, and cali-
Size, shape, and elemental composition should all be brated using a sensitometric curve. As opposed to ion
considered. Practically speaking, phantoms made from chambers, which require extensive corrections in order
slabs of water-equivalent plastic represent an excellent to accurately measure dose in regions where CPE does
choice as they meet the conditions described above and not fully exist, film has few limitations in this regard
are relatively easy to acquire/construct. Liquid water and represents a suitable choice for measuring dose
may also be a reasonable option depending upon the at the surface of a phantom or in the build-up region.
delivery technique and whether or not the container For patient-specific QA, the film is most often oriented
holding the liquid water is taken into account. Different within a physical phantom at a plane bisecting the high
geometric and anthropomorphic phantoms have also dose region of the dose distribution. If the phantom can
been offered by several vendors and represent a more accommodate more than one piece of film, concurrent
realistic interpretation of the clinical setting. measurements can be made either along different paral-
Aside from ion chambers, other point detectors lel planes and/or in an orthogonal configuration.
previously used for patient-specific QA include TLDs, Depending upon the sensitivity and corresponding
OSLDs, and MOSFETs. All three represent small, low dynamic range of the film, the calculated monitor units
cost, integrating dosimeters, which are robust to fre- may require linear scaling to avoid a saturated response.
quent handling. TLDs and OSLDs have long been Early implementation with radiographic film (XV2,
used within anthropomorphic phantoms for exter- Eastman Kodak Co., Rochester, NY) showed a satura-
nal dosimetry audits and protocol credentialing [37]. tion limit near 100 cGy [40]. This was later improved
They are reusable, versatile, and capable of 2–3% accu- to approximately 600 cGy with the release of extended
racy when properly calibrated. MOSFETs share many dose range film (EDR2) by the same manufacturer [41].
of the same features of TLDs/OSLDs but provide an The difference between the two products is largely the
result of a reduction in the size of AgBr grains, which sensitometric a curve is created by exposing blank films
were decreased by 90% in the EDR2 film. The change to known dose levels. Ideally, calibration films should
in silver content also affected the energy dependence of be exposed/scanned with each QA measurement. To
the film, which had been a previous limitation requiring facilitate this process, some institutions have devel-
calibration (optical density to dose) at the same depth as oped calibration plans that simultaneously deliver strip
the intended measurement. patterns at known dose levels [44]. The sensitometric
To reveal the latent image imparted by a radiation conversion is the final step in obtaining a relative dose
exposure, radiographic film requires chemical process- distribution. This distribution can be compared, in rela-
ing. Prior to the wide scale adoption of digital imag- tive terms, with a dose plane extracted from the veri-
ing, film processors were prevalent in the radiation fication plan at the same depth and orientation as the
oncology clinic. Relevant aspects of film processer QA film. As noted previously, an ion chamber is typically
include the periodic assessment of radiographic films utilized to verify absolute dose. The ion chamber is most
exposed to the same dose to assess for constancy and often located above or below the film near the isocenter.
scheduled maintenance/cleaning to avoid streaking and In scenarios where an ion chamber may exhibit partial
jamming. In the present, digital imaging for patient volume effects, such as the case for stereotactic radio-
alignment has removed the primary need for film pro- surgery fields, the calibrated film may be used directly
cessors and radiographic film. For QA applications that in a comparison between absolute dose distributions. In
maintain the utilization of film, the conversion has been either case, the measured distribution is aligned with
made to radiochromic forms [42]. The most common the calculated distribution through registration marks
radiochromic film used for patient-specific QA is the that have been placed on the film. The two distributions
commercial product Gafchromic® EBT (International can then be compared using various similarity metrics
Specialty Products, Wayne, NJ). EBT, like other radio- such as the gamma analysis.
chromic film, is self-developing, largely unaffected by
interior room light over short periods of time, rela- 12.2.3.3 Detector Arrays
tively energy independent/tissue-equivalent, inherently To improve the efficiency of patient-specific QA, several
slow to expose, and provides excellent spatial resolu- vendors have introduced detector arrays, which allow
tion. Successive versions (EBT, EBT2, and EBT3) have for easy setup and rapid evaluation of the measured
improved uniformity and consistency, and reduced the dose. The first iteration of these devices was configured
dependence on scan orientation. The latter is a char- as 2D arrays using either diodes or vented ion cham-
acteristic of radiochromic film, which is traditionally bers (see Table 12.1). The spacing between the detectors
scanned using a multichannel flatbed scanner. Both a is typically 7 mm, though some devices utilize a 10 mm
consistent orientation and a centralized position within spacing toward the periphery. In either scenario, the spa-
the flatbed scanner are important when scanning cali- tial resolution is significantly reduced in comparison to
bration and verification films. It is also important to film. This may present problems when measuring dose
wait a consistent length of time before scanning, and for in small fields or fields with very steep dose gradients.
the waiting period to be long enough (>24 h) to allow One solution is to acquire two measurements with the
for color formation to reach a plateau. Depending on device shifted during the second acquisition by half the
how the film is made, there will be different absorption width of the detector spacing. The two measurements
patterns for different color spectra. EBT film is typically can then be combined to double the number of detectors
read out using three color channels, red, green, and blue. within the field. Recently, detector arrays intended for
The red channel represents the component most sensi- use in stereotactic applications have been released. The
tive to ionizing radiation and is widely used for single detector spacing for these devices has been reduced to
channel dosimetry. Multichannel dosimetry is more 2.5 mm, which may be suitable for many cases.
complex, as it requires model-based corrections, but Detector arrays are often placed within or beneath
affords the opportunity to improve uniformity and/or slabs of water-equivalent plastic to better simulate
extend the dynamic range of the film [43]. patient dosimetry. Several vendors provide device-
Once scanned, film requires the use of a sensito- specific phantoms for this purpose, while others con-
metric curve to convert from optical density to dose. A figure their products with inherent buildup. Like film,
TABLE 12.1 Devices and Tools Used for Patient-Specific QA

Manufacturer Device Type Notes
Various Ion chamber Point detector V <0.125 cm3, location should avoid loss of
CPE
LANDAUER OSLD Point detector
BEST medical MOSFET Point detector
Various TLD Point detector
Various Film Point detector/spatial Gafchromic EBT3 is more popular, scanning
distribution and calibration required
Sun Nuclear MapCHECK 3 Detector array (2D planar) 1527 diodes, 26 × 32 cm2, 7.07 mm spacing
Sun Nuclear SRS MapCHECK Detector array (2D planar) 1013 diodes, 7.7 × 7.7 cm2, 2.47 mm spacing
Sun Nuclear ArcCHECK Detector array (2D helical) 1386 diodes arranged in a helical pattern
PTW OCTAVIUS 1500 Detector array (2D planar/ 1405 vented ion chambers, 27 × 27 cm2,
rotational) 7.07 mm spacing, ability to rotate w/gantry
PTW OCTAVIUS 1000 SRS Detector array (2D planar) 977 liquid filled ion chambers, 10 × 10 cm2,
2.5 mm spacing
ScandiDos Delta4 PHANTOM+ Detector array (2D 1069 diodes total (2 planes), 20 × 20 cm2,
orthogonal) 5 mm central spacing, 10 mm outer spacing
IBA MatriXX Detector array (2D planar) 1020 vented ion chambers, 24.4 × 24.4 cm2,
7.62 mm spacing
ScandiDos Delta4 DISCOVER Transmission chamber 4040 diodes, 1.5 mm spacing, link with
phantom-based IMRT QA
IBA Dolphin Transmission chamber 1513 ion chambers, 40 × 40 cm2, 5 mm
spacing
PTW DAVID Transmission chamber Multiwire ion chamber, real-time, no link to
pre-treatment IMRT QA
Varian Medical Systems Portal dosimetry EPID based
Sun Nuclear PerFRACTION EPID based/log file analysis
Mobius Medical Systems Mobius 3D Log file analysis
detector arrays provide a measurement of dose within orientation. Then for a given gantry angle, the 3D dose
a phantom analogous to dose within a patient. Another distribution is derived based upon depth-dependent
similarity is that calibration is required to convert attenuation/scatter factors, which are used to scale the
response to dose, and additionally to ensure a uniform measured dose [47]. Here, the dose is scaled along ray
response among the detectors. In contrast to film, 2D lines, which pass through the 2D measurement plane.
arrays (diode versions in particular) have been shown The composite dose from all angles provides the full 3D
to exhibit nontrivial angular dependence [45, 46]. As a dose distribution. A slightly different approach is uti-
result, the devices are often irradiated using an orienta- lized by the Delta4 PHANTOM (ScandiDos, Uppsala,
tion perpendicular to the beam. This can be achieved by Sweden), which includes two orthogonal detector
either setting the gantry angles of the verification plan planes. For this device, measurement-guided correction
to zero or by rotating the QA device in synchronization factors are applied to the dose distribution as calculated
with the gantry. The latter technique can further be bro- by the TPS. The correction factors are based upon the
ken down into methods that affix the 2D array to the dose measured at each detector and then applied along
gantry or EPID panel, or methods that rotate the array ray lines, which pass through said detectors [48]. The
itself using a motor-driven platform (see Octavius 4D). ArcCHECK (Sun Nuclear, Melbourne FL) represents
With the wide scale adoption of arc delivery, new yet another approach, whereby detectors are arranged
devices have been released to better address this issue in a helical fashion. The helical design maintains a near-
while also collecting a broader sample of the 3D dose perpendicular orientation for arc subfields while also
distribution. The aforementioned Octavius 4D (PTW, providing a novel method for reconstructing the 3D
Freiburg, Germany) represents one such device, which dose distribution (phantom or patient) based upon
utilizes a rotating platform to maintain an enface entrance and exit dose. Here, the diode measurements
are used as corrections factors to scale a pre-calculated, files capture with high fidelity the actual MLC leaf posi-
relative dose distribution [49]. tioning [32, 57]. The latter point likely depends upon the
mechanism being used to report the leaf position, the
12.2.3.4 EPID-Based Analysis tolerance within which it operates, and its robustness to
When utilized for patient-specific QA, the EPID avail- failure. Generally speaking, mechanisms that monitor
able on modern gantry-based delivery systems combines outputs (i.e., optical or magnetic sensors) appear more
the high spatial resolution of film with the efficiency and aptly suited than those that monitor inputs (i.e., step-
convenience of detector arrays [50]. Several versions of per motor position). Either way, log file analysis affords
EPID-based QA exist including in-house programs, leg- many potential benefits including the ability to auto-
acy software such as EPIDose™ (Sun Nuclear), vendor- mate QA, the ability to separate treatment planning
integrated systems such as Portal Dosimetry (Varian errors from treatment delivery errors, and the option to
Medical Systems, Palo Alto, CA), and new cloud-based provide continual monitoring over a course of treatment
platforms such as EPIbeam™ (DOSIsoft, Paris, France) due to the fact that log files are created each time the
and PerFraction™ Fraction 0 (Sun Nuclear) [51–55]. Like plan is delivered.
detector arrays, the EPID panel requires calibration in
order to ensure a uniform response. Typically, this is 12.2.3.6 Transmission Detectors
achieved via a dark field calibration, which adjusts for Thin transmission detectors are a relatively new class of
background noise and a flood field calibration, which device designed primarily for treatment monitoring but
accounts for variation in pixel sensitivity. In contrast to may also be configured as a tool for patient-specific QA.
detector arrays, the output of the QA measurement is Currently, there are three commercial systems available.
an indirect indicator of the in-phantom dose distribu- The first system, DAVID (PTW, Freiburg, Germany),
tion. As such, a comparison must be made either with utilizes a multiwire ion chamber design where each
a predicted panel response (Portal Dosimetry), or the wire is aligned beneath an individual leaf pair of the
panel response must be converted to an in-phantom MLC [58]. During the initial delivery, the charge accu-
dose plane (EPIDose™ and EPIbeam™). Both methods mulated within each wire creates a histogram, which is
require extensive commissioning. A third approach as subsequently used as a baseline for future deliveries. The
implemented by PerFraction™ utilizes the detector panel system, thus, does not perform patient-specific QA, but
as way to determine the position of individual leaves instead functions as a sophisticated monitoring device
during a dynamic delivery. This information is then to ensure an accurate delivery.
combined with data extracted from machine log files The second system, Delta4 DISCOVER (ScandiDos,
(output, gantry angle, energy, etc.) to reconstruct the Uppsala, Sweden), utilizes 4040 p-type diode detectors
incident energy fluence. The fluence can then be forward and operates much like the DAVID system in the way
projected through a CT dataset to calculate a 3D dose it creates a baseline measurement [59]. The system goes
distribution. further, however, by providing an option to link the ini-
tial baseline to a measurement made using the Delta4
12.2.3.5 Log File Analysis PHANTOM. This linking allows the Delta4 DISCOVER
For many systems, the log files described above also to predict phantom measurements for subsequent frac-
include time-resolved data related to MLC leaf position- tions and analyze the results using the same software
ing. This information can be used in a similar fashion tools as used for the initial patient-specific QA.
to derive the incident energy fluence and subsequent 3D The third system, Dolphin (IBA, Schwarzenbruck,
dose distribution. A number of in-house and commer- Germany), incorporates 1513 vented ion chambers with
cial programs utilize this method (Mobius 3D, Varian a 5 mm spacing at isocenter [60]. Dolphin serves as both
Medical Systems, Palo Alto, CA) or provide it as an a patient-specific QA device and an on-line monitor-
option (PerFraction™) [54–56]. The use of log files for ing device. In either case, there is no need for a separate
patient-specific QA has been a topic of much debate for QA tool. The Dolphin system is similar to a number of
reasons concerning the independent nature of the tech- others in the way it performs a forward-based dose cal-
nique, whether it constitutes a “physical measurement” culation using a derived energy fluence. To find said flu-
as required by many payers, and whether or not the log ence, an ideal energy fluence is modified based upon the
measured detector response and the way in which that validation of new techniques and may serve a role in
response differs from the ideal response as predicted the future as a way to validate processes related to the
by COMPASS, an associated software tool. The derived use of adaptive radiotherapy.
energy fluence can then be provided as an input, along
with the simulation CT, to a dose calculation engine 12.3 PERIODIC QA
using a collapsed cone convolution algorithm. The cal- 12.3.1 Rationale
culation results can be compared using a gamma analy- In addition to patient-specific QA, periodic tests need to
sis or DVH-based metrics. be done to ensure that the linear accelerator components
are functioning as they should. While patient-specific
12.2.3.7 3D Dosimeters QA will catch an error in a particular treatment plan, it
As noted for many of the devices above, there is an can be difficult to determine what aspect of the plan is
ongoing transition from 2D to 3D dosimetry for the actually causing the error. For example, is a plan failing
measurement and analysis of patient-specific QA. due to an output calibration, inconsistent flatness and
In each case, a novel approach has been developed symmetry or from MLC misalignment? For this reason,
to derive the 3D dose distribution based, essen- the periodic QA on a linac is vital to understanding how
tially, on a collection of 2D measurements. These your machine is performing over time and for trouble-
approaches include correction-based interpolation, shooting intermittent problems.
measurement-guided dose reconstruction, 3D back- For linacs delivering IMRT and VMAT treatment
projection/calculation, and 3D forward-projected/ plans, one good resource for periodic QA was published
calculation. In contrast to such methods are 3D by the NCS in 2015 [21]. This document focuses specifi-
dosimeters, which inherently record a high resolu- cally on different tests for VMAT. In terms of periodic
tion, 3D dose distribution through changes that occur linac QA, AAPM TG-142 is the current standard [19].
within the composition of a 3D volume. Traditionally As the focus of this chapter is specifically IMRT and
known as “gel dosimetry” due to the fact that the vol- VMAT QA, this section will focus on the QA tests and
ume comprises 80–90% water and 5% gelatin, the type equipment pertinent to IMRT and VMAT QA only.
of change which occurs depends upon the remaining
material, which may either induce a polymerization 12.3.2 Acceptance and Commissioning
process (polymer-based gels) [61] or an oxidation 12.3.2.1 Tests
process (Fricke gels) [62]. These changes are actual- When a new linear accelerator is installed, it is impor-
ized through an optical-CT or MRI-based acquisition tant to understand what type of procedures will be per-
from which a 3D response map can be derived. The formed on that machine in order to ensure the proper
response map can then be converted to dose through a components are examined. For a linear accelerator that
calibration procedure. The advantages of gel dosime- will be used for IMRT (especially VMAT), the number
try include the 3D nature of the measurement, energy of specific components to check greatly increases from
and dose-rate independence, high sensitivity and lin- standard 3D conformal therapy due to complexity of the
ear response, high spatial resolution, a near-tissue delivery. In addition, this is the time where you can run
equivalence, and the fact that gels can be prepared your initial QA tests to establish baseline values that you
in anthropomorphic forms with various densities. can use for your subsequent periodic QA.
Limitations include a sensitivity to time, temperature The complexity of IMRT delivery requires significant
and preparation method, the cost of the materials, testing before safely treating patients. During VMAT
the availability of MRI or optical-CT for processing, delivery, the gantry is in motion along with the MLC.
the need for frequent calibrations, the labor require- Therefore, testing these individual components in addi-
ments of the process, and the delay between measure- tion to their interplay with each other is vital. The NCS
ment and readout. At this time, the limitations of the [21] recommended tests examine the basic mechanical
method combined with the availability of attractive properties of the linac including the accuracy of both the
alternatives prevent 3D dosimeters from achieving gantry and collimator at all angles. Due to the dynamic
widespread use as tool for patient-specific QA. 3D delivery in VMAT, they also recommend checking the
dosimeters do, however, offer unique solutions for the gantry speed.
Again, due to the dynamic nature of VMAT in both PiPsPro to process these images or the film or in-house
mechanics and output, the linac output must be tested code needs to be written.
at different gantry angles to ensure proper delivery at all
times. First, it is recommended to check that the output 12.3.3 Daily QA
is linear and reproducible at the nominal dose rate. The 12.3.3.1 Tests
recommended tests for flatness and symmetry are more The purpose of daily QA is to catch any significant
detailed. Additionally, it is recommended to check at the changes in linac functionality that could impact patient
lowest dose rate at all gantry angles during delivery and treatment. The tests need to be simple and efficient so
at different dose rates to simulate an actual delivery. that treatments can begin on time but they also need
The final component recommended to be tested to catch the serious errors that could reach the patient.
is the MLC. The initial test should be to test the MLC AAPM TG-142 is a good resource for appropriate daily
leaf speed. The MLC position during a dynamic deliv- QA tests [19]. The tests are broken down into dosimetry,
ery should also be tested. Finally, tests that combine the mechanical, safety, and imaging, and tolerance limits
mechanical, output, and MLC should be run. These tests are provided based on the delivery type. The dosimetry
include output checks with varying dose rate while the test is designed to catch a significant variation in out-
gantry is rotating and varying gantry speed and dose put, meaning a change greater than 3%. The mechanical
rate. tests check the laser and ODI accuracy and that the col-
limator is at the appropriate position. The safety check
12.3.2.2 Equipment ensures that the patient can be seen and heard, the radi-
Although the tests for acceptance and commissioning ation monitors are functioning, and that the interlocks
are complex, the equipment needed to run the tests is are working. The imaging tests are to check that the col-
simple. One essential component is a scanning water lision interlocks are functioning and that the imaging
tank, usually three-dimensional. The water tank allows and treatment isocenters are coincident to within 2 mm
for beam characterization and for obtaining point for both MV and kV imaging.
measurements.
For the mechanical acceptance and commissioning 12.3.3.2 Equipment
tests, a standard level is sufficient to QA the gantry and The daily QA should be quick and efficient. Typically,
collimator angles. For the output measurements, an ion a detector array containing ion chamber and diodes is
chamber and an electrometer are required (discussed used to measure the output, energy, flatness, symmetry,
in Chapter 2). Depending on the test, a 1D or 3D water and field size simultaneously. SunNuclear DailyQA3,
tank can be used or solid water is also an option. PTW QuickCheck, and IBA StarTrack are examples of
The flatness and symmetry tests require some form such devices. A separate imaging phantom is used for
of a 1D or 2D linear array to acquire profiles. A gantry the imaging tests where an image is taken and com-
mount is also helpful in order to acquire the profiles pared to a DRR or CT image to assess isocenter coinci-
at different gantry angles although the cardinal angles dence. The QUASAR PentaGuide is an example of such
can be acquired by using solid water to support the a phantom.
device on the treatment couch. These linear arrays
can be composed of ion chambers or diodes. There 12.3.4 Monthly QA
are several different commercial options including 2D 12.3.4.1 Tests
arrays such as the SunNuclear Profiler (ion chamber The purpose of monthly QA is to catch degradations in
or diode), the IBA StarTrack, or the PTW StarCheck. machine performance over time before they can reach
These companies also offer 1D options as well like the the patient or cause machine malfunctions that inter-
SunNuclear WaterProof Profiler, the PTW LA48, and rupt the clinical workflow. Monthly QA tests tend to be
the IBA LDA-99. more involved and more precise and accurate than daily
The more complex tests require either the use of film QA tests. Again, AAPM TG-142 is a good resource along
or the EPID to acquire an image of the dynamic deliv- with the NCS report for the appropriate tests to run on
eries. In addition, there are software packages such as a monthly basis [19, 21]. For the dosimetric tests, output
Mobius DoseLab, RIT RITG142, or Standard Imaging constancy should be within 2% of the baseline at all dose
rates and the flatness and symmetry should be within test and a moving window test and four cardinal gantry
1% of the baseline. The NCS report also recommended angles [19].
to check flatness and symmetry on a monthly basis
[21]. The mechanical tests are to ensure that the gan- 12.3.5.2 Equipment
try and collimator rotation is within 1 degree. Specific The equipment required for annual QA is similar to that
MLC tests are also recommended for leaf speed and for monthly. The only extra piece of equipment would
positioning. The leaf speed should be compared to the be a 1D or 3D water tank to measure PDDs and to do
baseline value acquired during commissioning. A picket TG-51 [63].
fence test can be used for leaf positioning and multiple
gantry angles should be tested. The NCS report also 12.4 SUMMARY
recommends checking leaf positioning accuracy on a This chapter presents a comprehensive review of the
monthly basis [21]. The imaging tests are the same as the dosimeters and devices needed for proper patient-spe-
daily QA. cific QA, including a description of each device, the
acceptance testing and commissioning process, and
12.3.4.2 Equipment performing periodic QA tests. Patient-specific QA is a
The equipment needed to run the monthly tests is basi- critical component of IMRT and VMAT treatments.
cally the same that is used for acceptance and commis- Although currently there are many IMRT QA tools,
sioning. For the mechanical tests, a level is sufficient to every QA tool has limitations that need to be assessed
QA the gantry and collimator angles. For the output prior to its use for safe and quality treatments.
measurements, an ion chamber, electrometer, and solid
water are enough to collect the appropriate information. REFERENCES
The flatness and symmetry tests require some form 1. A. L. Boyer and C.X. Yu, “Intensity-modulated radiation
of a 1D or 2D linear array to acquire profiles. The MLC therapy with dynamic multileaf collimators,” Seminars
in Radiation Oncology 9(1), 48–59 (1999).
tests require either the use of film or the EPID to acquire
2. D. R. Puri, W. Chou, N. Lee, “Intensity-modulated
an image of the dynamic deliveries. The leaf speed can radiation therapy in head and neck cancers:
be tested by analyzing machine log files or with an ion Dosimetric advantages and update of clinical
chamber and a dynamic sweep of the leaves over the results,” American Journal of Clinical Oncology
chamber. 28(4), 415–423 (2005).
3. K. Otto, “Volumetric modulated arc therapy: IMRT
in a single gantry arc,” Medical Physics 35(1), 310–317
12.3.5 Annual QA (2008).
12.3.5.1 Tests 4. J. L. Bedford and A. P. Warrington, “Commissioning
The purpose of annual QA is to guard against major mal- of volumetric modulated arc therapy (VMAT),”
functions that might occur in machine components not Radiotherapy & Oncology 93(2), 259–265 (2009).
5. D. Rangaraj, S. Oddiraju, B. Sun, L. Santanam, D. Yang,
frequently checked before they can reach the patient or S. Goddu, L. Papiez, “Fundamental properties of the
cause machine malfunctions that interrupt the clinical delivery of volumetric modulated arc therapy (VMAT)
workflow. AAPM TG-142 recommends that a full TG-51 to static patient anatomy,” Medical Physics 37(8), 4056–
protocol is run to bring the absolute output to within 4067 (2010).
1% [19, 63]. The beam quality should also be within 6. M. Teoh, C. H. Clark, K. Wood, et al., “Volumetric mod-
ulated arc therapy: A review of current literature and
1% of the baseline value. Output constancy with dose
clinical use in practice,” British Institute of Radiology
rate and gantry angle and MU linearity should be veri- 84, 967–996 (2011).
fied. NCS recommends that linearity beyond 1000 MU 7. W. A. Woon, P. B. Ravindran, P. Ekayanake, Y. Y. F. Lim,
be tested. Flatness and symmetry should be within 1% “Validation of delivery consistency for intensity-modu-
of the baseline and small field output factor should be lated radiation therapy and volumetric-modulated arc
therapy plans,” Journal of Medical Physics 43(2), 119–
verified to be within 2% of the commissioned values [21].
128 (2018).
For the MLC, the transmission values should be mea- 8. A. D. Low, J. M. Moran, L. Dong, M. Oldham,
sured to ensure they match the planning system. “Dosimetry tools and techniques for IMRT,” Medical
TG-142 also recommended to do a segmental IMRT Physics 38(3), 1313–1338 (2011).
9. L. Masi, F. Casamassima, R. Doro, P. Francescon, 22. M. Miften, A. Olch, D. Mihailidis, J. Moran, T. Pawlicki,
“Quality assurance of volumetric modulated arc ther- A. Molineu, H. Li, K. Wijesooriya, J. Shi, P. Xia, N.
apy: Evaluation and comparison of different dosimetric Papanikolaou, D. A. Low, “Tolerance limits and meth-
systems,” Medical Physics 38, 612–621 (2011). odologies for IMRT measurement-based verification
10. C. De Wagter, “QA-QC of IMRT – European perspective,” QA: Recommendations of AAPM Task Group No. 218,”
in Image-Guided IMRT, edited by T. Bortfeld, R. Schmidt- Medical Physics 45(4), e53–e83 (2018).
Ullrich, W. De Neve, and D. E. Wazer, pp. 117–128 23. X. Wang, S. Spirou, T. LoSasso, et al., “Dosimetric veri-
(Springer-Verlag, Berlin, Heidelberg, New York, 2006). fication of intensity-modulated fields,” Medical Physics
11. R. Kulasekere, J. M. Moran, B. A. Fraass, P. L. Roberson, 23(3), 317–327 (1996).
“Accuracy of rapid radiographic film calibration for 24. C. S. Chui, S. Spirou, T. LoSasso, “Testing of dynamic
intensity-modulated radiation therapy verification,” multileaf collimation,” Medical Physics 23(5), 635–641
Journal of Applied Clinical Medical Physics 7(2), 86–95 (1996).
(2006). 25. B. Nelms and J. Simon, “A survey on planar IMRT QA
12. G. A. Ezzell, et al., “IMRT commissioning: Multiple analysis,” Journal of Applied Clinical Medical Physics
institution planning and dosimetry comparisons, a 8(3), 76–90 (2007).
report from AAPM Task Group 119,” Medical Physics 26. J. Moran, M. Dempsey, A. Eisbruch, et al., “Safety con-
36(11), 5359–5373 (2009). siderations for IMRT,” Practical Radiation Oncology
13. N. L. Childress, M. Salehpour, L. Dong, C. Bloch, R. 1(3), 190–195 (2011).
A. White, I. I. Rosen, “Dosimetric accuracy of Kodak 27. A. Hartford, J. Galvin, D. Beyer, et al. “American
EDR2 film for IMRT verifications,” Medical Physics College of Radiology (ACR) and American Society
32(2), 539–548 (2005). for Radiation Oncology (ASTRO) practice guideline
14. J. Son, T. Baek, B. Lee, D. Shin, S. Y. Park, J. Park, Y. K. for intensity-modulated radiation therapy (IMRT),”
Lim, S. B. Lee, J. Kim, M. Yoon, “A comparison of the American Journal of Clinical Oncology 35, 612–617
quality assurance of four dosimetric tools for intensity (2012).
modulated radiation therapy,” Radiology and Oncology 28. ACR practice parameters for intensity modulated radia-
49(3), 307–313 (2015). tion therapy (IMRT) (2016).
15. M. Hussein, E. J. Adams, T. J. Jordan, C. H. Clark, A. 29. ASTRO model policies, intensity modulated radiation
Nisbet, “A critical evaluation of the PTW 2D-ARRAY therapy (IMRT) (2015).
seven29 and OCTAVIUS II phantom for IMRT and 30. G. Ezzell, J. Galvin, D. Low, et al., “Guidance document
VMAT verification,” Journal of Applied Clinical Medical on delivery, treatment planning, and clinical imple-
Physics 14(6), 274–292 (2013). mentation of IMRT: Report of the IMRT subcommittee
16. M. Hussein, P. Rowshanfarzad, M. A. Ebert, A. Nisbet, of the AAPM radiation therapy committee,” Medical
C. H. Clark, “A comparison of the gamma index analy- Physics 30, 2089–2115 (2003).
sis in various commercial IMRT/VMAT QA systems,” 31. R. Alfredo, C. Siochi, A. Molineu, “Patient-specific QA
Radiotherapy & Oncology 109(3), 370–376 (2013). for IMRT should be performed using software rather
17. J. F. M. Colodro, A. S. Berná, V. P. Puchades, D. R. than hardware methods,” Medical Physics 40(7), 070601
Amores, M. A. Baños, “Volumetric-modulated arc ther- (2013).
apy lung stereotactic body radiation therapy dosimetric 32. N. Childress, Q. Chen, Y. Rong, “Parallel/opposed:
quality assurance: A comparison between radiochro- IMRT QA using treatment log files is superior
mic film and chamber array,” Journal of Medical Physics to conventional measurement-based method,”
42(3), 133–139 (2017). Journal of Applied Clinical Medical Physics 16(1), 4–7
18. M. A. Silveira, J. F. Pavoni, O. Baffa, “Three-dimensional (2015).
quality assurance of IMRT prostate plans using gel 33. D. A. Low, W. B. Harms, S. Mutic, J. A. Purdy, “A tech-
dosimetry,” Physica Medica 34, 1–6 (2017). nique for the quantitative evaluation of dose distribu-
19. E. E. Klein, J. Hanley, J. Bayouth, F. F. Yin, W. Simon, tions,” Medical Physics 25, 656–661 (1998).
et al., “Task Group 142 report: Quality assurance of 34. D. A. Low, and J. F. Dempsey, “Evaluation of the gamma
medical accelerators,” Medical Physics 36(9), 4197–4212 dose distribution comparison method,” Medical Physics
(2009). 30, 2455–2464 (2003).
20. NCS, “Code of practice for the quality assurance 35. J. M. Steers, and B. A. Fraass, “IMRT QA: Selecting
and control for intensity modulated radiotherapy.” gamma criteria based on error detection sensitivity,”
Netherlands Commission on Radiation Dosimetry Medical Physics 43(4), 1982–1994 (2016).
Report 22 (http://radiationdosimetry.org) (2013). 36. J. Van Dyk, R. B. Barnett, J. E. Cygler, P. C. Shragge,
21. NCS, “Code of Practice for the Quality Assurance “Commissioning and quality assurance of treat-
and Control for Volumetric Modulated Arc Therapy.” ment planning computers,” International Journal
Netherlands Commission on Radiation Dosimetry of Radiation Oncology, Biology, Physics 26, 261–273
Report 24 (http://radiationdosimetry.org) (2015). (1993).
37. A. Molineu, N. Hernandez, T. Nguyen, et al., 52. D. Bailey, L. Kumaraswamy, M. Bakhtiari, et al., “EPID
“Credentialing results from IMRT irradiations of an dosimetry for pretreatment quality assurance with
anthropomorphic head and neck phantom,” Medical two commercial systems,” Journal of Applied Clinical
Physics 40(2), 022101 (2013). Medical Physics 13(4), 82–99 (2010).
38. C. Chuang, L. Verhey, P. Xia, “Investigation of the use 53. B. Nelms, K. Rasmussen, W. Tome, “Evaluation of a
of MOSFET for clinical IMRT dosimetric verification,” fast method of EPID-based dosimetry for intensity-
Medical Physics 29(6), 1109–1115 (2002). modulated radiation therapy,” Journal of Applied
39. D. Low, “Quality assurance of intensity-modulated Clinical Medical Physics 11(2), 140–157 (2010).
radiotherapy,” Seminars in Radiation Oncology 12(3), 54. S. Bresciani, M. Poli, A. Miranti, et al., “Comparison
219–228 (2002). of two different EPID-based solutions performing pre-
40. N. Dogan, L. Leybovich, A. Sethi, “Comparative evalu- treatment quality assurance: 2D portal dosimetry ver-
ation of Kodak EDR2 and XV2 films for verification sus 3D forward projection method,” Physica Medica 52,
of intensity modulated radiation therapy,” Physics in 65–71 (2018).
Medicine and Biology 47, 4121–4130 (2002). 55. E. Hseih, K. Hansen, M. Kent, et al., “Can a commer-
41. A. Olch, “Dosimetric performance of an enhanced dose cially available EPID dosimetry system detect small
range radiographic film for intensity-modulated radia- daily patient setup errors for cranial IMRT/SRS?,”
tion therapy quality assurance,” Medical Physics 29(90), Practical Radiation Oncology 7, e283–e290 (2017).
2159–2168 (2002). 56. C. Nelson, B. Mason, K. Robinson, et al., “Commissioning
42. A. Diroomand-Rad, C. Blackwell, B. Coursey, results of an automated treatment planning verification
“Radiochromic film dosimetry, recommendations of system,” Journal of Applied Clinical Medical Physics
AAPM radiation therapy committee task group no. 55,” 15(5), 57–65 (2014).
Medical Physics 25(11), 2093–2115 (1998). 57. A. Agnew, C. Agnew, M. Grattan, et al., “Monitoring
43. A. Micke, D. Lewis, X. Yu, “Multichannel film dosim- daily MLC positional errors using trajectory log files
etry with nonuniformity correction,” Medical Physics and EPID measurements for IMRT and VMAT deliv-
38(5), 2523–2534 (2011). eries,” Physics in Medicine and Biology 59(9), N49–N63
44. A. Shukla, A. Oinam, S. Kumar, et al., “A calibration (2014).
method for patient specific IMRT QA using a single 58. G. Karagoz, F. Zorlu, M. Yeginer, et al., “Evaluation of
therapy verification film,” Reports of Practical Oncology MLC leaf positioning accuracy for static and dynamic
and Radiotherapy 18, 235–240 (2013). IMRT treatments using DAVID in vivo dosimetric sys-
45. J. Hosang, V. Keeling, D. Johnson, et al., “Interplay tem,” Journal of Applied Clinical Medical Physics 17(2),
effect of angular dependence and calibration field size of 14–23 (2016).
MapCHECK 2 on RapidArc quality assurance,” Journal 59. T. Li, T. Matzen, F. Yin, et al., “Diode-based transmis-
of Applied Clinical Medical Physics 15(3), 80–92 (2014). sion detector for IMRT delivery monitoring: A valida-
46. Y. Shimohigashi, F. Araki, H. Tominaga, et al., “Angular tion study,” Journal of Applied Clinical Medical Physics
dependence correction of MatriXX and its application 17(5), 235–244 (2016).
to composite dose verification,” Journal of Applied 60. J. Thoelking, Y. Sekar, J. Fleckenstein, et al.,
Clinical Medical Physics 13(5), 198–214 (2012). “Characterization of a new transmission detector for
47. B. Allgaier, E. Schule, J. Wurfel, “Dose reconstruction in patient individualized online plan verification and its
the OCTAVIUS 4D phantom and in the patient without influence on 6 MV X-ray beam characteristics,” Journal
using dose information from the TPS,” in PTW (2013). of Medical Physics 26(3), 200–208 (2016).
48. R. Sadagopan, J. Bencomo, R. Martin, et al., 61. M. Maryanski, J. Gore, R. Kennan, et al., “NMR relax-
“Characterization and clinical evaluation of a novel ation enhancement in gels polymerized and cross-
IMRT quality assurance system,” Journal of Applied linked by ionizing radiation: A new approach to 3D
Clinical Medical Physics 10(2), 104–119 (2009). dosimetry by MRI,” Magnetic Resonance Imaging 11,
49. B. Nelms, D. Opp, J. Robinson, et al., “VMAT QA: 253–258 (1993).
Measurement-guided 4D dose reconstruction on a 62. R. Kelly, K. Jordan, J. Battista, “Optical CT reconstruc-
patient,” Medical Physics 39(7), 4228–4238 (2012). tion of 3D dose distributions using the ferrous-ben-
50. W. Van Elmpt, L. McDermott, S. Nijsten, et al., “A litera- zoic-xylenol (FBX) gel dosimeter,” Medical Physics 25,
ture review of electronic portal imaging for radiother- 1741–1750 (1998).
apy dosimetry,” Radiotherapy & Oncology 88, 289–309 63. P. R. Almond, P. J. Biggs, B. M. Coursey, et al., “AAPM’s
(2008). TG-51 protocol for clinical reference dosimetry of high-
51. A. Esch, T. Depuydt, D. Huyskens, “The use of an energy photon and electron beams,” Medical Physics
aSi-based EPID for routine absolute dosimetric pre- 26(9), 1847–1870 (1999).
treatment verification of dynamic IMRT fields,”
Radiotherapy & Oncology 71(2), 223–234 (2004).
Chapter 13
Area and Individual

Radiation Monitoring
Nisy Elizabeth Ipe
Shielding Design, Dosimetry & Radiation Protection
CONTENTS
13.2 Protection and Operational Dose Quantities 186
13.3 Active and Passive Monitoring 188
13.3.1 Active Monitoring 188
13.3.2 Passive Monitoring 188
13.3.2.1 Individual Monitoring 188
13.4 Area Monitoring 188
13.4.1 Calibration of Area Monitors 188
13.4.1.1 Photon Calibration 189
13.4.1.2 Neutron Calibration 189
13.4.2 Desirable Properties of Area Monitors or Radiation Survey Meters 190
13.4.2.1 Sensitivity 190
13.4.2.2 Energy Dependence 190
13.4.2.3 Directional Dependence 190
13.4.2.4 Measurement Range 190
13.4.2.5 Response Time 190
13.4.2.6 Linearity of Response 190
13.4.2.7 Calibration and Long-Term Stability 190
13.4.2.8 Discrimination in Mixed Fields 190
13.4.2.9 Measurement Uncertainties 191
13.5 Photon Monitoring Techniques 191
13.5.1 Active Monitoring 191
13.5.1.1 Gas-Filled Detectors 191
13.5.1.2 Solid-State Detectors 194
13.5.2 Passive Monitoring 195
13.5.2.1 Individual Monitoring 195
13.5.2.2 Thermoluminescent Dosimeters 196
13.5.2.3 Optically Stimulated Luminescence Dosimeters 196
13.5.2.4 Film Dosimeters 197
13.5.2.5 Electronic Personal Dosimeters 197
185
13.6 Neutron Monitoring Techniques 198

13.6.1 Active Neutron Monitoring 198
13.6.1.1 Fluence Meters 198
13.6.1.2 Dose-Equivalent Meters 199
13.6.1.3 Neutron Spectrometers 202
13.6.2 Passive Neutron Monitoring 202
13.6.2.1 Film 202
13.6.2.2 TLDs 203
13.6.2.3 Solid-State Nuclear Track Detectors 203
13.6.2.4 Bubble Detectors 204
13.7 Continuous, Pulsed, And Mixed Radiation Fields 204
13.8 Conclusions 207
13.9 Acknowledgments 207
Notes 207
References 207
13.1 INTRODUCTION For the purposes of this chapter, external radiation

Radiation exposure can be external or internal. monitoring includes the following:
However, since sealed radioactive sources and par-
ticle accelerators are typically used in radiotherapy, 1. Area monitoring – the measurement of radiation
all exposures can be considered external. External levels in adjacent areas outside shielded radiother-
radiation doses to an individual result from expo- apy rooms and accelerators, or outside shielded
sure to external sources of ionizing radiation, that radioactive sources;
is, radiation sources outside the body [1]. Gamma 2. Individual monitoring – the measurement of doses
rays, x-rays, high-energy beta particles, protons, and received by individuals working outside shielded
neutrons are considered “penetrating radiation” and radiotherapy rooms and accelerators or in the
present an external exposure hazard. Conversely, vicinity of the shielded radioactive sources.
low-energy beta particles and alpha particles are
relatively “non-penetrating” and present less of an This chapter describes the monitoring of external radia-
external radiation exposure hazard. Neutrons are tion in radiotherapy and does not include radiation moni-
produced in particle therapy facilities, therapy linear toring inside therapy treatment rooms during treatment.
accelerators (linacs), fast neutron therapy facilities, The radiation monitors used for measuring external radia-
and boron neutron capture therapy (BNCT) facili- tion levels are typically referred to as area monitors, and
ties. Particle therapy facilities (heavy ions) produce include both portable and fixed radiation detectors or
neutrons with maximum energies in the GeV range survey meters. Fixed radiation detectors are area monitors
[2]. Proton therapy facilities produce neutrons with that are installed at a particular location. The monitors
energies as high as the primary proton beam energy used for recording the dose equivalents received by indi-
(~250 MeV). Typically, either cyclotrons or synchro- viduals working with radiation are referred to as personal
trons are used in these facilities. Fast neutron therapy dosimeters or individual dosimeters. Both area monitors
facilities produce neutrons with energies between and individual monitors can be either active or passive.
50 MeV and 70 MeV. These neutron therapy beams are
produced by reactors, cyclotrons, and linear accelera- 13.2 PROTECTION AND OPERATIONAL
tors. BNCT uses epithermal neutrons for therapy. The DOSE QUANTITIES
neutron source for epithermal radiation is generated External radiation monitoring is performed for the
from nuclear reactors and accelerator-based neutron purposes of radiation protection, that is, primarily to
sources (ABNS). demonstrate compliance with the shielding design or
Area and Individual Radiation Monitoring ◾ 187
TABLE 13.1 Radiation Weighting Factors Recommended by ICRP Publication 103 [6]
Radiation Type Energy Range (E) Radiation Weighting Factor (w R )
Photons, electrons, and muons All energies 1
Neutrons <1 MeV  (ln( E ))2 

2.5 + 18.2exp  − 
 6
Neutrons 1 MeV to 50 MeV  (ln(2E ))2 

5 + 17exp  − 
 6
 (ln(0.04 E ))2 
Neutrons >50 MeV 2.5 + 3.25exp  − 
 6
Protons, other than recoil protons >2 MeV 2
Alpha particles, fission fragments, All energies 20
and heavy nuclei
regulatory dose limits [3]. Thus, the radiation mea- be related to the particle fluence and, in turn, by con-
surements must be expressed in terms of quantities in version coefficients to each other. The fluence, Φ, is the
which the dose limits are defined. The International number of particles dN (such as photons or neutrons),
Commission on Radiological Protection (ICRP) incident on a sphere divided by the cross-sectional area
defines the dose quantities. They are expressed of the sphere (dA). In other words, it is the total number
in terms of protection quantities measured in the of particles per unit area with which a material is irradi-
human body. The dose equivalent was specified in ated. The unit is m−2 or cm−2.
ICRP Publication 21 [4]. ICRP Publication 60 [5] Note that the term “dose” might be used in a generic
introduced the concept of equivalent dose. ICRP sense throughout this document to refer to the various
Publication 103 [6] modified the weighting factors. dose quantities.
The radiation weighting factors, w R , for the protec- The operational quantities - ambient dose equivalent
tion quantities as recommended by ICRP Publication and directional dose equivalent, are defined for area
103 [6] are shown in Table 13.1. For neutrons, w R var- monitoring. The operational quantity - personal dose
ies with energy and therefore the computation for the equivalent, is defined for individual or personal moni-
protection quantities is made by integration over the toring. The definitions of operational quantities are
entire energy spectrum. taken from ICRU Report 51 [7], ICRU Report 57 [8] are
The dose equivalent, H, is the product of D and Q at as follows:
a point in tissue, where D is the absorbed dose and Q is The ambient dose equivalent, H*(d), at a point in a
the quality factor for the specific radiation at that point, radiation field, is the dose equivalent that would be pro-
thus H = DQ. duced by the corresponding expanded and aligned field,
The unit of the dose equivalent is joule per kilogram in the ICRU sphere (diameter = 30 cm, 76.2% O, 10.1%
(J kg−1), and its special name is Sievert (Sv). H, 11.1% C, and 2.6% N) at a depth, d, on the radius
The equivalent dose, HT , in a tissue or organ is given opposing the direction of the aligned field [7]. The ambi-
by HT = ∑ w R DT ,R , where DT ,R is the mean absorbed dose ent dose equivalent is measured in Sv. In the expanded
R
in the tissue or organ, T, due to radiation, R, and w R is and aligned field, the fluence and its energy distribution
the corresponding radiation weighting factor. The unit have the same values throughout the volume of interest
of equivalent dose is the Sievert. as in the actual field at the point of reference, but the
ICRP Publications 60 [5] and 103 [6] recommended fluence is unidirectional. The ambient dose equivalent is
the use of equivalent dose (HT) and effective dose (E) used for area monitoring.
as protection quantities. However, these quantities are The directional dose equivalent, H′(d, Ω), at a point
not directly measurable. For external individual expo- in a radiation field, is the dose equivalent that would be
sure, the accepted convention is the use of the following produced by the corresponding expanded field in the
operational quantities-ambient dose equivalent H*(d), ICRU sphere at a depth, d, on the radius in a specified
the directional dose equivalent H′(d, Ω), and personal direction, Ω [7]. The directional dose equivalent is mea-
dose equivalent Hp(d). The two sets of quantities might sured in Sv.
The personal dose equivalent, Hp(d), is the dose Typical passive monitors include film dosimeters,
equivalent in soft tissue, at an appropriate depth, d, thermoluminescent dosimeters (TLDs), or optically
below a specified point on the body. The personal dose stimulated luminescence dosimeters (OSLDs) for pho-
equivalent is measured in Sv. tons; and albedo dosimeters, nuclear track etch detec-
For the ambient dose equivalent and personal dose tors and bubble detectors for neutrons. Such dosimeters
equivalent, a depth of 10 mm is recommended for are insensitive to the time structure of the accelerator
strongly penetrating radiation. For weakly penetrating beam and can be used in pulsed fields. Electronic per-
radiation (photons below 15 keV and beta particles), a sonal dosimeters (EPDs) may also be used.
depth of 0.07 mm is recommended for the skin, and a The operational quantity for individual monitoring of
depth of 3 mm is recommended for the lens of the eye. external exposure is the personal dose equivalent Hp(d),
with the recommended depth d = 10 mm for strongly
13.3 ACTIVE AND PASSIVE MONITORING penetrating radiation and d = 0.07 mm for weakly pene-
Radiation monitoring can be active or passive. trating radiation (shallow or skin dose). Personal dosim-
eters are calibrated in these quantities.
13.3.1 Active Monitoring
The purpose of active monitoring is to obtain a cur- 13.4 AREA MONITORING
rent snapshot of the radiation environment [9]. This The measurement of radiation levels in adjacent areas,
information may be used to demonstrate compliance outside shielded radiotherapy rooms or shielded radio-
with regulatory requirements and verify the shielding active sources, can be performed with either active or
integrity of a new facility. It also allows the planning passive radiation monitors. Typical radiation instru-
of activities of radiation workers, so that they may be ments used for active area monitoring are either
carried out within the regulatory or facility-specified gas-filled detectors or solid-state detectors such as scin-
dose limits. tillators and semiconductor detectors [1]. The opera-
Active monitoring can be used to measure prompt tional quantities used for area monitoring are ambient
radiation, induced radiation (activation) and radia- dose equivalent and directional dose equivalent.
tion from radioactive sources. Active monitors provide
a direct display of the radiation levels and are usu- 13.4.1 Calibration of Area Monitors
ally equipped with additional functions such as alarm Area monitors must be calibrated against a reference
thresholds for dose or dose rate. While most active instrument that is traceable either directly or indirectly
monitors measure dose rate, some can be used in the to a national standards laboratory. Testing, calibration,
“integrate mode” to measure the cumulative dose over and calibration frequency requirements for portable
a defined period of time. Therefore, such monitors can radiation detectors can be found in ANSI Standard
also be used for passive monitoring. N323AB-2013 [10].
Area monitors should be calibrated in terms of the
13.3.2 Passive Monitoring
ambient dose equivalent, H*(d), or the directional dose
Passive monitors are used in both area monitoring and equivalent, H′(d, Ω), without the presence of a phantom,
individual monitoring. Individual monitors provide that is, “free in air” [11].
estimates of cumulative doses received by the indi- The calibration factor, N, is defined as the con-
vidual over a specified period of time, in terms of the ventional true value of the quantity the instrument is
operational quantities used for personnel dosimetry [9]. intended to measure, H, divided by the indication, M
The purpose of passive monitoring is to determine the (corrected, if necessary) given by the instrument [11].
cumulative dose over a specified period of time [9].
N = H /M (13.1)
13.3.2.1 Individual Monitoring
Individual or personal monitoring is the measurement The calibration factor is commonly incorporated into an
of the radiation doses received by individuals working instrument by adjustment of the electronic gain. Several
with radiation [1]. Individual monitoring may also be calibration points are evaluated, typically one on each
used to determine accidental exposures. range or decade. An attempt is made to achieve 10%
accuracy over the entire range for photon instruments conditions; h is the appropriate conversion coefficient,
and 30% for neutron rem meters. The calibration factor that is, dose equivalent per unit kerma. Thus, the dose
is normally only quoted for one reference radiation. If equivalent H is given by:
there is no unique factor applicable to the entire mea-
surement range of an instrument, the instrument is said H = hN R M R (13.5)
to have a nonlinear response.
The calibration factor N is dimensionless when both Reference instruments are calibrated free in air for the
the conventional and indicated values have the same range of reference radiation qualities, which are defined
units. A calibration factor of 1 implies that the instru- by the International Organization for Standardization
ment is perfectly accurate. (ISO)) [12]. These same reference qualities should be
The reciprocal of the calibration factor is equal to the used for the calibration of area monitors.
response under reference conditions. The instrument
response is applicable to the prevailing conditions. 13.4.1.2 Neutron Calibration
The response, R, of the area monitor (with respect to The response of an active detector such as a neutron
fluence or dose equivalent) typically varies with both the dose-equivalent meter is designed to be proportional
energy and directional distribution of the incident radi- to the dose equivalent and is therefore independent of
ation. Therefore, one can consider the response, R(E,Ω), neutron energy. Thus, dose-equivalent meters are use-
as a function of the energy E of the incident radiation ful in radiation fields where the neutron spectrum is
and of the direction (solid angle), Ω, of the incident not well characterized, or known. The underlying prin-
monodirectional radiation. R(E) and R(Ω) describe the ciple of operation for dose-equivalent meters is similar,
“energy dependence” and the “angular dependence,” although they may vary widely in size and geometrical
respectively, of the response. The solid angle Ω may also configuration [13].
be defined by the angle α between the reference orien-
tation of the instrument and the axis of the incident Neutrons are classified according to their energy as
monodirectional radiation. follows:
13.4.1.1 Photon Calibration Thermal: Ēn = 0.025 eV at 20°C. Typically En ≤ 0.5 eV

An ionization chamber with a measuring assembly is Intermediate: 0.5 eV < En ≤ 10 keV
usually used as a reference instrument for photons [1].
Reference instruments do not directly measure the dose Fast: 10 keV < En ≤ 20 MeV
equivalent, H, required for calibration of radiation sur- Relativistic: En > 20 MeV
vey instruments. Instead, basic radiation quantities such
as the air kerma (K) in air for photon radiation are mea- High-Energy Neutrons: En > 100 MeV
sured. Kerma is defined as the quotient of dEtr by dm
where dEtr is the sum of the initial kinetic energies of where En is the energy of the neutron and Ēn is the aver-
all the charged ionizing particles liberated by uncharged age energy of the neutron.
ionizing particles in a volume element of mass dm. A dose-equivalent meter’s response is shaped to fit an
appropriate fluence-to-dose equivalent conversion coef-
K = dEtr /dm (13.2) ficient over a particular energy range. The ICRP 21 [4]
fluence-to-dose equivalent conversion factors were used
H = NMR (13.3) in the older designs; and these instruments were referred
to as rem-meters. Similar conversion coefficients fac-
N = hN R (13.4) tors were also recommended by NCRP Report No. 38
[14]. The latter are the basis for the current federal and
where MR is the reading of the reference instrument various state regulations in the United States. The newer
corrected for influence quantities. NR is the calibration designs of dose-equivalent meters comply with the ICRP
factor (e.g., in terms of air kerma in air or air kerma effective dose recommendations. The operational quan-
rate in air) of the reference chamber under reference tity appropriate for dose meter calibration is the ambient
dose equivalent (H*(10)), which is defined for a known A dose-equivalent meter’s energy response should
neutron spectrum as follows: vary as the quantity [H*(10)/(Kair)air], as shown below:
(13.8)
∫
N H * = [H *(10)/M ] = [H *(10)/( K air )air ]/[( K air )air /M]
H *(10) = hΦ ( E ) Φ( E )dE (13.6)
13.4.2.3 Directional Dependence
where hΦ(E) is the fluence-to-ambient dose equivalent The directional response of the instrument can be stud-
conversion function, and Φ(E) is the neutron fluence as ied by rotating the survey meter about its vertical axis.
a function of energy for a given neutron field. The dose- Survey meters should exhibit isotropic response, as
equivalent meter response, Rm, in that field is given by required for measuring ambient dose equivalent, within
the equation below: ±60° to ±80° with respect to the reference direction of
calibration. They typically have a much better response
∫
Rm = CrΦ ( E ) Φ( E )dE (13.7) for higher photon energies, that is >80 keV [1]
13.4.2.4 Measurement Range

where r Φ (E) is the dose-equivalent meter’s response
The useful dose equivalent range for radiation survey
function in units of counts per unit fluence, and C is
meters is µSv/h to mSv/h. More sensitive instruments
the calibration constant in units of Sievert per count.
are capable of measuring dose equivalents as low as
As long as r Φ (E) has a similar energy response to that of
1 nSv/h.
hΦ (E), the dose-equivalent meter measurement is con-
sidered accurate. The traditional energy response of the
13.4.2.5 Response Time
dose-equivalent meter in terms of counts per unit dose
The response time of the survey meter is defined as the
equivalent is given by the ratio r Φ (E)/hΦ (E). However,
RC time constant of the measuring circuit. In this case,
r Φ (E) does not match hΦ (E) over much of the energy
R is resistance of the decade resistor and C the capaci-
range. Therefore, some detectors either over respond or
tance of the circuit. High R, hence high RC values are
under respond in certain energy regions. Usually, most
used for low dose equivalent ranges, resulting in slow
dose-equivalent meters have a very large over-response
indicator movements. The meter will stabilize only after
in the intermediate energy region. Therefore, they
about 3–5 time constants.
give an adequate measure of dose equivalent between
100 keV and 6 MeV [15]. Thus, it is important to have
13.4.2.6 Linearity of Response
a rough idea of the spectrum, in order to rely on the
The response of survey meters should be linear with the
instrument readings.
dose rate. Survey meters should read full scale when
subjected to dose rates of about ten times the maximum
13.4.2 Desirable Properties of Area Monitors scale range. However, some instruments suffer from sat-
or Radiation Survey Meters
uration as described later.
13.4.2.1 Sensitivity
The sensitivity S is defined as the inverse of the calibra- 13.4.2.7 Calibration and Long-Term Stability
tion coefficient N. A high sensitivity is desirable. Survey meters are required to be calibrated in a standards
dosimetry laboratory. In the United States, the regulatory
13.4.2.2 Energy Dependence frequency is once a year. They must also be calibrated
Survey meters are often used in situations in which upon repair or sudden change in response. The long term
the radiation field is complex or unknown. Further, stability of survey meters should be checked at regular
they are calibrated at a given energy spectrum or beam intervals using an appropriate radioactive source.
quality. Therefore, these survey meters should have
a low energy dependence over a wide energy range. 13.4.2.8 Discrimination in Mixed Fields
In the past, photon survey meters were designed to The survey meter should be able to distinguish between
exhibit a flat energy response that follows exposure different types of radiation in mixed radiation fields. For
or air kerma in air. example, end-window Geiger–Mueller (GM) counters
have a buildup cap that can be removed when measur- the chamber is ionized. The pulse height becomes inde-
ing beta particles; and which can be left in place when pendent of the primary ionization or the energy of the
measuring gamma rays. interacting particles. When the electric field is strong
enough to cause a breakdown in the chamber, continu-
13.4.2.9 Measurement Uncertainties ous discharge takes place.
Measurement uncertainties include uncertainty asso- Gas-filled detectors are designed to operate in one
ciated with the calibration factor (Type A); and uncer- of the following regions: The ionization region, propor-
tainties due to energy dependence, angular dependence, tional region or GM region [1]. Regions of recombina-
and variations from calibration conditions in user fields tion and of limited proportionality are typically not
(Type B). The two types of uncertainties are added in used for survey meters. However, recombination cham-
quadrature to obtain the measurement uncertainty. bers are used to measure the quality factor for radiation
fields. Operation in the continuous discharge region will
13.5 PHOTON MONITORING TECHNIQUES result in damage to detectors.
Photon monitoring techniques include active and pas- The design of the survey meter depends upon the
sive methods. type, energy, and nature of the radiation field as well as
its specific application [1]. Air ionization survey meters
13.5.1 Active Monitoring are quite common and are usually vented to the atmo-
Active methods include the use of gas-filled detectors sphere. Sealed ionization chambers are typically pres-
and solid-state detectors. surized and are usually filled with a non-electronegative
gas such as argon.
13.5.1.1 Gas-Filled Detectors Due to the low mobility of the negative ions formed
Gas-filled detectors are typically cylindrical in shape by electron attachment, these ions would increase the
[1]. They are constructed with an outer wall made of collection time, thus limiting the dose rate that could be
plastic and occasionally tissue equivalent composition monitored. Therefore, noble gases are generally used in
plastic, and a central electrode. The wall and electrode these detectors.
are insulated from each other. Voltage is applied across Beta-gamma survey meters with thin end windows
the two electrodes. The various regions of operation of are used to detect weakly penetrating radiation. The β
a gas-filled detector as a function of the applied voltage efficiency is almost 100% for β particles entering the
are shown in Chapter 2 of this book (see Figure 2.1 in detector; however, the γ efficiency is only a few percent.
Ref. [16]). The GM counters used to detect gamma rays are
The first region is the recombination region, where much smaller than the ionization chamber detectors,
the response gradually increases with voltage. Because because of their higher sensitivity.
the electric field in the detector is very weak, the elec- Detectors can operate either in the “current” mode
trons from the cathode are drawn slowly toward the ion or the “pulse” mode. Proportional and GM counters are
chamber. Therefore, the positive ions have an oppor- normally operated in the “pulse” mode.
tunity to recombine with the free electrons before
they reach the anode or collecting electrode. In the 13.5.1.1.1 Ionization Chambers The number of primary
ion chamber region, the electric field is strong enough ions of either sign collected in the ionization region is
to overcome the recombination losses and all the elec- proportional to the energy deposited by the charged
trons reach the anode. Since all the freed electrons are particles in the detector volume [1]. Particle discrimi-
collected, the resultant current is proportional to the nation can be used in this region because of the differ-
amount of radiation that causes the ionization. In the ences due to the linear energy transfer (LET). Build-up
proportional counting region, the electric field has caps are used for high-energy photons (>100 keV) in
increased sufficiently so as to accelerate the electrons, order to improve the detection efficiency. Some types
which then cause further ionization in the gas. The next of ionization chambers have removable caps or covers
region is the limited proportional region. In the GM that enable the measurements of very soft x-rays. The
region, the electric field is so strong that a single ioniz- ionization chamber is the most useful photon survey
ing event creates an electron avalanche. The entire gas in meter because it is almost energy-independent (usually
within ± 10% of unity) between 30 keV and a few MeV. The 451P ion chamber is a 230-cubic centimeter
However, one cannot measure the dose rates close to volume air ionization chamber pressurized to 6 atmo-
the background level, because, the lower detection limit spheres. The plastic chamber wall is 200 mg cm−2 thick.
is about 1 μSv/h. Since the ionization chamber survey The collecting voltage is 90 V DC. It detects beta parti-
meter measures a very weak current of the order of fem- cles above 1 MeV, and gammas and x-rays above 25 keV.
toamperes (fA) when placed in a field of several μSv/h, it It can measure exposure/dose rate or accumulated expo-
takes several minutes until the detector becomes stable sure/dose using the integrate mode. The operating range
after being switched on [17]. Ionization chamber detec- for the 451P-RYR (standard chamber) is 0–5 R/h. It is
tors can operate in the “pulse” mode or in the “current” also available as a dose-equivalent chamber (451P-DE-
mode, depending upon the electronics used. Ionization SI-RYR) with the operating range of 0–50 mSv/h.
chambers operating in the current mode are more suit- The energy dependence is within 20% between
able for higher dose rate measurements. They can also 50 keV and 1 MeV for 451P-RYR, and within 20%
be used in pulsed fields such as those encountered with between 30 keV and 1 MeV for 451P-DE-S1-RYR.
radiotherapy accelerators. However, they may be sen- Figure 13.2 shows the typical energy dependence for
sitive to electromagnetic interference caused by stray the 451P-RYR.
fields from RF cavities [3]. Figure 13.3 shows the 451B model ion chamber
Ionization chamber survey meters are also the most survey mete, which is a 349 cubic centimeter volume
suitable and reliable detectors for the measurement of air ionization chamber. The phenolic chamber wall is
ambient dose rate due to residual radioactivity. Detectors 246 mg cm−2 thick. The chamber window is composed
are often available with removal ionization chamber win- of 1.7 mg cm2 mylar, 0.025 mm thick. The wall is made
dows, to enable the measurement of beta ray dose for the conductive by the application of graphite. The collection
estimation of skin dose [17]. A wide range of dose equiva- potential is −63 V. The chamber is vented to air through
lent rates can be covered with ionization chamber-based the desiccant. The 451B model has a slide cover made of
survey meters (1 µSv/h–1 Sv/h) through the use of decade phenolic 440 mg cm−2 thick. The cover is provided as a
resistances, larger detector volumes or detector gases under beta shield and also serves as an equilibrium thickness
higher pressures. Figure 13.1 shows the Fluke Biomedical for photon measurements. It protects the mylar window;
451P pressurized μR ion chamber survey meter. and when open allows the detection of alpha particles
above 7.5 MeV, beta particles above 100 keV, and gam-
mas above 7 keV. It can measure exposure/dose rate or
accumulated exposure/dose using the integrate mode.
According to the manufacturer, the operating range
for the 451B (standard chamber) is 0–50 R/h, while the
operating range for the 451B-DE-SI (dose-equivalent
chamber) is 0–500 mSv/h. With the beta shield open,
the 451B-DE-SI can measure skin dose at H*(0.07), and
deep dose H*(10) with the beta shield closed.
Figure 13.4 shows the energy responses of the 451B
ion chamber survey meter. The energy dependence is
within 20% between 50 keV and 1 MeV for 451B, and
within 20% between 30 keV and 1 MeV for 451P-DE-S1-
RYR (dose-equivalent ion chamber).
13.5.1.1.2 GM Counters GM survey meters are widely

used to measure very low radiation levels because
of their very large charge amplification (nine to ten
FIGURE 13.1 Fluke Biomedical 451P pressurized μR ion orders of magnitude). They are also suitable for leak
chamber survey meter. (Reproduced with permission from testing and detection of radioactive contamination.
Fluke Biomedical, Everett, Washington.) A GM survey meter with a thin window has almost
FIGURE 13.2 Energy response for 451P-RYR. (Reproduced with permission from Fluke Biomedical, Everett, Washington.)
100% sensitivity to the incoming beta rays, and it is very GM detectors have a strong energy dependence at low
useful in classifying radioactive materials [17]. A survey photon energies. They are useful for detecting the pres-
meter having an extendable rod with a small GM coun- ence of radiation, whereas ionization chambers are use-
ter installed at its tip can be used for remote high dose ful for more precise measurements.
rate measurements. The dead time is defined as the minimum time
required between the separations of two events in
order that they may be recorded as two pulses [18].
The dead time may be determined by the detector or
the electronics. At high counting rates, a true event
will be lost because it occurs too quickly following a
preceding event. Dead time losses can become severe
when high counting rates are encountered. GM detec-
tors have very long dead times, ranging from tens
to hundreds of milliseconds. Therefore, GM detec-
tors are not used when accurate measurements are
required; and at high count rates (greater than a few
hundred counts per second). They cannot be used in
pulsed fields unless the count rate of the detector is
significantly below the pulse rate; and the dead time
is insignificant when compared to the detection rate.
Otherwise, they become saturated and count only the
repetition rate [19].
FIGURE 13.3 Fluke 451B ion chamber survey meter with beta 13.5.1.1.3 Proportional Counters Proportional coun-
slide. (Reproduced with permission from Fluke Biomedical, ters are always operated in the pulse mode and in the
Everett, Washington.) region where, the amplification is about 100 fold [1, 18].
FIGURE 13.4 Energy dependence of 451B ion chamber. (Reproduced with Permission from Fluke Biomedical, Everett,
Washington.)
Therefore, proportional counters are more sensitive the scintillator. Some survey meters use photodiodes in
than ionization chambers and can be used to mea- place of PMTs. NaI(Tl) scintillation survey meters with
sure low dose rates. The amount of charge collected correction circuits for energy dependency give accurate
from each interaction is proportional to the amount results of ambient dose rate, similar to an ionization
of energy deposited in the gas of the counter by the chamber [17]. The lower detection limits are sufficient
interaction. Like GM counters, they cannot be used in for background measurements. They can also be used
pulsed fields. for the measurement of radioactivity. Sodium iodide
and cesium iodide exhibit poor energy response for the
13.5.1.2 Solid-State Detectors measurement of dose equivalent. However, the energy
Solid state detectors include scintillators and response can be improved through the use of compen-
semiconductors. sation circuits. Scintillation survey meters are usually
insensitive to photons with energies below 50 keV and
13.5.1.2.1 Scintillators Scintillation detectors are based are not appropriate for low-energy x-ray fields.
on light emission. Certain phosphors (organic crystals High-purity germanium (Ge) detectors have an
and inorganic crystals doped with activator atoms) emit excellent energy resolution and are commonly used for
scintillations upon absorption of radiation [1]. Organic photon spectrometry in research work. Since the Ge
crystals include high-atomic number phosphors, and detector must be cooled down to liquid-nitrogen tem-
are primarily used for the measurement of gamma rays; perature, it is not suitable for routine measurements.
while plastic scintillators are mostly used to measure Handheld scintillation survey meters designed for pho-
charged particles, photons, and fast neutrons. ton spectral measurements are commercially available,
Solid organic materials such as anthracene, stilbene, such as InSpector™ 10001 [20] and identiFINDER™2 [21].
and plastic scintillators, as well as thallium-activated Handheld survey meters with cerium-doped lantha-
inorganic phosphors such as sodium iodide (NaI(Tl)) num bromide (LaBr3(Ce)) scintillators are also avail-
or cesium iodide (CsI(Tl)), are used as scintillators. The able. These survey meters have better energy resolution
light pulse is converted into an electric pulse by a pho- than the conventional thallium-doped sodium iodide
tomultiplier tube (PMT), which is optically coupled to (NaI(Tl)) scintillator. Conversion from the light-output
distribution to the photon energy spectrum requires an the lower average energy required to produce an ion pair
unfolding process. in solid detector materials compared with air (factor of
Scintillation-based systems are more sensitive than 10); and the higher density of the solid detector materi-
GM counters because of higher gamma conversion als compared with air (factor of 1000) [1]. The detector
efficiency and charge amplification (PM tube works by dimensions for high-energy electrons or gamma rays
amplifying the electrons generated by a photocathode can be made much smaller than the equivalent gas-filled
exposed to a light signal). They are used for surveys at detector.
very low dose rates such as contamination monitoring
and lost source detection surveys. However, they can 13.5.2 Passive Monitoring
also be used at higher dose rates levels because their Passive area monitors are similar to those used for indi-
resolving time is ≤ a few microseconds [1]. vidual monitoring. However, they provide an estimate
Handheld photon spectrometers may be used for of potential doses received by the individual over a given
nuclide analysis of residual activity [17]. However, com- period of time; and can be used for retrospective dosim-
plicated spectra cannot be resolved because of their etry. The most widely used photon monitors are OSLDs,
limited energy resolution. High-purity germanium (Ge) TLDs, and photographic emulsion film dosimeters.
detectors can be used to resolve complicated spectra. Such dosimeters can be used in pulsed radiation fields
Lithium iodide crystals, doped with europium detect because they are insensitive to the time structure of the
thermal neutrons; and therefore, can be used in neutron radiation field.
spectrometers as described later. Plastic scintillation EPDs are direct-reading dosimeters and show both
detectors can be used to measure recoil proton spectra the instantaneous dose rate and the accumulated dose
in real time, which provide neutron spectral informa- at any point in time.
tion in the energy range of 2 MeV to 20 MeV [3].
13.5.2.1 Individual Monitoring
13.5.1.2.2 Semiconductors A major limitation of scin- As previously described, the operational quantity, per-
tillators is their relatively poor energy resolution [18]. sonal dose equivalent Hp(d) with the recommended
Semiconductors used as radiation detectors provide depth d = 10 mm for strongly penetrating radiation, and
a much larger number of carriers for a given incident d = 0.07 mm for weakly penetrating radiation, is used
radiation event, than any other detector. Therefore, they for the individual monitoring of external exposure.
provide the best energy resolution.
Bulk conductivity detectors are formed from intrin- 13.5.2.1.1 Calibration Personal dosimeters are irradiated
sic semiconductors of very high bulk resistivity (e.g., for calibration while mounted on standard phantoms,
cadmium sulfide (CdS) or cadmium selenide (CdSe)). in order to account for the backscatter contributions of
They act like solid-state ionization chambers on expo- the human body [1]. The standard phantoms are usually
sure to radiation [1]. composed of ICRU tissue. In some cases, water phantoms
Extrinsic (i.e., doped with trace quantities of impuri- and polymethyl methacrylate (PMAA) phantoms are
ties such as phosphorus or lithium) semiconductors such also used. Different phantom configurations are used for
as silicon or germanium are used to form junction detec- whole body dosimeters (slab), wrist or ankle dosimeters
tors. On application of a reverse bias, these detectors act (pillar), and finger dosimeters (rod). The calibration pro-
as a solid-state ionization chamber in detecting radiation. cedure for whole body dosimeters is as follows:
PIN diodes are diodes with a wide, undoped intrinsic
semiconductor region between a p-type semiconductor 1. Air kerma in air (Kair)air is measured in a reference
and an n-type semiconductor region. The p-type and field, using a reference ionization chamber that is
n-type regions are typically heavily doped because they calibrated by a standards laboratory.
are used for ohmic contacts. In contrast to the ordinary
p-n diode, the wide intrinsic region makes a PIN diode 2. The reference instrument reading is converted to
a good photo-detector. [Hp(d)]slab using theoretical values of the ratio, hkHp
The sensitivity of solid-state detectors is about 104 (available for various beam qualities); where hkHp =
times higher than that of gas filled detectors, because of [Hp(d)/(Kair)air]slab.
3. The dosimeter to be calibrated is placed at the cali- doses are determined by measuring the light output
bration point on a phantom and its reading M is under different filters. The results are then compared
obtained. with results obtained from calibrated dosimeters, which
have been exposed to known doses under well-defined
4. The calibration factor in terms of the personal
conditions.
dose equivalent for the dosimeter is given by
Dosimeters that have high-atomic number TLDs are
NHp = Hp(d)/M.
not tissue equivalent. Therefore, they require filters to
match their energy response to that of tissue. Dosimeters
13.5.2.1.2 Desirable Characteristics of Individual
with low-atomic number TLDs do not require such fil-
Monitors Personal dosimeters should have high sensitiv-
ters. Due to its tissue equivalency, LiF TLD exhibits
ity and be capable of measuring doses as low as 10–30 µSv
acceptable energy dependence characteristics. However,
[1]. They should not have significant energy dependence.
CaSO4:Dy shows significant energy dependence and its
The response should be linear in the dose range of 10 µSv
energy response can be reduced by appropriate dosim-
to 10 Sv, thus covering both normal and accidental expo-
eter design.
sures. They must be iso-directional, that is, the angular
Due to their small size, TLDs are convenient for use
response must vary as the directional dose equivalent,
in extremity dosimetry. TLDs exhibit fading, that is,
H′(d, Ω). They should also be capable of discriminat-
the signal decreases with time after irradiation due to
ing against different types of radiation in mixed fields.
the spontaneous emission of light at room tempera-
Unfortunately, no single dosimeter meets all these criteria.
ture. The dose range covered by TLDs is 100 µSv to
10 Sv [23].
13.5.2.2 Thermoluminescent Dosimeters
Ionizing radiation can produce electron-hole pairs in an 13.5.2.3 Optically Stimulated
inorganic crystal [22]. In TLDs, the crystal material and Luminescence Dosimeters
impurities are such that the electrons and holes remain The principle of optically stimulated luminescence
trapped at the activator sites at room temperature. The (OSL) is similar to that of thermoluminescence, with
number of trapped electrons and holes depends upon the the exception that laser light is used instead of heat, to
radiation exposure. Upon heating, the trapped electrons release the trapped energy. Optically stimulated lumi-
and holes migrate and combine emitting photons with nescent dosimeters contain a thin layer of aluminum
energies of a few electron volts. A TLD reader is used to oxide doped with carbon (Al2O3:C). The aluminum
automatically heat the TLD and measure the light out- oxide is stimulated with selected frequencies of laser
put as a function of temperature – referred to as a glow light producing luminescence. The light output is pro-
curve. Several peaks appear as traps at different energy portional to the radiation dose [23].
levels are emitted. The total light output is given by the Figure 13.5 shows the Landauer3 Luxel+ [24] dosim-
area under the glow curve; and can be used to deter- eter badge which measures radiation exposure due to
mine the radiation dose. The TLD can be annealed by x-ray, gamma, and beta radiation with OSL technology.
heating it to high temperatures, so that all the traps are
emptied. It can be then reused. Among the many passive
dosimeters, TLDs are most commonly used in radio-
therapy. The dose response is linear over a wide range of
doses used in radiotherapy, but the response increases
in the higher dose region, thus exhibiting supralinearity
[23]. The most commonly used TLD materials include
LiF:Ti,Mg (Lithium fluoride doped with titanium and
magnesium), CaSO4:Dy (calcium sulfate doped with
dysprosium), and CaF2:Mn (calcium fluoride doped
with manganese) [1].
TLDs can be used with filters to discriminate against FIGURE 13.5 The Luxel+ dosimeter badge. (Reproduced
different types of radiation. Beta, x-ray, and gamma ray with permission from Landauer, Inc. Glenwood, IL.)
According to the manufacturer, it is an integrated, self- 13.5.2.5 Electronic Personal Dosimeters

contained packet. It incorporates a thin strip of Al2O3:C EPDs measure both instantaneous dose rates and accu-
(aluminum oxide doped with carbon) sandwiched mulated doses [1]. Therefore, they are used for supple-
between a multi-element filter pack that is heat sealed. It mental individual monitoring such as:
is covered by a laminated, light-tight paper wrapper. All
of these components are RF-sealed inside a temper-proof 1. Direct readout of the dose or dose rate at any time.
plastic blister pack. Luxel+ dosimeter is not affected by
2. Tracking of daily doses during various activities.
heat, moisture, and pressure as long as the clear blister
packaging is uncompromised. These dosimeters are 3. Tracking of doses during special situations such as
very sensitive and can be used to measure photon doses radioactive source loading, surveys, and handling
in the range of 10 µSv to 10 mSv over a photon energy of radiation incidents or emergencies.
range of 5 keV to 40 MeV. They can measure beta ray
doses in the range of 100 µSv to 10 mSv over an energy EPDs based on miniature GM counters or silicon
range of 150 keV to 10 MeV. The dosimeters can be rean- detectors are available. EPDs with energy-compen-
alyzed several times without losing sensitivity and may sated detectors have an energy dependence within
be used for up to a period of one year. ±20% over the photon energy range of 30 keV to
The dosimeter can be restimulated with laser 1.3 MeV. They provide an instantaneous display
numerous times to confirm the accuracy of a radia- of the cumulative dose at any time. They also have
tion dose measurement. A full reanalysis is automati- automatic ranging facilities and provide both visual
cally performed for every measurement yielding a and audio indications, thus monitoring immediate
dose in excess of 5 mSv. The badge can also incorpo- changes in the radiation field.
rate the neutron detector, Neutrak® 144, described in The RAD-60R4 EPD uses a 4.5 mm2 PIN Si-Diode,
Section 13.6.2.3. energy-compensated with copper filtration. This EPD
displays R on the display, but this model actually mea-
13.5.2.4 Film Dosimeters sures personal dose equivalent, Hp(10) in mrem [25].
Film badges consist of special emulsion photographic The PD-10i5 EPD utilizes an energy–compensated
film placed in a light-tight wrapper. The latter is GM tube. The specified energy response of that unit is
enclosed in a case or holder with windows behind tissue-equivalent to within ±25% from 55 keV to 6 MeV.
appropriate filters [1]. Film is non-tissue equivalent; There is a 70% under response at 40 keV. This unit display
therefore, a filter system is used to flatten the energy scales automatically between μR, mR, and R. According
response. A single filter is adequate for photons of to Meier, the EPDs were irradiated with 99mTc, 131I, and
18F radionuclides with emission energies at 140 keV, 364
energy above 100 keV, but multiple filters are required
for lower energy photons. keV, and 511 keV, respectively. The authors report that
The badge holder creates a distinct pattern on the the energy responses of RAD-60R and the PD-10i units
film, which indicates the type and energy of the inci- significantly diverge from unity at 140 keV, but trend
dent radiation. The film optical density under different toward unity as the incident photon energy increases to
filters is read, and the results are compared with calibra- 511 keV. In the energy ranges that were studied, both
tion films that have been exposed under known condi- EPD models were shown to have significant increases in
tions (doses and radiation types). The optical density is reading in the presence of backscatter from a PMMA
a measure of the cumulative dose. Film can be used to phantom.
determine cumulative doses from photons, beta parti- Thermo Scientific™6 EPDs measure ionizing radia-
cles, and neutrons (described in Section 13.6.2.1). Film tion in real time [26]. They are equipped with both
dosimeters can measure doses in the range of 0.1 mSv audible and visual alarms; and can be used to mea-
to 10 Sv. sure gamma radiation, beta radiation, neutron radia-
Films are affected by heat, liquids, and excessive tion, and x-rays. Various models are available.The
humidity; and the latent image on undeveloped film EPD-TruDoseTM measures H p(10) within ±5% for
137Cs and H (0.07) within ±15% for 90 Sr/90Y. It can
fades with time. Thus, the usage period is limited to p
about to three months under ideal conditions. measure photons in the energy range of 16 keV to
10 MeV. It has a dose rate range of 1 µSv/h to 10 Sv/h equivalent conversion coefficients (hΦ(E)) [8, 29].
and a dose range of 1 µSv–10 Sv. Because these coefficients depend strongly on the neu-
tron energy, the neutron spectrum must be known or
13.6 NEUTRON MONITORING TECHNIQUES approximated a priori [30]. Typically, most of the neu-
tron detectors are calibrated against standard neutron
Neutron monitoring techniques inside treatment rooms
sources such as Pu-Be (Plutonium–Beryllium) with Ēn =
have been described in the literature [27, 28]; and are
4.2 MeV, Am-Be (Americium–Beryllium) with Ēn = 4.5
not discussed in this chapter. Only, neutron monitoring
MeV or 252Cf with Ēn = 2.2 MeV where Ēn is the aver-
outside shielded rooms and accelerators are discussed.
age neutron energy. Therefore, correction factors must
Radiotherapy facilities requiring external neutron
be applied to account for the changes in neutron spectra
monitoring include therapy linacs, fast neutron therapy,
from calibration to actual field conditions.
BNCT and particle therapy facilities.
The most common thermal neutron detectors used in
In neutron monitoring, the quantities of interest are
fluence meters are the 10B-enriched BF3 (boron trifluo-
neutron fluence, neutron dose equivalent (usually ambi-
ride) detector, the 3He (helium-3) counter tube, and the
ent dose equivalent) or dose equivalent rate, and the 6LiI(Eu) (europium doped lithium-6) scintillator [3]. The
neutron spectrum as a function of energy [7].
reactions that take place in the above-mentioned detec-
Neutron monitoring techniques in radiotherapy con-
tors are as follows:
sist of both active and passive methods. Active methods
include the use of dose-equivalent and fluence meters;
BF3 proportional counter: 10B(nth,α)7Li, Q = 2.31 MeV,
while passive methods include the use of TLDS, solid-
σ = 3,840 barns,
state nuclear track detectors (SSNTDs) and bubble
detectors. 3He proportional counter: 3He(nth,p)3H, Q = 0.765
MeV, σ = 5,330 barns,
13.6.1 Active Neutron Monitoring 6 Li(Eu) scintillator: 6Li (nth,α)3H, Q = 4.78 MeV, σ =
Active neutron monitoring is based on slowing down 940 barns,
or moderating fast neutrons until they reach thermal
energies. The thermal neutrons are then detected by a where Q is the Q-value (kinetic energy released) and
thermal neutron detector. The geometry and configu- σ is the cross section for the thermal neutron reac-
ration of the instrument is designed to measure either tion. The cross sections vary roughly as En −1/2 , where
dose equivalent (referred to as dose-equivalent meter) En is the neutron energy. Therefore, at 1 MeV, the
or fluence (fluence meter). Outside the shielded therapy cross section is about four orders of magnitude lower
linac rooms, neutron and photon fluences are consid- than that at thermal energies. All these bare detec-
erably lower than inside the treatment room [27]. In tors are sensitive only to thermal neutrons. When
addition, the neutron pulse is spread over several hun- used with a moderator, higher energy neutrons can
dred microseconds due to moderation in the shielding be detected.
material. Therefore, active monitors may be used as The most commonly used neutron f luence meter
long as photon pulse pile-up and dead times effects on is a version of the BF3 long counter [27, 31]. The bare
their responses are taken into consideration. Neutron BF3 counter detects only thermal neutrons. The sen-
detection for radiotherapy facilities is spread over many sitivity of the counter is proportional to volume,
decades of energy from thermal energies up to GeV. pressure, and degree of enrichment in 10B. When
Unfortunately, no single detector can accurately mea- used in conjunction with a moderator (typically
sure the neutron dose equivalent or fluence over such a cylindrical), the fast neutrons are thermalized, and
wide energy range. detected. By enclosing the moderator in 0.5 mm of
cadmium, the thermal neutrons can be completely
13.6.1.1 Fluence Meters eliminated. The thickness of hydrogenous modera-
Neutron fluences can be used to determine dose equiva- tor is chosen so that a fairly f lat response is obtained
lent; however, this requires the use of fluence-to-dose per unit f luence for neutron energies up to several
MeV. Appropriate f luence-to-dose equivalent factors The counter tube is surrounded by a borated plas-
can be applied to the known f luence, based on an tic sleeve which minimizes its over-response in the
average energy, to obtain dose equivalent [29, 32]. 10 keV to 100 keV range. The response at thermal ener-
The use of the BF3 detector requires knowledge of gies is increased because of holes drilled in the sleeve.
the neutron spectrum. It is particularly suitable for use However, the directional response of the rem-meter
when neutron dose-equivalent rates are below measur- is impacted by its cylindrical moderator geometry.
able levels on the dose equivalent meter. Relative varia- According to Cosack and Leisecki [34], a change in
tions of the neutron field with time can be monitored response with instrument orientation of as much as
with the moderated BF3 detector. The ratio of the dose- 35% has been observed for neutron energy of 1 MeV.
equivalent meter and the moderated BF3 detector read- When the rem-meter is exposed with its side oriented
ings provides a rough estimate of the average energy of to a source of thermal neutrons, a 65% underestima-
the neutron spectrum. All the problems associated with tion in the true dose equivalent has been observed. A
dose-equivalent meters (described in the next section) modified version with a spherical polyethylene mod-
also apply to these detectors. erator and a cadmium layer was designed by Hankins
The 3He proportional counter is more sensitive and in order to improve the directional dependence [35]. It
more stable than the BF3 counter, but it is much more is currently marketed by Thermo Fisher Scientific as
expensive [27]. The 6Li(Eu) scintillator has a very high the model NRD. However, the high-energy response of
sensitivity but a very poor photon rejection. Therefore, the NRD decreases steadily at energies above 7 MeV.
it is difficult to use in mixed photon–neutron fields such At these energies, the neutron fluence is considerably
as those encountered at therapy facilities. It is impor- reduced for therapy linacs. Therefore, the rem-meter
tant to note that for all active monitors, manufacturers can still be used. However, it cannot be used for parti-
normally state photon rejection for steady fields and not cle therapy and fast neutron therapy facilities, because
pulsed fields. of the higher-energy neutrons that are encountered at
such facilities.
13.6.1.2 Dose-Equivalent Meters Figure 13.6 shows a Thermo Eberline ASP/2e NRD
Typically, most commercial dose-equivalent meters Neutron Survey Meter – a portable battery operated
consist of a neutron moderator such as polyethylene or instrument. The detector is a 22.9 cm (9 inch) in diam-
some other hydrogenous material, surrounding a ther- eter, cadmium-loaded polyethylene sphere with a BF3
mal neutron detector. Dose-equivalent meters incorpo- tube in the center. The instrument is available with dis-
rating BF3 proportional counters are more commonly play either in rem or Sv. According to the manufacturer,
used outside the shielding in therapy facilities because the energy response closely follows the theoretical dose
of their excellent photon rejection and low cost. In the equivalent curve for neutrons over the energy range from
BF3 detector, the thermal neutrons are captured in the 0.025 eV (thermal) to about 10 MeV. It measures dose
boron via the 10B(nth,α)7Li reaction. Both the alpha equivalent rates in the range of 1–100 mSv/h. The BF3
particle and recoil 7Li nucleus produce large pulses in tube provides gamma rejection up to 500 R/hr depend-
the proportional counter. The large pulses being orders ing on the applied voltage (1600–2000 V). The sensitivity
of magnitude higher, can be easily discriminated from is 45 cpm/(mrem/h) or 2700 counts per mrem. The dead
the smaller pulses produced by photons in mixed fields. time is 10 µs (nominal). It has a directional response
The energy response of a moderated detector is deter- within 10%. Its response time is programmable from
mined mainly by its size and geometrical configura- 0–255 milliseconds. The counting instrument is the
tion [3]. Model ASP-2e which has a dual analog/digital display.
The commonly used Andersson–Braun (AB) rem- It has a rate meter that can be used on an integrating or
meter has a cylindrical polyethylene moderator sur- scaler mode. It has a count range of 1–1.3 million counts
rounding a BF3 counter tube [33]. Its energy response per minute.
closely follows the NCRP 38 dose equivalent conver- The Fluke Biomedical7 Neutron Survey Meter Model
sion function (from thermal to about 10 MeV), except 190N, is based upon the classical Andersson–Braun
at intermediate energies where it over responds. rem-meter design. A polyethylene cylinder 24 cm long,
FIGURE 13.7 FHT 762 Wendi-2 Wide Energy Neutron

FIGURE 13.6 Thermo Fisher Scientific NRD Asp/2e Neutron Detector. (Courtesy of Arash Darafsheh.)
Survey Meter.
equivalent rates are a strong function of angle at these

and 21.6 cm in diameter surrounds the BF3 tube. The facilities [36].
tube is filled with 96% enriched 10B. The instrument is Figure 13.7 shows the Wide Energy Neutron Detection
available with display in either rem or Sv. According to Instrument Wendi-2 (also known as WENDI-II) dose-
the manufacturer, the instrument has a dose equivalent equivalent meter.
range from 0 μSv/h to 0.75 Sv/h, and it can integrate The Wendi-2 was originally designed at Los Alamos
doses from 0 μSv to 10 Sv. It has a gamma rejection up to National Laboratory [13]. A cylindrical polyethylene
500 Roentgens/h for 137Cs. The directionality is less than moderator assembly surrounds a 3He counter. Tungsten
20% in orthogonal directions. powder surrounds the counter at an inner radius of 4
The maximum neutron energy for a 15 MV linear cm, thus, generating neutrons with energies above 8
accelerator is about 10 MeV [27]. Since the response of MeV and absorbing neutrons with energies below sev-
the neutron dose-equivalent meters (AB, NRD, 190N) eral keV. The interaction of high-energy neutrons with
falls off rapidly at energies above 6 MeV, these detectors the tungsten material causes neutron multiplication and
are not necessarily accurate over the energy range of energy degrading reactions such as (n, 2n), thus improv-
interest in radiotherapy linac facilities, but are adequate ing the sensitivity to high-energy neutrons which
for monitoring purposes. extends well beyond 20 MeV to 5 GeV. The instrument
The Andersson–Braun design suffers from angular exhibits increased sensitivity, with a sensitivity to bare
dependence due to the lack of spherical symmetry [13]. 252Cf that is 12 times higher than the NRD sensitivity.
In addition, all instruments using a pure polyethylene Additionally, the energy response for Wendi-2 closely
moderator suffer from a lack of high-energy response. follows the theoretical ambient dose equivalent per unit
Because an ordinary rem-meter is practically insensi- fluence function [H*(10)/Φ] above 0.1 MeV. Its energy
tive to neutrons of energies above 15 MeV, it under- response at 500 MeV is approximately 15 times higher
estimates the dose equivalent by as much as a factor than that of the NRD and Andersson–Braun meters.
of 3 when used outside a shielding at particle therapy According to the manufacturer [37], the Wendi-2
facilities [17]. Therefore, such rem-meters cannot be has an excellent energy response in the normal energy
used in fast neutron therapy where neutron ener- range up to 15 MeV and closely follows the ambient
gies extend to 70 MeV; and particle therapy facilities dose equivalent up to 5 GeV. The linearity of response
where maximum neutron energies for particle therapy is within 20%. It can measure neutron ambient dose
facilities are in the GeV range. Further, neutron dose equivalent in the energy range of thermal to 5 GeV. The
dose equivalent rate range is 0.01 μSv/h to 100 mSv/h an 80-cm thick concrete shield in the CT7 reference
for 252Cf neutrons. The angular dependence is within exposure location. The neutron spectral fluence at this
20% in all directions. The neutron sensitivity (252Cf) is location is characterized by a low-energy peak with an
0.84 cps/(µSv/h) and the photon sensitivity is 1 to energy of about 0.4 eV, an intermediate region between
5 µSv/h at 100 mSv/h of 662 keV gamma rays from 137Cs. the thermal and the evaporation peak located at of about
No pulse pile-up needs to be considered for gamma dose 1 MeV; and a high-energy peak at about 100 MeV.
rates up to 1 Sv/h. The results show that the Wendi-2 agrees well within
Figure 13.8 shows the calculated energy response its uncertainties; and within one sigma of the H*(10)
relative to H*(10), normalized for a bare 252Cf calibra- value calculated by Monte Carlo calculation using the
tion, for the following rem meters: WENDI-II (side), FLUKA code. The conventional rem counters were in
Hankins-NRD, and AB (side) [13]. good agreement within their uncertainties and under-
An intercomparison of the performance of various estimated H*(10) as measured by the extended range
active neutron monitors, including Wendi-2 was car- instruments and as predicted by FLUKA [40]. The non-
ried out in at the CERN-EU reference field (CERF) facil- neutron part of the stray field accounts for about 30% of
ity [38]. The CERF mixed radiation field is produced the total H*(10).
by a positive hadron beam comprised of 2/3 protons Measurements performed in quasi-monoenergetic
and 1/3 positive pions, with a momentum of 120 GeV neutron fields indicate that H*(10) measurement results
c−1 impinging on a copper target placed inside an irra- of the Wendi-2 rem-meter can be reproduced satisfacto-
diation cave [39]. The measurements were made behind rily by FLUKA for energies of a few hundred MeV [41].
FIGURE 13.8 Energy response relative to H*(10), normalized for a bare 252 Cf calibration, for the following rem meters:
WENDI-II (side), Hankins-NRD, and AB (side) [13].
The energy response function of the Wendi-2 rem-meter discrimination to reject the photon signal. Therefore,
and the ambient dose equivalent per unit fluence as a the use of a scintillation spectrometer is not practical
function of energy showed good agreement. According for routine neutron measurements around radiotherapy
to the authors, the Wendi-2 can be considered to be facilities. Further, it is also affected by the pulsed nature
an appropriate device to measure H*(10) neutron dose of the neutron field.
behind shielding for primary proton beam energies of On the other hand, time-of-flight spectroscopy is a
about 250 MeV and about 400 MeV. laboratory method, not suitable for field applications.
Thus, Wendi-2 can be used for neutron monitoring of The best measurements of neutron spectra are obtained
fast neutron therapy and particle therapy facilities. with the time-of-flight spectrometers. When a neutron
is created or when it first enters the detector, a signal
13.6.1.3 Neutron Spectrometers is produced at the point of creation. The time taken by
The neutron spectrum can be determined by using ther- the neutron to get to a detector at some distance away is
mal neutron detectors inside a series of hydrogenous measured. The time taken to travel a given distance can
spheres of varying diameters [42]. It is possible to cal- then be used to determine the neutron energy. The time-
culate the total spectrum by taking all the response of-flight method may not be applicable for external radi-
functions and folding them into a series of equations, ation monitoring at radiotherapy facilities because of
because the amount of moderation varies in each of the scattering of the majority of neutrons after they are
these spheres. This method is referred to as the Multi- created. It is also not clear that a suitable “start signal”
Sphere or Bonner–Sphere method. The process is labori- may be obtained in such fields. Apart from being sus-
ous, as it requires software, calibration, a large number ceptible to pulsed field effects, these are expensive detec-
of spheres and long measurements times. Bonner sphere tors, and require difficult measurements. Therefore, they
spectrometers suffer from the same energy response are not very useful for external neutron monitoring at
limitations that constrain rem-meters due to their pure radiotherapy facilities.
polyethylene composition.
In addition, successful unfolding of the neutron spec- 13.6.2 Passive Neutron Monitoring
trum requires some knowledge of the actual spectrum The various types of passive monitors include film,
since there is no unique solution for a set of integral TLDs, SSNTDs, and bubble detectors. When using pas-
equations. This approach works pretty well for nuclear sive detectors, it is important to take several measure-
power plants but would prove impractical for therapy ments with the same detector type at a given location to
installations. reduce random errors.
The Bubble Technology Industries, Inc., ROSPEC
[43] is a rotating neutron spectrometer designed specifi- 13.6.2.1 Film
cally for the spectral measurement of degraded fission Photographic emulsions or film are divided into two cat-
neutrons, and detects neutrons in the energy range of egories: radiographic film and nuclear emulsions [18]. In
thermal to 4.5 MeV. It is capable of generating accu- radiographic films, a general darkening of the emulsion
rate spectral and dosimetric data simply and routinely occurs due to the cumulative effect of multiple individ-
in minutes or hours. Data analysis is performed via a ual interactions. Thus, an image is recorded of the trans-
notebook computer, with on-line display of individual mitted intensity of a radiation beam. Radiographic film
or all counter pulse height distributions, and an unfold- can also be used for thermal neutron monitoring [1].
ing program for generation of neutron spectra. Fluence, The cadmium window in the film holder absorbs ther-
kerma, maximum dose equivalent,ambient dose equiva- mal neutrons. The latent image from capture gamma
lent H*(10) and dose rate are calculated, and stored on radiation blackens the film below this window after
a hard disk. processing.
Alternatively, a scintillation spectrometer with either Nuclear emulsions are much thicker and differ in
a plastic or liquid scintillator can be used [44]. However, composition from radiographic films. They are best
the interference from photons in a mixed photon–neu- suited for recording single particle tracks. Nuclear track
tron field can be significant. A considerable amount emulsions are used for fast neutron monitoring. Recoil
of time may be required in adjusting the pulse shape protons are produced by neutrons interactions with the
hydrogen nuclei in the emulsion and surrounding mate- dose equivalent response, and the lower limit of detec-
rial. The latent image created by the recoil protons, leads tion depend on the material, the type and thickness of
to darkening of the film along their tracks after process- the radiator, and the etching process that is used. These
ing. Nuclear emulsions can be loaded with special target detectors suffer from directional dependence. CR-39
material such as boron or uranium to allow the detec- detectors can be used for neutron monitoring inside and
tion of thermal neutrons. outside the therapy linac treatment room but not inside
the primary photon beam due to the following photon-
13.6.2.2 TLDs induced effects [46, 47].
Various techniques are available such as using the body
as a moderator to thermalize neutrons (similar to albedo 2 D(γ,n)p, Eth = 2.23 MeV, (Eth = threshold energy)
dosimeters). The thermal neutrons are then detected (0.02% of hydrogen is deuterium)
using LiF enriched with 6Li. The reaction of thermal 16 O(γ,α)12C, Eth = 7.2.MeV
neutrons in 6Li provides enhanced sensitivity.
According to Mirion Technologies (GDS), Inc.,8 the 12 C(γ,α)8Be → 2α, Eth = 7.4.MeV
Genesis TLD Dosimeter [45] is a 4-element TLD with
three 7LiF:Mg,Cu,P (TLD 700H) and one 6LiF:Mg,Cu,P The Landauer3 Neutrak 144® is one of the commercially
(TLD 600 H) TLDs. The TLD 700H measures betas, available CR-39 detectors suitable for use at radiother-
gammas, and x-rays. The TLD 600H responds accu- apy facilities as a personnel dosimeter. There are two
rately to betas (0.251 MeV to 5 MeV), gamma rays neutron dosimeters [48]: a fast neutron dosimeter; and
and x-rays (5 keV to 6 MeV), and neutrons (200 keV a combination fast, intermediate and thermal neutron
to 6 MeV). Neutrons with energies from thermal to dosimeter. The fast neutron dosimeter uses a polyethyl-
6 MeV are detected by the TLD. The neutron energy ene radiator with CR-39. Recoil protons resulting from
can be extended to 20 MeV with the optional use of neutron interactions in the dosimeter are recorded. This
CR-39 (described in the next section). The response dosimeter is used for monitoring personnel working
of each element is adjusted by the application of its with unmoderated or moderately shielded fast neutron
own unique correction factor. These dosimeters mea- sources such as 252Cf and Am-Be. Landauer uses a mate-
sure deep dose, lens of eye, and shallow doses. Unlike rial that is exclusive to them and is not the typical per-
other TLD products, the Genesis Ultra TLD has virtu- formance seen by others, which generally has an energy
ally no fading characteristics. Due to the increase in response threshold of 100–150 keV.
signal response a minimum reportable dose as low as The combination neutron dosimeter measures fast,
0.01 mSv is available for the dosimeter. The useful dose intermediate, and thermal neutrons. The left area of
range is 0.01 mSv to 10 Gy. the chip uses a polyethylene radiator for fast neutrons.
The right area uses a boron loaded Teflon® radiator for
13.6.2.3 Solid-State Nuclear Track Detectors fast, intermediate, and thermal neutrons. The alpha
SSNTDs such as the polymer CR-39 (poly allyl diglycol particles resulting from thermal neutron interactions
carbonate) are used for individual monitoring. Recoil with the boron radiator are recorded in the dosimeter.
nuclei from neutron interactions with its constituent This dosimeter is used for personnel monitoring outside
atoms (hydrogen, carbon, and oxygen) leave submicro- shielded high-energy accelerators producing neutrons
scopic damage trails or tracks in the detector. A suitable with energy less than or equal to 40 MeV.
etching process, such as chemical etching, electrochem- The CR-39 is etched in a chemical bath for 15 hours
ical etching, or a combination is used to reveal these to enlarge the tracks; and the tracks are then counted
tracks. The track density can then be related to the neu- in an automatic counter. The track density is a measure
tron dose equivalent through calibration. The response of the neutron dose equivalent. The fast neutron dose
of CR-39 can be enhanced through the use of a hydrog- equivalent is measured by counting the tracks generated
enous radiator in contact with it. Additional protons are as a result of the proton recoil with the polyethylene
generated within the radiator. CR-39 can detect neutrons radiator, while the thermal neutron dose equivalent is
over a wide energy range from about 100 keV to 20 MeV measured by counting the alpha tracks generated with
when combined with an irradiator. The energy range, the boron radiator.
13.6.2.4 Bubble Detectors

Bubble detectors consist of tiny superheated droplets
that are dispersed throughout a firm elastic polymer
contained in a small sealed tube. The detector can be
sensitized by unscrewing the cap. The interaction of
neutrons with the droplets results in the production
of secondary charged particles. The charged particles
deposit energy and cause the droplets to vaporize, pro-
ducing bubbles which remain fixed in the polymer. The
bubbles can be counted either by eye or in an automatic
reader, as long as the detector is not saturated with
bubbles. A simple method of counting is to project the
image of the bubble detector via a video camera on to a
TV screen, and then count by eye or score bubbles on a FIGURE 13.9 BD-PND Bubble Detectors. (Courtesy of
clear transparency sheet placed over the TV screen [46, Bubble Technology Industries Inc., Chalk River, Ontario,
49]. The bubble detector should be rotated several times Canada.)
and counted again. The number of bubbles is a measure
of the neutron dose equivalent. Another method would Low: 0.093 to 0.019 bubbles/µSv; Medium: 0.94–1.8
be to take photographs of the detector with the bubbles, bubbles/µSv and High: 1.9–3.7 bubbles/µSv.
magnify the photographs, and count the bubbles [27]. The energy response of the BD-PND is shown in
Bubble detectors are easy to use, have a range of avail- Figure 13.10 [51, 52].
able sensitivities, are reusable, integrating and allow By varying the formulation, bubble detectors with
instant visible detection of neutrons. They also have different thresholds have been made. The bubble detec-
an isotropic response. Bubble detectors can be used for tor spectrometer or the “BDS” can be used for spectral
neutron monitoring inside and outside the therapy linac measurements. There are six thresholds – 10 keV, 100
treatment room but not inside the primary photon beam keV, 600 keV, 1 MeV, 2.5 MeV, and 10 MeV. A total of
due to photon-induced effects [46, 47]. The primary dis- 18 detectors (three of each threshold) are used to make
advantage is that the sensitivities vary within a given the spectral measurement. The measured spectrum
batch, and occasionally spurious results are obtained. In is derived by unfolding the response of the different
addition, there may be a loss of linearity at the higher threshold detectors. A recompression chamber is used
doses, which according to the manufacturer may be to zero the detectors, thus allowing for its reuse.
attributed to a change of pressure within the gel when Figure 13.11 shows the response per unit fluence of
there are a large number of bubbles present. Finally, it the BDS. For the BDS-2500, the response on the graph
is difficult to get good statistics with a single detector. has been reduced by a factor of 10; hence, the notation
Therefore, it is important to use a minimum of three 2500/10.
detectors at each measurement location. The BDT (bubble detector thermal) uses a 6Li com-
Figure 13.9 shows a photograph of bubble detec- pound dispersed throughout the polymer and a special
tors available from Bubble Technology Inc.9 The formulation to detect preferentially α particles from the
bubble detectors are normally calibrated by the manu- 6Li(n,α)3H reaction [51]. The thermal to fast neutron
facturer against an Am-Be source. The BD-PND (Bubble ratio is approximately 10:1. The energy range is thermal.
Detector-Personal Neutron Dosimeter) is a temperature- The dose range is 1.1–112 µSv. The sensitivity range is
compensated bubble detector and therefore, useful 1.8–3.6 bubbles/µSv.
for monitoring, since no temperature corrections are
required. According to the manufacturer, the BD-PND 13.7 CONTINUOUS, PULSED,
[50] has an approximate energy threshold of 100 keV AND MIXED RADIATION FIELDS
with a reasonably flat dose equivalent response from In radiotherapy, one encounters both continuous and
about 200 keV to greater than 15 MeV. The dose range pulsed radiation fields. Brachytherapy and radioactive
is from 1 µSv to 5 mSv. There are three sensitive ranges: sources provide continuous radiation, while therapy
FIGURE 13.10 BD-PND normalized response per unit fluence (closed circles – upper curve) and response per unit dose
equivalent (closed diamonds – lower curve). Conversion from fluence-to-dose equivalent based on NCRP Report No. 38 [51].
FIGURE 13.11 Normalized response of BDS at 6 different threshold energies. The number 2500/10 means that on the graph,
the response of the BDS-2500 has been reduced by a factor of 10 [51].
linear accelerators (linacs) and accelerator-based BNCT, the Geiger–Muller and proportional counters, tend to
fast neutron therapy, and particle therapy facilities can become saturated in such fields, and count only the rep-
produce pulsed radiation. In particle therapy facilities, etition rate. The dead time usually lies between 10 −8 s
the beams from a synchrotron and a synchrocyclo- and 10 −4 s.
tron are considered pulsed, while the beam from an Ionization chambers are less influenced as long as they
isochronous cyclotron is considered continuous [53]. are operated with a voltage that is sufficient to overcome
However, the extracted beam from the synchrotron recombination losses. Ionization chambers operating in
is considered continuous, when the particles are the current mode are more suitable for higher dose rate
extracted slowly [17]. measurements. Therefore, they are more widely used
Therapy linacs produce mixed radiation fields com- than GM or proportional counters in pulsed fields.
prised of both photons and neutrons [27]. For example, Scintillation detectors may become non linear at high
the photon fluence inside the primary beam of a therapy dose rates in pulsed fields, because photomultipliers
linac is at least 1000 to 4000 times higher than the cor- cannot handle the high instantaneous dose rates.
responding neutron fluence. Outside the primary beam, For neutron detectors that are moderated, it has been
the photon fluence is at least 10 to 100 times higher than observed that the measured readings are higher than the
the neutron fluence. repetition rate due to scattered radiation within the room;
Typically, therapy linacs are operated at repetition and coupled with the fact that neutron moderation allows
rates that vary from 100 to 400 Hz with pulse widths of an event to be detected even after the pulse has termi-
1 µs to 10 µs [27]. Peak electron currents may vary from nated [27]. It may be possible to adjust the accelerator to a
20 to 120 mA per pulse in the photon mode, and 3 to low output per pulse, but the photon-to-neutron dose rate
15 mA per pulse in the electron mode. Higher peak cur- may not remain the same at the lower output. Further,
rents may be used in the flattening filter-free mode (FFF). in mixed photon–neutron fields, the neutron detectors
The fraction of operating time during which the beam may have photon-induced reactions [47], which cannot
is on, is called the duty factor (DF) [19, 54]. The duty fac- be separated from the neutron interactions. This interfer-
tor is the product of the pulse width and the repetition ence should be considered when measuring neutrons in
rate or frequency. The peak or instantaneous intensity, the primary photon beam of a therapy linac. Thus, only
Ip, is given by: passive neutron detectors should be used in the primary
photon beam, provided that they do not experience any
I av
Ip = (13.9) photon-induced reactions.
DF
For pulsed radiation, with a pulse repetition rate of
where Iav is the average intensity. P Hz, if the dead time, TD, is shorter than the pulse width
The duty factor for a beam with a pulse width of 1 µs TP, the corrected counts Ccorr, is related to the measured
and a repetition rate of 100 Hz is equal to 1 × 10−4. This counts, C, by the following expression [55]:
very small duty factor imposes a severe limitation on
the radiation detector, as in this case, the peak intensity C
Ccorr = (13.10)
is 10,000 times the average intensity. This intense pulse 1− CTD
PTP (1 − 2TTDP )
usually overwhelms any active detector which detects
particle events electronically. As a result, the measured If TD is >> than TP
readings are only an indication of the repetition rate. A
detector that usually responds well to an average dose rate Ccorr = C − P ln(1 − C / P ) (13.11)
spread out in time may not be able to cope with such high
dose rates. For a moderated thermal neutron detector, the effective
The resolving time or dead time is the time required dead time is determined by the moderation time, and
by the detector to return to its normal state, after reg- not the pulse width for the neutrons. Therefore, the pulse
istering a pulse. It is the minimum time interval by width is replaced by the moderation time in the above
which two pulses must be separated in order to be equations. For well-shielded facilities, the neutron dose
detected as separate pulses by the counter and its acces- equivalent rates are fairly low and may be of the order
sories. Detectors, which have long dead times such as of 1 μSv/h or so. Therefore, integrated measurements
of dose equivalent are preferable to instantaneous dose 4. ICRP, Data for Protection Against Ionizing Radiation from
equivalent rate measurements. External Sources. Supplement to ICRP Publication 15.
International Commission on Radiation Protection
Publication 21(Pergamon Press, Oxford, 1973).
13.8 CONCLUSIONS 5. ICRP, 1990 Recommendations of the International
Photon and neutron area monitoring and individual Commission on Radiological Protection. ICRP
monitoring are discussed. Examples of some commer- Publication 60. Ann. ICRP 21 (1-3). (Pergamon Press,
UK, 1991).
cial monitors (available at the time of writing this chap-
6. ICRP, The 2007 Recommendations of the International
ter) are provided. However, with time, some of these Commission on Radiological Protection. ICRP
monitors may become obsolete. Publication 103. Ann. ICRP 37 (2-4) (Elsevier Science,
UK, 2007).
13.9 ACKNOWLEDGMENTS 7. ICRU, Quantities and Units in Radiation Protection
Dosimetry. International Commission on Radiation
Sincere gratitude is expressed to Richard H. Olsher and Units and Measurements Report 51 (International
Kamran Vaziri for their review of this chapter and for Commission on Radiation Units and Measurements,
their suggestions and comments. Bethesda, 1993).
8. ICRU, Conversion Coefficients for Use in Radiological
Protection against External Radiation. International
NOTES Commission on Radiation Units and Measurements
1 Mirion Technology Inc., 3000 Executive Pkwy #222, Report 57 (International Commission on Radiation
Units and Measurements, Bethesda, 1998).
San Ramon, CA 94583, USA.
9. L. Moritz, “Radiation measurements and monitor-
2 ICx Radiation Inc., 100 Midland Road, Oak Ridge, TN ing,” in Topics in Accelerator Health Physics, edited by
37830, USA. J. D. Cossairt, V. Vylet, J. W. Edwards (Medical Physics
3 Landauer, Inc., 2 Science Road, Glenwood, IL 60425- Publishing, Wisconsin, 2008).
1586, USA. 10. ANSI, American National Standard for Radiation
4 RADOS Technology Oy, Turku, Finland. Protection Instrumentation Test and Calibration,
5 Scientific Applications International Corporation, Portable Survey Instruments N323AB-2013 (American
McLean, VA, USA. National Standard Institute, Washington, 2013).
6 Thermo Fisher Scientific, 168 Third Avenue Waltham, 11. IAEA, Calibration of Radiation Monitoring Instruments.
MA 02451, USA. International Atomic Energy Agency Report 16
7 Fluke Biomedical, 6045 Cochran Rd., Cleveland, OH (International Atomic Energy Agency, Vienna, 2000).
44139, USA. 12. ISO, ISO 4037: Parts 1–3: Radiological protection - X
and gamma reference radiation for calibrating dose-
8 Mirion Technologies (GDS), Inc.2652 McGaw Avenue |
meters and doserate meters and for determining their
Irvine, CA 92614, USA.
response as a function of photon energy (International
9 Bubble Technology Industries Inc., Chalk River, Organization for Standardization, 2019).
Ontario, Canada. 13. R. H. Olsher, H. H. Hsu, A. Beverding, et al., “WENDI:
An improved neutron rem meter,” Health Physics 79,
REFERENCES 170–181 (2000).
14. NCRP, Protection Against Neutron Radiation. National
1. G. Rajan and J. Izewska, “Radiation monitoring instru- Council on Radiation Protection and Measurements
ments,” in Review of Radiation Oncology Physics. A Report No. 38 (National Council on Radiation Protection
Handbook for Teachers and Students, edited by E. and Measurements, Washington, 1971).
Podgorsak (International Atomic Energy Agency, 15. D. Rogers, “Why not to trust a neutron rem meter,”
Vienna, 2005), pp. 101–120. Health Physics 37, 735–742 (1979).
2. N. E. Ipe, “Radiological aspects of particle therapy 16. L. A. DeWerd and B. R. Smith, “Ionization chamber
facilities,” in Shielding Design and Radiation Safety at instrumentation,” in Radiation Therapy Dosimetry:
Particle Therapy Facilities, Particle Therapy Cooperative A Practical Handbook, edited by A. Darafsheh (CRC
Group Report 1, edited by N. E. Ipe (Particle Therapy Press, Boca Raton, FL, 2021), Chapter 2, pp. 19–30.
Cooperative Group, Villigen, 2009), pp. 24–79. 17. Y. Uwamino and G. Fehrenbacher, “Radiation moni-
3. NCRP, Radiation Protection for Particle Accelerators. toring,” in Shielding Design and Radiation Safety of
National Council on Radiation Protection and Charged Particle Therapy Facilities. PTCOG Report 1.
Measurements Report No. 144 (National Council on Particle Therapy Cooperative Group, edited by N. Ipe
Radiation Protection and Measurements, Bethesda, (Particle Therapy Cooperative Group, Villigen, 2010),
2003). pp. 176–200.
18. G. Knoll, Radiation Detection and Measurement. Second 31. A. O. Hanson and J. L. McKibben, “A neutron detec-
Edition (John Wiley & Sons, New York, 1989). tor having uniform sensitivity from 10 keV to 3 MeV,”
19. R. C. McCall and N. E. Ipe, “The response of sur- Physics Review 72, 673–677 (1972).
vey meters to pulsed radiation fields,” in Seventh 32. NCRP, Neutron Contamination from Medical Electron
International Congress of the International Radiation Accelerators, National Council on Radiation Protection
Protection Association. Radiation Protection Practice and Measurements Report No. 79 (National Council
Vol. 1 (International Radiation Protection Association, on Radiation Protection and Measurements, Bethesda,
Sydney, 1988). 1984).
20. Mirion Technologies (Canberra) Inc., canberra.com 33. I. O. Andersson and J. A. Braun, “A neutron rem counter
[Online] (2017). Available at: http://www.canberra.com/ with uniform sensitivity from 0.025 eV to 10 MeV in neu-
products/hp_radioprotection/pdf/In1K-SS-C38987.pdf tron dosimetry vol. II,” in Proceedings of an International
[Accessed 28 October 2018]. Atomic Agency Symposium. IAEA STI/PUB/69
21. Thermo Fisher Scientific, thermo.com/rmp. [Online] (International Atomic Energy Agency, Vienna, 1963).
(2007). Available at: http://www.thermo.com.cn/ 34. M. Cosack and H. Leisecki, “Dose equivalent survey
Resources/200802/productPDF_30096.pdf [Accessed meters,” Radiation Protection Dosimetry 10(1–4), 111–
15 March 2020]. 119 (1985).
22. J. Turner, Atoms, Radiation, and Radiation Protection 35. D. E. Hankins and J. R. Cortez, Directional Response
(Pergamon Press, New York, 1986). and Energy Dependence of Four Neutron Rem meters.
23. J. Izewska and G. Rajan, “Radiation dosimeters,” in Los Alamos Scientific Laboratory Report LA-5528 (Los
Review of Radiation Oncology Physics. A Handbook Alamos Scientific Laboratory, Los Alamos, 1974).
for Teachers and Students, edited by E. Podgorsak 36. N. E. Ipe, “Secondary radiation production and shield-
(International Atomic Energy Agency, Vienna, 2005), ing at proton therapy facilities,” in Proton Therapy
pp. 71–99. Physics, 2nd Edition, edited by H. Paganetti (CRC Press,
24. Landauer, Inc., Landauer.com. [Online] (2018). Available Boca Raton, 2018), pp. 207–240.
at: https://www.landauer.com/sites/default/files/product- 37. Thermo Fisher Scientific, www.thermofisher.com.
specification-file/Luxel%2B_1.pdf [Accessed 15 March [Online] (2018). Available at: https://www.thermo-
2020]. fisher.com/order/catalog/product/FHT762WENDI2
25. J. Meier and S. C. Kappadath, “Characterization of [Accessed 15 March 2020].
the energy response and backscatter contribution for 38. M. Carseana, M. Helmecke, J. Kubancak, et al.,
two electronic personal dosimeter models,” Journal of “Instrument intercomparison in the high-energy mixed
Applied Clinical Medical Physics 16(6), 423–434 (2015). field at the CERN-EU reference field (CERF) facility,”
26. Thermo Fisher Scientific, thermofisher.com. [Online] Radiation Protection Dosimetry 161(1–4), 66–72 (2014).
(2019). Available at: https://www.thermofisher.com/ 39. A. Mitaroff and M. Silari, “The CERN-EU high-energy
order/catalog/product/EPD161081000 [Accessed 15 reference field (CERF) facility for dosimetry at commer-
March 2020]. cial flight altitudes and in space,” Radiation Protection
27. NCRP, Structural Shielding Design and Evaluation Dosimetry 102, 7–22 (2002).
for Megavoltage X- and Gamma-Ray Radiotherapy 40. A. Ferrari, P. Sala, A. Fasso, et al., FLUKA: A Multi-
Facilities. National Council on Radiation Protection Particle Transport Code. CERN Yellow Report CERN
and Measurements Report No. 151 (National Council 2005-10. INFN/TC 05/11. SLAC-R-773, (CERN, Geneva,
on Radiation Protection and Measurements, Bethesda, 2005).
2005). 41. L. Jagerhofer, E. Feldbaum, D. Forkel-Wirth, et al.
28. AAPM, Neutron Measurements around High Energy “Characterization of the WENDI-II REM counter for its
Radiotherapy Machines. American Association of application at MedAustron,” Progress in Nuclear Science
Physicists in Medicine Report No. 19 (American Institute and Technology 2, 258–262 (2011).
of Physics, Inc., New York, 1986). 42. R. L. Bramblet, R. I. Ewing, T. W. Bonner, “A new type
29. ICRP, Conversion Coefficients for Radiological of neutron spectrometer,” Nuclear Instrument Methods
Protection Quantities for External Radiation Exposures. 9, 1–12 (1960).
International Commission on Radiation Protection 43. Bubble Technology Industries, Inc., bubbletech.ca.
Publication 116, Ann. ICRP 40(2–5) (International [Online] (2018). Available at: http://bubbletech.ca/prod-
Commission on Radiation Protection, Bethesda, 2010). uct/rospec/ [Accessed 8 November 2018].
30. R. Nath, K. W. Price, H. R. Holeman, “Mixed pho- 44. R. C. McCall, Neutron Measurements. Stanford Linear
ton-neutron field measurement,” in Proceedings Accelerator Center Publication 2262 (Stanford Linear
of Conference on Neutrons from Electron Medical Accelerator Center, Menlo Park, 1981).
Accelerators. April 9–10, 1979. National Bureau of 45. Mirion Technology (GDS), Inc., mirion.com. [Online]
Standards Special Publication 554 (U.S. Government (2014). Available at: https://mirion.app.box.com/s/
Printing Office, Washington, 1979). u55f753y2qv1x940v5vf [Accessed 26 October 2018].
46. N. E. Ipe and D. D. Busick, BD-100: The Chalk River 51. H. Ing, R. A. Noulty, T. D. Mclean, “Bubble detectors
Nuclear Laboratories’ Neutron Bubble Detector. – A maturing technology,” Radiation Measurements 27,
Stanford Linear Accelerator Center PUB 4398 (Stanford 1–11 (1997).
Linear Accelerator Center, Menlo Park, 1987). 52. B. J. Lewis, M. B. Smith, H. Ing, et al., “Review of bub-
47. N. E. Ipe, S. Roesler, S. B. Jiang, C. M. Ma, Neutron ble detector characteristics and results from space,”
Measurements for Intensity Modulated Radiation Radiation Protection Dosimetry 150(1), 1–21 (2012).
Therapy. CD ROM Proc. of 2000 World Congress 53. M. Schippers, “Proton beam production and dose
on Medical Physics and Biomedical Engineering, delivery techniques,” in Principles and Practice,
July 23-28. Also SLAC PUB 8433 (Stanford Linear American Association of Physicists in Medicine. Medical
Accelerator Center, Menlo Park, 2000). Physics Monograph No. 37, edited by I. J. Das and H.
48. Landauer, www.landauer.com. [Online] (2017). Available Paganetti (Medical Physics Publishing, Madison, 2015),
at: https://www.landauer.com/sites/default/files/product- pp. 129–164.
specification-file/Neutrak_0.pdf [Accessed 15 March 54. N. E. Ipe and R. C. McCall, The Stanford linear accelera-
2020]. tor center pulsed X-ray facility. [Also SLAC-PUB-3966
49. N. E. Ipe, D. D. Busick, R. W. Pollock, “Factors affecting Revised]. Health Physics 52(4), 463–468 (1987).
the response of the bubble detector BD-100 and a com- 55. IAEA, Radiological Safety Aspects of the Operation
parison of its response to CR-39,” Radiation Protection of Electron Accelerators. International Atomic Energy
Dosimetry 23, 135–138 (1988). Agency Technical Report Series No. 188 (International
50. Bubble Technology Industries Inc., bubbletech.ca. Atomic Agency, Vienna, 1979).
[Online] (2018). Available at: http://bubbletech.ca/prod-
uct/bd-pnd/ [Accessed 26 October 2018].
Chapter 14
Monte Carlo Techniques

in Medical Physics
Ruirui Liu and Tianyu Zhao
St. Louis, Missouri
Milad Baradaran-Ghahfarokhi
St. Louis, Missouri
Vanderbilt University Medical Center
Nashville, Tennessee
CONTENTS
14.1 Introduction of Monte Carlo Simulation 211
14.2 Applications of Monte Carlo Simulation In Medical Physics 213
14.2.1 Application in Patient Dose Calculation 213
14.2.2 Application in Dosimetry Instruments 214
14.2.3 Application in Microdosimetry 215
14.2.4 Application in Nuclear Medicine 217
14.2.5 Applications for X-ray Imaging 217
14.3 Monte Carlo Codes and Systems in the Public Domain 218
14.3.1 Geant4 218
14.3.2 MCNP 220
14.3.3 EGS4/EGSnrc 220
14.3.4 Penelope 220
14.3.5 FLUKA 221
14.4 Limitations of Monte Carlo Simulation 221
14.5 Summary 221
References 222
14.1 INTRODUCTION OF MONTE its frequency. Jacob Bernoulli concluded in his book, Ars
CARLO SIMULATION Conjectrandi, that “this method was not new nor special,”
Monte Carlo (MC) simulation, also known as random and he continued “everyone knows, in order to make pre-
sampling technique or statistical experimental method, diction of one type of phenomenon, one time or two times
was first introduced in early 17th century when it was real- observation is not enough, instead, much more times of
ized how to determine the probability of an event based on observation is needed. The more the observation, the less
211
The probability of needle crossing the line is given by:
p =
∫∫ g (x ,θ ) f (x ) f (θ )dx dθ
1 2
π l sinθ
1 1
=
∫
0
π
dθ
∫ a dx
0
(14.4)
2l
=
πa
Given the uniform distribution of x and θ , they can be

FIGURE 14.1 Position of the needle between the parallel sampled as multiplications of their range and two inde-
lines. pendent random numbers,
risk that the goal fails” [1]. The modern form of MC sim-
x = aξ1
ulation was introduced in the late 1940s by construct- (14.5)
ing a Markov chain [2] following the development of θ = πξ 2
computers. It was firstly applied in developing of nuclear
weapons [3]. In the following paragraph, an example is where ξ1 , ξ 2 are random numbers in (0,1). By the law of
presented to explain the basic idea of MC simulation. large numbers, the integral described in Equation 14.4
One well known example for illustrating how MC can be approximated by taking the arithmetic mean of
simulation works is taken from the work by Comte de independent samples of g ( x ,θ ),
Buffon (1777) [4], who proposed a MC-like method to
determine π by constructing an experiment repeatedly N
tossing a needle onto a ruled sheet of paper and recording pN =

1
N ∑ g ( x ,θ ) i i (14.6)
the probability of the needle crossing either of the lines. i =1
Given the coordinate system as shown in Figure 14.1,

Therefore π can be approximated by
the center of the needle x and the angle between the nee-
dle and the parallel lines θ follow a uniform distribution: 2l
π≈ . (14.7)
pN a
1
 , 0≤x ≤a
f1 ( x ) =  a (14.1) From perspective of probability theory, the idea of MC
 0, other method is to approximate the true value by running a

large number of sampling [5]. Using MC method, the
and integration could be taken as the mean of the random
variable g ( r ) which follows specific probability density
 1 function f ( r ), which is
 , 0 ≤θ ≤ π
f 2 (θ ) =  π (14.2)
 0, other
 〈g 〉 =
∫ g (r ) f (r )dr (14.8)
with x ∈[0, a] and θ ∈[0, π ].

The needle crosses the line when x ≤ l sinθ , 0 ≤ x ≤ a, With N samples of parameter r, r1 , r2 , …, rN , estimated
which can be described by a random variable g ( x ,θ ), 〈 g 〉 is given by the arithmetic mean of Zi = g (ri ) as
N
 1, x ≤ l sinθ
g ( x ,θ ) = 
 0, other
(14.3) Zˆ N =
1
N ∑Z
i =1
i (14.9)
Monte Carlo Techniques in Medical Physics ◾ 213
By the law of large numbers, Zˆ N converges to 〈 g 〉 if modeling components of treatment units, radiation
Z1 , Z 2 ,…, Z N are independent from each other and have sources, and patient structures, MC simulation can
finite mean ( E( Z ) < ∞), accurately estimate the dose distribution in heteroge-
neous media, including patient geometry [6]. The main
 1 N  practical limitation of using MC simulation for patient
lim P 
N →∞  N

∑Z − E(Z ) < ε  = 1
i =1
i (14.10) dose calculation is the high demand for computational
power to achieve acceptable accuracy in the calculated
dose. Due to this limitation, MC calculations are mainly
Equation 14.10 implies that, the estimated integration
used for forward planning (forward dose calculation)
results in converges to a true integration, once the size
[7] as the rapidly increasing computational burden from
of sample set is large enough [5].
inverse planning always outpaces the improvement of
According to the central limit theorem, if random
central processing unit (CPU) performance. Recently,
variables Z1 , Z 2 ,…, Z N are independent but follow the
it has been shown that graphic processing units (GPUs)
same distribution, we have
have great potential over conventional CPUs architec-
x ture to speed up the dose calculations. Instead of cen-
 N ˆ  1 t2
∫e
−
lim P Z N − E( Z ) < x  = 2 dt (14.11) tralized streamline of computation with fewer cores
N →∞  σ  2π
−x responsible to tasks in a sequential order, GPU, which
where σ is the standard deviation of variable Z given by consists of thousands of small cores performing mul-
tiple tasks concurrently, is able to increase the speed of
MC simulations dramatically [8].
σ 2 = E( Z − E( Z ))2 =
∫ (Z(t) − E(Z)) f (t )dt (14.12)
2
The modern GPU is a highly data-parallel proces-
sor, which is optimized to run excessive floating point
where f (t ) is the distribution of t. It is achievable to
arithmetic output for simulations using a single pro-
reduce the error between the true and estimated value
gram multiple data (SPMD) model [8]. On a single GPU,
below a desired level with a probability of 1−α , where α
the SPMD model operates by launching thousands of
denotes the significance level.
threads running the same program (the so-called “the
x
t2 kernel”) working on different data. GPU can rapidly
 xσ  1
P  Zˆ N − E( Z ) <
∫e
−
= 2 dt = 1 − α (14.13) switch between high numbers of different threads to
 N  2π
−x ensure the hardware is busy at all the processing times,
which can significantly hide memory latency. On the
If α = 0.05, then the following inequality holds:
other hand, using CUDA™ (Compute Unified Device
1.96σ Architecture) programs, which are written based on
Zˆ N − E( Z ) < (14.14) the C programming language, to employ the paral-
N
lelism of the GPU can provide very fast implementa-
It is worth noting that the error of MC method is a sta- tions of standard mathematical functions to calculate
tistical error. Since σ in Equation 14.14 is not known, patient dose [9].
an estimated value of σ is typically used to quantify the For photon dose calculations, Badal et al. intro-
statistical error. The estimated σ can be calculated as duced a GPU-based MC code that simulates photon
transport in a voxelized geometry [9]. For their simula-
2
N
 1 N
 tions, the accurate physics models were developed from
σˆ =
1
N ∑
i =1
Z −
i
2
N
∑i =1
Zi 

(14.15) PENELOPE MC package using the CUDA™ program-
ming model (NVIDIA Corporation, Santa Clara, CA).
They reported up to 27-fold speedup over a single core
14.2 APPLICATIONS OF MONTE CARLO CPU for the same calculations.
SIMULATION IN MEDICAL PHYSICS Fang et al. have presented a parallel MC algorithm
14.2.1 Application in Patient Dose Calculation accelerated by GPU for modeling time-resolved photon
The MC method has been demonstrated to be a use- migration in arbitrary 3D turbid media [10]. Similar to
ful tool for dose calculation in radiation therapy. By Badal et al. the code was implemented in CUDA, and
benchmarked under various parameters, such as selec- by Hissoiny et al. [17]. It was found that influence of
tion of random number generation (RNG), memory Lorentz force from the magnetic field acting on the
access pattern and thread number. Using a low-cost charged particles, i.e., electrons, should be taken into
graphic card in their method, a speed-up ratio over consideration by the dose calculation engine in the
300 has been reported with 1792 parallel threads over planning stage. According to the Hissoiny et al. GPU-
a conventional CPU. In a study on the development of based MC calculations agreed well with measurement
an accurate and efficient dose calculation engine for on a 2%/2 mm gamma analysis. Moreover, execution
patient dose calculation designed for online adaptive time of less than 20 seconds was achieved in a prostate
radiotherapy performed by Gu et al., a finite size pen- case phantom for a 2% statistical uncertainty, indicat-
cil beam (FSPB) algorithm was created on GPU with ing that GPU-based MC calculations is a promising
a 3D-density correction method [11]. The developed candidate for accurate and fast dose calculations for the
algorithm was tested for 10 intensity-modulated radia- hybrid MRI–Linac modality.
tion therapy (IMRT) treatment plans (5 head-and-neck On the patient-specific CT/CBCT imaging dose cal-
cases and 5 lung cases). Better than 2% of accuracy was culation, since patient-specific imaging dose may be
reported for the majority of the calculation voxels while proposed for the purpose of dose management, Jia et al.
the dose calculation was accomplished well within 1 have successfully developed a MC dose calculation code
second. Jia et al. have studied their GPU-based MC cal- on GPU architecture on the NVIDIA CUDA platform
culation package for coupled electron–photon transport for fast and accurate estimation of the x-ray imaging
[12]. Their results demonstrated speedup factors of about dose received by a patient during a CT or CBCT scan
5.0–6.6 times with an NVIDIA Tesla C1060 GPU card [18]. They found speedup of about 400 and 76.6 times
over a 2.27 GHz Intel Xeon CPU processor in the range over EGSnrc MC code when the simulations were per-
of energies used typically in radiotherapy. They also formed in a homogeneous water phantom and Zubal
have discussed about the factors that adversely restrict phantom, respectively. Despite the inhomogeneities
the efficiency of such simulations, such as the memory in the Zubal phantom, imaging dose calculation was
access pattern and no cashing of global memory on reported to complete in about 17 s with statistical uncer-
GPU. Men et al. proposed a GPU-based ultrafast IMRT tainty of 0.4%.
plan optimization for online adaptive radiotherapy For accurate dose calculation in clinical practice,
techniques in order to handle the inter-fraction varia- commercial treatment planning systems (TPS) imple-
tion of the patient’s geometry [13]. Using an NVIDIA menting MC dose calculation are available. Cygler
Tesla C1060, they achieved speedup factors of 20–40 et al. implemented clinically a commercial MC-based
with accuracy compared to the simulation running on TPS for electron beams. The software they used was
an Intel Xeon 2.27 GHz CPU. developed by MDS Nucletron based on Kawrakow’s
For low-energy brachytherapy, MC simulation has VMC++algorithm [19]. The MC module was integrated
been used widely to determine the dosimetric param- with Theraplan Plus™ TPS [20]. The adoption of MC for
eters for various source designs [14]. Compared to the dose calculation has increased over the last decade in
deterministic formalisms which usually take assump- some more popular commercial TPSs. Varian’s Eclipse
tions simplifying patient anatomy, heterogeneity, fidu- electron MC (eMC) deals with electron transport using
cial seeds orientation and inter-seed attenuation, MC MC calculations [21]. Elekta’s Monaco implements pho-
simulation has demonstrated excellent accuracy in the ton and eMC algorithms for patient’s dose calculation
calculated dosimetry [15]. Hissoiny et al. evaluated the and Monaco beam models for the Versa HD Linac for
capability of GPU-based MC calculations which ulti- routine clinical use [22]. RayStation TPS uses MC simu-
mately resulted in fast and accurate results for routine lation for both electron [23], and proton dose calcula-
planning [16]. The study was able to reproduce the dosi- tions [24].
metric parameters of the radiation sources to within
1.25% from other literature, depending on the seed type, 14.2.2 Application in Dosimetry Instruments
material, or radionuclide. MC simulation of dosimetry instruments, i.e., cavity
GPU-based fast dose calculation in magnetic fields of ion chambers, plays a central role in the calculations
a MRI–Linac (linear accelerator) has also been studied of ionizing radiation transport in dose determination.
Moreover, beyond different generations of radiation to radiation due to its high density, which is approxi-
dosimetry protocols, MC techniques have been shown mately 1800 times higher than that of air in the sensitive
to be the essential part on improving the accuracy of the volume of ion chambers [36]. Using EGSnrc MC code,
calculated absorbed dose. Consequently, a large amount Wang and Rogers, studied the depth response of diode
of research work has been devoted to these specific detector response in both 6 and 18 MeV electron beams
applications of MC simulation packages. [37]. Their results showed that, the water/diode dose
In clinical practice, the determination of absorbed ratio does not depend on field size and energy for the
dose is performed using a cavity ion chamber placed in broad incident electron beam. Such MC simulations
water phantom, based on a reference dosimetry setup have allowed studies to be performed in dosimetric situ-
which is defined by the geometry of interest and radia- ations that could not be done by other means, with the
tion source [25, 26]. Over the last decades, efforts have same level of accuracy.
been made to improve the accuracy (0.1% or better) of Cranmer-Sargison et al. have implemented a MC-
MC methods to calculate the response of an ion cham- based small field dosimetry formalism for a compre-
ber in a radiation beam according to the dosimetry pro- hensive set of diode detectors, including IBA (IBA
tocols [27–29]. For each dosimetric situation, this level of Dosimetry GmbH, Schwarzenbruck, Germany) ste-
accuracy can be obtained by assessing different correc- reotactic field diode as well as the PTW (PTW The
tion factors using MC simulation. Among the dosimetry Dosimetry Company, Freiburg, Germany) T60008,
related factors, beam quality conversion factor has the T60012, T60016, and T60017 field diodes [36]. Based on
largest source of uncertainty [26]. Muir et al. performed their simulations, calculated output factors in water and
MC simulations to calculate beam quality conversion output ratios at various depths were sensitive to source
factors for large sets of ion chambers used clinically in parameterization in small filed dosimetry.
radiation therapy [30]. They found excellent agreement In another study conducted by Crop et al., the effects
(up to 0.13% for the entire data set) between measure- of small field sizes, position in penumbra, spot size and
ment and MC calculated factor. This exceptional agree- measurement depth on perturbation factors such as cen-
ment observed in their study provided great confidence tral electrode, wall, air–wall interface and volume for
in adopting MC calculated factor for updated dosimetry micro-ionization chambers were investigated [38]. Their
protocols. Buckley and Rogers focused upon the wall results showed that small electron spot size on the target
correction factor for thimble ionization chambers [31], and sharp penumbra led to larger variation for the per-
which accounts for the dosimetric disturbance from the turbation factors.
material difference between the chamber wall and the Francescon et al. employed BEAMnrc MC code to
surrounding phantom, where the Bragg–Gray assump- calculate total scatter factors in CyberKnife radiosurgery
tion could break down. MC calculation as an invaluable system. In their work, simulations were also performed
tool for such investigations showed that, for a cylindrical to calculate correction factors for the microchambers
Farmer type chamber with graphite wall, the factor was by means of a consistency check between measurement
significantly sensitive (up to 2.5%) to the depth for a and simulation. They stated that, the MC method rep-
6 MeV electron beam [31]. resented an “ideal” dosimeter for such studies, since it
Using MC simulations, Bouchard et al. presented a simulated energy deposition per each radiation particle
calculation method of dose uncertainties caused by in a given material and their results could be applied to
introducing error deliberately, including machine iso- the Cyberknife radiosurgery systems in clinical practice
center and detector position in the coordinate system of [39, 40].
the dosimetry phantom [32]. This proposed MC method
can be beneficial during radiation beam calibration, 14.2.3 Application in Microdosimetry
determination of beam quality correction factors as well From the beginning, the aim of microdosimetry has been
as the measurements of output factors. the study of the stochastic nature in the energy deposi-
Some investigators used MC to model response of diode tion by ionizing radiation and its consequences in the
detectors to radiation in dosimetry protocols [33–35]. biological response mechanism. The damage induced by
Compared to the ionization chambers, doped silicon in radiation in biological structures, such as cells which is
diodes is much smaller in size with higher sensitivity in the order of micron scale, depends crucially on the
TABLE 14.1 The Major Characteristics of Different MC Track Structure Codes in Microdosimetry
Code Interactions Particle Type Energy Range
KURBUC Elastic scattering, excitation, ionization Proton, alpha 10 eV–10 MeV
PARTRAC Elastic scattering, excitation, ionization Electron, photon, and ions 10 eV–10 MeV
NOREC Elastic scattering, Excitation, ionization Electron 7.4 eV–1 MeV
Geant4-DNA Elastic scattering, excitation, ionization Electron, proton, alpha, and light ions 0 eV–106 MeV
spatial distribution of energy deposition events gen- DNA damages. The major characteristics of these codes
erated by the energetic ions or secondary electrons. are tabulated in Table 14.1.
Modeling the radiation-induced effects in a microscopic KURBUC was developed for water vapor medium
volume, such as cellular structure and DNA molecule, and can handle light ions such as protons (1 keV to
is the domain of microdosimetry or nanodosimetry, 1 MeV) and alpha particles (1 keV to 8 MeV). In addi-
where the two terms being often used interchangeably tion, the free radical species from water radiolysis in
[41]. Microdosimetric quantiles, which describes the the chemical stage are included in this code. NOREC
energy deposition information in microscopic volumes, is a code modified from the original OREC (Oak ridge
are used to explain the model differences of the radi- electron transport code) with updated electron elastic
ation-induced effects on biological targets at the same scattering cross sections. The chemical track remains
average energy deposited per unit mass. Due to the capa- similar as KURBUC but there is no biological phase
bility of modeling the stochasticity of energy deposition simulation [48]. PARTRAC was originally developed for
of radiation in cell, MC methods are used to simulate track structure of electrons, but the photon and ion can
the microdosimetric quantities, such as linear energy be tracked in the simulation now. The chemical track
and specific energy. Linear energy is the quotient of the includes the formation of diverse chemical species simi-
energy imparted to the volume, by a single particle or lar to KURBUC and NOREC. A DNA double-strand
radioactive decay inside it, and the volume mean chord. break (DSB) repair module, based on non-homolo-
The specific energy is defined as the energy imparted gous end-joining (NHEJ), has been added in the code.
divided by the mass of the specific region [42]. Both linear Recently, with the new developments of PARTRAC
energy and specific energy are stochastic variables and code, heterochromatin and euchromatin structures
commonly are described using probability distribution. were modeled to account for variations in DNA dam-
MC codes can be coupled to models that describe cell age yields, complexity and repair between these regions.
damage and biological effects caused by radiation based The mitochondrial DNA damage also can be simulated
on microdosimetric quantities [43]. A MC code address- using PARTRAC [49, 50]. Last but not least, Geant4-
ing this research domain should be capable of simulat- DNA is an open source simulation toolkit that aims to
ing the particle track structure over track lengths in the model the effects of radiation on biological systems at
order of nanometers which is compatible with the DNA cellular and DNA levels. Geant4-DNA includes physi-
size. This requirement implies that the cut off energy cal, chemical and biological models to simulate cellular
should go down to 10–100 eV to take soft energy-loss and subcellular damage induced by ionizing radiation.
collision into account. It requires detailed knowledge of Geant4-DNA introduces a new set of physical models to
the electronic properties of the target, to ensure the cor- increase the definition of physical models at lower par-
rect treatment of energy loss occurrences of a magnitude ticle energies, which is quite beneficial to the DNA dam-
comparable to electronic binding energies [41]. age quantification. Currently, Geant4-DNA handles the
One of the important applications of microdosim- transport of electrons, protons, hydrogen nuclei, alpha
etry simulation is quantifying the radiation-induced nuclei (with various charges) and four heavier ions (car-
DNA damage. Here we simply introduce some popular bon, nitrogen, oxygen, and iron ions).
simulation tools for quantifying DNA damages. Event- Those simulation codes have been adopted as investi-
by-event MC codes such as KURBUC [44], PARTRAC gation tool in many microdosimetry applications, such
[45], NOREC [46], and Geant4-DNA [47] represent the as modeling cellular dose, radiation-induced DNA dam-
preferred method of computing distributions of micro- ages of cell, study of radiosensitization of gold nanopar-
dosimetric quantities for quantifying radiation-induced ticle aided radiotherapy, etc.
Stewart et al. [51] calculated the single-event distri- accurate modeling of low energy photon interactions in
butions, site-hit probabilities and the frequency-mean different regions of the model employing Geant4-DNA
specific energy per event of a volume in size to a mam- physics within the cell, and Livermore physics within
malian cell or cell nucleus using PENELOPE. Rollet gold, respectively. DBSCAN is used to directly quantify
et al. [52] have calculated frequency distributions of DNA DSBs after irradiation. They obtained the radio-
deposited energy in a mini TEPC (with sizes equivalent sensitization factors of GNP with different sizes and
to 1 and 2 μm) due to photons using the FLUKA MC different distances from the cell nucleus. Several car-
code. They found a reasonably good agreement between bonatizations of these two conditions were obtained by
the simulated data with those obtained experimentally. using their simulation model.
Using Geant4, Cortés-Giraldoand Carabe [53] calcu-
lated dose-averaged linear energy transfer (LET) maps 14.2.4 Application in Nuclear Medicine
in voxelized geometries irradiated with proton beams. MC method has been widely used in nuclear medicine
They carried out microdosimetry calculations with the [64, 65] with a broad range of applications, such as gen-
aim of deriving reference dose-averaged LET values erating reference S-values in MIRD (Medical Internal
from microdosimetric quantities. With suitable scoring Radiation Dose) models, etc. [64]. MC method was ini-
method corresponding to a step-by-step computation tially used for reference dosimetry based on reference
of LET, their simulation gave the best agreement with models proposed by ICRP or the MIRD committee.
estimations from microdosimetry calculations on dose- The dosimetric parameters, i.e., S-value, were calcu-
averaged LET. lated in some dosimetry software like MIRDOSE [66]
Due to the fast development of Geant4-DNA proj- or OLINDA [67]. In the Oak Ridge National Laboratory,
ect and its open-source oriented characteristics, many Snyder et al. firstly used MC method to assess the frac-
works have been done to quantify DNA damages using tion and electron energy emitted from radionuclides
Geant4-DNA [54–56]. Besides the Geant4-DNA physics, in source tissues, deposited in target tissues [68, 69].
it is worth noting that there are also other physics mod- According to these simulations, with the anatomic
els for track structure simulation, such as the models information provided by MRI or CT the three dimen-
discussed in papers [57–60], which also have wide appli- sional voxelized patient phantom could be obtained, and
cations in the calculation of DNA damages. In 2011, with MC simulation the voxel dosimetry introduced in
Francis et al. [61] simulated the DSB using a clustering MIRD Pamphlet No.17 could be achieved [68]. Using
algorithm DBSCAN (Density-Based Spatial Clustering MC method, the patient dose in radionuclide therapy
of Applications with Noise) based on the energy depo- could be accurately calculated with voxelized phantom
sition information in cell nucleus obtained by Geant4. or mesh-type computational phantom [70].
Since then, DBSCAN quantification of DNA DSB
yield, based on Geant4 radiation transport simulation, 14.2.5 Applications for X-ray Imaging
has become a popular method in modeling radiation- In diagnostic radiology, the use of the MC method to
induced DNA damages [61]. Liu et al. [62] have devel- simulate radiation transport has become the most accu-
oped a computational model for quantifying cell dose rate means of predicting the diagnostic x-ray spectra
and DSB number in a multicellular system by simulating even in complex geometries, owing to incorporation of
the radiation transport in 2D and 3D cell culture. The proper interaction cross section data and more accurate
radiation transport simulation for cells was conducted physics modeling. Ay et al. have used MCNP MC code
using Geant4 package with the Geant4-DNA physics to to study x-ray spectra in diagnostic radiology and mam-
obtain the cellular dose and cellular DSBs yield. Using mography [71]. They focused on the simulation of vari-
the method they developed, it is possible to obtain the ous target and tube filter combinations to investigate the
cellular dose and DNA damage simultaneously. Liu et effect of tube voltage, target material and filter thick-
al. [63] investigated the radiosensitization properties of ness. They stated that, although using MC to derive x-ray
gold nanoparticles (GNPs) using a simple Geant4 cell spectra was time consuming, it could provide detailed
model considering a realistic cell geometry and a clus- information about particles’ interaction with differ-
tering algorithm to characterize the number of DSBs. ent parts of a diagnostic x-ray tube, including target
In their model, a mixed-physics approach was taken for and filter. They recommended using MC for x-ray tube
design and development of new target/filter combina- information for 2D mapping of patient position and
tions to improve image quality in diagnostic radiology dosimetric verification of treatments. Chin et al. have
and mammography [71]. On the other hand, consider- developed such a MC-based tool of EPIDs using the
ing the patient dose in diagnostic radiology, depending BEAMnrc/DOSXYZnrc code [78]. Their model showed
on both the attenuation characteristics of the absorbing good agreement with measured data for an inhomoge-
media and the energy spectrum of the incident x-rays, neous phantom, where the root-mean-square (RMS)
correction factors are needed to convert dose from of the difference was under 2% within the field. This
model-based dose calculations to absorbed dose-to- MC-based tool was capable of handling arbitrary gantry
medium. Pawlowski et al. have presented a method to angles, voxel-by-voxel phantom description and realistic
accurately account for the medium-dependent effect in particle transport throughout the geometry. Cone beam
model-based dose calculations for diagnostic x-ray [72]. CT (CBCT), commonly installed on Linacs for image
They used MC simulation to derive empirical medium- guided radiation therapy, adds additional imaging
dependent correction factors as a function of the amount dose to the prescribed radiation dose given to patients.
of bone that an x-ray beam must penetrate to reach a Although, the AAPM Task Group 75 report makes rec-
voxel. According to their results, their approach resulted ommendations on estimating and reducing imaging
in mean dose errors of less than 3% when tested for their dose from CBCT [79], calculation of the imaging dose
patients, providing a good alternative accurate dose cal- during treatment planning would allow for a better
culation method for kV x-rays. prediction of the total dose to target and organs at risk.
Computed tomography (CT) is a high-dose diagnos- Downes et al. have used MC simulations to calculate
tic technique that accounts for approximately 3–5% of all patient dose for a kilovoltage CBCT unit [80]. Based on
diagnostic examinations. However, it has been reported the MC results, they reported that the 3D dose distribu-
to be responsible for 34–45% of the total collective dose tion from CBCT in patient was complex. They suggested
from all diagnostic procedures in western countries [73, using full 3D MC calculation, instead of employing
74]. Therefore, calculation of patient CT dose is impor- numerical factors, to estimate CBCT exposure, espe-
tant, in order to estimate the patient risk for stochas- cially at regions at the center of the body.
tic effects, mainly carcinogenesis, and to optimize the
scanning protocols. Avilés Lucas et al., have developed 14.3 MONTE CARLO CODES AND
MC computational model of CT to investigate the effect SYSTEMS IN THE PUBLIC DOMAIN
of various physical factors such as bowtie filter as well as Several general-purpose MC codes have been developed
the size, shape and position of a patient phantom on the for radiation transport calculation, which are commonly
surface air kerma length product, the peak surface air used in medical physics, such as, Geant4, MCNP, EGS4/
kerma, the air kerma length product within a phantom EGSnrc, PENELOPE, and FLUKA. In this section we
and the energy imparted [75]. Based on their MC simu- briefly introduce these MC codes.
lations, they highlighted the importance of calculating
doses to organs by taking into account their size and 3.1 Geant4
position within the CT gantry. Geant4 is a freely available software for performing MC
In interventional radiology using fluoroscopically simulations of the interactions of energetic particles in
guided procedures, there have been several investiga- matter [81]. Geant4 is an acronym for “GEometry ANd
tions dealing with estimation of the dose to patient skin Tracking.” The Geant4 was originally developed by
and internal organs [76]. Although various methods CERN for the purpose of simulating the interaction of
including direct and indirect measurements have been high energy particles in large accelerators. The physics
examined, MC simulations have been shown to provide process covers a comprehensive range, including electro-
accurate results for the calculations of the dose to skin magnetic, hadronic and optic processes, a large number
and other sensitive organs. of particles, materials and elements, over a wide energy
In radiotherapy practice, for complex conformal and range starting from a few eV to TeV [81]. A variety of
IMRT, electronic portal imaging devices (EPIDs) are cross sections are used in Geant4, for instance, EPDL97,
well proven as important tools for geometric verification EEDL, and EADL for low energy electromagnetic inter-
[77]. They also provide near-instantaneous important actions. More detailed information about the physics
models used in Geant4 could be referred to their official hadron therapy [89–91]. For modeling the patient dose
report [82]. Since its development Geant4 has contin- in radiotherapy, Geant4 allows the modeling of com-
ued to evolve and expand. The original toolkit imple- plex geometries of treatment equipment and also the
mented methods for handling the fundamental aspects patient anatomy by constructing voxelized phantom
of physical simulation: geometry, materials, particle based on DICOM images [92]. Geant4 also has been
interactions, particle generation, and particle tracking. used in internal dosimetry simulations [93, 94], and
Moreover, it is capable of generation and storing event radiodiagnostic applications [95]. In recent years, many
and tracking data, and detector tracking visualization. simulation applications have been developed for radia-
From these building blocks, the code has expanded tion biology studies using Geant4. An international
widely. Its object-oriented nature allows the end-user to collaborative projects, the so-called Geant4-DNA, has
make use of the existing framework by instantiating use- been launched to build an experimentally validated
ful class systems while maintaining a unique and per- simulation platform for the modeling of DNA damage
sonal application framework. Users can build their own induced by ionizing radiations, with the help of modern
simulation code by using the user’s classes provided in computing tools and techniques [96]. Geant4-DNA, an
Geant4. Since most of the users’ classes are designed in extension to Geant4 developed under the framework of
base classes, users can directly instantiate the classes for Geant4-DNA project, which is an open-source simula-
their usage. There are two kinds of user classes includ- tion toolkit that aims to model the effects of radiation on
ing user initialization classes and user action classes. biological systems at cellular and DNA levels. Geant4-
User initialization classes are used during the initializa- DNA includes physical, chemical and biological models
tion phase (simulation initialization), while user action to simulate cellular and subcellular damage induced by
classes are used during the simulation (invoked actions ionizing radiation. Geant4-DNA introduces a new set
for controlling radiation transport). User initialization of models to increase the resolution of physical model-
should call directly G4RunManager class by invoking ing at lower particle energies. Certain processes handled
SetUserInitialization function, while user classes should vaguely by standard physics list in Geant4, are treated by
be defined in G4VUserActionInitialization class. The a model with increased details. Very low energy electron
three necessary initialization classes and five optional electromagnetic processes have been introduced, which
user action base classes are listed in Table 14.2 with their is quite beneficial to the DNA damage quantification
description. since DNA damages are largely induced by the depos-
Because of its advanced functionality, Geant4 ited energies of low energy electrons. More fundamen-
has become a popular simulation toolkit in various tal development and applications of Geant4-DNA were
domains in the field of medical physics [83]. Geant4 presented in papers [97, 98]. Due to the fast development
is used in radiotherapy simulations, such as external of Geant4-DNA project and its open-source oriented
beam treatment [84–86], brachytherapy [87, 88], and characteristics, it has been used by many investigators to
TABLE 14.2 Base Classes Used for Building the Geant4 Simulation
Class Type User Classes Functionality of Classes
Necessary G4VUserDetectorConstruction Defining the material and geometrical setup of the
initialization classes detector. Several other properties, such as detector
sensitivities and visualization attributes, are also
defined in this class.
G4VUserPhysicsList Defining all the particles, physics process and cut off
parameters.
G4VUserPrimaryGenerationAction Generating the primary vertices and particles.
Optional user action G4UserRunAction For actions at the beginning and end of every run.
base classes G4UserEventAction For actions at the beginning and end of every event.
G4UserStackingAction For customizing access to the track stacks.
G4UserTrackingAction For actions at the creation and completion of every
track.
G4UserSteppingAction For customizing behavior at every step.
Source: From [81].
quantify DNA damages [54, 56, 62, 63, 99–101]. Besides patient dose calculation [113, 114], internal dosimetry
the Geant4-DNA physics, it is worth mentioning that [115, 116], hadron therapy [117, 118]. MCNP is also been
there are also other physics models for track structure used in simulating the microdosimetry of cell, and DNA
simulation, such as the models discussed in [57–60], DSB in order to quantify the relative biological effective-
which also have wide applications in the calculation of ness (RBE) for photons, neutrons, and light ions [119].
DNA damages.
There are also some MC simulation packages that are 14.3.3 EGS4/EGSnrc
based on Geant4, such as TOPAS (TOlkit for PArticle The EGS (Electron Gamma Shower) MC code is a gen-
Simulation) [102] and GATE [95]. The radiation trans- eral purpose charged-particle transport simulation sys-
port simulations in TOPAS and GATE all depend on tem which was originally developed at Stanford Linear
Geant4, but they have their own methods to implement Accelerator Center (SLAC) to simulate higher energy
simulations. TOPAS was originally designed to cre- electromagnetic cascades, and it was used in large accel-
ate a specialized user-friendly tool for proton therapy, erator centers for many years (SLAC, CERN, etc.) [109].
and it has been quite widely used in medical physics ESG was initially used for high energy electron simu-
simulations [103, 104]. TOPAS also has extended its lation, later the low-energy electron transport in EGS
functionality to conduct radiobiology simulation with was improved by inclusion of user-defined restrictions
TOPAS-nBio toolkit [105]. GATE also has wide applica- in step sizes. This resulted in a more accurate EGS4 ver-
tions in radiotherapy [86, 106]. sion which added electron transport down to 1 keV and
Rayleigh scattering to original photon physics in the
14.3.2 MCNP code [109]. EGS system evolved into two independent
The MCNP (initially stood for MC Neutron Photon, systems, EGSnrc maintained by the National Research
now stands for Monte Carlo N-Particle transport) code Council of Canada (NRCC), and EGS5 maintained by
system developed by Group X-6 is the workhorse at the the Higher Energy Accelerator Research Organization
Los Alamos National Laboratory for neutron, photon, (KEK, Japan). EGSnrc is a general purpose MC pack-
and coupled neutron-photon calculations using the MC age for simulation of coupled electron-photon transport
method [107, 108]. MCNP’s history could be dating back in matter, with energy range from 1 keV to 10 GeV [27,
to the formulation and first uses of the MC method at 120]. EGS primarily uses cross-section libraries such as
Los Alamos during the 1940s by Fermi, Metropolis, EPDL97 [121] and XCOM [122] for photon and electron
Richtmyer, Ulam, and von Neuman [107]. The MCNP radiation transport simulation. A very detailed history
is, by far, one of the most widely used MC codes that has about EGS system development was given in a report by
been available to the radiation physics community. Most SLAC [123]. After the release of EGS, it has been used
of its users were initially connected to military projects in a wide variety of applications, particularly in medi-
and nuclear power engineering and industry [109]. The cal physics, radiation measurement studies, and indus-
cross-section data for photons and neutrons are taken trial development [123]. In medical physics, EGSnrc
from several sources, some of them dating back to 1970s, is one of the most widely used packages in radiation
expanded with the evaluated data from ENDF, and fur- therapy. Particularly, several MC packages were devel-
ther completed with the EPDL for high energies. The oped based on ESGnrc, such as BEAMnrc for modeling
energy ranges of simulated particles are from 10 −5 eV to radiotherapy sources [124], DOSXYZnrc for calculating
20 MeV for neutrons with data up to 150 MeV for some dose distributions in a rectilinear voxel phantom [125].
nuclides, 1 keV to 1 GeV for electrons, and 1 keV to 100 BEAMnrc can produce a phase-space output of the
GeV for photons. In the recent version, MCNP6 includes beam at any specified plain in geometry, and the output
considerable developments in low-energy electron and data calculated by BEAMnrc can be used as input file for
photon transport, aiming at simulations down to ener- DOSXYZnrc. EGSnrc also has been used in brachyther-
gies in the few eV range [109]. MCNP has been used apy dose calculations for clinical applications [126, 127].
successfully to solve many problems in the field of medi-
cal physics, and an extensive review is given by Solberg 14.3.4 PENELOPE
et al. [110]. In radiotherapy, MCNP has been used in PENELOPE was originally developed for low energy
many applications, such as in brachytherapy [111, 112], radiation, and it is maintained at the University of
Barcelona (Spain), and its name is an acronym for most challenging aspects of any MC model. Moreover,
“PENetration and Energy Loss of Positrons and having chosen the proper physics model and hav-
Electrons.” It simulates electron-photon showers in the ing finalized the geometry simulation of the problem,
energy range from 50 eV to 1 GeV [109]. PENELOPE uses appropriate cross-section tables must be selected and
photon cross-section data from the EPDL97 [128]. In assigned to each material in the simulation geometry.
PENELOPE, photon transport is simulated by means of Generally, if any of these steps were not performed in
the standard, detailed simulation scheme. Electron and detail with reasonable accuracy, the MC method might
positron histories are generated on the basis of a mixed not be considered as the golden method for dose calcu-
procedure, which combines detailed simulation of hard lation. Furthermore, in clinical applications, for radia-
events with condensed simulation of soft interactions tion transport problems with complex geometry and
[129]. The main application of PENELOPE includes varying material properties, MC-based dose quantifica-
radiotherapy and nuclear medicine, radiation dosim- tion requires long computational times. In other words,
etry, electron microscopy (SEM, electron-probe micro- detailed simulation of the problem at hand may signifi-
analysis), detector response, x-ray generators, etc. [130]. cantly decrease the efficiency of a calculation. Therefore,
In radiotherapy simulations, PENELOPE has been used there is a compromise between accuracy and efficiency
to conduct Linac simulations [131, 132]. PENELOPE of any MC simulation. For years, MC methods have been
also could be used in simulating the positron emission considered to be a rather sluggish technique compared
tomography (PET) imaging [133]. The ability to trans- with analytical or deterministic methods. Nonetheless,
port electrons and positions to energies as low as 100 eV more advanced computing power, aiming at decreasing
makes PENELOPE a suitable MC simulation system for computational time and simulating realistic geometries,
microdosimetry [51]. have been recently accessible. Moreover, as stated ear-
lier, developing the general-purpose graphics processing
14.3.5 FLUKA unit (GPU) has been found to be a promising solution by
The FLUKA code is a general purpose MC code for tremendously increasing the speed of MC simulations.
simulation of the interactions and transport of hadrons, On the other hand, all-purpose MC codes such as
heavy ions, and electromagnetic particle from few keV MCNP, Geant4, EGS4, and FLUKA usually require
to cosmic ray energies in whichever material [134]. It either extensive general programming knowledge (for
was developed originally at CERN in the 1960s, its name Geant4) or detailed understanding of their specific
is an acronym for “FLUktuierende KAskade” [109], and programming rules (for MCNP, EGS4, etc.), and are
it is used extensively at CERN for all beam-machine not suited well to the inexpert user for medical phys-
interactions, radioprotection calculations and facility ics applications. Whereas, MC codes can be much easier
design of forthcoming projects [135]. Beside its hadron to use, when packaged with a user-friendly simulation
physics, FLUKA can also handle electromagnetic inter- environment. In this regard, graphical user friendly MC
action simulations with a wide energy range: the pro- packages such as TOPAS, and GATE can be a valuable
gram can transport photons and electrons from 1 PeV tool on making simulation available to the inexperi-
down to 1 keV [134]. FLUKA uses EPDL97 library for enced programmer, thus facilitating MC use in radia-
photon electron shower [136]. FKUKA is widely used for tion therapy research.
studies related to both basic research and applications in
radiation protection and dosimetry, radiotherapy [137], 14.5 SUMMARY
radiobiology [138], cosmic ray calculations [139]. MC simulations are an exceptionally powerful tool not
only in external beam radiotherapy but also for model-
14.4 LIMITATIONS OF MONTE ing brachytherapy and diagnostic imaging systems. In
CARLO SIMULATION this chapter, we discussed and highlighted the capa-
Although MC method can model complex systems by bilities of Geant4, MCNP, EGS4/EGSnrc, PENELOPE,
providing flexible and exceptionally powerful tech- and FLUKA code for modeling and analyzing radiation
niques, it is important to consider the limitations that transport for clinical applications. Generally, majority
constrain its use for simulation of radiation transport. of the MC studies in medical physics have been con-
Simulation accuracy of the physics model is one of the ducted using the MCNP and EGS/EGSnrc-based codes
(including the user interface BEAM). However, other 7. E. Alerstam, T. Svensson, S. Andersson-Engels,
codes such as PENELOPE and Geant4 are experienc- “Parallel computing with graphics processing units
for high-speed Monte Carlo simulation of photon
ing an increasing use. Main limitations of MC meth-
migration,” Journal of Biomedical Optics 13(6), 060504
ods are the lack of efficiency of calculation and storage (2008).
of large amounts of data. Meanwhile, considering 8. E. Alerstam, W. C. Yip Lo, T. D. Han, J. Rose, S.
more advanced computing power and also generation Andersson-Engels, L. Lilge, “Next-generation accelera-
of GPU-based desktop computers that are less expen- tion and code optimization for light transport in turbid
sive compared to cluster of CPUs, these limitations are media using GPUs,” Biomedical Optics Express 1(2), 658
(2010).
now less of an issue. For implementation of the new 9. A. Badaland A. Badano, “Accelerating Monte Carlo
technology in radiotherapy and diagnostic imaging, simulations of photon transport in a voxelized geome-
MC-based techniques can play a critical role as speed try using a massively parallel graphics processing unit,”
and accuracy will be of the essence. In the near future, Medical Physics 36(11), 4878–4880 (2009).
MC methods may replace conventional analytical dose 10. Q. Fang and D. A. Boas, “Monte Carlo simulation of
photon migration in 3D turbid media accelerated by
calculation techniques for patient treatment planning
graphics processing units,” Optics Express 17(22), 20178
and also may be used to face new challenges in patient (2009).
treatment, such as LET-based dose painting in proton 11. X. Gu, U. Jelen, J. Li, X. Jia, S. B. Jiang, “A GPU-based
therapy, designing complex beam delivery techniques finite-size pencil beam algorithm with 3D-density cor-
for modulated electron therapy, and design and dosim- rection for radiotherapy dose calculation,” Physics in
Medicine and Biology 56(11), 3337–3350 (2011).
etry in MR–Linac systems. These advances in cancer
12. X. Jia, X. Gu, J. Sempau, D. Choi, A. Majumdar, S. B.
treatment require implementation of commercial Jiang, “Development of a GPU-based Monte Carlo dose
MC-based TPS as they provide significant improve- calculation code for coupled electron-photon trans-
ment in dose calculation accuracy over the conven- port,” Physics in Medicine and Biology 55(11), 3077–3086
tional algorithms. (2010).
For the studies on detector design and relative dosim- 13. C. Men, X. Jia, S. B. Jiang, “GPU-based ultra-fast direct
aperture optimization for online adaptive radiation
eters, for their dose response in advanced complex dose therapy,” Physics in Medicine and Biology 55(15), 4309–
delivery schemes, MC simulations will continue to 4319 (2010).
play an essential role in all these aspects of dosimetry. 14. R. Nath, L. L. Anderson, G. Luxton, K. A. Weaver, J.
According to the previous publications, the MC method F. Williamson, A. S. Meigooni, “Dosimetry of intersti-
is the only approach that can accurately model complex tial brachytherapy sources: Recommendations of the
AAPM Radiation Therapy Committee Task Group No.
dose delivery systems. 43,” Medical Physics 22(2), 209–234 (1995).
15. G. Landry, et al., “Sensitivity of low energy brachy-
REFERENCES therapy Monte Carlo dose calculations to uncertainties
in human tissue composition,” Medical Physics 37(10),
1. L. Pei and X. Zhang, Monte Carlo Method and its 5188–5198 (2010).
Application in Particle Transport (Science Press, Beijing, 16. S. Hissoiny, B. Ozell, P. Després, J. F. Carrier,
1980). “Validation of GPUMCD for low-energy brachyther-
2. W. R. Gilks, Markov Chain Monte Carlo in Practice apy seed dosimetry,” Medical Physics 38(7), 4101–
(Chapman and Hall/CRC, 1995). 4107 (2011).
3. A. F. Bielajew, “History of Monte Carlo,” in Monte 17. S. Hissoiny, A. J. E. Raaijmakers, B. Ozell, P. Després,
Carlo Techniques in Radiation Therapy (CRC/Taylor & B. W. Raaymakers, “Fast dose calculation in magnetic
Francis, Boca Raton, 2013), pp. 3–15. fields with GPUMCD,” Physics in Medicine and Biology
4. M. Aigner, G. M. Ziegler, M. Aigner, G. M. Ziegler, 56(16), 5119–5129 (2011).
“Buffon’s needle problem,” in Proofs from THE BOOK 18. X. Jia, H. Yan, X. Gu, S. B. Jiang, “Fast Monte Carlo
(Springer, Berlin Heidelberg, 2014), pp. 175–178. simulation for patient-specific CT/CBCT imaging dose
5. X. Shuyan, Application of Monte Carlo Method in calculation,” Physics in Medicine and Biology 57(3), 577–
Experimental Nuclear Physics (Atomic. Energy Press, 590 (2012).
Beijing, 2006). 19. I. Kawrakowand M. Fippel, “VMC++, a fast MC algo-
6. C. M. Ma, et al., “A Monte Carlo dose calculation tool for rithm for radiation treatment planning,” in The Use
radiotherapy treatment planning,” Physics in Medicine of Computers in Radiation Therapy (Springer, Berlin
and Biology 47(10), 1671–1689 (2002). Heidelberg, 2000), pp. 126–128.
20. J. E. Cygler, et al., “Clinical use of a commercial Monte 34. E. Mainegra-Hing, N. Reynaert, I. Kawrakow, “Novel
Carlo treatment planning system for electron beams,” approach for the Monte Carlo calculation of free-air
Physics in Medicine and Biology 50(5), 1029–1034 (2005). chamber correction factors,” Medical Physics 35(8),
21. S. L. Lawrence, N. H. M. van Lieshout, P. M. Charland, 3650–3660 (2008).
“Assessment of Eclipse electron Monte Carlo output 35. K. Zink and J. Wulff, “On the wall perturbation cor-
prediction for various topologies,” Journal of Applied rection for a parallel-plate NACP-02 chamber in clini-
Clinical Medical Physics 16(3), 99–106 (2015). cal electron beams,” Medical Physics 38(2), 1045–1054
22. J. E. Snyder, D. E. Hyer, R. T. Flynn, A. Boczkowski, D. (2011).
Wang, “The commissioning and validation of Monaco 36. G. Cranmer-Sargison, S. Weston, J. A. Evans, N. P.
treatment planning system on an ElektaVersaHD linear Sidhu, D. I. Thwaites, “Implementing a newly proposed
accelerator,” Journal of Applied Clinical Medical Physics Monte Carlo based small field dosimetry formalism for
20(1), 184–193 (2019). a comprehensive set of diode detectors,” Medical Physics
23. J. Y. Huang, D. Dunkerley, J. B. Smilowitz, “Evaluation 38(12), 6592–6602 (2011).
of a commercial Monte Carlo dose calculation algo- 37. L. L. W. Wang and D. W. O. Rogers, “Monte Carlo study
rithm for electron treatment planning,” Journal of of Si diode response in electron beams,” Medical Physics
Applied Clinical Medical Physics 20(6), 184–193 (2019). 34(5), 1734–1742 (2007).
24. A. N. Schreuder, et al., “Validation of the RayStation 38. F. Crop, et al., “The influence of small field sizes,
Monte Carlo dose calculation algorithm using a realis- penumbra, spot size and measurement depth on
tic lung phantom,” Journal of Applied Clinical Medical perturbation factors for microionization chambers,”
Physics 20(12), 127–137 (2019). Physics in Medicine and Biology 54(9), 2951–2969
25. “A protocol for the determination of absorbed dose (2009).
from high energy photon and electron beams,” Medical 39. P. Francescon, S. Cora, C. Cavedon, “Total scatter fac-
Physics 10(6), 741–771 (1983). tors of small beams: A multidetector and Monte Carlo
26. J. E. Rodgers, A. Niroomand-Rad, M. Lundsten, study,” Medical Physics 35(2), 504–513 (2008).
“Comment on clinical implementation of the AAPM’s 40. P. Francescon, S. Cora, C. Cavedon, P. Scalchi,
Task Group-51 protocol for calibration of high-energy “Application of a Monte Carlo-based method for total
photon and electron beams,” Medical Physics 27(8), scatter factors of small beams to new solid state micro-
1995–1996 (2000). detectors,” Journal of Applied Clinical Medical Physics
27. I. Kawrakow, “Accurate condensed history Monte Carlo 10(1), 147–152 (2009).
simulation of electron transport. II. Application to ion 41. S. Chauvie, et al., “Geant4 physics processes for micro-
chamber response simulations,” Medical Physics 27(3), dosimetry simulation: Design foundation and imple-
499–513 (2000). mentation of the first set of models,” IEEE Transactions
28. E. Poon, J. Seuntjens, F. Verhaegen, “Consistency test of on Nuclear Science 54(6), 2619–2628 (2007).
the electron transport algorithm in the GEANT4 Monte 42. A. Kellerer, “Fundamentals of microdosimetry,” in The
Carlo code,” Physics in Medicine & Biology 50(4), 681 Dosimetry of Ionizing Radiation Volume I (Academic
(2005). Press, London, 1985), pp. 78–162.
29. J. Sempau, E. Acosta, J. Baró, J. M. Fernández-Varea, F. 43. T. T. Böhlen, M. Dosanjh, A. Ferrari, I. Gudowska,
Salvat, “An algorithm for Monte Carlo simulation of cou- “Simulations of microdosimetric quantities with the
pled electron-photon transport,” Nuclear Instruments and Monte Carlo code FLUKA for carbon ions at therapeu-
Methods in Physics Research Section B: Beam Interactions tic energies,” International Journal of Radiation Biology
with Materials and Atoms 132(3), 377–390 (1997). 88(1–2), 176–182 (2012).
30. B. R. Muir, M. R. McEwen, D. W. O. Rogers, “Measured 44. T. Liamsuwan, D. Emfietzoglou, S. Uehara, H.
and Monte Carlo calculated kQ factors: Accuracy and Nikjoo, “Microdosimetry of low-energy electrons,”
comparison,” Medical Physics 38(8), 4600–4609 (2011). International Journal of Radiation Biology 88(12),
31. L. A. Buckley and D. W. O. Rogers, “Wall correction fac- 899–907 (2012).
tors, Pwall, for thimble ionization chambers,” Medical 45. W. Friedland, M. Dingfelder, P. Kundrát, P. Jacob,
Physics 33(2), 455–464 (2006). “Track structures, DNA targets and radiation effects
32. H. Bouchard, J. Seuntjens, I. Kawrakow, “A Monte Carlo in the biophysical Monte Carlo simulation code
method to evaluate the impact of positioning errors on PARTRAC,” Mutation Research – Fundamental and
detector response and quality correction factors in non- Molecular Mechanisms of Mutagenesis 711(1–2), 28–40
standard beams,” Physics in Medicine and Biology 56(8), (2011).
2617–2634 (2011). 46. M. Dingfelder, R. H. Ritchie, J. E. Turner, W. Friedland,
33. D. T. Burns, “A new approach to the determination of H. G. Paretzke, R. N. Hamm, “Comparisons of cal-
air kerma using primary-standard cavity ionization culations with PARTRAC and NOREC: Transport of
chambers,” Physics in Medicine and Biology 51 (4), 929– electrons in liquid water,” Radiation Research 169 (5),
942 (2006). 584–594 (2008).
47. S. Incerti, et al., “The Geant4-DNA project,” 60. D. Emfietzoglou, G. Papamichael, H. Nikjoo, “Monte
International Journal of Modeling, Simulation, and Carlo electron track structure calculations in liquid
Scientific Computing 01(02), 157–178 (2010). water using a new model dielectric response function,”
48. I. El Naqa, P. Pater, J. Seuntjens, “Monte Carlo role in Radiation Research 188(3), 355–368 (2017).
radiobiological modelling of radiotherapy outcomes,” 61. Z. Francis, C. Villagrasa, I. Clairand, “Simulation
Physics in Medicine and Biology 57(11), R75–R97 of DNA damage clustering after proton irradiation
(2012). using an adapted DBSCAN algorithm,” Computer
49. W. Friedland, et al., “Comprehensive track-structure Methods and Programs in Biomedicine 101(3),
based evaluation of DNA damage by light ions from 265–270 (2011).
radiotherapy-relevant energies down to stopping,” 62. R. Liu, T. Zhao, M. H. Swat, F. J. Reynoso, K. A. Higley,
Scientific Reports 7, 45161 (2017). “Development of computational model for cell dose
50. W. Friedland, E. Schmitt, P. Kundrát, G. Baiocco, A. and DNA damage quantification of multicellular sys-
Ottolenghi, “Track-structure simulations of energy tem,” International Journal of Radiation Biology 95(11),
deposition patterns to mitochondria and damage to 1484–1497 (2019).
their DNA,” International Journal of Radiation Biology 63. R. Liu, T. Zhao, X. Zhao, F. J. Reynoso, “Modeling
95(1), 3–11 (2018). gold nanoparticle radiosensitization using a cluster-
51. R. D. Stewart, W. E. Wilson, J. C. McDonald, D. J. Strom, ing algorithm to quantify DNA double-strand breaks
“Microdosimetric properties of ionizing electrons in with mixed-physics Monte Carlo simulation,” Medical
water: A test of the PENELOPE code system,” Physics in Physics 46(11), 5314–5325 (2019).
Medicine and Biology 47(1), 79–88 (2002). 64. D. Villoing, S. Marcatili, M. P. Garcia, M. Bardiès,
52. S. Rollet, et al., “Monte Carlo simulation of mini TEPC “Internal dosimetry with the Monte Carlo code GATE:
microdosimetric spectra: Influence of low energy elec- Validation using the ICRP/ICRU female reference com-
trons,” Radiation Measurements 45(10), 1330–1333 putational model,” Physics in Medicine and Biology
(2010). 62(5), 1885–1904 (2017).
53. M. A. Cortés-Giraldoand A. Carabe, “A critical study 65. Habib Zaidi, George Sgouros, Therapeutic Applications
of different Monte Carlo scoring methods of dose aver- of Monte Carlo Calculations in Nuclear Medicine (CRC/
age linear-energy-transfer maps calculated in voxel- Taylor & Francis, Boca Raton, FL, 2003).
ized geometries irradiated with clinical proton beams,” 66. M. G. Stabin, “MIRDOSE: Personal computer soft-
Physics in Medicine and Biology 60(7), 2645–2669 ware for internal dose assessment in nuclear medicine,”
(2015). Journal of Nuclear Medicine 37(3), 538–546 (1996).
54. R. M. Abolfath, D. J. Carlson, Z. J. Chen, R. Nath, “A 67. M. G. Stabin, R. B. Sparks, E. Crowe, “OLINDA/EXM:
molecular dynamics simulation of DNA damage induc- The second-generation personal computer software for
tion by ionizing radiation,” Physics in Medicine and internal dose assessment in nuclear medicine,” Journal
Biology 58(20), 7143–57 (2013). of Nuclear Medicine 46(6), 1023–1027 (2005).
55. M. Dos Santos, C. Villagrasa, I. Clairand, S. Incerti, 68. A. Maria, “Monte Carlo simulation in radionuclide
“Influence of the DNA density on the number of clus- therapy dosimetry,” Biomedical Journal of Scientific &
tered damages created by protons of different energies,” Technical Research 15(1), 11102–11107 (2019).
Nuclear Instruments and Methods in Physics Research 69. W. S. Snyder, H. L. Fisher, M. R. Ford, G. G. Warner,
Section B: Beam Interactions with Materials and Atoms “Estimates of absorbed fractions for monoenergetic
298, 47–54 (2013). photon sources uniformly distributed in various organs
56. N. Lampe, et al., “Mechanistic DNA damage simula- of a heterogeneous phantom,” Journal of Nuclear
tions in Geant4-DNA part 1: A parameter study in Medicine 3(Suppl.), 7–52 (1969).
a simplified geometry,” Physica Medica 48, 135–145 70. L. M. Carter, et al., “PARaDIM – A PHITS-based Monte
(2018). Carlo tool for internal dosimetry with tetrahedral mesh
57. W. Friedland, M. Dingfelder, P. Kundrát, P. Jacob, computational phantoms,” Journal of Nuclear Medicine
“Track structures, DNA targets and radiation effects 60(12), 1802–1811 (2019).
in the biophysical Monte Carlo simulation code 71. M. R. Ay, M. Shahriari, S. Sarkar, M. Adib, H. Zaidi,
PARTRAC,” Mutation Research 711(1–2), 28–40 “Monte Carlo simulation of x-ray spectra in diagnos-
(2011). tic radiology and mammography using MCNP4C,”
58. H. Nikjoo, S. Uehara, D. Emfietzoglou, F. A. Cucinotta, Physics in Medicine and Biology 49(21), 4897–4917
“Track-structure codes in radiation research,” Radiation (2004).
Measurements 41(9–10), 1052–1074 (2006). 72. J. M. Pawlowski andG. X. Ding, “A new approach to
59. H. Nikjoo, D. Emfietzoglou, T. Liamsuwan, R. Taleei, account for the medium-dependent effect in model-
D. Liljequist, S. Uehara, “Radiation track, DNA damage based dose calculations for kilovoltage x-rays,”
and response –a review,” Reports on Progress in Physics Physics in Medicine and Biology 56(13), 3919–3934
79(11), 116601 (2016). (2011).
73. J. T. M. Jansen, P. C. Shrimpton, J. Holroyd, S. Edyvean, 86. L. Grevillot, T. Frisson, D. Maneval, N. Zahra, J. N.
“Selection of bone dosimetry models for application in Badel, D. Sarrut, “Simulation of a 6 MV Elekta Precise
Monte Carlo simulations to provide CT scanner-spe- Linac photon beam using GATE/GEANT4,” Physics in
cific organ dose coefficients,” Physics in Medicine and Medicine and Biology 56(4), 903–918 (2011).
Biology 63(12), 125015 (2018). 87. D. Granero, J. Vijande, F. Ballester, M. J. Rivard,
74. P. C. Shrimpton and S. Edyvean, “CT scanner dosim- “Dosimetry revisited for the HDR 192Ir brachyther-
etry,” British Institute of Radiology 71(841), 1–3 apy source model mHDR-v2,” Medical Physics 38(1),
(1998). 487–494 (2011).
75. P. Avilés Lucas, D. R. Dance, I. A. Castellano, E. Vañó, 88. H. Afsharpour, et al., “ALGEBRA: ALgorithm for
“Monte Carlo simulations in CT for the study of the the heterogeneous dosimetry based on GEANT4 for
surface air kerma and energy imparted to phantoms BRAchytherapy,” Physics in Medicine and Biology
of varying size and position,” Physics in Medicine and 57(11), 3273–3280 (2012).
Biology 49(8), 1439–1454 (2004). 89. G. A. P. Cirrone, et al., “Hadrontherapy: A Geant4-
76. P. Alaei, B. J. Gerbi, R. A. Geise, “Lung dose calculations based tool for proton/ion-therapy studies,” Progress
at kilovoltage x-ray energies using a model-based treat- in Nuclear Science and Technology 2(0), 207–212
ment planning system,” Medical Physics 28(2), 194–198 (2011).
(2001). 90. M. De Napoli, et al., “Carbon fragmentation measure-
77. M. G. Herman, et al., “Clinical use of electronic por- ments and validation of the Geant4 nuclear reaction
tal imaging: Report of AAPM Radiation Therapy models for hadrontherapy,” Physics in Medicine and
Committee Task Group 58,” Medical Physics 28(5), 712– Biology 57(22), 7651–7671 (2012).
737 (2001). 91. G. A. P. Cirrone, et al., “The GEANT4 toolkit capability
78. P. W. Chin, E. Spezi, D. G. Lewis, “Monte Carlo simula- in the hadron therapy field: Simulation of a transport
tion of portal dosimetry on a rectilinear voxel geometry: beam line,” Nuclear Physics B – Proceedings Supplements
A variable gantry angle solution,” Physics in Medicine 150 (1–3), 54–57 (2006).
and Biology 48(16), (2003). 92. V. Giacometti, S. Guatelli, M. Bazalova-Carter, A. B.
79. M. J. Murphy, et al., “The management of imaging Rosenfeld, R. W. Schulte, “Development of a high res-
dose during image-guided radiotherapy: Report of the olution voxelised head phantom for medical physics
AAPM Task Group 75,” Medical Physics 34(10), 4041– applications,” Physica Medica 33, 182–188 (2017).
4063 (2007). 93. E. Amato, A. Italiano, F. Minutoli, S. Baldari, “Use of
80. P. Downes, R. Jarvis, E. Radu, I. Kawrakow, E. Spezi, the GEANT4 Monte Carlo to determine three-dimen-
“Monte Carlo simulation and patient dosimetry for a sional dose factors for radionuclide dosimetry,” Nuclear
kilovoltage cone-beam CT unit,” Medical Physics 36(9), Instruments and Methods in Physics Research Section A:
4156–4167 (2009). Accelerators, Spectrometers, Detectors and Associated
81. S. Agostinelli, et al., “GEANT4 –a simulation toolkit,” Equipment 708, 15–18 (2013).
Nuclear Instruments and Methods in Physics Research 94. S. D. Kost, Y. K. Dewaraja, R. G. Abramson, M. G.
Section A: Accelerators, Spectrometers, Detectors and Stabin, “VIDA: A voxel-based dosimetry method for
Associated Equipment 506(3), 250–303 (2003). targeted radionuclide therapy using Geant 4,” Cancer
82. Geant4Collaboration, Physics Reference Manual, Release Biotherapy and Radiopharmaceuticals 30(1), 16–26
10.5 (2019). [Online]. Available from: http://geant4- (2015).
userdoc.web.cern.ch/geant4-userdoc/UsersGuides/ 95. S. Jan, et al., “GATE: A simulation toolkit for PET
PhysicsReferenceManual/fo/PhysicsReferenceManual. and SPECT,” Physics in Medicine and Biology 49(19),
pdf (Accessed: 24June2019). 4543–4561 (2004).
83. L. Aichambault, et al., “Overview of Geant4 96. S. Incerti, et al., “The Geant4-DNA project,”
applications in medical physics,” IEEE Nuclear International Journal of Modeling, Simulation, and
Science Symposium Conference Record 3, 1743–1745 Scientific Computing 1(2), 157–178 (2010).
(2003). 97. M. A. Bernal, et al., “Track structure modeling in liq-
84. B. A. Faddegon, I. Kawrakow, Y. Kubyshin, J. Perl, J. uid water: A review of the Geant4-DNA very low energy
Sempau, L. Urban, “The accuracy of EGSnrc, Geant4 extension of the Geant4 Monte Carlo simulation tool-
and PENELOPE Monte Carlo systems for the simula- kit,” Physica Medica 31(8), 861–874 (2015).
tion of electron scatter in external beam radiotherapy,” 98. S. Incerti, M. Douglass, S. Penfold, S. Guatelli, E. Bezak,
Physics in Medicine and Biology 54(20), 6151–6163 “Review of Geant4-DNA applications for micro and
(2009). nanoscale simulations,” Physica Medica 32(10), 1187–
85. E. Poon and F. Verhaegen, “Accuracy of the photon 1200 (2016).
and electron physics in GEANT4 for radiotherapy 99. M. Dos Santos, C. Villagrasa, I. Clairand, S. Incerti,
applications,” Medical Physics 32(6), 1696–1711 “Influence of the DNA density on the number of clus-
(2005). tered damages created by protons of different energies,”
Nuclear Instruments and Methods in Physics Research 114. H. Yoriyaz. et al., “Monte Carlo MCNP-4B – based
Section B: Beam Interactions with Materials and Atoms absorbed dose distribution estimates for patient-
298, 47–54 (2013). specific,” Journal of Nuclear Medicine 42(4), 662–669
100. Z. Francis, C. Villagrasa, I. Clairand, “Simulation of (2001).
DNA damage clustering after proton irradiation using 115. A. Bitar, et al., “A voxel-based mouse for internal dose
an adapted DBSCAN algorithm,” Computer Methods calculations using Monte Carlo simulations (MCNP),”
and Programs in Biomedicine 101(3), 265–270 (2011). Physics in Medicine and Biology 52(4), 1013–1025
101. N. Lampe, et al., “Mechanistic DNA damage simu- (2007).
lations in Geant4-DNA Part 2: Electron and pro- 116. A. R. Prideaux, et al., “Three-dimensional radiobiologic
ton damage in a bacterial cell,” Physica Medica 48, dosimetry: Application of radiobiologic modeling to
146–155 (2018). patient-specific 3-dimensional imaging-based internal
102. J. Perl, J. Shin, J. Schümann, B. Faddegon, H. Paganetti, dosimetry,” Journal of Nuclear Medicine 48(6), 1008–
“TOPAS: An innovative proton Monte Carlo platform 1016 (2007).
for research and clinical applications,” Medical Physics 117. S. D. Randeniya, P. J. Taddei, W. D. Newhauser, P. Yepes,
39(11), 6818–6837 (2012). “Intercomparision of Monte Carlo radiation transport
103. M. Testa, et al., “Experimental validation of the TOPAS codes MCNPX, GEANT4, and FLUKA for simulating
Monte Carlo system for passive scattering proton ther- proton radiotherapy of the eye,” Nuclear Technology
apy,” Medical Physics 40(12), 1–16 (2013). 168(3), 810–814 (2009).
104. J. Schümann, H. Paganetti, J. Shin, B. Faddegon, 118. J. Seco and F. Verhaegen, Monte Carlo Techniques in
J. Perl, “Efficient voxel navigation for proton therapy Radiation Therapy (CRC/Taylor & Francis, Boca Raton,
dose calculation in TOPAS and Geant4,” Physics in 2013).
Medicine and Biology 57(11), 3281–3293 (2012). 119. R. D. Stewart, et al., “Rapid MCNP simulation of DNA
105. A. McNamara, et al., “Validation of the radiobiol- double strand break (DSB) relative biological effectiveness
ogy toolkit TOPAS-nBio in simple DNA geometries,” (RBE) for photons, neutrons, and light ions,” Physics in
Physica Medica 33, 207–215 (2017). Medicine and Biology 60(21), 8249–8274 (2015).
106. E. Rault, S. Staelens, R. Van Holen, J. De Beenhouwer, S. 120. I. Kawrakow, “Accurate condensed history Monte
Vandenberghe, “Fast simulation of yttrium-90 brems- Carlo simulation of electron transport. I. EGSnrc, the
strahlung photons with GATE,” Medical Physics 37(6), new EGS4 version,” Medical Physics 27(3), 485–498
2943–2950 (2010). (2000).
107. R. A. Forster and T. N. K. Godfrey, “MCNP – a general 121. D. E. Cullen, J. H. Hubbell, L. Kissel, “EPDL97: The
Monte Carlo code for neutron and photon transport,” in Evaluated Photon Data Library,” UCRL-504006, 1–35
Monte-Carlo Methods and Applications in Neutronics, (1997).
Photonics and Statistical Physics (Springer-Verlag, 122. M. J. Berger, J. H. Hubbell, S. M. Seltzer, J. Chang, J. S.
2006), pp. 33–55. Coursey, R. Sukumar, D. S. Zucker, K. Olsen, XCOM:
108. J. F. Briesmeister, “MCNPTM–A General Monte Carlo Photon Cross Section Database (NIST Standard Reference
N-Particle Transport Code: Manual,” December Database 8 (XGAM)) (Gaithersburg, MD, 2010).
(2000). 123. H. Hirayama, et al., The EGS5 Code System (2005).
109. P. Andreo, D. Burns, A. Nahum, J. Seuntjens, F. H. 124. D. W. O. Rogers, B. Walters, I. Kawrakow, BEAMnrc
Attix, The Monte Carlo Simulation of the Transport Users Manual (2009).
of Radiation Through Matter (WILEY-VCH, 2017), 125. D. W. O. Rogers, B. Walters, I. Kawrakow, DOSXYZnrc
pp. 349–396. User Manual (2005).
110. T. D. Solberg, et al., “A review of radiation dosimetry 126. R. E. P. Taylor, G. Yegin, D. W. O. Rogers,
applications using the MCNP Monte Carlo code,” “Benchmarking BrachyDose: Voxel based EGSnrc
Radiochimica Acta 89(4–5), 337–355 (2001). Monte Carlo calculations of TG-43 dosimetry param-
111. T. D. Bohm, P. M. DeLuca, L. A. DeWerd, eters,” Medical Physics 34(2), 445–457 (2007).
“Brachytherapy dosimetry of125I and103Pd sources 127. R. E. P. Taylor and D. W. O. Rogers, “EGSnrc Monte
using an updated cross section library for the MCNP Carlo calculated dosimetry parameters for 192Ir and
Monte Carlo transport code,” Medical Physics 30(4), 169Yb brachytherapy sources,” Medical Physics 35(11),
701–711 (2003). 4933–4944 (2008).
112. R. Wang and X. A. Li, “Monte Carlo dose calculations of 128. S. J. Ye, I. A. Brezovich, P. Pareek, S. A. Naqvi,
beta-emitting sources for intravascular brachytherapy: “Benchmark of PENELOPE code for low-energy pho-
A comparison between EGS4, EGSnrc, and MCNP,” ton transport: Dose comparison with MCNP4 and
Medical Physics 28(2), 134–141 (2001). EGS4,” Physics in Medicine and Biology 49(3), 387–397
113. A. Mostaar, M. Allahverdi, M. Shahriari, “Application (2004).
of MCNP4C Monte Carlo code in radiation dosim- 129. F. Salvat, PENELOPE2014: A Code System for Monte
etry in heterogeneous phantom,” Journal of Radiation Carlo Simulation of Electron and Photon Transport
Research 1(3), 143–149 (2003). (NEA, Barcelona, Spain, 2015).
130. F. Salvat, J. M. Fernandez-Varea, E. Acosta, J. Sempau, 134. G. Battistoni, et al., “The FLUKA code: Description and
Penelope – A Code System for Monte Carlo Simulation of benchmarking,” AIP Conference Proceedings 896 (May),
Electron and Photon Transport (Nuclear Energy Agency 31–49 (2007).
of the OECD (NEA): Organisation for Economic 135. T. T. Böhlen, et al., “The FLUKA Code: Developments
Co-Operation and Development – Nuclear Energy and challenges for high energy and medical applica-
Agency, Barcelona, Spain, 2001). tions,” Nuclear Data Sheets 120, 211–214 (2014).
131. J. Sempau, A. Badal, L. Brualla, “A PENELOPE-based 136. F. Cerutti, A. Ferrari, A. Mairani, P. R. Sala, “New
system for the automated Monte Carlo simulation of developments in FLUKA,” CERN-Proceedings, 469–475
clinacs and voxelized geometries-application to far- (2013).
from-axis fields,” Medical Physics 38(11), 5887–5895 137. K. Parodi, et al., “The FLUKA code for application of
(2011). Monte Carlo methods to promote high precision ion
132. A. Baumgartner, A. Steurer, F. Josef Maringer, beam therapy,” CERN-Proceedings 2, 509–516 (2010).
“Simulation of photon energy spectra from Varian 138. A. Mairani, et al., “The FLUKA Monte Carlo code cou-
2100C and 2300C/D Linacs: Simplified estimates with pled with the local effect model for biological calcula-
PENELOPE Monte Carlo models,” Applied Radiation tions in carbon ion therapy,” Physics in Medicine and
and Isotopes 67(11), 2007–2012 (2009). Biology 55(15), 4273–4289 (2010).
133. S. Espãa, J. L. Herraiz, E. Vicente, J. J. Vaquero, M. 139. Fasso, et al., “The FLUKA code: Present applications
Desco, J. M. Udias, “PeneloPET, a Monte Carlo PET and future developments,” in Computing in high Energy
simulation tool based on PENELOPE: Features and val- and Nuclear Physics Conference Proceedings, La Jolla,
idation,” Physics in Medicine and Biology 54(6), 1723– CA, 24-28 March 2003.
1742 (2009).
II
Brachytherapy
229
Chapter 15
Brachytherapy Dosimetry
Christopher L. Deufel
Mayo Clinic
Rochester, Minnesota
Wesley S. Culberson
University of Wisconsin – Madison
Madison, Wisconsin
Mark J. Rivard
Brown University & Rhode Island Hospital
Providence, Rhode Island
Firas Mourtada
Helen F. Graham Cancer Center & Research Institute
Newark, Delaware
CONTENTS
15.2 NIST/PSDL Calibration and Measurements 232
15.2.1 Low-Energy LDR Brachytherapy Standards 233
15.2.2 High-Energy LDR Brachytherapy Standards 234
15.2.3 High-Energy HDR Brachytherapy Standards 234
15.2.4 Beta Particle Brachytherapy Standards 235
15.2.5 Electronic Brachytherapy Standards 235
15.2.6 Retired Brachytherapy Standards 235
15.3 ADCL and SSDL Calibrations and Measurements 235
15.3.1 ADCL Brachytherapy Calibrations 236
15.4 Source Vendor Calibrations and Measurements 238
15.4.1 Well Chamber Measurement Specifics 238
15.5 Dosimetry Investigators Measurements and Calculations 238
15.5.2 Measurements 239
15.5.3 Calculations 239
231
15.6 Societal Recommendations 239

15.6.2 Consensus Data 240
15.6.3 Data Dissemination 240
15.6.4 Potential Future Directions 240
15.7 Clinical User’s Calibrations (Assays), Measurements, and Calculations for Photon Emitting Sources 240
15.7.1 Clinical User Calibrations 240
15.7.2 Clinical User Measurements 242
15.7.3 Dose Calculations 244
15.7.3.1 Correction-Based Dose Rate Calculations: TG-43 244
15.7.3.2 Model-Based Dose Calculation Algorithms 245
15.7.3.3 Dose Calculations for Permanent and Temporary Implants 246
15.8 Clinical User’s Calibrations (Assays), Measurements, and Calculations for Beta Emitting Sources 246
15.8.1 Clinical User Calibrations 246
15.8.2 Clinical User Measurements 246
15.8.3 Dose Calculations 246
15.9 Future Directions 247
15.9.1 Model-Based vs. TG-43 247
15.9.2 In Vivo Quality Control 247
15.9.3 New Dosimeters 247
15.9.4 Safety and Learning 248
References 248
15.1 INTRODUCTION sources. This chapter describes methods to measure

Radiotherapy can be delivered using external beam or and determine the brachytherapy source strength
brachytherapy. The main advantage of brachytherapy is and associated dosimetric parameters while ensuring
the close proximity of the radiation source to the target traceability to the national dosimetry standards. First,
(tumor or benign), when this option is viable. Delivery we present on the national standards (NIST/PSDL),
options include interstitial, intracavitary, endolumi- and the secondary laboratory (ADCL/SSDL/National
nal, intraoperative, or surface (i.e., plesiotherapy). Metrology Institute) establishments. Then, we describe
Brachytherapy has a long history since the early 1900s the required calibration process and requirements at the
and the most recent decade has exploded with new time of manufacturing by the vendor. Dosimetry tools
radiation sources, advanced image-guided treatment used for research and development to investigate new
planning, and improved delivery devices. Cancers of the sources are presented. End user (clinical physicist) cali-
cervix, prostate, breast, and skin have been most com- brations (assays), measurements, and calculations for
monly cured using brachytherapy as monotherapy or photon- and beta- emitting sources are provided, and
as a boost to external beam radiotherapy (EBRT) [1–5]. most current societal recommendations are presented
Other disease sites have also benefited from this spe- with future directions discussed.
cial radiotherapy modality such as ocular melanoma,
endorectal, nasopharynx, lung cancer, and keloids [6– 15.2 NIST/PSDL CALIBRATION AND
9]. Hence, brachytherapy advancements have enhanced MEASUREMENTS
its essential role for the treatment of malignant and Source-strength standardization in brachytherapy is
nonmalignant disease with radiation. The radiation- crucial for ensuring accurate dose deliveries and clini-
emitting sources used for this treatment modality are cal trial consistency. The strength of each brachytherapy
radionuclide- or electronic-based and may incorpo- source used for treatment needs to be characterized by
rate low (<2 Gy h−1) or high (>1 Gy min−1) dose rates. a well-defined and measurable quantity, with high accu-
There are unique physics challenges that one must racy and precision and a low uncertainty. Since mea-
realize to ensure accurate and precise dosimetry of such suring the source strength of a brachytherapy source
Brachytherapy Dosimetry ◾ 233
is not trivial due to generally low photon energies and concepts and quantities of interest. Air-kerma strength
steep dose gradients, there have been substantial efforts (SK) is the quantity used to determine source strength
over the past several decades to standardize the quanti- for photon-emitting brachytherapy sources. SK is the
ties, units, and measurement techniques. The organiza- source-strength metric recommended by the AAPM
tion in the U.S. responsible for maintaining most of the through Task Group (TG) 43 report and its associated
brachytherapy source-strength physical measurement updates [10, 11]. SK is defined by the following equation,
standards is called the National Institute of Standards
and Technology (NIST). This is a primary standards SK = K air d 2 (15.1)
dosimetry laboratory (PSDL) and serves as a govern-
ment-supported laboratory for developing, maintaining, where K air is the determined air-kerma rate, d is the dis-
and improving primary radiation calibration standards. tance between the source and the “reference distance”
Globally, there are many PSDLs in the world and the of 1 m. The AAPM TG-32 report recommends that
International Atomic Energy Association (IAEA) and the units for SK are represented by U in μGy m2 h−1 or
International Bureau of Weights and Measures (BIPM) cGy cm2 h−1 [12]. For beta-emitting sources, the quan-
are responsible for providing support and maintaining tity used to determine the source strength is absorbed
their consistency. A PSDL is the top of the calibration dose rate to water (ADW). The location of this quantity
chain. One step further down the calibration chain is varies by beta source type. For example, the ADW for
the secondary standards dosimetry laboratory (SSDL). planar ophthalmic applicators is on the surface and for
In the U.S. the SSDLs are not a part of the IAEA net- beta line sources or line-source arrays it is at a depth of
work; instead, they are accredited by the American 2-mm in water [13].
Association of Physicists in Medicine (AAPM) and are
called Accredited Dosimetry Calibration Laboratories 15.2.1 Low-Energy LDR Brachytherapy Standards
(ADCLs) instead of SSDLs. In principle, the AAPM The primary standard at NIST is a large-volume large-
ADCLs are similar to SSDLs. The purpose of SSDLs and angle free-air ionization chamber (FAC) called the
ADCLs is to disseminate the standards maintained at wide-angle free-air chamber (WAFAC) for low-energy
the PSDLs by way of transferring calibration traceability low dose rate (LDR) brachytherapy sources such as 125I,
to end-users. 103Pd, and 131Cs. The WAFAC is a cylindrically sym-
Traceability is the concept that describes the chain metric free-air ionization chamber shown by the basic
that links a measurement of a quantity in the field to schematic in Figure 15.1 and detailed in the publication
an established standard. This chain of measurements is by Seltzer et al. in 2003 [14]. The principal operation of
transferred from the primary standard to a local clini- the WAFAC is to determine the air ionization per unit
cal user by the process of repeated cross calibrations. At mass, or exposure, which is then converted to air-kerma
the top of the chain are the primary standard measure-
ment devices, maintained by NIST or sometimes by an
alternate laboratory as an interim national standard (as
is the case for HDR 192Ir) in the U.S. A primary standard
is defined as an instrument that does not need a cali-
bration. It directly realizes the quantity of interest. For
example, free-air chambers (FACs), calorimeters, and
certain well-defined ion chambers may be considered
primary standards. In brachytherapy dosimetry, there
are several different quantities of interest, determined
by a variety of basic measurement devices at NIST. The
different source-strength quantities and measurement
devices maintained by NIST or interim standards labo- FIGURE 15.1 Basic schematic of the NIST WAFAC showing
ratories are reviewed in this section. the source, aluminum filter, defining aperture and collecting
Before presenting the different primary brachy- volume. V represents the voltage of the polarizing electrode.
therapy standards, it is useful to define the calibration V/2 represents the voltage of the middle electrode.
rate and then air-kerma strength, SK , of the source. A re-entrant chamber has been established, a 226Ra source
beam-defining aperture is placed 30 cm from the source (with a half-life of 1,600 years) is used to verify response
and the center of volume for the ionization chamber is constancy.
located approximately 50 cm from the source. The mea-
surement is then corrected to the “reference distance” 15.2.3 High-Energy HDR Brachytherapy Standards
of 1 m. The determination of air-kerma rate is accom- Presently, there is no NIST standard for the air-kerma
plished through a derivative technique where the air strength of high dose rate (HDR) 192Ir sources. For
volume of the chamber is changed multiple times and these types of sources, secondary standards are main-
the net change in ionization current is measured for tained by ADCLs and are referred to as interim stan-
each chamber volume change. dards. There are two basic techniques to determine the
It should be noted that determination of the air-kerma air-kerma strength of HDR 192Ir sources as they do not
strength by means of a large-volume FAC measurement have a suitably shielded facility. The first method is to
is challenging due to the low signal-to-noise ratio from position a chamber at a fixed distance from the source.
the weak nature of the LDR sources. Each measurement Although this sounds straightforward, it is difficult to
takes hours to perform, and multiple measurements position the source at a precise location due to its abil-
are acquired and averaged to keep the statistical uncer- ity to move within the guide tube, resulting in relatively
tainties below 1% (k = 1). The combined uncertainty high uncertainties if this method is used. In Europe, this
estimates for 125I, 103Pd, and 131Cs encapsulated brachy- is the preferred method at PSDLs such as Physikalisch-
therapy seeds are approximately 1.5%, but depend on Technische Bundesanstalt (PTB), but in their imple-
each individual measurement and repeatability of the mentation they use a sophisticated custom source holder
measurements [15]. without a guide tube. For the ADCLs in the U.S., the
preferred method involves measurement of air kerma
15.2.2 High-Energy LDR Brachytherapy Standards at multiple distances from the source and solving for
For measurement of high-energy LDR brachytherapy unknown quantities of room scatter and positioning
sources such as 192Ir and 137Cs with average energies offsets by simultaneously solving several equations [20].
greater than 50 keV, the AAPM and the European The typical number of measurement distances is seven,
Society for Radiotherapy and Oncology (ESTRO) resulting in the technique being referred to as the seven-
have issued joint recommendations [16]. A procedure distance technique as shown schematically in Figure 15.2.
was established in 1979 at NIST to utilize known-volume This technique takes advantage of the highly accurate
spherical graphite-walled ionization chambers for change in distance from the source from a precision lead
determining the air-kerma rate at source-to-chamber screw and stepper motor system.
distances between 50 cm and 100 cm [17]. The wall A small (3.6 cm3) spherical ionization chamber, such
thickness of each chamber was designed to meet as an Exradin A3 (Standard Imaging, Middleton, WI)
Bragg–Gray cavity theory conditions and the air-
kerma rate was determined from cavity theory first
principles [18]. The formula to convert the basic
quantity of ionization current to air-kerma rate is
described in Chapter 16 of the 2009 AAPM Summer
School Proceedings by Mitch and Soares [19]. Once
the air-kerma rate is determined, SK is determined
similarly as before (Equation 15.1).
Due to the low signal-to-noise ratio in the spheri-
cal graphite-walled ionization chambers from the 192Ir
and 137Cs sources, sometimes multiple seeds (up to
50) were measured simultaneously [17]. The spherical
graphite-walled chamber measurement result is then FIGURE 15.2 Schematic of the HDR 192Ir seven-distance
transferred to a spherical re-entrant ionization cham- measurement technique established at the University of
ber. Once the calibration coefficient for this spherical Wisconsin ADCL.
is normally used for HDR 192Ir air-kerma strength mea-

surements [21]. Since a NIST-traceable calibration for
192Ir is not available for these chambers, a linear interpo-
lation method is used to derive the calibration for HDR

192Ir. The two beam qualities used to interpolate the 192Ir
calibration coefficient are 137Cs and M250 x-ray beam,

both of which are available at NIST and the ADCLs.
FIGURE 15.3 Schematic of the NIST eBT calibration
This method works well since the average energy of the standard.
gamma rays emitted from 192Ir is about midway between
the two interpolated energies. The chambers used are
selected to have a flat energy response over this range, vaginal cancers. Although the AAPM has not issued
yielding a variation of only a couple of percent over formal recommendations for eBT dosimetry, NIST has
the range of energies in the interpolation. More details developed an air-kerma based eBT standard, which was
on the specific equations for the determination of air- detailed in a publication by Seltzer et al. in 2014 [23]. A
kerma strength and the equation solver are presented in schematic of the basic setup is shown in Figure 15.3. In
great detail by Stump et al. in 2002 [20]. a specially designed facility at NIST, a FAC and a high-
purity germanium detector (HPGe) have been mounted
15.2.4 Beta Particle Brachytherapy Standards on opposite sides of a source holder that can accommo-
For many years, the NIST standard for beta particle date a Xoft Axxent (Xoft, a subsidiary of iCAD, Inc, San
sources used in medical applications has been an ambi- Jose, CA, USA) eBT source inside its cooling catheter.
ent air extrapolation chamber based on the design by The spectrum of each source is measured by the HPGe
Loevinger [22]. Very small gap sizes are utilized, since detector and the air-kerma rate is determined from the
the range of beta particles is on the order of millime- Lamperti FAC. To average out over source anisotropy,
ters in air. Monte Carlo simulations are used to correct the air-kerma rate is measured at the four cardinal
each measurement for divergence and chamber mate- angles around the source (by rotating the source about
rials. Planar beta-particle brachytherapy sources, such its long axis) and then averaged to determine a mean
as those used for ocular radiotherapy, are measured value.
with a 4-mm diameter electrode. Seed- and line-based
sources are measured with a 1-mm diameter electrode 15.2.6 Retired Brachytherapy Standards
and the sources are embedded in water-equivalent plas- NIST has retired many of its measurement standards
tic phantoms. A larger electrode increases the detec- over the years. The reasons for a standard being retired
tor signal, and is therefore used with planar sources may vary, but are mostly due to sources no longer being
whose source strength completely covers the electrode. manufactured or lack of resources at NIST to main-
In-phantom measurement depths for the seed- and line- tain the standard. Most notably in recent years was
based sources are usually 2 mm. For all beta-particle the extrapolation chamber standard for beta-emitting
measurements at NIST, the source distributions (across 90Sr/90Y ophthalmic applicators [24, 25]. Consequently,
the face of the source) are mapped out by stepping the the ADCLs are not able to maintain NIST-traceability
1-mm electrode around the source to find the region of and have also discontinued calibrations for 90Sr ophthal-
maximum signal. mic applicators.
15.2.5 Electronic Brachytherapy Standards 15.3 ADCL AND SSDL CALIBRATIONS

Electronic brachytherapy (eBT) involves the treatment AND MEASUREMENTS
of disease with miniature x-ray tubes that mimic the NIST maintains the primary standards for brachy-
size and geometry of sealed sources. The number of therapy sources, however, the demand for source mea-
clinical users and manufacturers of eBT devices have surements from clinical users far exceed the resources
increased in the last two decades. Applications of eBT available at NIST or the other PSDLs. It is for this reason
range from surface treatments of skin disease, to inter- that a network of secondary standards laboratories has
nal treatments with applicators for breast, cervical and been established. These labs are called SSDLs or ADCLs
in the United States, as described in an earlier section. the determination of a calibration coefficient for the
ADCLs have been an integral part of the calibration customer’s well-type ionization chamber based on a
chain since their inception in 1971. All three ADCLs NIST-traceable calibration chain. The specifics of each
currently provide brachytherapy source calibrations to calibration vary depending on the NIST-traceable
AAPM members. These ADCLs are: quantity and the method of transfer, but ultimately
the customer’s equipment is returned with a calibra-
1. University of Wisconsin (Madison, WI) tion certificate with a unique calibration coefficient for
their chamber and source type. Each source model will
2. MD Anderson (Houston, TX)
render a unique calibration coefficient in the customer’s
3. K & S Associates (Nashville, TN) well chamber, even if the source radionuclide itself is the
same. For example, a model 200 103Pd seed (Theragenics
The purpose of the ADCLs is to provide precise cali- Corp., Buford, GA) will not yield the same calibration
brations of customer equipment (for the purposes of coefficient as the model 2335 103Pd seed (Best Medical,
brachytherapy, these are predominantly well cham- Springfield, VA). The reason for this is due to the inter-
bers and electrometers), to reduce the time and costs nal characteristics of each seed, yielding differing emis-
for these calibrations that it would otherwise take if the sion spectra and angular distributions. One should not
customer chose to have NIST calibrate their equipment, assume that two different seed models would have the
and to provide technical resources (such as answering same calibration coefficients simply because they share
questions on equipment selection) for AAPM mem- the same isotope. The next sections provide details on
bers. As time has progressed, an AAPM subcommit- the calibrations of well-type ionization chambers for the
tee, the Calibration Laboratory Accreditation (CLA) different brachytherapy source types. Electrometer cali-
Subcommittee, has been developed to oversee the brations are performed separately from well chamber
ADCLs. The CLA is comprised of technical experts and calibrations and are not described within this chapter.
representatives from NIST. Members of the CLA serve
as auditors for onsite ADCL visits to the clinical users 15.3.1 ADCL Brachytherapy Calibrations
around the Unites States at least once every two years. Most recommended brachytherapy source-strength
The following sections provide an overview of the cali- quantities are based on air-kerma calibration standards
brations provided by the ADCL network. Not every ADCL at NIST. The techniques and measurement standards
is accredited to provide every type of calibration, and the employed by NIST are extremely complicated and not
accreditation scope for each lab is listed on their website practical for clinical user measurements. For this reason,
for reference. One unique characteristic of the ADCL well-type ionization chambers (often called well cham-
network, compared with the IAEA-based SSDLs, is the bers) are used by the clinical user instead. Well chambers
AAPM requirement for proficiency testing. It is required are commercially available instruments and are the most
that all ADCLs periodically inter-compare their calibra- common source-strength measurement device used by
tion results and present their results for CLA review. ADCL clinical brachytherapy physicists and calibration laborato-
source measurement comparisons must agree within 2% ries. A well chamber can be calibrated by an ADCL with
for long half-life photon-emitting sources, 3% for short a high degree of accuracy and precision, and transported
half-life photon-emitting sources, 3% for photon-emitting easily by mail. Well chambers can be either air-communi-
intravascular brachytherapy (IVBT) sources, and 6% for cating or sealed (pressurized), with the air-communicat-
beta-emitting IVBT sources. The well-chamber calibra- ing type being the more common. A basic schematic of
tions must agree within 3% for photon-emitting calibra- a well chamber is shown in Figure 15.4, where the seed is
tions, and 6% for beta-emitting IVBT calibrations. These held securely inside the well chamber with a source holder.
constancy checks help to ensure standardization within The ionization current is read out with an electrometer at
the ADCL network, and distinguishes the ADCLs from the other end of the triaxial cable.
SSDLs, which do not have such requirements. The calibration technique for well chambers is rela-
The process of calibrating customer’s equipment tively straightforward in an ADCL. The clinical physicist
at the ADCL for brachytherapy sources is similar for specifies the source model(s) for which they are seeking
all brachytherapy source types. The process involves calibration, and supplies the source holder(s) that they
FIGURE 15.4 Basic schematic of a well-type ionization cham-

ber used by a clinical physicist and sent to an ADCL for brachy-
therapy seed source-strength measurements.
use. The ADCL calibration should be performed with

the end-user’s source holder rather than the ADCL’s FIGURE 15.5 Schematic of a typical source holder for
source holder since they can differ structurally and brachytherapy well chamber calibrations and measurements.
might change the source position, attenuation, or scatter The well chamber surrounding the source is indicated by dot-
and thereby alter the measured calibration coefficient. ted lines.
The resulting value from an ADCL well chamber cali-
bration is the air-kerma strength calibration coefficient, holder, and a standoff spacer is used as a pedestal for
NSK, and has units of μGy m2 h−1 A−1. In North America, the seed to sit upon during the measurement. The seed
NSK is corrected to 22°C and 101.325 kPa. The exception is dropped into the catheter with a pair of tweezers for
to an air-kerma strength calibration is for eBT source LDR sources, while an HDR source is moved into posi-
where air-kerma rate at 50 cm in air is the established tion inside a guide tube. The general design of the source
quantity. The eBT calibration coefficient is NAKR@50cm holder is shown in Figure 15.5.
to reflect the Lamperti FAC-based NIST standard. As Source holders for eBT calibrations are slightly dif-
stated before, the source holder is an integral part of the ferent since the miniature x-ray tube and its cooling
reported calibration coefficient. The calibration coef- catheter need to be stabilized during the calibration. A
ficient assumes that the source is positioned for the specialized source holder with a metal extension tube on
maximum response inside the well chamber, also called top has been developed for eBT air-kerma well chamber
the sweet spot. This position is measured at the time of measurements. These are commercially available and
calibration and is reported on the calibration certificate. provided by the manufacturer during eBT acceptance
The clinical physicist should place the source being mea- testing and commissioning. As for LDR seeds and HDR
sured in the same position as that used for calibration. sources, these holders should be sent for calibration con-
A variety of source holders are available for calibra- currently with the well chamber.
tion of brachytherapy well chambers, and the correct Uncertainties for each of the ADCL calibration are pro-
source holder should be used for each source to ensure vided with the calibration certificate and include the NIST-
reproducible results and safe handling of the sources. traceable uncertainties as well as any additional uncertainties
The basic design of all holders is similar, with the source incurred during the calibration transfer to the end-user.
being suspended at the cylindrical axis (center) and Uncertainties for photon-emitting sources are approxi-
sweet spot within the well chamber. To accomplish this, mately 2.5% (k = 1) and approximately 8% (k = 1) for beta-
a small-diameter plastic catheter (typically Kapton® in emitting sources (mainly due to uncertainties of the NIST
order to minimize attenuation) is secured within the primary measurements) [25, 26].
15.4 SOURCE VENDOR CALIBRATIONS ionization currents are high. For well chamber measure-
AND MEASUREMENTS ments of LDR brachytherapy sources, charge is often
Manufacturers of each brachytherapy source model measured over a specified time and then divided by a
are responsible for providing end users with a certifi- time increment to arrive at the same value (i.e., current).
cate for source strength that is NIST-traceable, accurate, It is useful for the manufacturer to employ quality
and precise. The manufacturer is essentially at the same assurance (QA) checks of the well chamber to verify its
level in the traceability chain as the clinical physicist. response is stable with time. This is normally achieved
Manufacturers utilize well-type ionization chambers by use of another long-lived source such as a 137Cs nee-
to determine the source strength and are responsible dle, 90Sr source, or a linear accelerator (irradiation of the
for calibrating their equipment at least every two years. entire chamber). In addition to QA checks, leakage of
Each individual source is measured by the manufac- the chamber should be measured to check for instabili-
turer before sending it to a clinical physicist, and the ties in the electrical signal pathways.
manufacturer’s results are communicated via a source In summary, there is a robust framework of brachy-
strength certificate. This certificate will inform the end therapy source calibration standards available to
user of the mean source strength or individual source end users in North America. The source calibration
strengths for the many ordered sources, such as would framework enables traceable calibrations of end-user
be the case for a permanent seed prostate implant where instrumentation the strength of brachytherapy sources
up to 102 seeds might be implanted. consistently.
In addition to providing the customer with the
mean or individual source strength, the manufacturer 15.5 DOSIMETRY INVESTIGATORS
is responsible for periodically (approximately once per
MEASUREMENTS AND CALCULATIONS
year) initiating a proficiency test with NIST and the 15.5.1 Rationale
ADCLs to ensure that the manufacturing processes Single-source brachytherapy dose distributions are
have not changed to a point where the calibration coef- needed for determining brachytherapy dosimetry
ficient of a well-type ionization chamber is affected. parameters as used in clinical brachytherapy treatment
Proficiency tests first require at least five sources be sent planning systems (TPSs). However, evaluation of single-
to NIST, and then circulated through the ADCLs, before source brachytherapy dose distributions is challenging
returning to NIST for a final measurement. Results of for medical physicists, and much more difficult than
each proficiency test are communicated to the ADCL for evaluating EBRT dose distributions. The problem
directors as well as the AAPM CLA for review. is due to lack of an established dosimetry protocol such
as AAPM TG-51 as for EBRT [27], and lack of available
15.4.1 Well Chamber Measurement Specifics resources in a typical radiotherapy clinic to perform
For determination of the air-kerma strength of a source such evaluation. Brachytherapy dosimetry is further
by a well chamber measurement, the calibration coeffi- complicated by its high dose gradients and lower photon
cient for the well chamber is used along with the ioniza- energies in comparison to EBRT. Lacking standardized
tion current and associated corrections according to the methods and equipment for evaluating brachytherapy
following equation: dose distributions, physicists must develop their own
dosimetry protocol and devise measurement or calcula-
SK = I ⋅ N e ⋅ ktp ⋅ N SK ⋅ Aion ⋅ Pion (15.2) tion methods.
Over the years as new brachytherapy sources have
where I is the measured ionization current, N e is the elec- entered the marketplace, a cadre of physicists (some-
trometer calibration coefficient, ktp is the correction for times termed dosimetry investigators) have embraced
temperature and pressure, N SK is the air-kerma strength this challenge, evaluated single-source brachytherapy
calibration coefficient, Aion is the correction for recom- dose distributions, and determined the brachytherapy
bination at the time of calibration, and Pion is the cor- dosimetry parameters as used in TPSs [11]. The field of
rection for recombination at the time of measurement. brachytherapy dosimetry has evolved through scientific
Note that this equation involves a current measurement, advancements learnt from one source that have bene-
which is the case for HDR 192Ir measurements where the fited later investigation of another source, largely from
careful documentation of the evaluation techniques in material where the uncertainty in the medium correc-
the peer-reviewed literature. Furthermore, the AAPM tion is low, even though the magnitude of the medium
has provided explicit guidance to dosimetry investiga- correction may be greater than a more water-equivalent
tors for performing dose measurements and calcula- plastic [31]. For high-energy sources, differences in
tions, see sections V.D. and V.E. of Ref. [11], respectively. medium composition make little difference when
Given the necessary initial investment in equipment assessing the dose distribution, and dosimetry investi-
and knowledge, dosimetry investigators have repeated gators will then select the phantom composition based
the process of dose evaluation for new brachytherapy on convenience. However, dosimetry for high-energy
sources as they become available for clinical use. In fact, sources is more subject to sensitivity of radiation scatter
part of the process in making sources available for gen- to the experimental setup. Dosimetry investigators need
eral clinical use requires the dosimetry parameters to be to provide sufficient backscattering material so that full
determined [28, 29]. scatter conditions are approximated for dose at a given
From the context of dosimetry, brachytherapy sources distance from the source [32]. For LDR sources, seeds
may be categorized in a 2-by-2 matrix of low- or high- having the highest possible source strength are ordered
energy photon emitters or as LDR or HDR sources. As so that the radiation signal is maximized and to pro-
applicable to measuring the dose distribution, sources of vide a long timeframe for which the source can be used
lower energy and lower dose rate are harder to evaluate since their decay half-lives are 9.7, 17.0, and 59.4 days for
due to increased sensitivity of results because of detec- 103Pd, 125I, and 131Cs, respectively [11]. For HDR sources,
tor and phantom medium composition and due to lower dose at distant locations are sometimes of interest and
detector signal-to-noise, respectively. These sources the dosimetry investigator is challenged trying to char-
would include LDR seeds of 103Pd, 125I, and 131Cs with acterize its distribution over a large range of distances
photon energies ranging from 0.02 to 0.03 MeV and indi- that may be beyond the dynamic-range capabilities of a
vidual source strengths 10 U or less. Conversely, brachy- single exposure for a radiation detector.
therapy sources of higher energy and higher dose rate are
easier to evaluate due to decreased sensitivity of results 15.5.3 Calculations
because of detector and phantom medium composition The gold standard for calculating single-source brachy-
and due to higher detector signal-to-noise, respectively. therapy dose distributions is through Monte Carlo sim-
These sources would include HDR sources of 192Ir and ulations of radiation transport [33], and is applicable
60Co with photon energies ranging from 0.4 to 1 MeV for all brachytherapy source types. Unlike dosimetry
and individual source strengths exceeding 104 U. measurements, the Monte Carlo technique can simulate
things impossible to produce in reality like individual
15.5.2 Measurements photons from nuclear disintegration (instead of the poly-
Since dose measurements are sensitive to the aforemen- energetic emissions common to radionuclide decay, for
tioned source differences, dosimetry investigators must gleaning detector energy-response corrections); exceed-
customize their techniques and equipment to the type ingly small detectors (having negligible dose gradients
of source being measured [30]. Low-energy sources and absolute spatial accuracy); and unusual material
require special attention since their dose gradients are dimensions, compositions, and densities (for ascertain-
typically steeper than high-energy sources, due to the ing sensitivity of simulation results to assumptions of
fact that they are more easily attenuated, and the high- geometric design).
energy sources are typically used within HDR brachy-
therapy applicators where patient tissue is positioned 15.6 SOCIETAL RECOMMENDATIONS
farther from the source. Fortunately, low-energy seeds 15.6.1 Rationale
are often used for volume or surface implants where the The field of brachytherapy dosimetry currently affords
multi-seed geometry grossly diminishes the dose gradi- global consistency to the brachytherapy community
ent. For low-energy sources, substantial corrections are through use of dosimetry parameters in TPSs based
needed to adjust the dose measured in a (typically) plas- on the AAPM TG-43 dose calculation formalism [11].
tic phantom to the reference material of liquid water. An However, there are subtle issues around which dosim-
approach developed over the years is to select a phantom etry parameters to use and how TPSs perform TG-43
dose calculations with specific algorithms [34]. Based sources and establishing similar dosimetric and calibra-
on choice of dosimetry parameters or algorithm imple- tion standards as for radionuclide-based sources. Such
mentation, dose calculations could vary by more than standards would be complicated by the low-energy HDR
50% [35]. Consequently, professional societies like nature of eBT sources sharing strengths and weaknesses
the AAPM, ESTRO, ABS (American Brachytherapy of current standards which are largely low-energy LDR
Society), and ABG (Australasian Brachytherapy Group) or high-energy HDR. Extension of such standards could
have issued reports aiming for standardization. These also include the applicators associated with the eBT
standards include data resolution requirements and sources. An additional aspect of eBT sources is that they
data interpolation (and extrapolation) methods. do not explicitly fit the TG-43 dose calculation formu-
lism, which is based on source strength specification.
15.6.2 Consensus Data The TPS would require modification or a work-around
Given the dynamic state of the literature on brachyther- to fit within the current formulism [38].
apy dosimetry data for a given source model, profes- Another potential future direction of societal recom-
sional societies have developed a process through which mendations on brachytherapy dosimetry is adoption
publications containing data are reviewed for meeting of a new source strength metric, dose rate to water at
minimum standards, and then such data are formally 1 cm on the source transverse plane (Dw,1cm), to replace
disseminated to the community. Currently and largely the product of air-kerma strength (SK) and the dose-rate
led by the AAPM, formal committees have examined constant (Λ) [39]. This approach would have a smaller
published dosimetry parameters for new source mod- uncertainty than the current technique where uncer-
els. The evaluation methods are described elsewhere in tainties in the SK × Λ product are determined separately.
explicit detail [31]. The AAPM works harmoniously with With Dw,1cm being a direct measure of the source out-
other professional societies so that there are no conflict- put in the medium of interest and for a specific source,
ing consensus data. However, there is not yet an estab- not being traced to a measurement (or calculation) per-
lished process to revise the consensus data when new formed by a dosimetry investigator at a much earlier
dosimetry parameters are published or improvements date, the uncertainty for LDR seeds and HDR sources
are made to the dosimetry techniques. could potentially be lower. However, diminishment in
the overall uncertainty by this amount is not deemed
15.6.3 Data Dissemination clinically relevant given it is smaller than the current SK
Consensus data are made available online through the measurement uncertainty performed by the end-user.
posting on the Brachytherapy Source Registry managed The general feeling in the community is that resources
jointly by the AAPM and the Imaging and Radiation should be put toward addressing material heterogene-
Oncology Core (IROC) Houston QA Center. The reg- ities for improved brachytherapy dose calculation accu-
istry is available online and updated regularly given racy [34].
changes to source vendor contact information, inclu-
sion of additional related supporting publications for a 15.7 CLINICAL USER’S CALIBRATIONS
specific brachytherapy source model, and addition/sub- (ASSAYS), MEASUREMENTS, AND
traction of sources based on compliance with AAPM CALCULATIONS FOR PHOTON
standards for calibration intercomparison frequency EMITTING SOURCES
and accuracy [33, 36, 37]. The HDR 192Ir consensus data 15.7.1 Clinical User Calibrations
on the AAPM/IROC Houston registry is actually man- The accurate measurement of LDR and HDR sources
aged in the form of web links to the ESTRO website, is of the utmost importance, and is required or rec-
where all data is downloadable and in a standardized ommended by the US Food and Drug Administration
format. (FDA), AAPM, and ABS. According to AAPM Task
Group Reports 40, 56, and 138, the clinical medical
15.6.4 Potential Future Directions physicist has the responsibility to verify the manufac-
There are several new directions that societal recom- turer-provided source strength measurements using
mendations for brachytherapy dosimetry may take in ADCL-traceable instrumentation [13, 40, 41]. Although
the near future. These could include evaluation of eBT rare, large deviations in source strength have been
reported. In some instances seeds had no radioactivity, to be used (e.g., eye plaques) [37]. When loose seeds are
and in other instances the source strength was grossly ordered in combination with stranded sources, it is rec-
misstated by the vendor calibration certificate. Clerical ommended to assay 5% or 5 seeds, whichever is fewer.
errors by the manufacturer have also resulted in a clinic The clinical user calibration should agree with the man-
receiving the wrong sources or paperwork. ufacturer to within 3%, and an investigation should be
Well chambers are the most common instruments performed for greater deviations. It is recommended
used for air-kerma strength (U) source calibration of to contact the vendor if deviations are greater than 5%.
LDR and HDR photon sources, as well as for beta-emitting Individual seed variations greater than 6% should be
sources. A NIST traceable calibration factor for a well discussed with the radiation oncologist. Independent
chamber may be obtained by sending the chamber to an third-party calibration may also be available for all of
ADCL. Alternatively, a clinic may order a source cali- the sources for a fee. The clinical user should review the
brated by an ADCL, insert the source into the chamber, third-party results and maintain the documentation
and transfer the ADCL calibration to their well chamber. for the United States Nuclear Regulatory Commission
Nuclear medicine chambers are a less desirable alterna- (NRC) inspection.
tive for source calibration, where disadvantages include For HDR 192Ir sources, the NRC regulation Title 10
that they are generally calibrated for source apparent CFR requires calibration of the source by the clinical
activity (not air-kerma strength), and pre-determined user prior to first use, following repair that requires
settings are for nuclear medicine radionuclides and not removal of the source or major repair of the compo-
brachytherapy sources. The source geometry may not be nents associated with the source exposure assembly, or
adequately affixed and pressurized chambers may expe- at intervals not exceeding 3 months, which is a typical
rience gas leakage. Furthermore, when a nuclear medi- period for source exchanges. The clinical user calibra-
cine well chamber is used for both unsealed and sealed tion should agree within 5% with the manufacturer
radionuclides, there is a risk of contamination or recali- certificate [42]. The typical clinical setup is to place a
bration by the nuclear medicine team. thin plastic catheter inside of a plastic source holder
Commissioning tests for a well chamber include that fits into the well chamber as shown in Figure 15.6.
identification of the sweet-spot location and size, mea- The source holder helps to center the plastic catheter
surement of any change in the chamber response as a in the middle of the chamber. The catheter may be
function of the orientation angle of the seed holder, mea- secured to the source holder using a small piece of
surement of the polarity effect, linearity, measurement tape. After identifying the “sweet spot” of the cham-
of recombination effects, and calibration for the sources ber, where the chamber response is greatest, through
that will be used clinically. The user should also estab- a stepwise measurement (use the afterloader treatment
lish the assay procedure, including performing a con- console to step source in ≤5mm steps) of the output
stancy check prior to usage with another known source, as a function of location inside the catheter, the user
typically with a source of comparable energy and a long measures the well chamber current produced by the
half-life. With a photon energy of approximately 60 keV radioactive source. This current reading is converted
and half-life of 433 years, 241Am is commonly used for to a dose rate using an ADCL calibration coefficient
constancy tests. Other long-lived sources such as 137Cs (cGy cm 2 h −1 nA−1 or U/nA).
and 90Sr may also be used. Constancy checks are useful The electrometer used in brachytherapy calibrations
for detecting leaks in sealed well chamber systems, as should have a range that is appropriate to the type of
well as problems with the collection region, cables, or sources being measured. Some low-activity source mea-
electrometer. The well chamber and electrometer system surements produce very little ionization current in the
voltage and leakage should also be assessed prior to the well chamber. The electrometer should have a resolu-
measurement. tion as low as 10 fA for LDR sources or up to 200 nA for
For LDR sources, the AAPM Report 98 from the Low HDR sources [43]. The signal resolution should be better
Energy Brachytherapy Source Calibration Working than 0.1%. An electrometer with a high and low range is
Group recommended that the end-user calibrate at least therefore recommended for brachytherapy calibrations,
10 sources or 10% of sources, whichever is greater, and and both settings should be calibrated by the ADCL at a
preferably all of the sources when small quantities are frequency not to exceed two years.
FIGURE 15.6 A typical clinical user calibration setup: (a) Electrometer (model MAX-4000 Plus, Standard Imaging, Inc.,
Middleton, WI), (b) note card of electrometer ADCL-calibration coefficients, (c) source holder and tape to secure plastic
catheter, (d) well chamber (model IVB1000, Standard Imaging, Inc., Middleton, WI), (e) note card of well chamber ADCL-
calibration coefficient, (f) triax cable connector for well chamber coupling to electrometer, (g) plastic treatment catheter and
transfer tube to HDR remote afterloader unit, (h) aneroid barometer, and (i) electronic thermometer.
15.7.2 Clinical User Measurements For example, the consensus data sets presented in the
The clinical user may perform dosimetric measure- TG-43 reports were generated using a blend of 2D refer-
ments in order to confirm or elucidate dosimetry, such ence-quality dose distribution data from Monte Carlo
as to verify the dose calculation parameters or the and experiments [11]. The experimental verification
impact of heterogeneities and treatment geometries, or has an essential role in the consensus process, and pro-
to perform in vivo verification of the treatment dose. vides independent validation prior to clinical use of a
In general, there are steep dose gradients in brachy- new brachytherapy source model. TLD have been most
therapy treatments and the clinical user will require commonly used for such experiments because their
high precision phantoms and detectors for these types diminutive size reduces the uncertainty from steep dose
of measurements. This section highlights examples of gradients, and well known protocols are available for
the types of experiments that the clinical user might calibration and energy dependence [71–73]. Although
conduct. Additional references are provided to give the film provides even greater resolution than TLD, care
reader more comprehensive guidance on measurement must be taken to properly calibrate the radiochromic
techniques and choices of apparatus for using ionization film for the resultant photon energies. For low-energy
chambers [44–48], radiochromic film [49–54], gel [55, photon-emitting radionuclides, the corrections can
56], thermoluminescent dosimeters (TLD) [49, 57], opti- exceed 10% [74]. Another example of source param-
cally stimulated luminescence dosimeters (OSLDs) [58], eter verification is the measurement of the profiles
diode [59, 60], metal-oxide-semiconductor field-effect and depth dose for a beta-emitting planar source con-
transistors (MOSFET) [61–64], scintillation plastics [57, taining 32P. Diodes and radiochromic film have been
65–68], or lithium formate pellets [69, 70]. used to cross-check Monte Carlo results, as well as to
Verification measurements may be performed to ver- evaluate how the dosimetric parameters change when
ify dose profiles, depth dose, and the source anisotropy. nonstandard geometries are encountered in the clinic
[75, 76]. Radiochromic film is an excellent dosimeter for an HDR 192Ir planar applicator that is placed upon the
beta emitters since the energy dependence of the film is skin may have one side facing the tissue and the other
smaller than it is for photon emitters [77]. However, the side facing air. The lack of tissue causes a reduction in
user should take care to properly radiologically scale any backscatter. Consequently, dose to the treatment site is
phantom materials as well as the materials within the reduced, with a reduction that varies as a function of
film [75]. Radiological scaling is used to adjust the actual depth and applicator geometry. Calcifications, bone,
depth in the medium to an equivalent depth in water. and lung tissues can also produce changes in the dose
Clinical user measurements may also establish the delivered to the treatment site. Several authors have
effect of non-water heterogeneities in the vicinity of the reported the impact of tissue heterogeneities on the
brachytherapy treatment. Heterogeneities may be built delivered dose [82–89].
into the applicator, present in the patient tissues (or Occasionally, external shields are used in brachy-
absence of tissues), or even applied external to the treat- therapy to reduce the dose to nearby healthy tissues.
ment site. Experimental measurement offers a method For example, lead shields may be used for intraoperative
to validate calculations or measure dose under condi- brachytherapy to protect nearby organs that are outside
tions that are difficult to model. of the treatment field [90]. Similarly, oral shields may be
Source applicators are typically constructed from used to protect uninvolved healthy tissue for brachy-
plastics or other materials. The heterogeneity effects are therapy of the head-and-neck region [91]. Shields may
relatively minor when the materials are nearly water- also be used in vaginal cylinders and cervical cancer
equivalent, but effects can be significant when dense applicators. Skin applicators may use conical shaped
materials such as high-Z metals are involved [78]. For tungsten applicators to focus the radiation and spare
example, a shielded vaginal cylinder that is currently uninvolved tissue [92]. The scatter in the applicator and
manufactured by both Varian and Nucletron includes the air cavity between the source and the skin can be
tungsten shields (Z = 74) for reducing the dose to unin- difficult to model, and depends on the size and the shape
volved healthy tissues, namely the rectum, bladder, or of the conical applicator. The impact of the lead shields
uninvolved vaginal mucosa. The clinical user may per- is best assessed by experimental measurement using a
form measurements in order to experimentally confirm variety of dosimeters, including diodes, ion chambers,
that the shields provide the expected attenuation magni- or TLDs [93–96].
tude and distribution. Radiochromic film has been used In vivo dosimetry is a type of clinical user measure-
to evaluate the dose distribution for the shielded vaginal ment that provides a record of the dose that is delivered
cylinder with a HDR 192Ir source [79]. A second example to the patient’s tissues. In vivo dosimetry is challenging
of heterogeneous source applicators is an eye plaque as for brachytherapy treatments because of the presence of
used for temporary LDR treatment of ocular melanoma. high dose gradients, and the detector energy sensitivity
COMS eye plaques have non-water equivalent Silastic corrections that may change as a function of distance
and Modulay components that have effective atomic from the source. The high dose gradients demand precise
numbers greater than water. Eye plaque dose distribu- positioning of the detector active volume with respect
tions have been measured by several authors [80, 81]. In to the radiation source(s). If patient dose is to be mea-
particular, the work by McCauley et al. demonstrated sured and not simply inferred, the detector must also
the use of radiochromic film for commissioning eye be inserted into the treatment region. In vivo dosimetry
plaque therapy and verifying the ~16% attenuation that may be used in the intraoperative setting to quantifying
is due to the Silastic insert that holds 125I seeds within the dose to the nearby healthy tissues or target [97]. In
the plaques [81]. McCauley et al. also demonstrated the intraoperative setting, in vivo dosimetry is particu-
how dosimetric measurements could be used to identify larly useful because often a single high-dose treatment
potential deviations in the thickness of the Silastic mate- fraction is delivered in the OR and it is typically not pos-
rial that could arise from manufacturing variations. sible to obtain a CT scan with the applicator in position.
Patient tissue heterogeneities can also impact deliv- Therefore, the physicist is unable to exactly recreate the
ered dose to the patient, which is more substantial for precise placement and curvature of the planar applica-
the lower energy radionuclides and also for tissues with tor. In vivo dosimetry has also been used to monitor
electron densities that differ from water. For example, the location and duration of source dwell positions in
HDR brachytherapy in order to prevent gross errors in

the execution of the treatment plan, such as an incorrect
connection of the treatment unit to the treatment cath-
eters or an incorrect catheter length used in the treat-
ment plan [98, 99]. Another approach is to place an in
vivo dosimeter directly into the patient tissue, such as
the rectum, urethra, or vaginal mucosa to monitor the
dose that is delivered near these critical structures dur-
ing treatments [61, 62, 100–102]. Chapter 11 of this book
provides further discussion on the exciting future of in
vivo dosimetry for brachytherapy.
15.7.3 Dose Calculations

Brachytherapy dose calculations for permanent and
temporary implants may be broadly categorized into FIGURE 15.7 Coordinate system used for brachytherapy
correction-factor based and model-based approaches. dosimetry calculations. A source of length L is centered at
Correction factor methods begin with a calibrated dose the origin, and the long axis of the source is oriented in the
under a set of standard conditions, and use correction z-direction. A reference point, P, is located at a radial dis-
tance, r, and an angle, θ, with respect to the origin. θ1 and θ2
factors to adjust the dosimetry for the actual conditions
represent the angles between the reference point and the ends
encountered in the patient treatment. Model-based of the source. The interior angle, β, is also denoted.
calculations are useful when modeling the dosimetric
impact of heterogonous materials, and can directly solve The standard conditions in this formalism are rep-
the Boltzmann transport equation for photons and elec- resented by the source strength, SK , as measured by a
trons using deterministic discrete ordinate methods, NIST-traceable calibration, and is intended to reflect the
collapsed cone convolution, or Monte Carlo radiation air-kerma rate in vacuo, K δ ( d ), without radiation scatter
transport simulations. and with an energy cutoff that is typically δ > 5 keV for
low-energy photon-emitting sources.
15.7.3.1 Correction-Based Dose Rate
Calculations: TG-43 SK = K δ ( d ) d 2 (15.5)
The AAPM TG-43 reports provide the most widely
used formalism for brachytherapy dose rate calcula- The dose-rate constant, Λ , converts the air-kerma
tions for photons, and can be classified as a correction strength into a dose rate to water in water at a distance
factor approach. The correction factors are established of 1 cm from the center of the source along the source
through a consensus dataset composed of theoretical transverse plane, θ = π /2. The dose-rate constant varies
and experimental results [10, 11]. The reader is encour- with source model, because of differences in the scatter-
aged to read the AAPM TG-43 reports for discussion ing and attenuation in water from various photon ener-
and examples of reference datasets. The dose rate calcu- gies. The distribution of radioactivity inside the source
lation equations for point- and line-source approxima- capsule also influences the value of Λ . Sources that con-
tion are specified in Equation 15.3 and Equation 15.4, centrate radioactivity near the capsule center will have
respectively. In Figure 15.7, θ 0 = π2 radians and r0 = 1 cm. higher dose rates at the reference position, and there-
fore higher Λ values due to the inverse-square law. Also,
G ( r ,θ ) sources with higher photon energies will generally have
D ( r ,θ ) = SK Λ P g P ( r )φan ( r )
GP ( r0 ,θ 0 ) higher Λ values.
(15.3)
(point source approximation) The geometry function, GP ( r ,θ ) or GL ( r ,θ ), mitigates
the influence of the inverse-square law on dose-rate cal-
G ( r ,θ ) culations, and is the dominant factor characterizing the
D ( r ,θ ) = SK Λ L g L (r )F (r ,θ )
GL ( r0 ,θ 0 ) dose-rate distribution. The angles β and θ have units of
(15.4)
(line source approximation) radians in the calculation.
1 therapy dose distribution [103–107]. This is due to the

G P ( r ,θ ) = (15.6)
r2 properties of a volumetric implant being dominated by
dose from the nearest sources, which by their proximity,
 β
 ,θ ≠0 implies that they are not attenuated by closer sources.
 Lrsin(θ ) An advantage of the TG-43 formalism is that the
G L ( r ,θ ) =  1 (15.7) correction factors (i.e., dosimetry parameters) may be
 ,θ =0
L2 interpreted to have some physical basis. Additionally,
 r2 −
 4 the geometry function dominates for characterizing
the dose-rate distribution and is based upon a simple
While it is convenient in brachytherapy TPSs to use a equation that is not subject to interpolation errors.
simple function such as a point- or line-source model The remaining correction factors exhibit less variation
for the geometry function, these simplifications do not and can be described in small tables with linear inter-
precisely capture the dose falloff. This is due to simplifi- polation without substantial loss of accuracy. A disad-
cation of the radioactivity spatial distribution as a point vantage of the TG-43 approach is that it assumes an
or line segment, as well as how it does not account for aqueous medium and does not properly account for
radiation attenuation and scatter in water, the reference material heterogeneities in the brachytherapy applicator
medium. or the patient. The assumption of an aqueous medium
The radial dose function is identified as, g P ( r ) or is good for most LDR and HDR applications; however,
g L (r ), depending upon which function is used to model the clinical team should carefully evaluate the impact of
the radioactivity distribution. The radial dose func- material heterogeneities when high density (e.g., dense
tion accounts for scatter and attenuation effects on the plastics or metals) or low density (e.g., air or lung) media
source transverse plane. It decreases more rapidly as a are present near or within the treatment site.
function of distance for lower energy sources because
they are more easily attenuated in water due to the pho- 15.7.3.2 Model-Based Dose Calculation Algorithms
toelectric effect. The preferred method for calculating dose under realistic
Brachytherapy source design is meant to be cylindri- conditions that include material heterogeneities is with
cally symmetric, but asymmetric with respect to polar model-based dose calculation algorithms (MBDCAs).
angle due to the internal components and self-attenua- Through varying methods, these approaches solve the
tion by the source material itself. Variations in the dose linear Boltzmann transport equation to calculate the
distribution as a function of polar angle around the dose. Monte Carlo methods employ a statistical approach
source are accounted for by the 1D anisotropy function, to solve the transport equation for a distribution of ini-
ϕ an ( r ), or 2D anisotropy function, F (r ,θ ). tial conditions, and provide a statistical average of the
TG-43 dose calculations rely on the principle of dose based on the sum of microscopic interactions. The
dose superposition. The dose delivered from a group of quality of Monte Carlo results is statistically dependent
sources is simply the sum of the doses delivered by each on the number of particle histories that are considered.
individual source. This approximation is excellent for a Discrete ordinate approaches are also available, which
high-energy HDR 192Ir source that is administered with bin the Boltzmann transport equation into a finite set as
a remote afterloader using a single dwelling source, as a function of predefined spatial grid, angles, and energy
well as historically for high-energy LDR sources such as histogram and apply radiological equations to the
137Cs tubes and 192Ir ribbons. However, it is important to
bulk materials. In either case, the quality of the results
consider that HDR sources are modeled with a certain requires accurate modeling of the source, geometry,
amount of source cable, any bending of the cable during and boundary conditions. The accuracy of the discrete
treatment delivery can lead to small differences between ordinates and collapsed cone approaches is dependent
the ideal dose and the delivered dose at locations proxi- on the type and amount of discretization that is used to
mal to the source. For low-energy brachytherapy where group the spatial dimensions and energy.
multiple sources are simultaneously placed inside the A detailed overview of calculation algorithms is found
treatment site, the inter-seed attenuation effects have in a pair of Medical Physics Vision 20/20 publications on
been reported to have a small effect on the radiation the use of Model-Based Dose Calculation Algorithms
(MBDCAs) by Rivard et al. [108, 109]. A description of radiochromic film may be cross calibrated by a device
MBDCA methods and recommendations for their clini- with a NIST-traceable calibration (e.g., via a linear
cal implementation is presented in the TG-186 report accelerator with electron energy comparable to the
[110]. Furthermore, a reference set of CT scans and plans source being assayed) and have been used with success.
is available for the commissioning of a MBDCA [111]. A depth dose curve should also be obtained for calcu-
The user may download reference datasets, calculate a lation of the dose at depths different than the calibra-
dose distribution for an applicator using their TPS, and tion depth. The IAEA also recommends that the user
compare the dose results against the Monte Carlo refer- tests the uniformity of the source in terms of their dose
ence data [112]. rate at the calibration distance. When the calibration
is performed at a depth in phantom and converted to
15.7.3.3 Dose Calculations for Permanent a water-equivalent depth, it is important to properly
and Temporary Implants scale the phantom materials. The IAEA TECDOC-1274
Once the initial dose rate, D , to a point in the treatment and ICRU Report 56 provide guidance for radiological
field is known, the dose to the patient for a permanently scaling for beta emitters, and Deufel et al. provides an
implanted source is calculated using the following example for the use of radiological scaling to calibrate a
formula: planar 32P source using radiochromic film [43, 75, 113].
Dose = D × τ (Dose from a permanent implant) (15.8) 15.8.2 Clinical User Measurements
Clinical measurements for beta emitters have a simi-
t1/2
Where τ represents the average life and is equal to ln(2) . lar role as their photon emitting counterparts. Clinical
For a temporary implant, the dose calculation is measurements may be performed in order to verify
modified by the amount of source strength that remains dosimetry parameters and the impact of nonideal clini-
after completion of the treatment and the sources are cal conditions on the delivered dose. Suitable detec-
removed. tors for beta emitters must have high spatial resolution,
since depth dose variations of 10% per 0.1 mm change
 − 
t
in the depth are possible. Clinical measurements using
Dose = D × τ  1 − e τ  (Dose from a temporary implant)
  radiochromic film have been used to assess the dose
(15.9) uniformity, dose profiles, and depth dose as a part of
pretreatment QA for 32P planar sources that are used
for the treatment of the spinal dura [75]. The impact
15.8 CLINICAL USER’S CALIBRATIONS of source size and material heterogeneities was also
(ASSAYS), MEASUREMENTS, AND evaluated in the same study. Radiochromic film dosim-
CALCULATIONS FOR BETA etry has also been applied to the measurement of the
EMITTING SOURCES dose distribution from a beta source for intravascu-
15.8.1 Clinical User Calibrations lar brachytherapy [114, 115]. Diodes and microMOS-
Calibration of beta-emitting sources, such as 90Sr and FETs are alternate dosimeters for beta emitters. Diodes
32P, is especially challenging due to the steepness of the
have been used to measure the depth dose from planar
depth dose curve. The IAEA TECDOC-1274 provides sources [76]. MicroMOSFETs have been used to evaluate
guidance for the calibration of beta emitting sources the impact of stents on the dose delivered during intra-
[43]. The IAEA recommends that beta-emitting pla- vascular brachytherapy [116, 117].
nar sources be calibrated at a distance of 1.0 mm mea-
sured from the surface of the source. Beta seed and line 15.8.3 Dose Calculations
sources should be calibrated at a distance of 2.0 mm For beta-emitting sources, simple correction-based
from the center of the source. Beta balloon shells and approaches may be used to calculate the dose rate to a
stents should be calibrated at a distance of 0.5 mm point. Dose profiles and a depth-dose table can be used to
from the surface of the shell or stent. Other detectors estimate the dose at a point from the center of the source.
may also be used for the calibration of beta-emitting ICRU Report 56 outlines a dose-kernel model that can
sources. Parallel-plate ionization chambers, diodes, and be used to perform dosimetry for a beta-emitting source
in a homogenous medium [113]. The approach assumes the “holy grail” for HDR brachytherapy dosimetry. A
a slab geometry and is based on the work by Cross et al. Medical Physics Vision 20/20 paper by Tanderup et al.
[118]. Janicki et al. extended the work of Cross et al. highlighted how in vivo dosimetry can be used for iden-
for a non-homogenous slab medium [119]. The reader tifying HDR afterloader malfunction, intra-fraction
is referred to Deufel et al. for a complete description of motion, applicator reconstruction errors, applicator
theoretical calculations and application to 32P planar length errors, source step size, and transfer tube connec-
sources that are used for brachytherapy treatments of tions [122]. In vivo dosimetry would also be valuable for
the spinal dura [75]. the QC of intensity modulated brachytherapy, in which a
Monte Carlo methods may also be used to calculate shielding material is used to asymmetrically block radia-
the dose for a beta-emitting source. This is essential tion from the source in order to create a dose distribution
when the medium cannot be approximated by simple that is more conformal to the target and thereby better
slab geometry or when the source cannot be approxi- spares healthy tissue. In vivo dosimetry could be used to
mated as planar. Beta-emitting eye plaques, intravas- ensure the position and function of any moving parts.
cular sources, and planar applicators for spinal dura Although researchers have made substantial prog-
treatments are good examples of the use of Monte Carlo ress and several prototypes have been used for the types
dosimetry techniques for beta dosimetry [76, 120, 121]. of QC suggested by Tanderup et al., in vivo dosimeters
have not been incorporated into commercial HDR
15.9 FUTURE DIRECTIONS treatment systems. Flat-panel based in vivo dosim-
15.9.1 Model-Based vs. TG-43 etry is a likely candidate for first-to-market for HDR
Each year, computers become faster and less expensive. brachytherapy, since it is a straightforward add-on to a
It is therefore worth considering whether MBDCAs will brachytherapy system and requires no modifications to
eventually entirely replace TG-43 as the primary dose cal- the HDR afterloader and placement of additional inva-
culation algorithm. As noted in Section 15.7.3.2, model- sive devices into the patient. The flat panel approach is
based dose calculations with HDR 192Ir brachytherapy an indirect measurement of the dose, since it records
generally result in minor dose differences to soft tissues the treatment dwell locations and time and uses the
when compared with the TG-43 standard approach. information to reconstruct the dose inside the patient.
However larger differences have been noted when there Gel, MOSFET, and scintillator detectors have focused
are substantial tissue heterogeneities near the treatment on directly measuring the dose in a clinically relevant
region, for example in an HDR partial breast brachy- region, such as the urethra, rectum, or vaginal mucosa.
therapy treatment, the TG-43 will overestimate the dose The future of brachytherapy dosimetry will hopefully
to skin that is in close proximity to air. The benefits of include in vivo tools for evaluating the delivered HDR
a more accurate dose distribution will likely mean that dose as it is being administered. An immediate chal-
MBDCAs will replace TG-43 once high- and low-energy lenge that all in vivo dosimetry systems face is the inter-
dose distributions can be rapidly calculated and thor- facing with the treatment delivery system. The in vivo
oughly validated. Of course, physicians must consider system ought to be designed such that it automatically
how to modify their prescription doses in circum- interrupts the treatment when a delivery deviation is
stances where there are large systematic differences (e.g., detected.
eye plaques), akin to the changes in the prescriptions
for lung treatments that occurred ~30 years ago when 15.9.3 New Dosimeters
heterogeneity corrections became standard for EBRT. New radiation dosimeters are exciting when they pro-
Physicists will also need to unify the method for how vide superior methods of calibration or improve the
the electron density of materials is calculated from CT verification of the dose delivery. Brachytherapy dosim-
numbers or assigned for dense materials. etry is challenging because there are high dose gradients
near the treatment region. These high gradients mean
15.9.2 In Vivo Quality Control that a small uncertainty in the position of the dosimeter
In vivo measurement of the dose delivered to the patient, can produce a large change in the measured dose. With
which includes real-time quality control (QC) that in vivo dosimeters, there is the added challenge that the
can detect errors in the treatment delivery, is perhaps direct measurement of the dose to a critical structure or
target may require a detector that can fit inside of a nar- Root cause analysis and FMEA are risk-analysis tools
row catheter or similar device. for estimating practice risk and identifying weak points
With regards to calibration dosimetry, a recent pub- in the process. Risk-analysis safety tools can be used to
lication on graphite calorimetry systems for the dosim- reduce the likelihood or severity of errors in the workflow.
etry of therapeutic x-ray beams invites the possibility The AAPM TG-100 report provides a description of risk-
of a calorimetry-based brachytherapy calibrator [123]. analysis methods that can be used in the design of a quality
Calorimetry is used by primary standard dosimetry management program [125]. For brachytherapy dosimetry,
laboratories because it offers a direct measurement of this approach might be applied to the design of an LDR
the energy delivered to medium, and therefore relies on source assay procedure, an HDR applicator commission-
fewer correction factors than an ionization chamber, ing procedure, or the design of a planning checklist.
diode, or other dosimeters.
With regards to in vivo dosimetry, researchers have REFERENCES
been working toward smaller detectors that can be eas-
1. N. Tselis, et al., “High dose rate brachytherapy as
ily placed inside of a catheter. Continued development of monotherapy for localised prostate cancer: Review of
MOSFET, diodes, and scintillators will hopefully provide the current status,” Clinical oncology (Royal College of
tools for directly measuring the dose in a clinically relevant Radiologists) 29(7), 401–411 (2017).
region, such as the urethra, rectum, or vaginal mucosa. 2. K. K. Lee, et al., “High-dose-rate vs. low-dose-rate
Researchers have also been striving to produce detector intracavitary brachytherapy for carcinoma of the
uterine cervix: Systematic review and meta-analysis,”
systems, internal or external to the patient, that can be Brachytherapy 14(4), 449–457 (2015).
used to monitor the location and duration of source dwells. 3. K. Ota, et al., “Review: The reemergence of brachy-
The continued development of new detectors will hope- therapy as treatment for non-melanoma skin cancer,”
fully make QC provided by in vivo dosimetry a routine and Journal of Dermatological Treatment 29(2), 170–175
straightforward part of the treatment delivery process. (2017).
4. J. Skowronek, M. Wawrzyniak-Hojczyk, K.
Ambrochowicz, “Brachytherapy in accelerated partial
15.9.4 Safety and Learning
breast irradiation (APBI) – review of treatment meth-
At its most fundamental level, the role of the medical ods,” Journal of Contemporary Brachytherapy 4(3), 152–
physicist is to ensure the safe and accurate delivery of 164 (2013).
radiotherapy. Recently, incident learning systems (ILS), 5. L. C. Mendez, et al., “Three-dimensional-guided peri-
neal-based interstitial brachytherapy in cervical can-
root cause analyses, and failure modes effect analy-
cer: A systematic review of technique, local control and
ses (FMEAs) have been employed to help predict and toxicities,” Radiotherapy & Oncology 123(2), 312–318
understand errors in the radiotherapy process. (2017).
With regards to brachytherapy dosimetry, ILS is a 6. A. Youroukou, et al., “The potential role of brachyther-
useful tool for the design of a quality management pro- apy in the irradiation of patients with lung cancer: A
gram. ILS provides a mechanism, for rational, evidence systematic review,” Clinical and Translational Oncology
19(8), 945–950 (2017).
based practice improvement. For example, the ILS can 7. H. Buckley, C. Wilson, T. Ajithkumar, “High-dose-rate
be used to evaluate whether use of the antiquated units brachytherapy in the management of operable rectal
of source strength curie (Ci) should be abolished. The cancer: A systematic review,” International Journal of
IAEA and NRC medical event reports detail how the Radiation Oncology, Biology, Physics 99(1), 111–127
simultaneous use of Ci and U for clinical applications (2017).
8. J. J. Echegaray, et al., “Iodine-125 brachytherapy for
has been a contributing cause of multiple medical events uveal melanoma: A systematic review of radiation dose,”
[43]. Deufel et al. reported on how the comprehensive Ocular Oncology and Pathology 3(3), 193–198 (2017).
use of an ILS can be used to reduce risk, improve com- 9. I. Goutos and R. Ogawa, “Brachytherapy in the adjuvant
munication, and improve the quality of written proce- management of keloid scars: Literature review,” Scars,
dures [124]. The ILS described by the authors was used Burns & Healing 3, 2059513117735483 (2018).
10. R. Nath, et al., “Dosimetry of interstitial brachytherapy
to record each mistake that was made in the radiother-
sources: Recommendations of the AAPM Radiation
apy treatment process, and the results were used to com- Therapy Committee Task Group No. 43. American
municate potential failure modes to team members and Association of Physicists in Medicine,” Medical Physics
to identify ways to improve the process. 22(2), 209–234 (1995).
11. M. J. Rivard, et al., “Update of AAPM Task Group No. (vol. 108, p. 337, 2003),” Journal of Research of the
43 Report: A revised AAPM protocol for brachyther- National Institute of Standards and Technology 109(2),
apy dose calculations,” Medical Physics 31(3), 633–674 301–301 (2004).
(2004). 27. P. R. Almond, et al., “AAPM’s TG-51 protocol for clini-
12. R. Nath, L. Anderson, D. Jones, C. Ling, R. Loevinger, J. cal reference dosimetry of high-energy photon and elec-
Williamson, W. Hanson, Specification of brachytherapy tron beams,” Medical Physics 26(9), 1847–1870 (1999).
source strength: A report by Task Group 32 of the American 28. J. Williamson, et al., “Dosimetric prerequisites for
Association of Physicists in Medicine (American Institute routine clinical use of new low energy photon inter-
of Physics, Inc., New York, NY, 1987). stitial brachytherapy sources. Recommendations of
13. R. Nath, et al., “Code of practice for brachytherapy phys- the American Association of Physicists in Medicine
ics: Report of the AAPM Radiation Therapy Committee Radiation Therapy Committee. Ad Hoc Subcommittee
Task Group No. 56. American Association of Physicists of the Radiation Therapy Committee,” Medical Physics
in Medicine,” Medical Physics 24(10), 1557–1598 (1997). 25(12), 2269–2270 (1998).
14. S. M. Seltzer, et al., “New national air-kerma-strength 29. M. J. Rivard, et al., “Supplement 2 for the 2004 update
standards for I-125 and Pd-103 brachytherapy seeds,” of the AAPM Task Group No. 43 Report: Joint recom-
Journal of Research of the National Institute of Standards mendations by the AAPM and GEC-ESTRO,” Medical
and Technology 108(5), 337–358 (2003). Physics 44(9), e297–e338 (2017).
15. W. S. Culberson, et al., “Large-volume ionization cham- 30. M. J. Rivard, et al., “Supplement to the 2004 update of
ber with variable apertures for air-kerma measurements the AAPM Task Group No. 43 Report,” Medical Physics
of low-energy radiation sources,” Review of Scientific 34(6), 2187–205 (2007).
Instruments 77(1) (2006). 31. R. M. Kennedy, et al., “Experimental and Monte Carlo
16. Z. Li, et al., “Dosimetric prerequisites for routine clini- determination of the TG-43 dosimetric parameters
cal use of photon emitting brachytherapy sources with for the model 9011 THINSeed brachytherapy source,”
average energy higher than 50 keV,” Medical Physics Medical Physics 37(4), 1681–1688 (2010).
34(1), 37–40 (2007). 32. J. Perez-Calatayud, et al., “Dose calculation for photon-
17. T. P. Loftus, “Standardization of Ir-192 gamma-ray emitting brachytherapy sources with average energy
sources in terms of exposure,” Journal of Research of the higher than 50 keV: Report of the AAPM and ESTRO,”
National Bureau of Standards 85(1), 19–25 (1980). Medical Physics 39(5), 2904–2929 (2012).
18. F. H. Attix, Introduction to Radiological Physics and 33. L. Beaulieu, et al., “Report of the Task Group 186 on
Radiation Dosimetry (Wiley, New York, NY, 1986), xxi, model-based dose calculation methods in brachy-
607 p. therapy beyond the TG-43 formalism: Current status
19. M. G. Mitch and C. G. Soares, “Primary standards and recommendations for clinical implementation,”
for brachytherapy sources,” in Clinical Dosimetry Medical Physics 39(10), 6208–6236 (2012).
Measurements in Radiotherapy, edited by D. W. O. 34. M. J. Rivard, J. L. Venselaar, L. Beaulieu, “The evolution
Rogers and J. E. Cygler (Medical Physics Publishing, of brachytherapy treatment planning,” Medical Physics
Madison, WI, 2009), Chapter 16, pp. 549–566. 36(6), 2136–53 (2009).
20. K. E. Stump, et al., “Calibration of new high dose rate 35. B. R. Thomadsen, et al., “Anniversary paper: Past and
192Ir sources,” Medical Physics 29(7), 1483–1488 (2002). current issues, and trends in brachytherapy physics,”
21. S. J. Goetsch, et al., “Calibration of Ir-192 high-dose-rate Medical Physics 35(10), 4708–4723 (2008).
afterloading systems,” Medical Physics 18(3), 462–467 36. L. A. DeWerd, et al., “Procedures for establishing and
(1991). maintaining consistent air-kerma strength standards
22. R. Loevinger and N. G. Trott, “Design and operation of for low-energy, photon-emitting brachytherapy sources:
an extrapolation chamber with removable electrodes,” Recommendations of the Calibration Laboratory
International Journal of Applied Radiation and Isotopes Accreditation Subcommittee of the American
17(2), 103–111 (1966). Association of Physicists in Medicine,” Medical Physics
23. S. M. Seltzer, M. O’Brien, M. G. Mitch, “New national air- 31(3), 675–681 (2004).
kerma standard for low-energy electronic brachytherapy 37. W. M. Butler, et al., “Third-party brachytherapy source
sources,” Journal of Research of the National Institute of calibrations and physicist responsibilities: Report of the
Standards and Technology 119, 554–574 (2014). AAPM Low Energy Brachytherapy Source Calibration
24. J. S. Pruitt, Calibration of Beta-Particle-Emitting Working Group,” Medical Physics 35(9), 3860–3865 (2008).
Ophthalmic Applicators (NBS Measurement Services, 38. L. A. DeWerd, et al., “A modified dose calculation
Gaithersburg, MD, 1987). formalism for electronic brachytherapy sources,”
25. J. S. Pruitt, C. G. Soares, and M. Ehrlich, Calibration Brachytherapy 14(3), 405–408 (2015).
of Beta-Particle Radiation Instrumentation and Sources 39. F. A. Siebert, J. L. Venselaar, M. Paulsen, T. Hellebust,
(NBS Special Publication, Gaithersburg, MD, 1988). P. Papagiannis, A. Rijnders, M. J. Rivard, “Dose-rate to
26. S. M. Seltzer, et al., “New national air-kerma-strength water calibrations for brachytherapy sources from the
standards for I-125 and Pd-103 brachytherapy seeds end-user perspective,” Metrologia 49, S249–S252 (2012).
40. G. J. Kutcher, et al., “Comprehensive QA for radia- 55. E. P. Pappas, et al., “On the use of a novel Ferrous
tion oncology: Report of AAPM Radiation Therapy Xylenol-orange gelatin dosimeter for HDR brachy-
Committee Task Group 40,” Medical Physics 21(4), 581– therapy commissioning and quality assurance testing,”
618 (1994). Physica Medica 45, 162–169 (2018).
41. L. A. DeWerd, et al., “A dosimetric uncertainty analysis 56. Y. Watanabe, et al., “Polymer gel dosimetry for the high-
for photon-emitting brachytherapy sources: Report of dose-rate brachytherapy using Ir-192 source,” Igaku
AAPM Task Group No. 138 and GEC-ESTRO,” Medical Butsuri 37(3), 173–176 (2017).
Physics 38(2), 782–801 (2011). 57. M. D. Belley, et al., “Real-time dose-rate monitoring
42. Commission, U.S.R. NRC 10CFR. [cited 11/15/2020] with gynecologic brachytherapy: Results of an initial
(2018). Available from: https://www.nrc.gov/reading- clinical trial,” Brachytherapy 17(6), 1023–1029 (2018).
rm/doc-collections/cfr/part035/part035-0633.html. 58. M. Rejab, et al., “Dosimetric characterisation of the opti-
43. International Atomic Energy Agency, IAEA-TECDOC- cally-stimulated luminescence dosimeter in cobalt-60
1274 Calibration of Photon and Beta Ray Sources Used high dose rate brachytherapy system,” Australasian
In Brachytherapy (2002). Physical and Engineering Sciences in Medicine 41(2),
44. A. Keyvanloo and H. S. Jans, “Technical Note: 475–485 (2018).
Minimizing geometrical uncertainties of cylindrical 59. J. Jeong, et al., “Impact of source position on high-dose-
well-type ionization chamber measurements: There rate skin surface applicator dosimetry,” Brachytherapy
is an optimal chamber length,” Medical Physics 45(8), 15(5), 650–660 (2016).
3962–3968 (2018). 60. N. Chofor, et al., “The mean photon energy EF at the
45. A. A. Schoenfeld, et al., “Reference conditions for ion- point of measurement determines the detector-specific
chamber based HDR brachytherapy dosimetry and radiation quality correction factor kQ,M in 192Ir brachy-
for the calibration of high-resolution solid detectors,” therapy dosimetry,” Zeitschrift für Medizinische Physik
Zeitschrift für Medizinische Physik 28(4), 293–302 (2017). 26(3), 238–250 (2016).
https://pubmed.ncbi.nlm.nih.gov/28969957/ 61. M. Carrara, et al., “In vivo rectal wall measurements
46. I. Iftimia, A. B. McKee, P. H. Halvorsen, “Varian HDR during HDR prostate brachytherapy with MOSkin
surface applicators – commissioning and clinical imple- dosimeters integrated on a trans-rectal US probe:
mentation,” Journal of Applied Clinical Medical Physics Comparison with planned and reconstructed doses,”
17(2), 231–248 (2016). Radiotherapy & Oncology 118(1), 148–153 (2016).
47. C. Candela-Juan, et al., “Design and characterization of a 62. J. Mason, et al., “Real-time in vivo dosimetry in high
new high-dose-rate brachytherapy Valencia applicator dose rate prostate brachytherapy,” Radiotherapy &
for larger skin lesions,” Medical Physics 43(4), 1639 (2016). Oncology 120(2), 333–338 (2016).
48. J. Metyko, W. Erwin, S. Landsberger, “Verification of 63. R. Phurailatpam, et al., “Characterization of commercial
I-125 brachytherapy source strength for use in radioac- MOSFET detectors and their feasibility for in-vivo HDR
tive seed localization procedures,” Applied Radiation brachytherapy,” Physica Medica 32(1), 208–212 (2016).
and Isotopes 112, 62–68 (2016). 64. M. Persson, J. Nilsson, A. Carlsson Tedgren, “Experience of
49. M. Aima, et al., “Dosimetric characterization of a using MOSFET detectors for dose verification measure-
new directional low-dose rate brachytherapy source,” ments in an end-to-end 192Ir brachytherapy quality assur-
Medical Physics (2018). ance system,” Brachytherapy 17(1), 227–233 (2018).
50. O. A. Awunor, “Assessment of a source position check- 65. Y. Watanabe, et al., “Automated source tracking with a
ing tool for the quality assurance of transfer tubes used pinhole imaging system during high-dose-rate brachy-
in HDR 192Ir brachytherapy treatments,” Brachytherapy therapy treatment,” Physics in Medicine and Biology
17(3), 628–633 (2018). 63(14), 145002 (2018).
51. M. Bellezzo, et al., “A novel rectal applicator for contact 66. F. Therriault-Proulx, L. Beaulieu, S. Beddar, “Validation
radiotherapy with HDR 192Ir sources,” Brachytherapy of plastic scintillation detectors for applications in
17(6), 1037–1044 (2018). low-dose-rate brachytherapy,” Brachytherapy 16(4),
52. C. L. Deufel, J. P. Mullins, M. J. Zakhary, “A quality 903–909 (2017).
assurance device for measuring afterloader perfor- 67. G. Kertzscher and S. Beddar, “Inorganic scintillation
mance and transit dose for nasobiliary high-dose-rate detectors based on Eu-activated phosphors for 192Ir
brachytherapy,” Brachytherapy 17(4), 726–731 (2018). brachytherapy,” Physics in Medicine and Biology 62(12),
53. M. Hermida-Lopez and L. Brualla, “Absorbed dose dis- 5046–5075 (2017).
tributions from ophthalmic (106) Ru/(106) Rh plaques 68. M. Eichmann and B. Thomann, “Air core detectors for
measured in water with radiochromic film,” Medical Cerenkov-free scintillation dosimetry of brachyther-
Physics 45(4), 1699–1707 (2018). apy beta-sources,” Medical Physics 44(9), 4900–4909
54. E. Martin, K. Sowards, B. Wang, “Investigation of a (2017).
source model for a new electronic brachytherapy tan- 69. E. Adolfsson, et al., “Measurement of absorbed dose
dem by film measurement,” Journal of Applied Clinical to water around an electronic brachytherapy source.
Medical Physics 19(5), 640–650 (2018). Comparison of two dosimetry systems: Lithium formate
EPR dosimeters and radiochromic EBT2 film,” Physics and 192Ir high-dose-rate brachytherapy,” Journal of
in Medicine and Biology 60(9), 3869–3882 (2015). Contemporary Brachytherapy 10(1), 73–84 (2018).
70. S. Mishra and T. P. Selvam, “Monte Carlo calculation of 84. D. Jacob, et al., “Clinical transition to model-based dose
beam quality and phantom scatter corrections for lith- calculation algorithm: A retrospective analysis of high-
ium formate electron paramagnetic resonance dosim- dose-rate tandem and ring brachytherapy of the cer-
eter for high-energy brachytherapy dosimetry,” Journal vix,” Brachytherapy 16(3), 624–629 (2017).
of Medical Physics 42(2), 72–79 (2017). 85. V. Peppa, et al., “On the impact of improved dosimetric
71. R. S. Sloboda and G. V. Menon, “Experimental determi- accuracy on head and neck high dose rate brachyther-
nation of the anisotropy function and anisotropy fac- apy,” Radiotherapy & Oncology 120(1), 92–97 (2016).
tor for model 6711 I-125 seeds,” Medical Physics 27(8), 86. C. A. Collins Fekete, et al., “Calcifications in low-dose rate
1789–1799 (2000). prostate seed brachytherapy treatment: Post-planning
72. T. D. Solberg, et al., “Dosimetric parameters of three dosimetry and predictive factors,” Radiotherapy &
new solid core I-125 brachytherapy sources,” Journal of Oncology 114(3), 339–344 (2015).
Applied Clinical Medical Physics 3(2), 119–134 (2002). 87. C. Furstoss, et al., “Monte Carlo study of LDR seed
73. A. S. Meigooni, et al., “Dosimetric characteristics of the dosimetry with an application in a clinical brachyther-
bests double-wall 103Pd brachytherapy source,” Medical apy breast implant,” Medical Physics 36(5), 1848–1858
Physics 28(12), 2568–2575 (2001). (2009).
74. C. G. Hammer, et al., “Experimental investigation of 88. S. M. Oliveira, et al., “Dosimetric effect of tissue hetero-
GafChromic((R)) EBT3 intrinsic energy dependence geneity for 125I prostate implants,” Reports of Practical
with kilovoltage x rays, (137) Cs, and (60) Co,” Medical Oncology and Radiotherapy 19(6), 392–398 (2014).
Physics 45(1), 448–459 (2018). 89. J. K. Mikell, et al., “Impact of heterogeneity-based dose
75. C. L. Deufel, et al., “Experimental and theoretical dosim- calculation using a deterministic grid-based Boltzmann
etry of the RIC-100 phosphorus-32 brachytherapy source equation solver for intracavitary brachytherapy,”
for implant geometries encountered in the intraoperative International Journal of Radiation Oncology Biology
setting,” Brachytherapy 14(5), 734–750 (2015). Physics 83(3), E417–E422 (2012).
76. G. N. Cohen, et al., “32P brachytherapy conformal 90. S. Lloyd, et al., “Intraoperative high-dose-rate brachy-
source model RIC-100 for high-dose-rate treatment therapy: An American Brachytherapy Society consen-
of superficial disease: Monte Carlo calculations, sus report,” Brachytherapy 16(3), 446–465 (2017).
diode measurements, and clinical implementation,” 91. T. W. Leung, et al., “High dose rate brachytherapy for
International Journal of Radiation Oncology, Biology, early stage oral tongue cancer,” Head & Neck. 24(3),
Physics 88(3), 746–752 (2014). 274–281 (2002).
77. J. G. Sutherland and D. W. Rogers, “Monte Carlo cal- 92. V. Pastor-Sanchis, et al., “Experimental validation of
culated absorbed-dose energy dependence of EBT and the Valencia-type applicators developed for the BEBIG
EBT2 film,” Medical Physics 37(3), 1110–1116 (2010). HDR afterloader Saginova,” Medical Physics 44(6),
78. M. J. Price, et al., “Dose perturbation due to the polysul- 3176–3176 (2017).
fone cap surrounding a Fletcher-Williamson colpostat,” 93. D. Granero, et al., “Radiation leakage study for the Valencia
Journal of Applied Clinical Medical Physics 11(1), 68–76 applicators,” Physica Medica 29(1), 60–64 (2013).
(2010). 94. P. T. Finger, et al., “Intraocular radiation blocking,”
79. G. Zwierzchowski, et al., “Film based verification of calcula- Investigative Ophthalmology & Visual Science 31(9),
tion algorithms used for brachytherapy planning-getting 1724–1730 (1990).
ready for upcoming challenges of MBDCA,” Journal of 95. M. Hira, et al., “Measurement of dose perturbation
Contemporary Brachytherapy 8(4), 326–335 (2016). around shielded ovoids in high-dose-rate brachyther-
80. P. Saidi, M. Sadeghi, C. Tenreiro, “Experimental mea- apy,” Brachytherapy 10(3), 232–241 (2011).
surements and Monte Carlo calculations for (103)Pd 96. A. S. Kirov, et al., “Measurement and calculation of het-
dosimetry of the 12 mm COMS eye plaque,” Physica erogeneity correction factors for an Ir-192 high dose-
Medica 29(3), 286–294 (2013). rate brachytherapy source behind tungsten alloy and
81. S. McCauley Cutsinger, C. Deufel, S. Corner, “An steel shields,” Medical Physics 23(6), 911–919 (1996).
experimental method for commissioning and quality 97. F. W. Hensley, “Present state and issues in IORT
assurance of COMS eye plaques,” Brachytherapy 17(4, Physics,” Radiation Oncology 12(1), 37 (2017).
Suppl.), S103 (2018). 98. A. Espinoza, et al., “The evaluation of a 2D diode array
82. S. M. Vahabi, M. Bahreinipour, M. Shamsaie in "magic phantom" for use in high dose rate brachy-
Zafarghandi, “Study on the dose modification factor therapy pretreatment quality assurance,” Medical
of strut adjusted volume implant (SAVI) with a (169) Physics 42(2), 663–673 (2015).
Yb source using MCNP4C,” Australasian Physical and 99. J. Poder, et al., “HDR brachytherapy in vivo source posi-
Engineering Sciences in Medicine 41(2), 445–450 (2018). tion verification using a 2D diode array: A Monte Carlo
83. I. Fotina, et al., “A comparative assessment of inhomo- study,” Journal of Applied Clinical Medical Physics 19(4),
geneity and finite patient dimension effects in 60Co 163–172 (2018).
100. E. L. Seymour, et al., “In vivo real-time dosimetric 112. Sources, Imaging and Radiation Oncology Core (IROC)
verification in high dose rate prostate brachytherapy,” Source Registry. Model-Based Dose Calculations.
Medical Physics 38(8), 4785–4794 (2011). Reference Data Sets (2018) [cited 11/15/2020]; Available
101. M. Carrara, et al., “Semiconductor real-time quality from: http://rpc.mdanderson.org/RPC/BrachySeeds/
assurance dosimetry in brachytherapy,” Brachytherapy Model_calculations.htm.
17(1), 133–145 (2018). 113. ICRU, Dosimetry of External Beta Rays for Radiation
102. D. Wagner, M. Hermann, A. Hille, “In vivo dosimetry Protection (International Commission on Radiation
with alanine/electron spin resonance dosimetry to eval- Units and Measurements, Bethesda, MD, 1997).
uate the urethra dose during high-dose-rate brachy- 114. R. Nath, et al., “Intravascular brachytherapy physics:
therapy,” Brachytherapy 16(4), 815–821 (2017). Report of the AAPM Radiation Therapy Committee Task
103. H. Afsharpour, et al., “A Monte Carlo study on the effect Group No. 60,” Medical Physics 26(2), 119–152 (1999).
of seed design on the interseed attenuation in perma- 115. D. M. Duggan, et al., “Radiochromic film dosimetry of
nent prostate implants,” Medical Physics 35(8), 3671– a high dose rate beta source for intravascular brachy-
3681 (2008). therapy,” Medical Physics 26(11), 2461–2464 (1999).
104. J. F. Carrier, et al., “Impact of interseed attenuation and 116. E. Drud, et al., “Influence of a commonly used stent
tissue composition for permanent prostate implants,” type on the dose distribution in endovascular brachy-
Medical Physics 33(3), 595–604 (2006). therapy,” Radiotherapy and Oncology 81, S262–S262
105. J. F. Carrier, et al., “Postimplant dosimetry using a (2006).
Monte Carlo dose calculation engine: A new clinical 117. E. Drud, et al., “Beta dosimetry with microMOSFETs
standard,” International Journal of Radiation Oncology, for endovascular brachytherapy,” Physics in Medicine
Biology, Physics 68(4), 1190–1198 (2007). and Biology 51(23), 5977–5986 (2006).
106. O. Chibani, J. F. Williamson, D. Todor, “Dosimetric 118. W. G. Cross, N. O. Freedman, P. Y. Wong, “Beta-ray
effects of seed anisotropy and interseed attenuation for dose distributions from point sources in an infinite
103Pd and 125I prostate implants,” Medical Physics 32(8), water medium,” Health Physics 63(2), 160–171 (1992).
2557–2566 (2005). 119. C. Janicki, et al., “Dose model for a beta-emitting stent
107. J. Mason, et al., “Monte Carlo investigation of I-125 in a realistic artery consisting of soft tissue and plaque,”
interseed attenuation for standard and thinner seeds in Medical Physics 26(11), 2451–2460 (1999).
prostate brachytherapy with phantom validation using 120. M. Hermida-Lopez, “Calculation of dose distributions
a MOSFET,” Medical Physics 40(3), 031717 (2013). for 12 106Ru/106Rh ophthalmic applicator models with
108. M. J. Rivard, L. Beaulieu, F. Mourtada, “Enhancements the PENELOPE Monte Carlo code,” Medical Physics
to commissioning techniques and quality assurance 40(10), 101705 (2013).
of brachytherapy treatment planning systems that use 121. J. M. DeCunha and S. A. Enger, “A new delivery system
model-based dose calculation algorithms,” Medical to resolve dosimetric issues in intravascular brachy-
Physics 37(6), 2645–2658 (2010). therapy,” Brachytherapy 17(3), 634–643 (2018).
109. M. J. Rivard, J. L. Venselaar, L. Beaulieu, “The evolution 122. K. Tanderup, et al., “In vivo dosimetry in brachyther-
of brachytherapy treatment planning,” Medical Physics apy,” Medical Physics 40(7), 070902 (2013).
36(6), 2136–2153 (2009). 123. I. J. Kim, et al., “Building a graphite calorimetry system
110. L. Beaulieu, et al., “Report of the Task Group 186 on for the dosimetry of therapeutic x-ray beams,” Nuclear
model-based dose calculation methods in brachy- Engineering and Technology 49(4), 810–816 (2017).
therapy beyond the TG-43 formalism: Current status 124. C. L. Deufel, et al., “Patient safety is improved with an
and recommendations for clinical implementation,” incident learning system—clinical evidence in brachy-
Medical Physics 39(10), 6208–6236 (2012). therapy,” Radiotherapy & Oncology 125(1), 94–100 (2017).
111. Y. Ma, et al., “A generic TG-186 shielded applicator for 125. M. S. Huq, et al., “The report of Task Group 100 of the
commissioning model-based dose calculation algo- AAPM: Application of risk analysis methods to radia-
rithms for high-dose-rate 192Ir brachytherapy,” Medical tion therapy quality management,” Medical Physics
Physics 44(11), 5961–5976 (2017). 43(7), 4209 (2016).
III
External Beam Radiation Therapy
253
Chapter 16
Photon Beam Dosimetry

of Conventional Medical
Linear Accelerators
Francisco J. Reynoso
Washington University in Saint Louis
Saint Louis, Missouri
CONTENTS
16.2 Photon Beam Dosimetry 256
16.2.1 Collimator Scatter Factor 256
16.2.2 Phantom Scatter Factor 258
16.2.3 Percent Depth Dose Curve 259
16.2.4 Tissue Air Ratio 260
16.2.5 Tissue Phantom Ratio and Tissue Maximum Ratio 261
16.2.6 Beam Dose Profiles and Off-Axis Ratios 262
16.3 Monitor Unit Calculations 262
16.3.1 SSD Calculations 262
16.3.2 SAD Calculations 263
16.4 Absolute Dosimetry Protocol: AAPM TG-51 263
16.4.1 Absorbed-Dose Calibration Factor N DQ , w 264
16.4.2 Charge Measurements Corrections 264
16.5 Summary 265
References 265
16.1 INTRODUCTION frequently dependent on manual dose calculations. It is

The primary aim of a course of radiation therapy is to cure largely not possible to directly measure doses in patients
or control disease while minimizing the damage to nor- and physical measurements in phantoms are the founda-
mal tissues. The response of tumors and normal tissues tion of all dose calculation approaches. Phantom-based
to radiation is highly variable and dose response curves beam characterization of linear accelerators (linacs) is a
can be quite steep. As a result, the optimal window for staple of Medical Physics practice and forms the basis of
radiation treatment is quite small and imposes a strong dosimetric quantities that characterize how dose changes
requisite for accurate and consistent radiation dose calcu- with depth, along the beam profile, and with the use of
lation and delivery. Modern radiotherapy techniques rely beam modifiers.
on computational algorithms using a treatment planning In the early years of linac manufacturing each
system (TPS) to calculate dose, and becoming much less machine possessed unique characteristics that made
255
it imperative to carefully characterize each linac sepa- of the field also results in the radial inverse-square law
rately, with reference data provided as a guide for data where the photon fluence decreases as 1/r 2 as the dis-
comparison. Modern linacs are manufactured to tight tance from the source increases.
specifications with highly reproducible dose character- The dose to any point within a phantom or patient
istics for machines of the same model and energy [1], contains two main components that determine the total
and manufacturer provided reference data is now rou- dose: the primary component and the scatter compo-
tinely used for TPS beam modeling with beam-matched nent. The primary photon component originates at the
machines sharing the same model [2–5]. The output source and is composed of the original photons that
characteristics of a linac are typically defined relative were generated at the target. The primary component
to an absolute point where the output of the machine is of dose is mainly dependent on the photon energy and
calibrated. The distinction between absolute and relative the depth within the phantom, and it is independent of
dosimetry should be made in order to understand the field size. The scatter component is the result of photons
different dosimetric quantities used to define the output scatter within the linac head and the phantom and can
of a linac. be further divided into a collimator scatter component
Absolute dosimetry protocols like AAPM Task Group and a phantom scatter component. The collimator scat-
51 (TG-51) [6] and IAEA Technical Report Series TRS- ter originates from photon scatter within the collima-
398 [7] carefully define very specific reference condi- tor, flattening filter, monitor unit (MU) chamber, and
tions to define the absolute dose or dose rate to a point. air. The phantom scatter component is the result of
These reference conditions are important in order to scatter within the phantom. The notion of a purely pri-
maintain charged particle equilibrium (CPE) and define mary dose component is mainly a conceptual one as it is
measurement conditions that are ideal for the accurate impossible to measure directly as collimator scatter dose
measurement of dose at a clinically relevant depth. The is always present. Similarly, phantom scatter only dose
dose characteristics throughout the rest of the patient is also impossible to measure directly and is a derived
will change significantly as the beam is attenuated and quantity from other measurements.
the points get farther away from the source. There are
also a wide range of clinical conditions encountered that 16.2.1 Collimator Scatter Factor
make it impossible to always define dose or dose rate in The beam output (dose rate in free space) measured
absolute terms. Relative external beam dosimetric func- in air depends on the field because as the field size is
tions characterize the output in relative terms that scale increased the collimator scatter increases. The scatter
the calibrated absolute dose under reference conditions component of dose that originates on the linac head and
to accommodate real clinical conditions. In this chapter, changes mostly with different collimator jaw settings
the general dosimetric functions of megavoltage beams can be quantified with a collimator scatter factor ( Sc )
are described, general beam parameters are defined, and that is measured in air and in the absence of full scat-
the details of AAPM’s TG-51 are covered. ter conditions to avoid the phantom scatter component.
The concept of collimator scatter factor is also called
16.2 PHOTON BEAM DOSIMETRY an in-air output ratio and was introduced to charac-
Megavoltage beams are characterized for their ability to terize how the incident photon fluence per MU changes
spare the skin and treat deep lesions by using accelerat- with jaw settings [8]. Figure 16.1 shows the position of
ing potentials that are in the megavoltage range which the MU chambers are above both the upper and lower
is now commonly 6–18 MV. This energy requirement is collimator jaws. This results scattered photons from the
most commonly produced in a linear accelerator that has jaws affecting the charge reading within the MU cham-
the general design characteristics similar to those shown bers and nominally changing the amount dose that is
in Figure 16.1. A continuous spectrum of megavoltage deposited per MU. This is the primary effect being char-
photons is produced at the target after pulsed electrons acterized by the collimator scatter factor and important
are accelerated to the desired energy in the accelerat- parameter that directly changes machine output.
ing waveguide. The point-like source of photons from The collimator scatter factor is defined in AAPM
the target produces a diverging field of photons that is TG-74 [8] as the ratio in free-space of primary collision
collimated within the linac head. This diverging nature KERMA (Kinetic Energy Released in Matter) between a
Photon Beam Dosimetry of Conventional Medical Linear Accelerators ◾ 257
FIGURE 16.1 General schematic of a modern clinical linear accelerator. Dose contributions from different components at any
given point within a patient or phantom.
given field size and a reference field size. This is the ratio radiological dimensions that approximate CPE are
of the output in air for a given field size and a reference also acceptable. However, electron contamination and
field size typically chosen to be 10 × 10 cm2. The ratio the alteration of the incident spectrum may introduce
can be defined in terms of dose to free space D fs defined errors. Therefore, the use of a miniphantom is recom-
as the dose to equilibrium mass of water measured in air mended with dimensions that allow for lateral and
under CPE conditions: longitudinal CPE but also maintain a position for the
ion chamber to be deep enough to filter out electron
D fs ( r ) contamination [9]. The values of both D fs and Sc both
Sc ( FS ) = (16.1)
D fs (rref ) increase with r because as the jaws open more fluence is
seen by the detector and less photons scatter back into
where r is the given field size and rref is the reference field the MU chambers. Typical measured values of collima-
size typically chosen to be 10 × 10 cm2. The values are usu- tor scatter factors range between 0.90 and 1.05 for most
ally measured for a wide range of square fields normal- clinical megavoltage beams with values below 1.0 for
ized to 1.0 for the reference field size. The measurement is field sizes below the reference field size (10 × 10 cm2) and
typically done using an ion chamber with a buildup cap above 1.0 for field sizes above. The backscatter from the
that is large enough to attain CPE (Figure 16.2). jaws is dependent on how close each pair of jaws is to
In high-energy applications water equivalent buildup the MU chambers so the amount of backscatter differs
caps can be become prohibitively large for even mod- from lower and upper jaws and results in the so called
erately small field sizes so a high-Z buildup cap with collimator exchange effect where Sc ( a, b ) ≠ Sc (b, a ) for
FIGURE 16.2 (a) Collimator scatter factor measurement

setup using a miniphantom and ion chamber setup. (b) Brass
buildup cap to allow for CPE and the measurement of smaller
fields in high-energy applications. (c) Miniphantom and ion
chamber setup using a vertical ion chamber alignment.
rectangular fields if the upper and lower jaw settings are FIGURE 16.3 Total output factor measurement setup, the
inverted. The consequence of this is that the machine depth of measurement is chosen to be deep enough to avoid
output for rectangular fields or blocked fields is not electron contamination and usually 5 cm for lower energies
always the same as that of its equivalent square fields like 6 MV and 10 MV, and 10 cm for higher energies like
and can differ by up to 2%. 15 MV and 18 MV. The phantom should be large enough to
approximate full scatter conditions and capture all relevant
16.2.2 Phantom Scatter Factor phantom scatter.
The scatter component of dose that originates in the

patient itself due to scattered photons can be quantified ( Sc , p ) that contains both the collimator and phantom
with a phantom scatter factor ( S p ), and defined as the scatter components, and dividing out the collimator
ratio of the output in water-equivalent media at a ref- scatter factor (Sc):
erence depth for a given field size to the same point at
the same depth for a reference field size (typically cho- Sc , p ( r )
S p ( FS ) = (16.2)
sen to be 10 × 10 cm2) and the same collimator settings Sc (r )
such that to capture phantom scatter changes alone.
This definition may be conceptually convenient in order where r is the given field size and the measurement
to separate collimator and phantom scatter but highly of Sc , p ( r ) is taken as the ratio of output for a refer-
impractical to measure. Under this definition the only ence field size taken at the same depth for a reference
way to measure this factor is using a very large collima- field size typically chosen to be 10 × 10 cm2. The mea-
tor setting (e.g., 40 × 40 cm2) and changing the size of surement is taken using a source-to-axis (SAD) setup
the phantom to adjust the field size that irradiates the where the measurement is taken at the machine iso-
phantom [10] (Figure 16.3). center at a depth that is deep enough to avoid electron
The most common way to determine phantom scatter contamination. For 6 MV and 10 MV beams the depth
factors is indirectly by measuring a total output factor is typically 5 cm whereas for higher energy beams like
16.2.3 Percent Depth Dose Curve

The quantities defined so far characterize different com-
ponents of the dose and how it changes with collimator
settings. The way that the dose changes with depth can
be addressed with several different functions and the
most complete and intuitive to understand is the PDD
curve. PDD is defined as the percentage of the maxi-
mum dose at a depth dmax as:
D (d )
PDD(d , r ) = × 100 (16.3)
D(dmax )
Figure 16.5 shows a representative depth dose curve and

how dose changes with depth for a 6 MV beam enter-
ing a phantom. PDD exhibits a dose buildup region
( ) for a 6 MV beam
FIGURE 16.4 Total output factor Sc , p where the dose increases monotonically until it reaches
measured at 5 cm depth and 95 cm SSD. The output factor is a maximum at dmax. As photons enter the patient or
normalized for a 10 × 10 cm2 and monotonically increases or phantom, high-speed electrons are ejected from the sur-
decreases with a change in field size due to a change in scat- face and subsequent layers. These electrons can deposit
tering conditions.
their energy a significant distance away from their site
of origin. This means that the electron fluence, and
hence absorbed dose, increase with depth until the ris-
15 MV and 18 MV beams a 10 cm depth is more appro- ing slope due to buildup of charged particles is balanced
priate. Figure 16.4 shows the monotonically increasing by the descending slope due to the attenuation of the
function of total output factor for a 6 MV photon beam photon beam. At greater depths, CPE is attained and the
measured at 5 cm depth at 95 cm source-to-surface dis- absorbed dose decreases as the photon beam is attenu-
tance (SSD). The total output changes rapidly for small ated. Beyond the depth of maximum dose, PDD values
fields starting out at around 0.765 for a 1 × 1 cm2 field increase with photon energy because higher energy
and quickly increasing to its normalized value of 1.00 beams have greater penetrating power. The field size
for a 10 × 10 cm2 field. For larger fields, the total output dependence of PDD can be attributed to the increase in
factor changes much more slowly and reaching a maxi- scattered radiation that contributes to the absorbed dose.
mum of 1.119 for a 40 × 40 cm2 field. The measurement Because the scattered dose is greater at larger depths
of output factors for small fields where there is a lack than at dmax, the PDD increases with increasing field size.
of lateral CPE (~3 × 3 cm2) is much more challenging The dependence on SSD of PDD is a direct result of the
than those for larger fields due not only to the lack of inverse square dependence of photon fluence emitted by
lateral CPE but also to the detector volume averag- a radiation source. In addition to the inverse square law,
ing, high dose gradients and source occlusions. Large PDD variation with depth is also governed by exponen-
variations in measured output of over 10% have been tial attenuation and scattering. Therefore, the variation
measured for small field and as such great care must of the PDD with depth is given by:
be taken when measuring small field outputs [11,12].
2
The field size where this becomes an issue depends on  SSD + dmax  − µ (d − dmax )
PDD ( d , r , SSD ) ~  e (16.4)
the photon beam energy but typically understood to  SSD + d 
be below ~3 × 3 cm2 or when the detector volume is
comparable to the field size. Detailed recommenda- where µ is the linear attenuation coefficient and shows
tions and measurement were made by the joint IAEA/ the exponential attenuation of the beam beyond dmax,
AAPM working group on small field dosimetry and and the dependence on SSD [10].
published on the IAEA TRS-483 [13]. The formalism is At beam entrance, kinetic energy is released in the
based on the published technique by Alfonso et al. [14]. medium as secondary electrons are released and as such
FIGURE 16.5 Percent depth dose curve for a typical 6 MV photon beam in water equivalent media. The shaded area repre-
sents the buildup region characteristic of indirectly ionizing radiation where secondary charge particles build up to deposit
dose. KERMA decays exponentially as the beam fluence attenuates within the medium and the two decreases by the same rate.
KERMA is at a maximum. The beam fluence attenuates depths where the PDD and KERMA curves remain
exponentially with depth and KERMA decreases by the mostly parallel the absorbed dose is proportional to
same rate. On the other hand, the dose absorbed in the KERMA under transient charged particle equilibrium
material behaves differently due to the indirectly ionizing (TCPE). The distance between the PDD and KERMA
nature of photon beams. At the surface, the dose from a curves beyond dmax represents the average distance that
megavoltage beam is very low and would be almost zero secondary electrons travel through the medium and
if it weren’t for electron contamination from the head. denoted Rave in Figure 16.5 [15].
This low entrance dose gives high-energy megavoltage
beams its skin sparing property that is typically below 16.2.4 Tissue Air Ratio
50% under most conditions. As secondary charged par- One of the oldest functions used to characterize how
ticles are released in the material, the absorbed dose dose changes with depth is the tissue-air ratio (TAR).
increases and the dose reaches a dose maximum (Dmax ) TAR was first introduced in 1953 to simplify dose calcu-
where the buildup of secondary charged particles is lation for rotational therapy in orthovoltage beams [16].
countered by the attenuation of the beam. For the 6 MV TAR can be defined as the ratio of dose to some depth
beam shown in Figure 16.5, this depth of maximum dose d in water-equivalent tissue to the dose in free space D fs
dmax occurs at around 1.5 cm. A few centimeters beyond for the same field size under SAD conditions.
Dmax the PDD and KERMA curves become parallel and
this occurs at a depth equal to the maximum range Rmax D ( d , rd )
TAR ( FS ) = (16.5)
of secondary electrons released in the medium. In the D fs (rd )
case of 6 MV beam like the one shown in Figure 16.5
this happens at around 3.0 cm for the maximum range where rd is the field size at depth. The quantity is con-
of a 6 MeV electron being released through photoelec- venient because it is measured under SAD conditions
tric interaction from the low yield of 6 MeV photons and not dependent on the SSD. However, similar to the
present in the 6 MV photon spectrum. At these deeper collimator scatter factor it is difficult to measure D fs
particularly for higher energies because of lack of CPE because the scatter contribution is not dependent on
for small fields and the added uncertainty of using the divergence on the beam, and only of the actual
high-Z buildup caps and miniphantoms. Therefore, field size at depth. This is a main difference with PDD
TARs are typically more useful for low-energy beams measurements that depend on the field size at the sur-
with an isocentric treatment geometry. TAR increases face. This behavior makes TMR a convenient choice
with field size because the increased phantom scatter for MU calculations because blocked field size don’t
component of the dose measurement at depth increases need to be corrected for divergence at depth. TPR and
faster than the increased collimator scatter component TMR decrease with depth as the beam is attenuated
of the dose in a free space measurement. through tissue, and increases with field size because
the increased phantom scatter component of the dose
16.2.5 Tissue Phantom Ratio and Tissue at depth increases (Figure 16.6).
Maximum Ratio TPR and TMR can be cumbersome to measure
Tissue phantom ratios (TPRs) overcome the limita- directly given the requirement to maintain an iso-
tions of the TAR for high-energy photon measure- centric geometric with increasing depth. This can be
ment. TPR is defined in an isocentric geometry as the accomplished in a water tank that can be raised while
ratio of dose at a given depth d and field size at depth r, simultaneously moving the ion chamber deeper into
to the dose at a reference depth dref and reference field the phantom to maintain an isocentric geometry. Other
size at depth r: techniques may include using a solid water phantom
and adding different layers of solid water to increase the
D (d , r ) depth of measurement. Modern scanning tank equip-
TPR ( FS ) = (16.6)
D(dref , r ) ment may include software and specially designed tank
platforms that can make TPR and TMR measurements
The special case where dref = dmax give rise to the tis- much more streamlined. TPR and TMR data are also
sue maximum ratio (TMR). Isocentric measurements commonly derived from PDD measurements correcting
like TAR and TMR are insensitive to changes in SSD for divergence and phantom scatter.
FIGURE 16.6 Isocentric measurement geometry for TPR and TMR measurements in a water equivalent phantom. The refer-
ence depth dref is taken to be dmax for TMR measurements.
FIGURE 16.7 A 6 MV beam dose profile at 10 cm depth for a 10 × 10 cm2 field size at the surface. The beam regions of a
megavoltage beam consist of the in-field region, the penumbra and the umbra. The typical width of the penumbral region for a
6 MV field is on the order of 6 mm. It should be noted flattening filters are typically optimized to flatten the field at 10 cm
depth like the profile shown.
16.2.6 Beam Dose Profiles and Off-Axis Ratios FWHM. Flattening filters are typically designed and
All quantities defined up to this point deal with how optimized to deliver a flat field at 10 cm depth. At shal-
dose changes with jaw collimator settings or with depth lower depths beam profiles may exhibit dose horns where
on the central axis (CAX). Dose variation away from the the dose off axis is higher than that at the CAX. This effect
CAX is characterized with dose profile measurements is more pronounced for higher energies at dmax. Typical
and off-axis output factors ratios (OARs). Dose profiles flatness values for modern linacs should stay below 3%
are an important part of TPS beam model commission- and typically below 2% for flattened beams. Symmetry
ing and validation, and characterize all off-axis aspects is another important beam parameter that reflects the
of the beam with respect to the CAX values like PDD beam symmetry by comparing the area under to profile
and output factors. The beam profile consists of three for each side of the curve. This quantity can typically
main regions: the in-field region, the penumbra, and the be directly measured using most modern scanning tank
umbra. The central region typically refers to the region software. Typical symmetry values for most modern lin-
of the field that is within 80% of the beam width, i.e., for acs are below 2% but routinely stay below 1%.
a 10 × 10 cm2 field the central 8 × 8 cm2 is considered the
central region (Figure 16.7). AreaL − AreaR
S= ×100 (16.8)
AreaL + AreaR
Dmax − Dmin
F= ×100 (16.7)
Dmax + Dmin
16.3 MONITOR UNIT CALCULATIONS
Field flatness and symmetry are two important param- 16.3.1 SSD Calculations
eters that are good indicators of machine performance Machines are usually calibrated to deliver 1 cGy/MU at
and quality assurance metrics. The flatness (F) is defined dmax for a reference field 10 × 10 cm2 at 100 SSD. Taking
as a percentage difference of the maximum (Dmax ) and into account all correction factors in an SSD setup, the
minimum (Dmin ) values across a profile within the cen- number of monitor units needed for a specific dose is
tral region of the field, defined as the inner 80% of the given by:
Dose 16.3.2 SAD Calculations

MU = (16.9)
PDD In an SAD setup there is no dependence on SSD and
D0 ·Sc ·S p ·Inv 2 · ·F ·TF ·WF ·OAF
100 hence the Mayneord F factor is not needed. As such the
number of monitor units needed for a specific dose is
where D0 = 1 cGy/MU, Inv 2 is the inverse square factor, given by:
F is the Mayneord F factor, TF is the tray factor, WF is
the wedge factor and OAF is the of axis factor. Any other Dose
correction factors can be included here too. It should be MU = (16.13)
D0 ·Sc ·S p ·Inv ·TMR·TF ·WF ·OAF
2
noted that the Mayneord F factor is needed to correct for
any changes in SSD. Using Equation 16.4 one can deter- The Inv 2 factor is still needed to correct the output fac-
mine how to correct the percent depth does for a change tor, therefore for an SAD setup the Inv 2 is given by:
in SSDs. Taking the ratio of PDD at two different SSDs
we can obtain the following expression:  100 
2
Inv 2 =  (16.14)
 SSD + d 
2 2
PDD(d , SSD2 )  SSD2 + dmax   SSD1 + d 
= (16.10)
PDD(d , SSD1 )  SSD1 + dmax   SSD2 + d  The collimator and phantom scatter factors Sc and S p
are determined in a similar fashion to MU calculations
The factor on the right-hand side is the Mayneord F fac- using PDDs and an SSD geometry. However, the isocen-
tor and corrects the PDD for a change in SSD. It should tric geometry means that the phantom scatter factor S p
be noted that the Mayneord F factor works reasonably is determined at depth and highlights the convenience
well for small field sizes since scattering is minimal of an isocentric geometry that does not dependent to
under such conditions. However, it overestimates the SSD and does not require further correction from the
change in PDD for large field sizes and very large SSD. Mayneord F factor.
This dependence on SSD makes PDD an inconvenient
choice in the clinic because corrections for SSD changes 16.4 ABSOLUTE DOSIMETRY PROTOCOL:
AAPM TG-51
can become quite cumbersome to apply routinely in a
clinical setting. The Inv 2 is needed to correct the dose AAPM’s TG-51 clinical reference dosimetry protocol
rate, hence the output factor, due to the change in flu- was prescribed for photon beams with nominal energies
ence. Therefore, for an SSD setup: between 60Co and 50 MV, and electron beams between
4 and 50 MeV. This is the third generation of such a
2 protocol and was introduced in 1999 replacing TG-21
 100 + dmax  protocols. TG-21 protocols improved errors in beam
Inv = 
2 (16.11)
 SSD + dmax  calibrations of up to 5% at the expense of much more
2 complex calculations which represent an increased
 SSD + dmax   100 + d  2 potential for errors in the clinic. A major advantage in
F =   (16.12)
 100 + dmax   SSD + d  TG-51 is that it requires absorbed dose to water calibra-
tion factors, making it simpler to implement and under-
The collimator and phantom scatter factors Sc and S p stand than earlier protocols. Standards for absorbed
take into account different aspects of scatter and are dose to water have an uncertainty of less than 1% in 60Co
handled differently for SSD MU calculations using and bremsstrahlung beams up to 25 MV. TG-51 takes
PDD. The collimator scatter factor Sc is determined advantage of this improved accuracy as it is based on
using the jaw collimator setting, while the phantom the use of ion chambers calibrated in terms of absorbed
scatter factor S p is determined by the portion of the dose to water in a 60Co beam. TG-51 protocols must be
field that contributes scatter dose to the point of inter- performed in a water phantom to ensure simplicity and
est or the blocked field size at the surface. This can dif- accuracy because the quantity of interest is absorbed
fer from the jaw delimited field based on SSD, machine dose to water.
design, the use of multi-leaf collimators, or other types The main purpose of the TG-51 protocol is to ensure
of photon blocks. uniformity of clinical reference dosimetry in external
radiation therapy with high-energy photons and elec- forgo the use of a lead foil for flattened photon beams to
trons. This is accomplished with a common starting reduce the potential for error.
point of an ion chamber calibration factor, which is For 60Co beams, %dd (10 )X is not needed as kQ = 1.000
directly traceable to national standards of absorbed dose by definition. In order to obtain the appropriate
to water maintained by Primary Standards Laboratories %dd (10 )X , the point of measurement needs to be cor-
(National Institute of Standards and Technology, NIST). rected for gradient effects and effective depth of mea-
This traceability is maintained via calibration factors surement. The point of measurement for a cylindrical
obtained from an Accredited Dosimetry Calibration chamber is on the CAX of the chamber and is always
Laboratory (ADCL). Furthermore, all data used in the placed at the reference depth when measuring dose at
TG-51 protocols apply only under certain, well-defined an individual point. The effective point of measurement
reference conditions and are specified individually is upstream of the point of measurement as second-
where it is relevant. These conditions include depth of ary electrons are predominantly forward directed. As
measurement, field size, SSD, and number of MUs. a result, depth-dose data with a cylindrical chamber is
The absorbed dose to water under reference condi- shifted to shallower depths by a distance proportional
tions is given by: to the radius of the cavity rcav . For photon beams and
cylindrical or spherical chambers, the shift is taken as
DwQ = MN DQ,w (16.15) 0.6rcav . Once the appropriate %dd (10 )X is obtained, kQ
can be extrapolated for range of %dd (10 )X values from
where DwQ is the absorbed dose to water in J/kg (Gy) at data available from AAPM’s TG-51 or equivalent pub-
the point of measurement of the ion chamber when it lished data.
is absent, M is the fully corrected electrometer reading
in coulombs (C) and N DQ,w is the absorbed-dose to water 16.4.2 Charge Measurements Corrections
calibration factor in a beam of quality Q. The electrometer reading has to be fully corrected for
ion recombination, polarity, and electrometer calibra-
16.4.1 Absorbed-Dose Calibration Factor N DQ,w tion effects, and also corrected to standard environmen-
The absorbed-dose calibration factor is obtained for ref- tal conditions of temperature and pressure and given by:
erence conditions in a 60Co beam and denoted N DQ,w . In
order to convert the absorbed-dose calibration factor for M = Pion PTP Pelec Ppol M raw (C ) (16.17)
a 60Co beam into the calibration factor for an arbitrary
beam of quality Q, it is multiplied by the quality conver- where M raw is the raw ion chamber reading in coulombs
sion factor kQ such that: (C ), PTP is the temperature-pressure correction, Pion cor-
rects for incomplete ion collection efficiency, Ppol cor-
N DQ,w = kQ N D60,Co
w (Gy /C ) (16.16) rects for any polarity effects, and Pelec takes into account
the electrometer’s calibration factor if the electrometer
This quality conversion factor, kQ is chamber specific and and ion chamber are calibrated separately.
a function of the photon component of the photon beam Polarity effects vary with beam quality and a variety
percentage-depth dose at 10 cm in a 10 × 10 cm 2 field on of conditions such as cable position. Thus it is necessary
the surface of a water phantom at an SSD of 100 cm or to correct for these effects by making measurements
%dd (10 )X . This notation specifies that this is the contri- each time clinical reference dosimetry is performed. To
butions due to photons only and any electron contami- calculate Ppol , an ion chamber’s raw reading is taken for
nation should be removed when applicable (>10 MV). both polarities applied and calculated as:
It has been shown that electron contamination can be
reduced to a negligible level by placing a 1 mm thick lead + − M−
Mraw raw
Ppol = (16.18)
foil below the accelerator head to determine %dd (10 )X 2 M raw
only. The foil should be removed for all other dose mea-
surement steps which can lead to potential confusion Typical values for Ppol should be below 1.003 for 6 MV
and potential operator error. Therefore, the latest rec- and can be a good indicator to verify the charge collec-
ommendations from the addendum to TG-51 [17] are to tion setup is working properly.
The electrometer correction factor is obtained from consistency of modern linacs, and the availability of
an ADCL and Pelec = 1.000 if the electrometer and ion reliable reference datasets. The understanding of how
chamber are calibrated as a unit. dose behaves inside the field and how it changes with
Calibration factors are always given in standard envi- different machine parameters is a fundamental skill to
ronmental condition of temperature at To = 22°C and the field of Medical Physics. The relationship between
pressure of Po = 760 mmHg thus electrometer readings dose and different collimator settings as well as the basic
must be corrected to standard environmental condi- irradiation conditions that are required to measure
tions using: these quantities were covered in this chapter. The basics
of hand calculated MU as well as the standard reference
273.2 + T 760 mmHg dosimetry protocol from AAPM was also covered.
PTP = × (16.19)
273.2 + 22 P
REFERENCES
where T is the ambient temperature in degrees C in the
water near the ion chamber and P is the pressure in 1. R. J. Watts, “Comparative measurements on a series
of accelerators by the same vendor,” Medical Physics
mmHg. The correction means that for PTP to change by
26(12), 2581–2585 (1999).
1%, the temperature must change by 3°C and pressure 2. D. Sjöström, U. Bjelkengren, W. Ottosson, C. F. Behrens,
must change by 7.6 mmHg. Humidity is assumed to be “A beam-matching concept for medical linear accelera-
within 20–80% for the temperature and pressure cor- tors,” Acta Oncologica 48(2), 192–200 (2009).
rection to be within 0.15%. 3. Z. Chang, Q. Wu, J. Adamson, et al., “Commissioning
The standard two-voltage technique is used to deter- and dosimetric characteristics of TrueBeam system:
Composite data of three TrueBeam machines,” Medical
mine the ion recombination correction Pion and involves Physics 39(11), 6981–7018 (2012).
measuring the charge produced by the ion chamber 4. C. Glide-Hurst, M. Bellon, R. Foster, et al.,
when two different bias voltages are applied to the “Commissioning of the Varian TrueBeam linear acceler-
detector. It should be noted that once voltage has been ator: A multi-institutional study,” Medical Physics 40(3),
changed, it is necessary to wait for the chamber read- 031719 (2013).
5. M. C. Glenn, C. B. Peterson, D. S. Followill, R. M.
ings to come to equilibrium. For pulsed or pulsed-swept Howell, J. M. Pollard-Larkin, S. F. Kry, “Reference data-
beams, Pion is calculated as set of users’ photon beam modeling parameters for the
Eclipse, Pinnacle, and RayStation treatment planning
VH systems,” Medical Physics 47(1), 282–288 (2020).
1−
V 6. P. R. Almond, P. J. Biggs, B. M. Coursey, et al., “AAPM’s
Pion (VH ) = H L (16.20) TG-51 protocol for clinical reference dosimetry of high-
M raw VH
− energy photon and electron beams,” Medical Physics
M raw
L VL 26(9), 1847–1870 (1999).
7. Absorbed Dose Determination in External Beam
It must be emphasized that Pion is a function of dose per Radiotherapy (International Atomic Energy Agency,
pulse in accelerator beams and will change if either the Vienna, 2001).
pulse rate for a fixed dose rate or the dose rate is changed. 8. T. C. Zhu, A. Ahnesjö, K. L. Lam, et al., “Report of
AAPM Therapy Physics Committee Task Group 74:
A properly functioning ion chamber and charge collection
In-air output ratio, for megavoltage photon beams,”
system should exhibit a correction factor Pion below 1.05. Medical Physics 36(11), 5261–5291 (2009).
Taking into account all conversion and correction 9. J. J. M. van Gasteren, S. Heukelom, H. J. van Kleffens,
factors, a raw electrometer reading M yields an absorbed R. van der Laarse, J. L. M. Venselaar, C. F. Westermann,
dose to water as “The determination of phantom and collimator scat-
ter components of the output of megavoltage photon
P P P P k N
60 Co beams: Measurement of the collimator scatter part
M raw 
ion TP elec pol
→ M 
Q D ,w
→ DwQ (16.21) with a beam-coaxial narrow cylindrical phantom,”
Radiotherapy and Oncology 20(4), 250–257 (1991).
16.5 SUMMARY 10. F. M. Khan and J. P. Gibbons, Khan’s the Physics of
Radiation Therapy (Wolters Kluwer Health, 2014).
The characterization of photon beam output from high- 11. I. J. Das, G. X. Ding, A. Ahnesjö, “Small fields:
energy linacs remains an important part of the field Nonequilibrium radiation dosimetry,” Medical Physics
of Radiation Therapy Physics despite the stability and 35(1), 206–215 (2008).
12. I. J. Das, M. B. Downes, A. Kassaee, Z. Tochner, “Choice 14. R. Alfonso, P. Andreo, R. Capote, et al., “A new formal-
of radiation detector in dosimetry of stereotactic radio- ism for reference dosimetry of small and nonstandard
surgery-radiotherapy,” Journal of Radiosurgery 3(4), fields,” Medical Physics 35(11), 5179–5186 (2008).
177–186 (2000). 15. F. H. Attix, Introduction to Radiological Physics and
13. H. Palmans, P. Andreo, M. S. Huq, J. Seuntjens, K. E. Radiation Dosimetry (Wiley, 2008).
Christaki, A. Meghzifene, “Dosimetry of small static 16. H. E. Johns and J.R. Cunningham, The Physics of
fields used in external photon beam radiotherapy: Radiology (Charles C. Thomas, 1983).
Summary of TRS-483, the IAEA–AAPM interna- 17. M. McEwen, L. DeWerd, G. Ibbott, et al., “Addendum
tional Code of Practice for reference and relative dose to the AAPM’s TG-51 protocol for clinical reference
determination,” Medical Physics 45(11), e1123–e1145 dosimetry of high-energy photon beams,” Medical
(2018). Physics 41(4) (2014).
Chapter 17
Dosimetric Considerations with

Flattening Filter-Free Beams
Jessica Lye
Australian Radiation Protection and Nuclear Safety Agency
Olivia Newton-John Cancer Wellness & Research Centre, Austin Health
Melbourne, Australia
Stephen F. Kry
Imaging and Radiation Oncology Core
The University of Texas MD Anderson Cancer Center
Houston, Texas
Joerg Lehmann
Calvary Mater Newcastle
University of Newcastle
Newcastle, NSW, Australia
University of Sydney
Sydney, NSW, Australia
CONTENTS
17.2 Characteristics of FFF Beams 268
17.2.1 Beam Shape 268
17.2.2 Dose Rate 269
17.2.3 Beam Quality 269
17.3 Reference Dosimetry in FFF Beams 270
17.3.1 Beam Non-Uniformity 270
17.3.2 Recombination 270
17.3.3 Energy Spectrum and Corrections with FFF Beams 271
17.4 Routine Clinical Measurements 271
17.4.1 Profile Shape 271
17.4.2 Dose Rate and Recombination 271
17.4.3 Energy Dependence 272
267
17.5 Other Clinical Considerations 273

17.5.1 Small Fields 273
17.5.2 Motion and Interplay Effects 273
17.5.3 Electron and Neutron Contamination 273
17.6 Summary 273
References 274
17.1 INTRODUCTION the forward-peaked shape is largely constant with depth

Flattening filter-free (FFF) beams have been used in because the spectrum is approximately constant across
clinical radiation therapy in specialized treatment the field [3, 4]. This is in contrast to flattened beams
machines such as Tomotherapy and Cyberknife for where the flattening filter differentially hardens the
many years. Now standard linear accelerators (linacs) beam which causes the shape of a flattened beam pro-
can include FFF beams as well as flattened beams. The file to change with depth (producing horns at shallow
new FFF beams are sufficiently different from conven- depths and rounded profile shoulders at depth).
tional flattened beams to require special considerations The issue of beam shape is relevant to general dosim-
and occasionally unique corrections [1, 2]. In particular, etry discussions because, in static fields, it means that
FFF beams are spatially non-uniform, have an elevated there is always a degree of dose gradient. The inherent
dose rate, and a larger spectral spread with a greater field gradient is small (particularly for the low FFF beam
lower energy component than conventional flattened energies most commonly available in clinical practice:
beams. These issues, described below, require attention 6 and 10 MV), so it is only relevant for relatively large
during dosimetric measurements. detectors. The gradient is also relatively linear so that
for many applications the non-uniformity will simply
17.2 CHARACTERISTICS OF FFF BEAMS average-out and provide a reasonable dose estimate at
17.2.1 Beam Shape the center of the detector. Nevertheless, for applications
As shown in Figure 17.1, FFF beams are dosimetrically such as primary beam calibration, where the detector is
peaked toward the central axis. This arises because of large and the gradient is non-linear, the shape of the FFF
the forward-peaked nature of the bremsstrahlung gen- beam introduces a dosimetric perturbation to the mea-
erated distribution and the lack of a flattening filter to surement. A non-uniformity correction may be neces-
force the distribution to be spatially uniform. However, sary to correct such conditions.
FIGURE 17.1 Typical profile of 10 × 10 cm2 FFF beams at 10 cm depth.

Dosimetric Considerations with Flattening Filter-Free Beams ◾ 269
17.2.2 Dose Rate 17.2.3 Beam Quality

The flattening filter absorbs more than half of the pho- Removing the flattening filter changes the energy spec-
tons generated in the bremsstrahlung target. Removal of trum of a beam compared to a flattened beam as the
the flattening filter therefore dramatically affects the dose filter preferentially removes lower energy photons. For
rate. It is important to differentiate changes in the instan- the beams discussed here, the flattening filter has been
taneous dose rate (also known as the dose per pulse) from replaced with a thin sheet of metal in the path of the
changes in the dose rate that manifest over a longer time beam, so some beam hardening occurs [1]. For Varian
period. An example of the latter is changing the pulse FFF beams the electron energy is not adjusted with the
repetition rate (rep rate) on the linac; because most linacs removal of the flattening filter and the effective energy
change rep rates by pulse dropping (with pulse shaping of the FFF beams are considerably lower than those
playing a relatively minor role). Changing the rep rate of the flattened beams with the same nominal energy.
affects the number of pulses being delivered, not the dose For Elekta FFF beams the electron energy is adjusted
in each pulse and therefore not the instantaneous dose so the effective energy and beam quality specifier
rate. Removal of the flattening filter, on the other hand, (TPR 20,10/%dd(10)x) remains similar. This effect, and the
increases the instantaneous dose rate as compared to magnitude of it, is apparent in Figure 17.2, which shows
flattened beams. This distinction can be important for the %dd(10)x specifier for 6 and 10 MV beams of Varian
dosimetry applications where dose-rate effects manifest. and Elekta accelerators that were equipped with both
Ion chambers, for example, have a signal collection time FFF and flattened beams. This data was self-reported
on the order of micro-seconds. The dose per pulse thereby each institution to IROC in the context of output
fore affects recombination, but the number/frequency of verification.
pulses (e.g., through changing the rep rate) does not.
FIGURE 17.2 Beam quality (%dd(10)x) for different clinical beams for Varian and Elekta accelerators for flattening filter-free
(FFF) beams and those with flattening filters (WFF).
FIGURE 17.3 In-air spectra from modelled Elekta linacs from a beam with the flattening filter (dashed line), and an FFF beam
(solid line). All spectra intensities have been normalized to 2 MeV, the approximate average energy.
Even in the case of the Elekta beams where the beam the correction. Even at that level, the noise will likely
quality specifier is matched between FFF and flat- require some smoothing before the calculation can be
tened beams, the spectrum itself is still quite different performed. Alternatively a curve can be fitted to the
between the two types of beams as shown in Figure 17.3. data and used for the calculation.
The FFF beams have a larger low energy component Correction factors have been reported to range from
and the maximum energy is higher in the FFF cases to about 0.4% for 6 MV to about 0.7% for 10 MV FFF beams
compensate for this low energy component. The differ- [5] for Farmer-type ion chambers with a length of ~2.3
ent spectral composition of FFF beams raises questions cm. This non-uniformity correction is large enough to
on the ability of a single beam quality specifier, such as warrant inclusion in the reference dosimetry, however
%dd(10)x to adequately define the energy of the beam the non-uniformity correction is insensitive to depth
and the related beam quality correction factor kQ in the and has no effect on the beam quality measurement [6].
context of dosimetry.
17.3.2 Recombination
17.3 REFERENCE DOSIMETRY IN FFF BEAMS
A substantial part of the appeal of FFF beams lies in
17.3.1 Beam Non-Uniformity
their higher dose rate compared to the same energy
In reference dosimetry with a 0.6 cm3 Farmer type ion- flattened beams. As described above, this corresponds
ization chamber, the pointed shape of the profile requires to an increase in the dose per pulse, unlike a change in
a geometric correction factor for optimal accuracy. This pulse repetition rate where the dose per pulse remains
need arises because the beam profile changes over the the same.
length of the collection volume, reducing in all directions The increased dose per pulse has potential dosimet-
from the maximum on the central axis. The non-unifor- ric consequences as it results in higher recombination
mity correction can be calculated from beam profiles. For for an ion chamber and therefore necessitates a larger
a Farmer-type chamber the correction may be approxi- recombination correction factor. This correction is typi-
mated by summing the profile over the long axis: cally around 1% at 10 cm depth [6, 7], but can approach
2% at dmax where the dose rate is higher [7].
D0
kn = To date, Farmer chambers evaluated in FFF beams
∑D i have been found to be “well behaved” in terms of recom-
bination. That is, the recombination can be described
where D0 is the central axis dose and Di is the value at accurately following the standard 2-voltage technique
each point i in the scan over the length of the chamber. which is recommended by calibration protocols such as
Depending on the chamber used, a profile with 1 mm Task Group 51 (TG-51) [8] and TRS-398 [9]. However, ion
resolution will suffice to provide reasonable accuracy on chambers do not always behave according to standard
recombination theory, and it is possible for the 2-voltage TPR 20,10 beam quality is there may be a small ~0.3% cor-
technique to calculate the recombination erroneously. rection recommended for FFF beams depending on the
As recommended in the TG-51 addendum [10], for any protocol followed in the user’s country [1, 8, 9].
substantially new beam (e.g., FFF) and a new type of ion
chamber, a clinic should (upon commissioning of the 17.4 ROUTINE CLINICAL MEASUREMENTS
system) verify that the ion chamber does in fact follow Routine clinical measurements, for example, beam scan-
recombination theory and can have its recombination ning, IMRT QA, etc. can be done with a variety of dif-
corrected using the standard 2-voltage technique. This ferent dosimeters, and measurements are taken at a wide
can be verified through the measurement of a Jaffé plot range of locations within the treatment field. These two
and looking for linearity in the relationship between 1/V issues (dosimeter and variable measurement location)
and 1/Q (for a pulsed beam) [11]. are discussed below in the context of the unique charac-
teristics of FFF beams as described in Section 17.2.
17.3.3 Energy Spectrum and Corrections
with FFF Beams 17.4.1 Profile Shape
Due to the softer beam spectrum, a different kQ is In general, beam non-uniformity corrections are not
expected for FFF beams than for flattened beams of needed when measurements are performed away from
the same nominal energy. The interesting question that the center of the beam as the approximately linear dose
has arisen is how to calculate kQ. Initial work into the gradient across the detector will cause the measured dose
effect of removing the flattening filter on kQ was based to accurately describe the dose to the center of the detec-
on Monte Carlo simulations. Xiong and Rogers [12] cal- tor volume. Even for measurements in the center of the
culated restricted stopping power ratios where flatten- beam, non-uniformity corrections are minimal for typi-
ing filters were completely removed and found that kQ cal detector sizes, being less than 0.1% for detectors with
factors determined for flattened and FFF beams could the longest dimensions less than 1 cm [9]. Figure 17.4
lead up to 0.2% errors when following AAPM TG-51 shows a Farmer 2571 and a CC13 chamber in scale to the
%dd(10)x or 0.7% when applying TPR 20,10. This study profile roll-off. Given that relative dosimetry is typically
suggested that TPR 20,10 was a less accurate beam quality conducted with such detectors (non-Farmer ion cham-
metric than %dd(10)x for FFF beams. A similar Monte bers, diodes, thermoluminescent dosimeters (TLD), opti-
Carlo study reported by Dalaryd et al. [13] was based on cally stimulated luminescent dosimeters (OSLD), and
more realistic clinical FFF beams where the flattening planar dosimeters such as film), no correction for profile
filter was replaced with a thin metal filter. Their study shape is generally warranted.
showed smaller differences for TPR 20,10 than predicted
by Xiong and Rogers and this was supported by Lye et al. 17.4.2 Dose Rate and Recombination
[6] who found differences smaller than 0.2% comparing The elevated dose rate associated with FFF beams leads
absorbed dose measurements based on TPR 20,10 (TRS- to increased recombination in ion chambers. This
398) or %dd(10)x (TG-51) for flattened and FFF beams at can reach almost 2% for some chambers [7]. Notably,
6 MV or 10 MV. because the dose rate changes as a function of depth,
The accuracy of kQ when following TRS-398 using recombination actually changes with depth as well; for a
TPR 20,10 for commercially available FFF beams was Farmer-chamber the recombination correction can vary
further confirmed by de Prez et al. [14] who performed by nearly 0.5% between dmax and 10 cm depth [7]. The
direct calorimetry measurements and concluded that recombination would similarly vary across the radiation
for farmer type chambers, differences less than 0.23% field as the dose rate decreases from a maximum on the
existed in kQ factors based on TPR 20,10 between the cor- central axis.
responding flattened and FFF beams. However, such a large recombination correction,
In short, kQ will be similar for an FFF beam and for and the corresponding sensitivity in recombination to
a flattened beam with the same %dd(10)x. Moreover, it the measurement location, is largely reserved for just
is acceptable to use either %dd(10)x or TPR 20,10 to deter- Farmer-type ion chambers. Smaller volume ion cham-
mine kQ according to the standard methodology used bers, as are more commonly used for relative dose mea-
for flattened beams. One consideration for users with surements, have less recombination because of their
FIGURE 17.4 Comparison of length of a farmer-type 2571 and CC13 thimbles (shown to scale) with respect to the profile roll-
off with a 10 FFF beam at 10 cm depth.
smaller volume. Recombination is therefore a smaller more prone to energy dependence effects. On the
concern, and correspondingly, recombination varies favorable side, FFF beams show a uniform spectrum
less with depth or lateral position. across the treatment field (unlike flattened beams), so
Other types of detectors typically exhibit smaller there is no variation in response within a given treat-
recombination effects than ion chambers. TLD and ment field.
OSLD have negligible dose-rate dependency over a Energy responses of different detectors have been
very wide range of dose rates [15]. Similarly, film has no studied at length, including for ion chambers, film,
measurable dose-rate dependence [16, 17]. No correc- TLD, OSLD, and diodes, although specific FFF versus
tion or special consideration is therefore warranted for flattened beam responses have received very little atten-
dose-rate effects for these detectors in FFF beams. On tion so far. Ion chambers show very little energy depen-
the other hand, diodes may show dose-rate dependence, dence except for chambers with steel central electrodes
which is the primary component of the “SSD depen- (typically microchambers) [19, 20]. Modern radiochro-
dence” of these detectors [18]. Consequently, it is not mic film shows minimal energy dependence across
possible to make universal statements about dose-rate megavoltage energies, and a flattened calibration would
effects. In general, if the dose-rate dependence for given generally suffice for an FFF beam [21–23]. Similarly,
detector is not well known, it is prudent to evaluate this LiF-based TLD (e.g., TLD-100) has an effective atomic
characteristic before using the detector in high dose rate number very close to water and therefore shows very
environments including FFF beams. The instantaneous little energy dependence throughout the MV energy
dose rate can be easily reduced by increasing the source- range. Because of this, it is unlikely that TLD would
to-detector distance or the depth of the measurement require unique calibration or corrections for mea-
in, for example, a water tank, to reduce the photon flux. surements in an FFF beam [15]. OSLD (Al2O3:C) has a
The measured dose can also be compared with the mea- slightly more distinct effective atomic number and may
sured dose of a dose-rate independent detector. Once show more variability to FFF beam energies. While a
the recombination effects are known, a clinical manage- difference in response of up to 1–2% may be expected
ment decision can be made (e.g., effects are corrected at different locations in the treatment field [24], both
for, or the effects are less than the uncertainty in the IROC and the Australian Clinical Dosimetry Service
measurements and are therefore neglected). (ACDS) use the same beam quality factor for flattened
and FFF beams at dmax under reference conditions (in
17.4.3 Energy Dependence their respective Level I remote audits) and achieve high
As discussed previously, FFF beams have a softer accuracy (within 1%). There are many types of diodes,
spectrum than flattened beams, and are therefore and these devices often have very pronounced energy
effects. Unique calibrations are likely required for these target and patient motion during the treatment cannot
devices when employed in FFF beams [18]. be predicted reliably for a specific patient. Shorter over-
For each of the detectors described above, studying all treatment (beam on) times (integrated over all frac-
the energy dependence is generally quite challenging. tions) can be associated with increased interplay effect
Practically therefore, the most reasonable approach to for a given modulation. Single fraction treatments might
determine a given detector’s response is through a cross be of concern depending on the specifics of the plan
calibration in an FFF beam. An ion chamber, having (length of beam on time) and the patient motion. The
very little energy dependence, is a good choice for per- current body of published literature is limited and based
forming such a cross calibration. In general, if a detector on this it is impossible to exclude the possibility of the
is being used that has an effective atomic number mark- interplay effect at high dose rates (short beam on times)
edly different from that of water (e.g., diode, MOSFET, for all possible scenarios.
etc.), there is a good chance that a specific FFF calibra-
tion coefficient (or correction factor) is needed. 17.5.3 Electron and Neutron Contamination
Neutron contamination is a concern for patient safety
17.5 OTHER CLINICAL CONSIDERATIONS and radiation shielding when beam energies >10 MV
17.5.1 Small Fields are employed. Most commercial implementations of
For small field measurements, a similar approach is FFF beams are 10 MV or less, so neutron contami-
taken with both FFF beams and conventional flattened nation is often not an issue. However, there are some
beams following the TRS-483 protocol [25]. The small FFF beams with energy >10 MV in clinical operation
field chamber is cross-calibrated with reference cham- (e.g., Siemens units [1]). For high energy FFF beams,
ber in a reference field with the FFF beam. When mea- neutron production has been found to be dramatically
suring the FFF small fields the appropriate field output reduced as compared to flattened beams [34–36]. This
correction factor kQfclin , f msr
clin ,Qmsr
is applied determined from occurs because the photon treatment is delivered more
the equivalent square field size. This has been specifi- efficiently so the neutron production is reduced (fewer
cally evaluated for FFF beams in recent studies of differ- photons are generated for the treatment which cre-
ent detectors [26, 27]. ate correspondingly fewer neutrons) and also because
removing the flattening filter removes a direct source
17.5.2 Motion and Interplay Effects of neutrons. These benefits will be offset if the energy of
For moving targets the high dose rates and associ- the FFF beam is raised to restore the PDD as neutrons
ated short treatment times of FFF beams require re- are produced primarily by the high-energy component
consideration of possible interplay between multileaf of the photon spectrum [37].
collimator (MLC) motion and tumor motion, in par- In addition to the abundance of low energy x-rays
ticular for hypo-fractionation (SBRT). Interplay effect, that comprise FFF beams, there is also the potential for
which can conceptually lead to over or underdosing if increased electron contamination [38]. To minimize the
tumor motion is synchronized with MLC motion, was contribution of electrons, filtration of a few mm of metal
evaluated in the context of the implementation of IMRT is placed in the accelerator head (in the same location in
and found to be negligible for standard fractionations the accelerator head as the flattening filter) [1].
[28]. Some theoretical and measurement-based stud-
ies have been reported on the topic including consid- 17.6 SUMMARY
erations of the shorter treatment times associated with FFF beams are spatially non-uniform, have an elevated
hypo-fractionated treatments [29–33]. These studies dose rate, and a larger spectral spread with a greater
report local dose discrepancies around 10% attributed lower energy component than conventional flattened
to interplay. However, such point doses difference need beams. These issues require attention during dosimetric
to be put in the context of the other uncertainties of dose measurements. A geometric correction factor is needed
delivery to a moving target. The actual impact of inter- for reference dosimetry with Farmer type chambers.
play effect depends on the MLC pattern, gantry motion, Recombination effects may become an issue in selected
the treatment (beam on) time and the patient motion. It ionization chambers. FFF beams have a softer spectrum
is non-trivial to assess interplay, especially as the actual which is more uniform across the treatment field. The
very low energy component as well as the increased 11. M. Boutillon, “Volume recombination parameter in
electron contamination is generally reduced with a thin ionization chambers,” Physics in Medicine and Biology
43, 2061–2072 (1998).
metal plate in the beam path. Detectors with energy
12. G. Xiong and D. W. Rogers, “Relationship between
dependent response should be cross calibrated before %dd(10)x and stopping-power ratios for flattening filter
use in an FFF beam. free accelerators: A Monte Carlo study,” Medical Physics
35(5), 2104–2109 (2008).
REFERENCES 13. M. Dalaryd, T. Knoos, C. Ceberg, “Combining tissue-
phantom ratios to provide a beam-quality specifier for
1. Y. Xiao, S. F. Kry, R. Popple, E. Yorke, N. Papanikolaou, flattening filter free photon beams,” Medical Physics
S. Stathakis, P. Xia, S. Huq, J. Bayouth, J. Galvin, 41(11), 111716-1 (2014).
et al.,“Flattening filter-free accelerators: A report from 14. L. de Prez, J. de Pooter, B. Jansen, T. Perik, F. Wittkamper,
the AAPM Therapy Emerging Technology Assessment “Comparison of kQ factors measured with a water
Work Group,” Journal of Applied Clinical Medical calorimeter in flattening filter free (FFF) and conven-
Physics 16(3), 5219 (2015). tional flattening filter (cFF) photon beams,” Physics in
2. G. Budgell, K. Brown, J. Cashmore, S. Duane, J. Frame, Medicine and Biology 63(4), 045023 (2018).
M. Hardy, D. Paynter, R. Thomas, “IPEM topical report 15. S. F. Kry, P. Alvarez, J. E. Cygler, L. A. DeWerd, R. M.
1: Guidance on implementing flattening filter free (FFF) Howell, S. Meeks, J. O’Daniel, C. Reft, G. Sawakuchi,
radiotherapy,” Physics in Medicine and Biology 61(23), E. G. Yukihara, D. Mihailidis, AAPM TG-191 report:
8360–8394 (2016). Clinical use of luminescent dosimeters: TLD and OSLD.
3. O. N. Vassiliev, U. Titt, F. Ponisch, S. F. Kry, R. Mohan, Report from the American Association of Physicists in
M. T. Gillin, “Dosimetric properties of photon beams Medicine (2018).
from a flattening filter free clinical accelerator,” Physics 16. L. Karsch, E. Beyreuther, T. Burris-Mog, S. Kraft, C.
in Medicine and Biology 51(7), 1907–1917 (2006). Richter, K. Zeil, J. Pawelke, “Dose rate dependence for
4. D. Georg, G. Kragl, S. Wetterstedt, P. McCavana, B. different dosimeters and detectors: TLD, OSL, EBT
McClean, T. Knoos, “Photon beam quality variations films, and diamond detectors,” Medical Physics 39,
of a flattening filter free linear accelerator,” Medical 2447–2455 (2012).
Physics 37(1), 49–53 (2010). 17. A. Niroomand-Rad, C.R. Blackwell, B.M. Coursey, K.P.
5. A. Sudhyadhom, N. Kirby, B. Faddegon, C. F. Chuang, Gall, J.M. Galvin, W.L. McLaughlin, A.S. Meigooni, R.
“Technical Note: Preferred dosimeter size and associ- Nath, J.E. Rodgers, C.G. Soares, “Radiochromic film
ated correction factors in commissioning high dose per dosimetry: Recommendations of AAPM Radiation
pulse, flattening filter free x-ray beams,” Medical Physics Therapy Committee Task Group 55,” Medical Physics
43(3), 1507–1513 (2016). 25, 2093–2115 (1998).
6. J. E. Lye, D. J. Butler, C. P. Oliver, A. Alves, J. Lehmann, 18. AAPM Task Group 62 report: Diode in vivo dosimetry
F. P. Gibbons, I. M. Williams, “Comparison between for patients receiving external beam radiation therapy.
the TRS-398 code of practice and the TG-51 dosim- Medical Physics Publishing/American Association of
etry protocol for flattening filter free beams,” Physics in Physicists in Medicine (2005).
Medicine and Biology 61(14), N362– N372 (2016). 19. B. R. Muir and D. W. O. Rogers, “Monte Carlo calcu-
7. S. F. Kry, R. Popple, A. Molineu, D. S. Followill, lations of k(Q), the beam quality conversion factor,”
“Ion recombination correction factors (P(ion)) for Medical Physics 37, 5939–5950 (2010).
Varian TrueBeam high-dose-rate therapy beams,” 20. J. R. Snow, J. A. Micka, L. A. DeWerd, “Microionization
Journal of Applied Clinical Medical Physics 13(6), chamber air-kerma calibration coefficients as a func-
3803 (2012). tion of photon energy for x-ray spectra in the range
8. P. R. Almond, P. J. Biggs, B. M. Coursey, et al., “AAPM’s of 20–250 kVp relative to Co-60,” Medical Physics 40,
TG-51 protocol for clinical reference dosimetry of high- 041711-1–041711-5 (2013).
energy photon and electron beams,” Medical Physics 21. H. Bekerat, S. Devic, F. DeBlois, K. Singh, A. Sarfehnia,
26(9), 1847–1870 (1999). J. Seuntjens, S. Shih, X. Yu, D. Lewis, “Improving
9. Technical Report Series No 398, Absorbed Dose the energy response of external beam therapy (EBT)
Determination in External Beam Radiotherapy: An GafChromicTM dosimetry films at low energies (≤100 keV),”
International Code of Practice for Dosimetry Based on Medical Physics 41, 022101 (2014).
Standards of Absorbed Dose to Water (International 22. C. Richter, J. Pawelke, L. Karsch, J. Woithe, “Energy
Atomic Energy Agency, Vienna, 2006). dependence of EBT-1 radiochromic film response for
10. M. McEwen, L. DeWerd, G. Ibbott, D. Followill, D. W. photon (10 kvp–15 MVp) and electron beams (6–18
O. Rogers, S. Seltzer, J. Seuntjens, “Addendum to the MeV) readout by a flatbed scanner,” Medical Physics 36,
AAPM’s TG-51 protocol for clinical reference dosim- 5506–5514 (2009).
etry of high-energy photon beams,” Medical Physics 23. M. J. Butson, P. K. N. Yu, T. Cheung, H. Alnawaf,
41(4), 041501-1 (2014). “Energy response of the new EBT2 radiochromic film to
x-ray radiation,” Radiation Measurements 45, 836–839 effect,” Physics in Medicine and Biology 54(13), 4311–
(2010). 4324 (2009).
24. S. B. Scarboro, D. S. Followill, J. R. Kerns, R. A. White, 32. M. Rao, J. Wu, D. Cao, T. Wong, V. Mehta, D. Shepard, J.
S. F. Kry, “Energy response of optically stimulated lumi- Ye, “Dosimetric impact of breathing motion in lung ste-
nescent dosimeters for non-reference measurement reotactic body radiotherapy treatment using intensity
locations in a 6 MV photon beam,” Physics in Medicine modulated radiotherapy and volumetric modulated arc
and Biology 57, 2505–2515 (2012). therapy [corrected],” International Journal of Radiation
25. Technical Report Series No 483, Dosimetry of Small Oncology Biology Physics 83(2), e251–e256 (2012).
Static Fields Used in External Beam Radiotherapy 33. X. Li, Y. Yang, T. Li, K. Fallon, D. E. Heron, M. S. Huq,
(International Atomic Energy Agency, Vienna, 2017). “Dosimetric effect of respiratory motion on volumet-
26. W. Lechner, H. Palmans, L. Sölkner, P. Grochowska, ric-modulated arc therapy-based lung SBRT treatment
D. Georg, “Detector comparison for small field out- delivered by TrueBeam machine with flattening filter-
put factor measurements in flattening filter free pho- free beam,” Journal of Applied Clinical Medical Physics
ton beams,” Radiotherapy & Oncology 109(3), 356–360 14(6), 4370 (2013).
(2013). 34. S. F. Kry, R. M. Howell, U. Titt, M. Salehpour, R. Mohan,
27. T. S. A. Underwood, B. C. Rowland, R. Ferrand, L. O. N. Vassiliev, “Energy spectra, sources, and shielding
Vieillevigne, “Application of the exradin W1 scintillator considerations for neutrons generated by a flattening
to determine Ediode 60017 and microdiamond 60019 filter-free Clinac,” Medical Physics 35(5), 1906–1911
correction factors for relative dosimetry within small (2008).
MV and FFF fields,” Physics in Medicine and Biology 60, 35. A. Mesbahi, “A Monte Carlo study on neutron and
6669–6683 (2015). electron contamination of an unflattened 18-MV pho-
28. T. Bortfeld, S. B. Jiang, E. Rietzel, “Effects of motion ton beam,” Applied Radiation and Isotopes 67(1), 55–60
on the total dose distribution,” Seminars in Radiation (2009).
Oncology 14(1), 41–51 (2004). 36. S. F. Kry, U. Titt, F. Ponisch, et al., “Reduced neutron
29. R. I. Berbeco, C. J. Pope, S. B. Jiang, “Measurement of production through use of a flattening-filter-free accel-
the interplay effect in lung IMRT treatment using EDR2 erator,” International Journal of Radiation Oncology
films,” Journal of Applied Clinical Medical Physics 7(4), Biology Physics 68(4), 1260–1264 (2007).
33–42 (2006). 37. R. M. Howell, S. F. Kry, E. Burgett, N. E. Hertel, D. S.
30. C. L. Ong, M. Dahele, B. J. Slotman, W. F. Verbakel, Followill, “Secondary neutron spectra from modern
“Dosimetric impact of the interplay effect during ste- Varian, Siemens, and Elekta linacs with multileaf col-
reotactic lung radiation therapy delivery using flatten- limators,” Medical Physics 36(9), 4027–4038 (2009).
ing filter-free beams and volumetric modulated arc 38. U. Titt, O. N. Vassiliev, F. Ponisch, L. Dong, H. Liu, R.
therapy,” International Journal of Radiation Oncology Mohan, “A flattening filter free photon treatment con-
Biology Physics 86(4), 743–748 (2013). cept evaluation with Monte Carlo,” Medical Physics
31. E. D. Ehler and W. A. Tomé, “Step and shoot IMRT to 33(6), 1595–1602 (2006).
mobile targets and techniques to mitigate the interplay
Chapter 18
Linac-Based SRS/SBRT Dosimetry

Karen Chin Snyder and Ning Wen
Henry Ford Health System
Detroit, Michigan
Manju Liu
William Beaumont Hospital
Royal Oak, Michigan
CONTENTS
18.1 Introduction to Linear Accelerator-Based SRS/SBRT 278
18.1.1 Common Treatment Platforms 278
18.1.1.1 Novalis Tx 278
18.1.1.2 Varian EDGE and Truebeam 279
18.1.1.3 Elekta Versa HD 280
18.1.2 Flattening Filter Free, High Dose Rate Modes 280
18.1.2.1 Ion Recombination Factor 280
18.1.2.2 Beam Spectrum 281
18.2 Physics of Small Field Dosimetry 281
18.2.1 Loss of LCPE 281
18.2.2 Source Size Occlusion 282
18.2.3 Field Size Definition for Linac SRS/SBRT 282
18.2.3.1 Jaws 282
18.2.3.2 Multi-Leaf Collimators 283
18.2.3.3 Stereotactic Cones 284
18.3 Detectors used in Characterization of Linear Accelerator Beams for Stereotactic Radiosurgery 285
18.3.1 Absolute Dose Calibration 285
18.3.1.1 Beam Quality 285
18.3.1.2 Perturbation and Volume Averaging Effects 285
18.3.1.3 Alfonso Formalism 286
18.3.2 Relative Output factors 287
18.3.2.1 Detectors used for Relative Dosimetry 287
18.3.2.2 Electrometers 289
18.3.2.3 Output Factors from Common Treatment Platforms 289
18.3.2.4 Detector Specific Correction Factors 290
18.3.3 Percentage Depth Dose and Profiles 292
18.3.3.1 Reference Detectors 293
277
18.3.3.2 Jaw-Based PDD and Profiles 294

18.3.3.3 Cone-Based Profiles 294
18.3.4 Tissue Phantom Ratio – TPR 295
18.3.4.1 TPR Measurement 295
18.3.4.2 TPR Conversion 295
18.3.5 MLC Characterization 296
18.3.5.1 Varian Eclipse TPS 297
18.3.5.2 Monaco TPS 297
18.3.5.3 Pinnacle TPS 297
18.4 Patient Specific QA Measurement Devices and Techniques 298
18.4.1 Multi-Detector Arrays 298
18.4.1.1 Film 298
18.4.1.2 Diode Arrays 298
18.4.1.3 Ion Chamber Arrays 298
18.4.1.4 Electronic Portal Imaging Device 299
18.4.2 3D Calculation on Patient Geometry 299
18.4.2.1 Log File Verification 300
18.4.2.2 Transmission Measurement and Back Projection 301
18.5 Recommended Periodical Quality Assurance 301
Acknowledgment 301
References 301
18.1 INTRODUCTION TO LINEAR 18.1.1 Common Treatment Platforms

ACCELERATOR-BASED SRS/SBRT Several commercial linac platforms exist that are used
Linear accelerators (linacs) have been used in radia- for radiotherapy treatments as well as some specialized
tion oncology as medical treatment devices since 1953, platforms used primarily for radiosurgery. C-arm linac-
when the first patient was treated at the Hammersmith based SRS and stereotactic body radiotherapy (SBRT) is
Hospital in London with a stationary linac and a treat- attractive because of the range and flexibility of treat-
ment head that could swivel. The original system was ment options. Whereas Gamma Knife is primarily used
simple, allowing for delivery of treatment of open to treat single fractionated intracranial lesions, a linac
fields, or field blocked to shield normal tissues with can have a wide variety of field sizes and fractionation
devices that were manufactured out of Cerrobend. schemes used to treat small as well as large lesions.
Linac-based stereotactic surgery occurred in 1980s Linacs may also be fitted with stereotactic accessories
after Leksell Gamma Knife had been used for sev- mounts such as tertiary collimators and cones that can
eral decades. Its implementation was hindered by the be used to define small fields used in radiosurgery.
mechanical uncertainty and stability of the multiple
moving components. 18.1.1.1 Novalis Tx
The use of f loor frames by several groups miti- The Novalis Tx (NTx) platform combines a Trilogy machine
gated the mechanical uncertainties of the movement (Varian Medical Systems, Palo Alto, CA, USA) with an
of the couch [1, 2]. In 1984, at the joint center for ExacTrac x-ray system (Brainlab AG, Munich, Germany).
radiation therapy (JCRT) at Massachusetts General The system has dual image-guided systems: the Varian on
Hospital, the Winston–Lutz test was developed that board imaging (OBI) system as well as the Brainlab ExacTrac
evaluated the accuracy of the mechanical and radia- localization (Figure 18.1(a)). The Varian OBI system can be
tion isocenters [3]. The Winston–Lutz test has still utilized to acquire cone beam CTs (CBCTs) for volumetric
been a common quality assurance test used in linac- imaging that is useful for imaging 3-dimensional soft tis-
based stereotactic radiosurgery (SRS) to determine sues. The Brainlab ExacTrac system consists of two kV x-ray
delivery precision. generators and two mounted aSi flat panel detectors. They
Linac-Based SRS/SBRT Dosimetry ◾ 279
FIGURE 18.1 Common C-arm linear accelerator treatment platforms. (a) Novalis Tx (Image courtesy of Varian Medical
Systems.); (b) Varian EDGE (Image courtesy of Varian Medical Systems.); (c) Elekta Versa HD (Image courtesy of Elekta.)
can be utilized for 2D planar x-ray correction as well as integrate a combination of image-guided systems such as
patient monitoring during treatment [4]. kV/MV, CBCT, as well as Calypso and an optical surface
The Novalis Tx machine can deliver flattened photon monitoring system (OSMS) for image guidance [5]. The
beams as well as electron beams. The photon beam used CBCT can be used for bony as well as soft tissue localiza-
for radiosurgery is a 6X-SRS high dose rate flattened tion, and kV planar images can be used to perform 2D/3D
beam. The high dose rate SRS beam has a flattening filter matches with 6 degrees of freedom correction.
separate from the 6X beam for the high dose rate mode The EDGE is a dedicated radiosurgery machine based
that is limited by field size but is designed for 1000 MU/ on the Varian Truebeam platform (Figure 18.1(b)). The
min dose rate. The Novalis Tx uses either cones or the EDGE has multiple photon energies, 6/10 MV flat-
high definition multi-leaf collimator (MLC) to collimate tened and two high intensity flattening filter-free (FFF)
and shape the beam. The high definition MLC (HD120 modes, 6X FFF and 10X FFF, as well as a 2.5 MV beam
MLC) contains 120 MLCs, 60 on each bank, with 32 used for imaging. The dose rate for the high dose rate
2.5 mm central leaves and 28 5 mm outer leaves with a radiosurgery beams are 1400 MU/min for the 6X FFF
maximum field size of 22 × 40 cm2. The cones are manu- beam and 2400 MU/min for the 10X FFF beam. The
factured by Brainlab and typically consist of diameters EDGE uses either cones or a high definition MLC to
of 4 mm, 6 mm, 7.5 mm, 10 mm, 12.5 mm, and 15 mm. collimate and shape the beam. The high definition MLC
(HD120 MLC) contains 120 MLCs, 60 on each bank,
18.1.1.2 Varian EDGE and Truebeam with 2.5 mm central leaves and 5 mm outer leaves with
The EDGE and Truebeam (Varian Medical Systems, a maximum field size of 22 × 40 cm2. The EDGE radio-
Palo Alto, CA) are streamlined treatment platforms that surgery package includes the Varian Integrated Conical
collimator Verification and Interlock (ICVI) system that volumetric modulated arc therapy (VMAT) planning, the
includes 7 conical collimators of diameters, 4 mm, 5 mm, beam does not need to be flat when MLCs modulate the
7.5 mm, 10 mm, 12.5 mm, 15 mm, and 17.5 mm. fluence to compensate for the beam profile [6].
The flattening filter is typically designed out of tung-
18.1.1.3 Elekta Versa HD sten and copper alloys to filter and attenuate low energy
The Elekta Versa HD (Elekta Oncology Systems, Crawly, photons. When the flattening filter is removed, the dose
UK) combines radiosurgery and conventional radiother- rate on the central axis can increase by a factor of 2–5
apy in a single solution (Figure 18.1(c)). Image guidance times the flattened dose rate, given a constant target
on the Versa HD is performed with the Symmetry 4D current [7]. Scatter dose from the head of the linac is
scan CBCT system which provides 4D image guidance to also reduced when the flattening filter is removed, which
assist with respiratory tracking and motion management. can reduce scatter dose from the head of the machine to
The Versa HD also utilizes the Clarity Autoscan system the patient [8]. FFF high dose rate beams have become
which uses ultrasound to monitor target anatomy. quite attractive in radiosurgery, allowing for delivery of
The Elekta Versa can be configured for multiple flat- greater amounts of dose in a single or few fractions in
tened and FFF photon beams as well as electrons. The high a shorter amount of time [9]. High dose rates allow for
dose rate modes for radiosurgery include a 6X FFF and 10X quicker treatments, thus minimizing the likelihood of
FFF photon beams with 1400 MU/min and 2400 MU/min, patient motion. When the flattening filter is removed,
respectively. The Versa HD uses either cones or MLCs to less of the primary beam is attenuated and scattered
collimate and shape the beam. It is equipped with an Agility from the flattening filter, resulting in a lower percentage
MLC high-definition, full field MLC. The MLCs replace the of low energy photons contributing to out of field dose
jaws in the orthogonal direction and consist of 160 MLCs, to the patient, thus reducing the chances of secondary
80 leaves on each bank, with a 5 mm leaf width that spans malignancies. However, removal of the flattening filter
over the entire 40 × 40 cm2 field. Cones of diameters of can lead to challenges in dosimetry due to the change in
5 mm, 7.5 mm, 10 mm, 12.5 mm, and 15 mm are common the beam properties and the response of the dosimeter to
cone sizes included. the change in energy as well as dose rate across the field.
Table 18.1 summarizes characteristic of commercially
18.1.2 Flattening Filter-Free, High Dose Rate Modes available C-arm linac FFF beams from the American
Linacs have been historically designed with flattening fil- Association of Physicist in Medicine (AAPM) Therapy
ters for 3D conformal treatments in order to flatten the Emerging Technology Assessment Work Group for FFF
radiation beam into a useable beam, typically at a refer- accelerators [6].
ence depth of 10 cm. In radiosurgery, small beams field
sizes are often not flat, and peaked in the center, due to the 18.1.2.1 Ion Recombination Factor
geometric penumbra and source occlusion. Furthermore When the flattening filter is removed in the linac, the
for intensity-modulated radiation therapy (IMRT) and dose rate is highest on the central axis and decreases
TABLE 18.1 Characteristics of Commercially Available FFF Beams

Varian Elekta Siemens
Nominal energy (MV) 6 FFF 10 FFF 6 FFF 10 FFF 7 FFF 11 FFF 14 FFF 17 FFF
Bremsstrahlung target material Tungsten Tungsten Tungsten
Approximate mean electron energy on target (MeV) 6.2 10.5 7 10.5 8.9 14.4 16.4 18.3
Filtration 0.8 mm Brass 2 mm Stainless steel 1.27 mm Al
Dmax (cm) 1.5 2.3 1.7 2.4 1.9 2.7 3.0 3.3
Dose at 10 cm depth (%) 64.2 71.7 67.5 73 68.5 74.5 76.5 78
Maximum dose rate on beam axis at Dmax (cGy/min) 1400 2400 1400 2400 2000 2000 2000 2000
Dose per pulse on beam axis at Dmax (cGy/pulse) 0.08 0.13 0.06 0.09/0.14a 0.13 0.13 0.13 0.13
Feedback/nonfeedback machine.
a
Note: Dosimetric quantities are given for a 10 × 10 cm2 field at 100 cm SSD unless otherwise noted. Dosimetric properties of the beam are
specified at depth of maximum dose (Dmax). Values were provided by the manufacturers and is summarized from “Flattening filter-free
accelerators: a report from the AAPM Therapy Emerging Technology Assessment Work Group” [6].
with distance from the central axis. This affects the quality specifier, as well as the differences in the ioniza-
amount of columnar ion recombination that occurs tion chamber perturbation correction for the specific
across the profile of the beam. In the TG-51 absolute beam quality. Xiong et al., showed that the relationship
dosimetry protocol, ion recombination is corrected between the %dd(10)x and the stopping power ratios
using the ion recombination factor, Pion. It is used to given in TG-51 are acceptable to define beam quality
correct for incomplete collection of the signal, when ion with or without the flattening filter, with a maximum
pairs recombine along an ionization track. The Pion cor- error of 0.4% [14]. Although, it is important to note that
rection factor is clinically measured and corrected using this is only valid for chambers with low Z or aluminum
a two-voltage method [10]. electrodes, that chambers with high Z electrodes should
The effect of ion recombination in high dose rate flat- not be used for reference dosimetry for FFF beams.
tening filter beams was investigated by Kry et al. using Similarly, for high energy beams greater than 10 MV,
different chambers to verify the validity of the two- TG-51 recommends adding 1 mm of lead, when measur-
voltage method of determining Pion in TG-51 [11]. The ing the beam quality, to reduce electron contamination.
ion-recombination correction for FFF beams is larger However, for FFF beams specified as 10 MV, it is unclear if
compared to those of standard flattened beams. The lead should be used. It is recommended in the TG-51 adden-
two-voltage technique was within 0.3% of the Jaffe-plot dum [12] to measure the %dd(10)x and %dd(10)Pb for all FFF
results of determining Pion. The recommended limit for beams, including those below the 10 MV limit from TG-51.
Pion, as suggested by TG-51, is Pion ≤ 1.05, and still holds
for FFF beams [12]. The choice of chamber is important 18.2 PHYSICS OF SMALL FIELD DOSIMETRY
and the Pion for the chamber may exceed the recom- Stereotactic treatments have been found to be effective
mended Pion correction. It is recommended for absolute in treating a large spectrum of intracranial and extra-
and reference dosimetry, that the detector be evaluated cranial lesions. Since lesions treated with SRS/SBRT are
for dose rate dependency. typically very small, it brings many challenges to the
dosimetric aspects of modeling the beam, calculating
18.1.2.2 Beam Spectrum and verifying the dose. A small field is typically defined
When the flattening filter is removed, the beam has a when the field size dimension is smaller than the lateral
softer photon spectrum than a beam flattened with a flat- range of the charged particles, which is also dependent on
tening filter. The FFF modes of the Elekta and Siemens the density of the medium and energy of the beam in the
machines do not use the same electron beam to create medium. For example, a field size of less than 3 × 3 cm2
the FFF and corresponding flattened beam. The electron in water is considered as small field size for 6 MV beam.
energy at the target is higher for an Elekta 6FFF beam to Besides lateral charged particle disequilibrium, other
more closely match the beam spectra at central axis to challenges in the small field dosimetry include the
the Elekta 6X flattened beam. On the Varian linac, the occlusion of x-ray target due to the collimation and the
electron energy impinging on the target is the same for comparable size of detectors to the beam dimensions. In
the Varian 6FFF as to the Varian 6X flattened beam, and this section, we will give a brief description of the phys-
thus the beam spectra of a Varian 6FFF beam is softer ics of small field dosimetry and focus on the practical
relative to the 6X flattened beam. aspects of beam data acquisition of small static fields,
The beam spectrum is important in characterization the detector consideration, beam modeling and patient
of the beam, specifically in its use for absolute dosim- specific quality assurance.
etry. AAPM Task Group 51 is a standard calibration
protocol used to calibrate the output of a linac [13]. The 18.2.1 Loss of LCPE
addendum to AAPM TG-51 gives additional data and Charge particle equilibrium (CPE) condition is defined
guidance for FFF beams, primarily in recommendations as the condition when the charged particles entering the
in how to characterize the beam spectrum. A softer cavity or sensitive volume of a detector are equal to the
photon spectrum will result in a different beam qual- charged particles exiting it. This equilibrium includes the
ity conversion factor, kQ. This is due to the difference in number, energy and angular properties of the charged
water-to-air stopping power ratios as a function of per- particles. In small fields, the photon beam loses lateral
centage depth dose at 10 cm depth (PDD10), or the beam CPE (LCPE) condition when the beam radius or
FIGURE 18.3 Schematic demonstrating location of primary

FIGURE 18.2 Ratio of dose to collision KERMA vs. field
collimator relative to the flattening filter, secondary collima-
radius for different energies. This is simulated by Monte Carlo
tors (jaws) and tertiary collimators (MLCs) and stereotactic
at 5 cm depth. The field radius is defined at 100 cm SSD for 4
cones for a Varian system.
MV to 24 MV and 80 cm SSD for 60Co. Loss of LCPE happens
at the field sizes that the ratio of dose to collision KERMA
is less than 1. (Copied with permission from Dr. Pavlos models the scatter radiation generated below the target,
Papaconstadopoulos [15].) contributes minimally to beam output in small fields
due to the blockage from the collimation.
half beam width is smaller than the maximum lateral
range of secondary electrons. rLCPE is defined as the mini- 18.2.3 Field Size Definition for Linac SRS/SBRT
mum radius of a photon beam field in which collision 18.2.3.1 Jaws
KERMA equals to absorb dose. rLCPE has been practically The radiation field can be defined by multiple collima-
used to quantitatively define small fields (Figure 18.2). tors on a linac, Figure 18.3. The primary collimator is a
Papaconstadopoulos performed Monte Carlo simula- conical tungsten block that collimates the beam to the
tions to illustrate the relationship between rLCPE and largest circular field size. It is located beneath the tar-
beam quality [15]. The loss of LCPE condition is more get before the flattening filter and monitor chamber. The
pronounced for higher beam energies and in lower den- secondary collimator typically consists of mobile jaws.
sity media, in which the charge particle ranges are longer. These are made of thick high-Z material used to colli-
mate the beam into square fields. In Elekta and Siemens
18.2.2 Source Size Occlusion machines, the secondary collimator consists of the MLC.
In Monte Carlo beam models, the source is defined as In Varian machines, the MLCs are tertiary collimators.
the electron intensity distribution on the target surface, The jaws define the square fields and the correspond-
whereas in analytical beam models, the photon fluence ing output of the field. The International Electrotechnical
distribution below the target, also known as the focal- Commission (IEC) provides two different definitions of
spot, is defined as the source. The primary photon beam field size: geometric and irradiation field size [16]. The
source is on the order of a couple millimeters (less than geometrical field size is defined as the geometric projec-
5 mm). When the field size is comparable to or smaller tion of the collimator, typically the light field, on a plane
than the dimension of the primary photon beam source, perpendicular to the beam axis. The irradiation field
the beam output decreases with decreasing field size size is defined as the dimensions of the full width at half
due to the partial source occlusion effect, where the col- maximum (FWHM) of the irradiated area in a plane
limation shields part of the source. The partial source perpendicular to the beam axis of the radiation beam.
occlusion effect also largely affects the detector response For large field sizes, where CPE conditions exist, the
due to the changes of particles spectrum and sharp local geometrical field size corresponds to the FWHM of the
absorb dose gradients. The extra-focal source, which irradiated field size. The positions of the jaws are usually
calibrated using a two point method at larger jaw sizes a 16% discrepancy in measured output of a 1 × 1 cm2
>5 cm. The position is set at two positions and is linearly field when the set field size deviated by 2 mm, and
interpolated across the entire field. When a linac is used decreased to 8% discrepancy when the field size deviation
to treat large fields, the field size is historically defined decreased to 1 mm [18].
as the full width at half maximum of the radiation field. On C-arm linacs, in which rectangular fields are
At smaller fields the correlation between the geomet- defined by lower and upper jaws, the measured out-
ric field size and irradiation field size begins to deviate, put factor can change for rectangular fields depending
Figure 18.4. When the field size is on the same order of on which jaw delineates the field due to the collimator
magnitude as the length of the lateral particle track, the exchange effect. The collimator exchange effect occurs
machine set jaw size is slightly larger than the machine when photons backscatter off the jaw closest to the mon-
set field size. This occurs because the penumbra from itor chamber, resulting in increased charge collection,
the field edges overlap with the field [17]. When the thus changing the delivered output. The profiles differ as
machine set jaw size is much smaller, the FWHM of well since the distance to the source between the X and
the resulting dose profiles is greatly overestimated. The Y jaws differ and the source has a finite size. Figure 18.5
effect in which the irradiated field size is greater than the illustrates the penumbra for an X and Y jaw profile for a
machine set field size is called apparent field widening. square field less than 1 cm.
The collimator calibration is important in small field,
especially for linac-based radiosurgery when the size of 18.2.3.2 Multi-Leaf Collimators
the field is used for both large and small field sizes. The MLCs are used to shape and modify the radiation treat-
accuracy of the jaw position becomes critical at small field ment beam. The field shape can be defined using static
sizes since it can drastically change the output and moni- leaf positions, or the MLCs can be modulated in both
tor unit settings on the machine. Sharpe et al. showed position and velocity to create the optimal fluence to
FIGURE 18.4 Demonstration of the change in geometric and irradiation field sizes for large field sizes (right) and for small
field sizes when the source is partially blocked and the field size is smaller than lateral charged particle track (left). (Reproduced
by kind permission of the Institute of Physics and Engineering in Medicine, York, UK. Small Field MV Photon Dosimetry,
IPEM Rep.103, 2010 [19].)
is to the patient, the sharper the geometric penumbra.

Most MLC designs now have curved leaf tips to reduce
the penumbra at both large and small field sizes.
Smaller MLCs are preferable for radiosurgery, when
small lesions are treated with high doses of radiation.
The smaller the leaf, the more conformal the dose dis-
tribution can be achieved, especially for small irregular
shaped lesions. It has been shown that the conformal-
ity of spine and intercranial SRS can be improved using
smaller HDMLCs for smaller lesions, relative to the
FIGURE 18.5 Illustration of the penumbral effect at very small
Millennium 120 MLC [22, 23].
fields (< 1 cm). Intensity profiles plotted for the same size aper-
ture formed by X × Y vs. Y × X jaws. The X (lower) jaws form
18.2.3.3 Stereotactic Cones
the smallest aperture [20]. (From Qin et al., Deriving detector-
specific correction factors for rectangular small fields using a Stereotactic cones are the historic method of collimat-
scintillator detector, JACMP, 17, 379–391, 2016.) ing the linac beam to small, circular field sizes to treat
small spherical or ellipsoidal lesions. They are typically
made of high density materials such as lead to atten-
deliver a specified dose distribution [21]. IMRT and uate the beam and covered with a layer of bronze or
VMAT are treatment delivery techniques that are used aluminum.
to create conformal dose distributions. Both IMRT and Elekta, Brainlab, and Varian have focused cones in
VMAT use the MLCs to create small segments to vary which the conical aperture in the center is focused back
the intensity across the radiation field. IMRT or VMAT to the radiation source in order to minimize the penum-
optimization allows the MLCs to conform around irreg- bra. The diameter of the cone is defined as the projection
ular shaped lesions, allowing for better sparing of organs of 50% of the dose at the isocenter of the machine.
at risk. When MLCs are used for IMRT or VMAT, small Several studies show that SRS planning with MLCs
irregular segments are often created by the MLCs to is advantageous in many situations, by being able to
achieve the desired dose distribution. This can lead to create more conformal distributions and thus have
difficulties with measurements and small field measure- better ability to spare dose to organs at risk. Using
ment techniques are often utilized to verify the delivery MLCs also has the ability to treat multiple lesions
of small MLC segments. simultaneously. In order to treat non spherical lesions
MLCs can be secondary collimators or tertiary col- with cones, multiple isocenters are used which leads
limators, succeeding the primary collimator that is used to increased treatment times. However, cones are still
to define the primary radiation field from the target. The advantageous relative to MLCs for shaping radiation
MLCs on the radiosurgery platforms discussed in the fields in some situations.
above section include the Varian High definition MLC An advantage of cones over MLCs includes good geo-
(HDMLC), the Elekta Agility MLC, and the Siemens metrical stability where MLC positional variability due
Artiste. to calibration or motor issues can result in dosimetric
The location of the MLCs can affect the penumbra and effects. Cones also have a lower transmission value rela-
transmission values. Example measurement techniques tive to the MLCs. For greater sparing of normal tissue
and results of common radiosurgery platforms are given especially when delivering large quantities of dose such
in the following section. When MLCs are the secondary as 80–90 Gy in treatment of trigeminal neuralgia, cones
jaws, such as the Elekta and Siemens machine design, are attractive, since there is less transmission through
the MLCs are located closer to the source and further the cone, resulting in less dose to the brainstem. Cones
from the patient surface. Because there is more room in are typically tertiary devices on a linac mounted to
the head of the machine when there is not a tertiary col- the gantry with a collimator mount. Since the cone is
limation device, the MLCs can be made thicker parallel mounted on the collimator, the smaller distance from
to the beam, thus reducing the leakage and transmission the isocenter to the collimation device results in a
dose through the MLC. However, the closer the MLC sharper geometric penumbra compared to the MLC.
18.3 DETECTORS USED IN The low energy photons scattered from the primary
CHARACTERIZATION OF LINEAR beam are reduced for a small field, therefore resulting
ACCELERATOR BEAMS FOR in a hardening of the energy spectrum. The change of
STEREOTACTIC RADIOSURGERY both photon and electron energy spectra of small fields
18.3.1 Absolute Dose Calibration results in a variation of stopping power ratio used for
Absolute dose calibration for linacs can be performed the dose calculation. However, this variation is generally
following the AAPM TG-51 [13] protocol for clinical ref- negligible. Ding et al. found that the differences in the
erence dosimetry of high-energy photons and electrons mean energy between 4 mm field and 10 × 10 cm2 field
or the International Atomic Energy Agency (IAEA) were over 20% for photons and 5% for electrons [25].
Technical Reports Series (TRS), IAEA TRS-398 [24]. This translated to a variation within 1% in the water-to-
The AAPM TG51 protocol is appropriate for megavolt- air stopping power ratio between a 4 mm cone and 10 ×
age photon energies in Co-60 beams to bremsstrahlung 10 cm2 square field.
beams up to 25 MV, with an uncertainty of less than
1%. The protocol uses ion chambers with absorbed dose- 18.3.1.2 Perturbation and Volume Averaging Effects
to-water calibration factors that can be traced back to a Detectors generally contain materials with density prop-
primary standard, such as NIST (national institute stan- erty that differ from water, which causes charged particle
dards) by the ADCL (accredited dosimetry laboratory) if fluence perturbation. Non-gaseous detectors are usually
located in the United States. used for the small field measurement. Such a detector
Absolute dosimetry should be performed in water. cannot be treated as a Bragg–Gray cavity since the cav-
In AAPM TG-51 the photon beam quality is defined as ity could disturb the charged particle fluence considering
%dd(10)x, which is the photon component of the photon the size and material of the detector. For example, high
beam percentage depth dose (PDD) at 10 cm depth in a density detectors such as silicon diodes and diamond
10 × 10 cm2 field at a source to surface distance of 100 cm. detectors over respond when measuring the output.
The beam is calibrated for a 10 × 10 cm2 field size, and dose Therefore, perturbation correction factors need further
can be calibrated with either a SSD setup or a SAD setup. quantification in the small field to determine absorbed
Chambers recommended for absolute calibration dose when applying Bragg–Gray cavity theory based on
should be waterproof, have good stability, as well as leak- medium-to-detector stopping power ratios.
age, polarity and ionization factors within limits. The Monte Carlo simulations have been used to investigate
specification of ionization chambers used for absolute the small field perturbations factors including the elec-
dosimetry in AAPM TG-51 is summarized in Table 18.2. trodes, the walls, and the detector materials [26–29]. The
electrode and wall perturbation factors were reported to
18.3.1.1 Beam Quality be close to unity, whereas the volume averaging effect and
The beam quality factor, kQ, published in the original density differences between the detector and the medium
AAPM TG-51 report was based on photon spectrums dominate the detector-specific correction factor. Due to
of flattened beams. The use of FFF beams in SRS may the complexity of understanding individual perturbation
change the kQ and should be verified. correction factors, which may be dependent on other, a
TABLE 18.2 Specifications for Ionization Chambers Used for Absolute Photon Dosimetry
Upper Limit of Specification
Chamber settling Should be less than a 0.5% change in chamber reading per monitor unit from beam-on, for a warmed up
machine, to stabilization of the ionization chamber
Pleak <0.1% of chamber reading (0.999 < Pleak < 1.001)
Ppol <0.4% of correction (0.996 < Ppol < 1.004)
Pion Linear with dose per pulse
Initial recombination Within 0.3% of unity
Pion, Polarity dependence <0.1% difference between opposite polarities should be
Chamber stability Should exhibit less than 0.3% change in calibration coefficient over the typical recalibration period of 2 years
Note: Values given are upper-limit values at the reference depth, not standard uncertainties.
Source: (Summarized from AAPM TG-51.)
correction factor generated from Monte Carlo simula- where %dd(10,10)x is the PDD at 10 cm depth in a water
tion is favored. However, there is still scientific interest to phantom for a field size of 10 × 10 cm2 at an SSD of
study the individual components. 100 cm. L is the length of the ionization chamber (in cm)
In FFF beams, due to the change in the beam profile, and SDD is the source-to-detector distance (which
a volume averaging correction factor can be used when equals the source-to-surface distance, SSD, plus the
using a cylindrical farmer type ionization chamber. The measurement depth), in cm. The assumption is made
non-uniformity of the beam, where it is peaked on the that the field size is defined at a distance of 100 cm from
central axis, can lead to an under-response when vol- the photon source.
ume averaged over the length of the ionization chamber.
Correction factors can be applied to mitigate the volume 18.3.1.3 Alfonso Formalism
averaging effect. Figure 18.6 from IAEA 483 demon- Alfonso et al. proposed a recommendation formalism
strates the amount of volume averaging over the length to determine absorbed dose using ionization chambers
f , fmsr
of the ionization chamber for common treatment plat- [31]. A new detector-specific correction factor kQclin clin , Qmsr
forms and corresponding FFF beams. was introduced, which relates to the perturbation and
IAEA 483 [30] recommends the following volume volume averaging effects. It accounts for the difference
f
factor correction factor (kvol )Qref which can be calculated between the ion chamber response in fmsr (machine spe-
empirically from %dd(10,10)x from Equation 18.1. cific reference field) and fclin (clinical radiation field that
needs dose measurement). The detector-specific correc-
( kvol )Qf
ref
( )
= 1 + 5.9 × 10−5 ⋅ %dd (10,10 )x − 3.38 × 10−3 ⋅ tion factor is defined in Equation 18.2.
2
 100  2 Dwfclin fclin
  ⋅L f f , Qclin /M Qclin
SDD  kQclin , msr
clin , Qmsr
= (18.2)
(18.1) Dwfmsr fmsr
, Qmsr /M Qmsr
FIGURE 18.6 Volume averaging effect for various FFF beams across the cavity length of the ionization chamber. (Reproduced
from IAEA publication: International Atomic Energy Agency, Dosimetry of Small Static Fields Used in External Beam
Radiotherapy, Technical Reports Series No. 483, IAEA, Vienna (2017) [30], with permission by the IAEA.)
where Dwfclin,Qclin is the absorbed dose to water in the clinical

The advantages of using Monte Carlo are its preci-
field, fclin, in the clinical beam of quality, Qclin. MQfclin is sion and accuracy when the material composition and
clin
the measured reading from the detector in the clinical geometry are known accurately. Otherwise, errors will
field, fclin, in the clinical beam of quality, Qclin. Dwfmsr ,Qmsr is
be propagated to the calculation of output correction
the absorbed dose to water in the machine specific refer- factors. The experimental approach is more practical
ence field, fmsr, in the machine specific beam of quality, and includes the detector-related effects even for the
Qmsr. MQfmsr is the measured reading from the detec- ones that are not easily simulated with Monte Carlo.
msr
tor in the machine specific reference field, fmsr, in the However, the dosimeters that are recommended as ref-
machine specific beam of quality, Qmsr. The field output erence may not be accessible in the clinic.
f , fmsr
factor (ΩQclinclin , Qmsr
), Equation 18.3, converts the absorbed
dose to water for the machine-specific reference field to 18.3.2 Relative Output Factors
absorbed dose to water for the clinical field. Output factors obtained during commissioning of lin-
acs used for SRS include the different output factors for
fclin , fmsr Dwfclin
,Q the method in which the field is defined: jaw, MLC, or
Ω Qclin , Qmsr = fmsr clin (18.3) cones. Typically jaw-based output factors are needed to
Dw , Qmsr
model the treatment beam.
The field output factor resembles the definition of tradi- Detectors may respond and result in different out-
tional output factor but explicitly for dose at a point. The put factors at different field sizes due to either the finite
equation for the detector-specific correction factor can size of the detector or energy response. Relative output
be rewritten as in Equation 18.4. factors are measured relative to a reference field size of
10 × 10 cm2. In order to obtain relative output factors,
fclin , fmsr Dwfclin fmsr
, Qclin /Dw , Qmsr at small field sizes with a relative detector, a daisy chain
k = (18.4)
Qclin , Qmsr
MQfclin /MQfmsr method is used. The daisy chain method involves mea-
clin msr
suring the charge detected at an intermediate field size,
As a result, the field output factor can be expressed as usually 3 × 3 or 4 × 4 cm2, and cross calibrated to a mea-
a function of the detector-specific correction factor, surement at 10 × 10 cm2, Equation 18.6.
Equation 18.5.
M FS ,detector M 3× 3, IC
fclin OFdetector = ⋅ (18.6)
fclin , fmsr M Qclin fclin , fmsr M 3× 3,detector M10×10, IC
Ω Qclin , Qmsr = fmsr ⋅k
Qclin , Qmsr (18.5)
M Qmsr
It is recommended by the AAPM as well as the
The detector-specific correction factors have been Radiosurgery Society (RSS) in the joint Medical Physics
extracted using Monte Carlo for a few linacs and a series Practice Guideline (MPPG) 9.a for SRS-SBRT (AAPM-
of small field detectors [32, 33]. RSS MPPG 9.a) [35] and several other governing bodies
An experimental approach has also been proposed that redundant detectors or multiple detectors be used
f , fmsr
to calculate kQclin clin , Qmsr
for different types of detectors. to measure small field output factors and the output fac-
Detectors with minimal corrections can be used as ref- tors compared between the two.
erences to help extracting out the detector specific fac- When obtaining relative output factors for small
tor for other types of dosimeters. Those approaches were field sizes, it is ideal to scan a cross- and in-line profile
based on the assumption that the detector specific cor- at depth to verify that the detector is aligned and posi-
rection for the reference detector is approximately unity, tioned at the center of the beam before obtaining output
f , fmsr
kQclin
clin , Qmsr
≈ 1. The field factor can be calculated by the factors.
ratio of reading in the clinical field of beam quality Qclin
and machine special field and beam quality. Example 18.3.2.1 Detectors Used for Relative Dosimetry
reference detectors that have been used in experimental Small field dosimetry is difficult due to the challenges
studies include Gafchromic film, thermoluminescence in detector choice and setup. The choice of detector
dosimeters, plastic scintillators, and polymer gels [34], all is important in accurately characterizing the beam.
of which are water equivalent and have high resolution. Many manufacturers have different solutions for small
TABLE 18.3 Different Relative Detectors Characteristics Obtained from the Manufacturers and Published Commissioning Data
Detector Active Volume Dimensions Material/Characteristics Use
Ion Chambers Length Inner Radius Wall Electrode
IBA Dosimetry CC01 0.01 cm3 3.6 mm 1 mm C552 Steel Scanning
IBA Dosimetry CC04 0.04 cm3 3.6 mm 2 mm C552 C552 Scanning, relative
output factors
IBA Dosimetry CC13 0.13 cm3 5.8 mm 3 mm C552 C552 Scanning, ROF, ABS
PTW Pinpoint 31016 3D 0.016 cm3 2.9 mm 1.45 mm PMMA/ graphite Al Scanning, ROF, ABS
(new type 31022)
PTW Pinpoint 31015 0.03 cm3 5.0 mm 1.45 mm PMMA/ graphite Al Scanning, ROF
PTW Pinpoint 31014 0.015 cm3 5.0 mm 1 mm C552 Al Scanning, ROF
PTW Semiflex 3D 31021 0.07 cm3 4.8 mm 2.4 mm PMMA/ graphite Al Scanning, ROF, ABS
Exradin A16 0.007 cm3 1.65 mm 1.2 mm C552 NA ROF
Diodes Active Volume Area Thickness n- or p-type Shielding Use
IBA SFD diode 0.017 mm3 0.6 mm ø 0.06 mm p-type No ROF, Scanning
IBA PFD diode 0.019 mm3 2 mm ø 0.06 mm p-type Yes ROF, Scanning
PTW Diode P 60016 0.03 mm3 1 mm2 circular 0.03 mm p-type Yes Scanning, ROF
Sun Nuclear EDGE Detector 0.019 mm3 0.8 mm square 0.03 mm n-type Yes Scanning, ROF
PTW 60012 0.03 mm3 1.13 ø 0.03 mm p-type No Scanning, ROF
PTW 60018 SRS Diode 0.3 mm3 1.13 ø 0.25 mm p-type No ROF
Plastic Scintillator Active Volume Diameter ø Thickness Material Use
Exradin W1 Scintillator 2.4 mm3 1.0 ø 3 mm Polystyrene ROF
Diamond Active Volume Diameter ø Thickness Material Use
PTW 60019 microdiamond 0.004 mm3 2.2 ø 0.001 mm Synthetic single crystal diamond ROF
(SCDD)
Acronyms: ROF = relative output factors, ABS = Absolute dosimetry, PMMA = poly(methyl methacrylate), C552 = C552 Shonka air-
equivalent plastic.
stereotactic detectors. Different types of detectors are from increased phantom and collimator scatter. In
available commercially. For measuring relative output small field sizes, the collimator scatter decreases result-
factors, the volume averaging effect is the major facing in a harder energy spectrum, thus minimizing the
tor for detector selection. The size of a suitable detec- over response found in larger field sizes.
tor is fulfilled when uniform radiation fluence is present Ion chambers can be mounted for high resolution
across the detector’s active volume. The energy depen- measurements depending on the size and shape of the
dence of the detector with field size is another factor ion chamber, the ion chamber can be mounted so that
to consider when measuring output factors. Table 18.3 the long axis of the chamber is parallel or perpendicu-
shows a variety of common detectors, their character- lar to the central axis of the beam. Most ion chamber
istics, as well as references from institutions that have measurements are performed so that the long axis is
published their output factor data for small jaw-based perpendicular to the axis of the beam. Ion chambers
linacs. Several studies have done a cross comparison of can be used parallel to the central axis of the beam, but
different detectors [36–39]. the readings may become dependent on polarity effects
from irradiation of the stem. Effects such as from stem,
18.3.2.1.1 Ion Chambers Ion chambers are tissue equiv- cable, polarity and leakage need to be carefully checked
alent, however due to the air cavity they have less sensi- and corrected.
tivity than solid state detectors. To compensate for the
lack of sensitivity, small field ion chambers may have 18.3.2.1.2 Solid State Detectors Silicon diodes can be
a high Z central electrode to increase the chamber’s advantageous due to their fast response time. They also
sensitivity. However, the high Z electrode results in an have higher sensitivity, due to the higher density relative
over-response at larger field sizes due to the increase in to air, and thus can be made smaller, and have higher
the photoelectric cross section to lower energy photons spatial resolution than some ion chamber. However
some of the disadvantages include dose rate, energy, and fiber because the hollow core fiber is not expected to cre-
depending on the encapsulation angular dependency. ate Cherenkov light [41]. Chromatic removal is another
Silicon diodes tend to over respond to low-energy scat- successful method, owing to the fact that the Cherenkov
tered radiation due to the larger photo-electric cross sec- light is mostly concentrated at the lower wavelengths
tion of silicon (Z = 14) relative to tissue. of the visible light spectrum, while the scintillation
Often times, a metal shield is used in a shielded diode spectrum peaks at a different wavelength. This method
to reduce the over response to low energy photons. consists of measuring the combined scintillation and
However, in FFF beams used in linac-based radiosur- Cherenkov light in two different spectral regions; the
gery, the ratio of low energy photons can change with dose is then calculated by using coefficients obtained
depth. Shielded diodes have shown to over respond to from calibration when maximum and minimum por-
FFF beams due to interaction of the low energy scat- tion of the fiber is irradiated [42].
tered photons; unshielded diodes are recommended to
be used in small field dosimetry. 18.3.2.2 Electrometers
Diode’s angular dependency character makes the Special electrometers may be necessary and the user
detector placement critical. It is advised that the detec- should verify compatibility when using different detec-
tor’s axis of symmetry must be aligned parallel to the tors. Some small detectors require electrometers with
beam axis. It is also recommended to radiograph the high sensitivity that are suitable for low signal readings.
detector at different angles before using in small field Other detectors, such as the Exradin W1 Scintillator,
measurements. This will help verify the location of sen- require an electrometer [42] with two channels in order
sitive volume against the outside marker provided by the to subtract the Cherenkov component of the acquired
manufacturer for chamber alignment. signal.
Diamond detectors are more tissue equivalent and
have a small active volume. However, they are expensive 18.3.2.3 Output Factors from Common
relative to the other commercially available detectors. Treatment Platforms
Natural diamond-based detectors are no longer com- Consistency of output factors may not be applicable
mercially available although artificial chemical vapor to very small field sizes. Studies have shown that
deposition (CVD) diamond-based detectors are com- the size of the source affects dosimetric parameters
mercially available (PTW 60019). such as the output factor and beam profiles. This
Plastic scintillator detectors (PSD) utilize the light may change on different accelerators and for differ-
emitted from the scintillating probe during its irradia- ent energies on the same accelerator model. For the
tion. The light is transferred through an optical fiber, 2 × 2 cm 2 field size, the standard deviation increased
and then converted and amplified in the photomultiplier up to 3.5% [43]. The development of a common set of
tube (PMT) to signals that are ultimately recorded. The data for small field size output measurements is dif-
PSDs are nearly water-equivalent with similar density ficult. Measurements using the correct detector still
and atomic number to water. They can be manufactured need to be performed for each individual accelerator;
very small while maintaining adequate sensitivity. Their however, published data can be used as a guide to help
response is linear to the absorbed dose. Owing to the identify any gross errors in commissioning of small
closely matched mass stopping power and mass energy field sizes.
absorption coefficient to water, their response is almost
independent of energy, so they can measure in both 18.3.2.3.1 Jaw and MLC Output Factors In 2012, the
large and small fields. Several studies have indicated the Radiological Physics Center (RPC), now known as the
correction factor for PSDs is close to unity. As discussed Imaging and Radiation Oncology Core (IROC), pub-
in detail in Chapter 9 of this book, the main problem lished a compilation of small output factors for com-
associated with using PSDs is the Cherenkov light gen- mon treatment platforms [44]. Although it does not
erated in the optical fibers, whose effect depends on the include FFF beams, it is a good resource to verify mea-
length of the irradiated fiber. Various methods have sured output factors during commissioning. The mea-
been proposed to solve this problem [40]. One solution surements were performed by the RPC on-site as part
is to use a hollow core fiber instead of a plastic transport of on-site audits, and compared their measured values
TABLE 18.4 Summary of Example Small Field RPC-measured and Institution Treatment Planning System-Calculated Small Field for 6
MV Beam from the Varian, Elekta, and Siemens Machine
Field Size (cm × cm) Varian Elekta Siemens
RPC Institution RPC Institution RPC Institution
10 × 10 1.000 1.000 1.000 1.000 1.000 1.000
6×6 0.921 0.929 0.930 0.934 0.914 0.920
(0.013) (0.004) (0.010) (0.009) (0.008) (0.008)
(0.9%, n = 64) (0.5%, n = 18) (0.7%, n = 13)
4×4 0.865 0.874 0.878 0.888 0.855 0.863
(0.018) (0.021) (0.015) (0.027) (0.010) (0.009)
(1.3%, n = 64) (1.3%, n = 22) (1.1%, n = 13)
3×3 0.828 0.841 0.842 0.848 0.820 0.825
(0.017) (0.025) (0.012) (0.009) (0.008) (0.011)
(1.7%, n = 62) (0.9%, n = 17) (1.3%, n = 13)
2×2 0.786 0.796 0.790 0.796 0.764 0.757
(0.019) (0.031) (0.007) (0.010) (0.010) (0.042)
(2.3%, n = 55) (1.6%, n = 17) (2.8%, n = 12)
Note: The values in square brackets and parentheses are the average absolute percent difference and standard deviation.
Source: Summarized from Followill et al., The Radiological Physics Center’s standard dataset for small field size output factors, 13, p. 3962,
2012.
to institution and treatment planning system calcu- 18.3.2.4 Detector Specific Correction Factors
lated output factors. Measurements were performed on Extensive research and publications exist for small field
Varian (n = 64), Elekta (n = 22), and Siemens (n = 10) measurements for different treatment platforms using a
linacs with varying energies from 6 to 18 MV. multitude of different detectors. Research has been per-
Output factors were acquired for MLC-shaped fields formed comparing different detectors experimentally,
(2 × 2 cm2, 3 × 3 cm2, 4 × 4 cm2, and 6 × 6 cm2) with as well comparison of detectors to theoretical methods
the secondary jaws set at 10 × 10 cm2. The measurement such as Monte Carlo [48]. Recently the IAEA 483 has
setup was 100 cm SSD with the point of measurement at compiled a set of small field output correction factors
10 cm depth in water, with relative output factors rela- from the literature for selected detectors. In general,
tive to a 10 × 10 cm2 field. The Exradin A16 cylindrical solid state detectors tend to overestimate small field out-
micro ionization chamber (0.007 cm3 sensitive volume) put factors, and ionization chambers tend to underesti-
was used. Across the different institutions, the RPC mate small field output factors.
measured output factors were reproducible from stan- Figure 18.7 shows the detector specific output correc-
f , fmsr
dard deviations ranging from 0.1% to 2.4% (Table 18.4). tion factor kQclin
clin , Qmsr
as a function of square field size for
a Sun Nuclear-EDGE shielded diode and a PTW-31014
18.3.2.3.2 Cone-Based Output Factors Example of Micro ionization chamber. Of note is for small field
measured cone-based output factors is given in sizes, the square small field correction is less than 1 and
Table 18.5. approaches 1 around 0.8 cm field size. This shows that
TABLE 18.5 Summary of Example Measured Cone-Based Output Factors for Brainlab, Elekta, and Varian Cones on a Variety of Varian
and Elekta Machines for Different Energies, Detectors, and Cone Sizes
Cone diameter (mm)
Machine Cone Energy Detector 4 5 (6*) 7.5 10 12.5 15 17.5 20
Varian STx Brainlab [45] 6X FFF SNC EDGE diode 0.689 0.790* 0.830 0.871 0.890 0.901 – –
10X FFF 0.566 0.699* 0.756 0.826 0.864 0.888 – –
Novalis Tx Brainlab [46] 6X-SRS Scanditronix SFD corrected with MC – 0.683 0.779 0.836 0.876 0.897 0.911 0.926
Elekta Versa Elekta [47] 6X FFF SFD – 0.564 0.648 0.706 0.740 0.770 – –
Varian EDGE Varian [5] 6X FFF SNC EDGE diode 0.607 0.671 0.755 0.800 0.827 0.848 0.859 –
10X FFF 0.516 0.589 0.700 0.769 0.815 0.847 0.872 –
FIGURE 18.7 Plots of detector specific output correction factor as a function of square field size for a (a) PTW-31014 ioniza-
tion chamber and (b) a Sun Nuclear-EDGE shielded diode. (Reproduced from IAEA publication: International Atomic Energy
Agency, Dosimetry of Small Static Fields Used in External Beam Radiotherapy, Technical Reports Series No. 483, IAEA,
Vienna (2017), with permission by the IAEA.)
the SunNuclear-EDGE detector over responds at smaller The IAEA compiled detector specific output cor-
field sizes. The correction factor for the PTW-31014 rection factors for fields collimated by an MLC or SRS
ionization chamber is greater than 1 and approaches cone for 6 and 10 MV flattened and FFF beams for com-
1 around 0.7 cm. The micro ionization chamber under monly used ionization chambers and solid state detec-
responds at small field sizes. tors. Tables 18.6 and 18.7 compile field output correction
f f
TABLE 18.6 Field Output Correction Factors kQclin
clin , Qmsr for Fields Collimated by an MLC or SRS Cone for 6 MV Flattened and FFF Beams
, msr
as a Function of Equivalent Square Field Size

Detector Equivalent Square Field Size (cm)
Ion Chambers 8.0 6.0 4.0 3.0 2.5 2.0 1.5 1.2 1.0 0.8 0.6 0.5 0.4
IBA Dosimetry CC01 1.002 1.004 1.007 1.008 1.008 1.009 1.011 1.013 1.018 1.027 1.047 – –
IBA Dosimetry CC04 1.000 1.000 1.000 1.000 1.000 1.002 1.009 1.022 1.041 – – – –
IBA Dosimetry CC13 1.000 1.000 1.000 1.001 1.002 1.009 1.030 – – – – – –
PTW Pinpoint 31016 3D 1.000 1.000 1.000 1.001 1.001 1.004 1.013 1.025 1.039 – – – –
PTW Pinpoint 31014 1.000 1.000 1.000 1.002 1.004 1.009 1.023 1.041 – – – – –
Exradin A16 1.000 1.000 1.000 1.000 1.001 1.003 1.008 1.017 1.027 1.043 – – –
Diodes
IBA SFD diode 1.008 1.017 1.025 1.029 1.031 1.032 1.030 1.025 1.018 1.007 0.990 0.978 0.963
IBA PFD diode 1.000 1.000 0.998 0.995 0.992 0.986 0.976 0.968 0.961 0.952 – – –
PTW Diode P 60016 1.000 1.000 0.999 0.995 0.991 0.984 0.970 0.956 – – – – –
Sun Nuclear EDGE Detector 1.000 1.000 1.000 0.999 0.998 0.994 0.986 0.976 0.966 0.951 – – –
PTW 60012 1.005 1.010 1.015 1.017 1.017 1.016 1.010 1.003 0.996 0.985 0.970 0.960 –
PTW 60018 SRS Diode 1.004 1.007 1.010 1.011 1.009 1.006 0.998 0.990 0.983 0.973 0.960 0.952 –
Plastic Scintillator
Exradin W1 Scintillator 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000
Diamond
PTW 60019 microdiamond 1.000 1.000 1.000 1.000 0.999 0.997 0.993 0.989 0.984 0.977 0.968 0.962 0.955
Source: Values summarized from IAEA 483 [30].
f f
TABLE 18.7 Field Output Correction Factors kQclin , msr
clin , Qmsr
for Fields Collimated by an MLC or SRS Cone for 10 MV Flattened and FFF Beams
as a Function of Equivalent Square Field Size. Values Summarized from IAEA 483 [30]
Detector Equivalent squire field size (cm)
Ion Chambers 8.0 6.0 4.0 3.0 2.5 2.0 1.5 1.2 1.0 0.8 0.6 0.5 0.4
IBA Dosimetry CC01 1.001 1.003 1.004 1.005 1.005 1.006 1.007 1.009 1.014 1.023 1.043 – –
IBA Dosimetry CC04 1.000 1.000 1.000 1.000 1.000 1.002 1.009 1.022 1.041 – – – –
IBA Dosimetry CC13 1.000 1.000 1.000 1.001 1.002 1.009 1.030 – – – – – –
PTW Pinpoint 31016 3D 1.000 1.000 1.000 1.001 1.001 1.004 1.013 1.025 1.039 – – – –
PTW Pinpoint 31014 1.000 1.000 1.000 1.002 1.004 1.009 1.023 1.041 – – – – –
Exradin A16 1.000 1.000 1.000 1.000 1.001 1.003 1.008 1.017 1.027 1.043 – – –
Diodes
IBA SFD diode 1.005 1.010 1.015 1.018 1.018 1.018 1.015 1.010 1.003 0.992 0.974 0.962 –
IBA PFD diode 1.000 1.000 0.998 0.995 0.992 0.986 0.976 0.968 0.961 0.952 – – –
PTW Diode P 60016 1.000 1.000 0.999 0.995 0.991 0.984 0.970 0.956 – – – – –
Sun Nuclear EDGE Detector 1.000 1.000 1.000 0.999 0.998 0.994 0.986 0.976 0.966 0.951 – – –
PTW 60012 1.003 1.006 1.009 1.010 1.010 1.008 1.002 0.994 0.986 0.976 0.960 0.951 –
PTW 60018 SRS Diode 1.002 1.004 1.006 1.006 1.004 1.000 0.992 0.984 0.976 0.966 0.953 – –
Plastic Scintillator
Exradin W1 Scintillator 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000
Diamond
PTW 60019 microdiamond 1.000 1.000 1.000 1.000 0.999 0.997 0.993 0.989 0.984 0.977 0.968 0.962 0.955
factors for the selected detectors discussed in Section PDD can change up to 12% when the gantry tilt devi-
18.3.2.1 and summarized in Table 18.3. ates from the motion of the detector by 0.0 to 1.0 degrees
The relative field output factor can be calculated [46]. The deviations in PDD show up in depths greater
with the field output correction factor for different than 7–8 cm. Since most output factors are typically
detectors utilizing the Alfonso formalism shown in measured at 5 cm depth, slight misalignment when
Section 18.3.1.3 in Equation 18.5. measuring output factor is less detrimental. However,
the alignment can significantly change the magnitude
18.3.3 Percentage Depth Dose and Profiles of the PDD for depths greater than 10 cm, Figure 18.8.
PDD and profiles (in plane and cross plane) are used The choice of detector is important for small field
to characterize the linac beam. PDD and profiles are profile measurements to accurately model the treat-
obtained using a scanning water phantom. In order to ment beam. The ideal detector should have high reso-
ensure the highest quality results, literature has cited lution that can accurately determine the penumbra of
that proper measurement techniques and methods can the beam, be fairly energy independent in response to
reduce measurement errors to within 1% [49]. low energy scattered photons, and have limited dose rate
Proper setup and quality assurance (QA) of the dependency in FFF radiosurgery beams.
scanning water tank include verification of movement Effects from volume averaging can change the shape
of the mechanical arms, including both accuracy and and size of the penumbra in cross- and in-plane pro-
linearity over the entire length of the scanning field. files. The use of diodes for scanning can also affect the
Manufactured scanning systems often offer annual pre- low dose tail region under the jaw. Kim et al. compared
ventative maintenance service which often checks the profiles using several detectors, including two diodes
accuracy and precision of the mechanical movements. (SFD, PFD) and two ion chambers (CC13 and CC01) [4].
Several tests can be performed that are described in For field sizes larger than 6 × 6 cm2, the SFD (stereo-
AAPM TG-106 to ensure high quality data collection tactic unshielded diode) overestimated the dose by ~2%
[49]. Scanned profiles for small fields are very sensitive relative to the CC13. Although the SFD had the high-
to the setup of the machine and scanning water tank. est resolution of the detectors, it tended to over respond
Khelashvili et al. demonstrated through a Monte Carlo to the low energy scattered photons in the tail region of
simulation of PDDs of a 5 mm cone that the measured the profiles. The CC13 scanning ion chamber resulted
profiles for a 6 MV beam on a Varian Truebeam and the

subsequent effect on the anisotropic analytical algorithm
(AAA) beam model in the Eclipse treatment planning sys-
tem [52]. They found that for the different detectors, the
size of the penumbra was not a large factor in the beam
model. Instead, the dose in the tail region of the affected
the beam model created the largest difference in the beam
model. When the PTW 60012 diode, an unshielded diode,
was used for scanning there was an over response in the
tail region and penumbra that became more pronounced
for larger field sizes and increased depth where the contri-
bution of low energy scattered photons increased. When
the beam profiles of the PTW 60012 were used to create
a beam model, the measured over-response resulted in an
overestimation of dose to nearby OARs, where scatter dose
FIGURE 18.8 Dependence of PDD for 5 mm conical cone on was the main contributor to the total dose.
alignment of central axis and detector motion. (Figure from
Khelashvili et al. [46].) 18.3.3.1 Reference Detectors
During scanning of PDD and profiles, two detectors are
in a larger blurring of penumbra compared to the other generally used. The primary detector is the scanning
detectors due to the volume averaging effect, Figure 18.9. detector attached to the mechanical arms of the scanner
The smaller scanning ion chamber, the CC01, resulted in and measures the profiles. The secondary detector is the
sharper penumbra but over responded in the tail regions reference detector that is used to remove the instanta-
of the profile, due to the over response of the higher steel neous fluctuation in output of the linac.
electrode to the lower energy scattered photons. When For small field measurements, the reference detector
measuring profiles with scanning detectors, the shorter should be carefully placed to prevent shadowing of the
dimension of the active volume is recommended to be primary scanning detector and perturbation of the pri-
placed perpendicular to the central axis to increase the mary field. For most fields, the reference detector can be
limiting resolution. placed on the edge of the primary radiation field above
Both scanning diodes and ion chambers can be used or below the primary detector (see Figure 18.10a). It is
for measurement of PDDs and profiles, however the type also recommended by AAPM TG-106 not to place the
of detector used can impact the measured profiles which reference detector outside the primary beam due to the
in turn can affect the modeling of the beam. Studies of low signal-to-noise ratio. For very small field measure-
unshielded diodes in FFF beams show over response in ments, the reference detector can be placed inside the
the profile of tails due to the relatively high photoelectric head of the machine, above the secondary and tertiary
cross section of silicon. Thus, shielding is often used to collimators, but this often requires some disassembling
reduce the over response [50]. of the head of the machine.
Diamond detectors [38] have shown significant New reference detectors have been designed that
reduction in response with dose rate, and have been minimize the effect of the primary detector and
shown to over-respond in the low dose regions in the increase efficiency. The Sun Nuclear Out-of-Field
tail regions of the profile. The Exradin W2 scintillator Reference detector is mounted on the outside of the
is an update to the Exradin W1 scintillator which was linac, on the top of the gantry (see Figure 18.10b). It
developed to be used for scanning as well as output fac- is a large parallel plate ion chamber, 39 cm3 in volume
tor measurement. The Exradin W1 could not be used that measures the head leakage from the linac. The
for scanning since it was unable to account for instanta- IBA StealthCHAMBERTM is a reference detector that is
neous Cherenkov radiation corrections [51]. mounted on the face of the collimator in the primary
A study by Gersh et al. compares the PTW 60012 diode, radiation field. In some linac designs, it attaches to the
EDGE diode and IBA CC13 ion chamber for scanning accessory mount (see Figure 18.10c). It is a transmission
FIGURE 18.9 PDD for (a) 5 × 5 mm2, (b) 10 × 10 mm2, and (c) 150 × 150 mm2 field sizes. (d) PDD at 100 mm depth for field
sizes ranging from 5 × 5 mm2 to 150 × 150 mm2. Profiles of (e) 5 × 5 mm2 and (f) 100 × 100 mm2 fields. Measurements were
performed with four detectors (SFD, PFD, CC01 and CC13). (Figure from Kim et al. [4].)
detector with an active area 72 mm in diameter. The than they used to be and now machines from the fac-
active area is constructed of carbon fiber and has a total tory are often fairly well matched for a given manufac-
attenuation equivalent to less than 0.5 mm AL. turer machine type. Although a user’s values for basic
dosimetric values such as Dmax, PDD, and penumbras
18.3.3.2 Jaw-Based PDD and Profiles
may not match exactly with published data, published
Commissioning of the linac regardless if it is used for data can be used as a baseline to compare measured data
SRS or conventional treatments begin with character- for a newly commissioned machine. This can help the
izing the treatment beam. This consists of measuring physicist verify that the correct detector and measure-
percent depth dose, cross-plane and in-plane profiles for ment techniques are used to help mitigate errors during
different field sizes. linac commissioning.
Table 18.8 summarizes published data for com-
mon SRS/SBRT FFF beams including the Elekta Versa, 18.3.3.3 Cone-Based Profiles
Varian EDGE, Truebeam and Novalis Tx machines. The relative output factors and thus dosimetric profiles
Machine tolerances from manufacturers are tighter of cones are dependent on the particular configuration
FIGURE 18.10 Reference detector placement. (a) Conventional placement on the in-side edge of the primary beam (Image
courtesy of Standard Imaging.) (b). Sun nuclear out-of-field reference detector and mounted location (Image courtesy of Sun
Nuclear Corporation.) (c) IBA stealth chamber mounted in accessory mount (Image courtesy of IBA Dosimetry.)
of the linac and cone. Depending on the divergence 18.3.4.1 TPR Measurement
of the cone, the source occlusion effect on output may In a TPR measurement, the detector remains at a fixed
become more significant for small field sizes. distance from the source, while the amount of water
Unlike jaw-based fields, PDD and profiles can be between the source and detector is varied. Several com-
measured with unshielded diode. PDD and profiles mercial add-ons to conventional mechanical scanning
measured with an unshielded SRS diode (Scanditronix) water tanks use water-draining systems to directly mea-
and were within Monte Carlo simulations of 1% or bet- sure TPR. The system slowly removes water from the
ter [46]. For cones, it is often difficult to avoid volume tank, and a water sensor monitors the water level track-
averaging when using detectors to scan. Film can be ing the source to surface distance. However, the TPR
used to validate cross- and in-plane profiles. measurement system is typically an add-on and not
always readily available at all clinics.
18.3.4 Tissue Phantom Ratio – TPR
Depending on the treatment planning system, tissue 18.3.4.2 TPR Conversion
phantom ratio (TPR) may be required in the treatment When direct TPR measurements are not available, the
planning system instead of the PDD. TPRs can be mea- TPR can be determined from conversion of measured
sured directly or can be converted from measured PDDs. PDD data. Conversion of PDD to tissue maximum ratio
TABLE 18.8 Dosimetric Data for Common SRS/SBRT Platforms

Penumbra
PDD 10 × PDD 10 × PDD 10 × (Cross Penumbra
Energy 10 cm2 10 cm2 10 cm2 Plane) (In Plane)
(MV) Dmax (cm) (5 cm) (10 cm) (20 cm) Pion Ppol (mm) (mm) Detector
Elekta 6 1.5 87 67.8 39.5 1.003 0.999 7.6 5.5 PTW Semiflex
Versa* 6 FFF 1.8 87.4 67.6 39.3 1.005 1.000 7.2 5.1 31010
10 2.1 90.8 72.8 45.7 1.003 1.000 8.1 5.6
10 FFF 2.4 91.9 72.9 45 1.006 0.998 7.3 5.8
18 3.0 95.4 78.3 51.7 1.006 0.999 8.0 6.7
Varian 6 FFF 1.35 84.2 63.0 34.4 1.007 NA Comparable to Varian Scanditronix
EDGE† 10 FFF 2.36 90.2 70.6 42.6 1.010 NA Truebeam CC04
Varian 6X 1.5 85.8 ± 0.2 66.2 ± 0.3 38.1 ± 0.3 1.003 ± 0.002 0.999 ± 0.002 6.82 ± 0.13 7.67 ± 0.15 EDGE
Truebeam‡ 10X 2.4 91.5 ± 0.1 73.5 ± 0.1 46.4 ± 0.2 1.004 ± 0.001 0.999 ± 0.002 7.42 ± 0.08 8.36 ± 0.20 Detector
15X 2.9 94.1 ± 0.2 76.6 ± 0.1 49.9 ± 0.2 1.005 ± 0.001 0.998 ± 0.002 7.56 ± 0.06 8.13 ± 0.11 (UTSW),
Scanditronix
6X FFF 1.5 84.3 ± 0.2 63.3 ± 0.1 34.5 ± 0.2 1.006 ± 0.002 0.999 ± 0.001 6.96 ± 0.22 7.65 ± 0.51
CC13
10 X FFF 2.4 90.4 ± 0.1 71.0 ± 0.1 42.9 ± 0.1 1.010 ± 0.004 0.999 ± 0.001 6.98 ± 0.12 8.08 ± 0.43 (HFHS),
Novalis Tx** 6X-SRS 1.5 86 66.4 37.7 1.000 1.005 4.96 NA Scanditronix
CC13
Note: Including depth of maximum dose (Dmax), percent depth dose at 5, 10, and 20 cm depth for a 10 × 10 cm2 field at 100 SSD, cross-plane
and in-plane penumbra widths (mm) from 20% to 80% for 10 × 10 cm2 field at 10 cm depth and detector used for scanning.
* Ganesh et al. University of Texas Health Science Center at San Antonio [53].
† Wen et al. Henry Ford Health Systems Detroit Michigan [5].
‡ Glide-Hurst et al. Henry Ford Health Systems, University of Colorado, UT Southwestern [54].
** Kim et al. Henry Ford Health Systems Detroit Michigan [4].
(TMR, a special case of TPR) can be calculated for con- A new term Dinf ( d , rd ), in Equation 18.9, is the depth
ventional radiotherapy using the formula described in dose curve corrected for inverse square law and repre-
Khan, Equation 18.7. sents a curve corrected for source-detector distance.
TMR ( d , rd ) = P ( d , rs , f ) 
( )
 S p rt   f + d  2
0
 (18.7)  f +d 
2
 S p ( rd )   f + t0  Dinf ( d , rd ) = P(d , rs , f ) ⋅ Sc , p (rs ) ⋅  (18.9)

 f + t0 
The method uses the inverse square law and phantom
scatter factors to convert PDD to TMR. However, for van Battum et al. fitted Dinf ( d , rd ) for each depth as
small fields, the phantom scatter is difficult to determine a function of r, field size, by a double exponential
and can introduce significant errors when converting function.
PDD to TMR. The phantom scatter is determined by cal-
culation from the measured total and collimator scatter Dinf ( d , rd ) = a1 + a2 ⋅ e ( − a3 ⋅r ) + a4 ⋅ e ( − a5 ⋅r ) (18.10)
factor. However, for small fields the ability to measure
small collimator scatter factor becomes problematic. And TMR can be calculated combining Equations 18.8–
van Battum et al. proposed a method of converting PDD 18.10 into Equation 18.11.
to TMR using a total scatter factor, in which the total scatter
factor is an approximation of the real total scatter factor [55].  Dinf (d , rd )   Dinf (d , rd ) 
TMRrd ( d ) =   =  (18.11)
TMR can be calculated in the method starting from  Dinf (t0 , rs* )   Sc , p (rs* ) 
Equation 18.7, but the phantom scatter Sp is approxi-
mated by the PDD and total scatter factor (Sc,p) 18.3.5 MLC Characterization
D   Sc , p ( rs )   f + d 
2 MLCs are characterized dosimetrically in the treatment
TMR ( d , rd ) =  d  = P ( d , rs , f ) ⋅   ⋅  planning system with several measured parameters. The
 Dt0   Sc , p ( rs* )   f + t0  parameters of the MLC that are input in the treatment
(18.8) planning system to be modeled may include inter- and
intra-leaf leakage, transmission, as well as leaf tip trans- 18.3.5.2 Monaco TPS
mission and penumbra. Ion chamber and films are often The Monaco treatment planning system uses a ven-
used to characterize the MLC [56]. Due to the small size dor-assisted modeling procedure, in which the vendor
of MLCs used for radiosurgery, films are useful in accu- provides a set of fields to be delivered on the end user’s
rately characterizing the edge and penumbra regions choice of phantom. After the fields are delivered, the
of the MLC. However, small field detectors as well as information is sent back to the vendor and a vendor-
multi-detector arrays are often used to verify, validate, created MLC model is created.
and modify different MLC modeling parameters to bet- The vendor provides an ExpressQA package for MLC
ter match measurements. Different treatment planning modeling that consists of eight QA fields. Two of the
systems require different measurements. Depending on fields simulate head and neck treatment ports, two are
the treatment planning system different factors can be simple open fields to check symmetry, and four QA
tweaked. fields are used to obtain specific dosimetric features of
the MLC. The four QA fields consist of: 3ABUT – a step-
18.3.5.1 Varian Eclipse TPS and-shoot plan where three consecutive segments are
In the Varian Eclipse treatment planning system, MLCs matched to create a uniform field; DMLC1 – a dynamic
for the compatible linacs are modeled using two user sweep field designed to produce the same uniform field
input parameters, the dosimetric leaf gap (DLG) and leaf as 3ABUT; 7SEGA – similar to 3ABUT but with smaller
transmission. The DLG is the gap between the radiation segments; and FOURL – a four segment step-and-shoot
and light field defined by the MLC. The DLG parameters plan that consists of nested L-shape patterns.
is used to incorporate the rounded leaf ends of the MLCs, Snyder et al. utilized the vendor-provided MLC plans
it also incorporates the minimal physical gap between to determine the effect on different parameters on the
the leaf banks that prevents collisions when the MLCs MLC model [58]. The 3ABUT field was used to investi-
meet. MLC plans can be obtained from the manufac- gate changes to the leaf offset, leaf tip leakage, leaf trans-
turer for measurements of these parameters. The Varian mission, and corner leakage parameters. The FOURL
provided MLC plans consist of 10 fields and an open field was used to investigate changes to the leaf trans-
field to reference the measurements. MLC transmission mission and groove width parameters. And finally, the
is measured with the same field size as the open field DMLC1 field (and slight modifications thereof) was used
with the bank of MLC closed over it, and repeated for to investigate changes to the interleaf leakage param-
the second bank. MLC transmission can be measured eter. They found additional adjustments improved the
using film or ion chamber. For film, a region of inter- point dose measurements as well as gamma pass rate for
est including transmission through the gap and between TG-119 and IROC phantom irradiation.
the MLC gaps should be averaged. For ion chamber,
chamber leakage should be verified, since the typical 18.3.5.3 Pinnacle TPS
measured dose in an open field is much higher than the MLCs are modeled in the Pinnacle treatment plan-
dose through the MLCs. The ion chamber should be ori- ning system using a Model set table. The parameters are
ented parallel to the leaf travel and measured in different shared by all photon energies. Some of the parameters
positions, between the leaf and through the leaf, and the that can be change in Pinnacle include the MLC offset
values averaged. and rounded leaf tip radius. The MLC offset is the dif-
The DLG is performed after the MLC transmission is ference in the nominal MLC position measured by the
measured because the measured MLC transmission is optical light field and the projected physical leaf edge at
used to correct the DLG measurements for MLC trans- isocenter. This takes into account the increase in trans-
mission. The DLG measurement uses seven dynamic mission through the rounded leaf ends.
sliding window fields of varying gaps (2, 4, 6, 10, 14, 16, Young et al. utilized film, point dose ion chamber
and 20 mm) [57]. An ion chamber is used for measure- measurements, as well as the SNC ArcCHECK diode
ment of the DLG. The DLG is defined as the extrapolated array to adjust the rounded leaf end parameters in the
gap that would give zero dose for the function y = ax + b, Pinnacle TPS to improve the small field dosimetry of
where y is the gap and x is the dose. The transmission VMAT plans with small segments, as well as improve
and DLG is performed for each energy. the accuracy of calculated out of field dose [59].
18.4 PATIENT SPECIFIC QA MEASUREMENT corrections should be performed to reduce the uncer-

DEVICES AND TECHNIQUES tainties. Although film dosimetry can be difficult and
Patient-specific measurements are ideal for SRS and time consuming, it is a good way to verify the penumbra
SBRT cases and are recommended by AAPM MPPG 5, and profiles for cones, because it has the least amount of
since IMRT and VMAT plans can be highly modulated. volume averaging compared to scanning using a detec-
Measurements can be performed on separate devices to tor or detector array device.
assess the delivered planar or volumetric dose. Several
devices are available commercially that are designed to 18.4.1.2 Diode Arrays
measure planar or 3D dose to verify the delivered dose. SRS MapCHECK by Sun Nuclear 2D array for SRS
Ideal measurement devices for SRS and SBRT should (Figure 18.11(a)). The detector active area is 77 × 77 mm2.
have small detectors and high spatial resolution in order It has 1013 diodes with a detector spacing of 2.47 mm.
to measure small fields and high dose gradients. Inserted into a stereoPHAN, the 2D array can be used
for static, rotational, non-coplanar, FFF, cone and MLC
18.4.1 Multi-Detector Arrays fields. Its predecessor, the MapCheck2, is a 2D array
Planar 2D measurement can be evaluated using 2D that is often used for conventional radiotherapy treat-
planar gamma analysis. The gamma analysis is evalu- ment that has a larger detector spacing of 7.07 mm and
ated using a dose difference and distance to agreement a larger detector array of 32 × 26 cm2 and also supports
criteria that was proposed by Low et al. [60]. Most con- high dose rate FFF modes (Figure 18.11).
ventional cases look at 3% dose difference and 3 mm dis- ArcCHECK is a 4D array by Sun Nuclear
tance criteria. For stereotactic surgery, tighter gamma (Figure 18.11(b)). The detectors are located in a concen-
criteria are usually proposed in order to better evaluate tric circle around the cylindrical phantom, so that the
the difference for dose distributions that may be small detectors are always perpendicular to the gantry for arc
and have sharp dose gradients. Gamma criteria of 2%/2 delivery, reducing any angular dependence. Consists of
mm and 3%/1 mm are commonly used in the litera- 1386 diodes, 0.019 mm3 in size, spaced 1 cm apart. The
ture, to better evaluate the dose difference at sharp dose detectors are arranged in a helical pattern so that the
gradients. detectors are not shadowed in the beams eye view (BEV)
and can measure both entrance and exit dose. There are
18.4.1.1 Film 221 detectors in a 10 × 10 cm2 field and 1386 detectors
Radiochromic film is a common measurement device in the full 21 × 21 cm2 phantom active area. For small
for SRS and SBRT, due to its high spatial resolution. A fields for SRS or SBRT, a merge feature can increase the
common radiochromic film model commercially avail- number of detected regions to 442 detectors within a
able is Gafchromic EBT3 film manufactured by Ashland 10 × 10 cm2 field. The 4D component of the array is the
Inc. Gafchromic has been shown in several studies its temporal resolution which allows additional verification
use for linac-based SRS/SBRT dosimetry. Gafchromic of the delivery of dose with respect to time.
EBT3 film has very good resolution, where the resolu- A hollow central cavity in the center of the array can
tion of the film is depending on the scanning resolution, be filled with multiple plugs of different density and
typically 150 dpi. Gafchromic film also is responsive measurement devices, such as ion chamber and film
over a large dose range in the clinical range of SRS/SBRT. to measure dose in the lesion. In conjunction with the
More recently, a newer Gafchromic EBT-XD model has measurement, 3DVH software can be used to estimate
been introduced by the manufacturer with an extended the delivered dose to the patient.
dynamic range up to 40 Gy.
Gafchromic film is labor intensive and requires more 18.4.1.3 Ion Chamber Arrays
time than other detector arrays for analysis because the The MatriXX FFF array by IBA dosimetry is an air-
readout of the film is not instantaneous. The response vented ion chamber array (Figure 18.11(c)). Its predeces-
of the film changes over the first several hours and pla- sor, the MatriXX Evolution, is also an ion chamber array,
teaus around 10 hours but can be up to 24 hours [61]. and although it is suitable, it is not designed to be used in
Film processing also requires a lot of effort. Studies have high dose rate flattening filter beams. The MatriXX FFF
shown heterogeneity across the scanner and film, and array consists of 1020 ionization chamber with an active
FIGURE 18.11 Examples of different manufactured detector arrays for radiosurgery applications. (a) Sun Nuclear SRS
MapCHECK (Image courtesy of Sun Nuclear Corporation.), (b) SunNuclear ArcCHECK (Image courtesy of Sun Nuclear
Corporation.), (c) IBA Dosimetry MatrixFFF (Image courtesy of IBA Dosimetry.), and (d) PTW Octavius1000 SRS (Image
courtesy of PTW.)
detector area of 24.4 × 24.4 cm2. The ion chambers are 18.4.1.4 Electronic Portal Imaging Device
0.032 cm3, with a detector spacing of 7.6 mm. The electronic portal imaging device (EPID) can also
The OCTAVIUS 1000 SRS is a liquid-filled ioniza- be used to measure the delivered fluence. The EPID is
tion chamber array by PTW (Figure 18.11(d)). The array often used to acquire MV images for localization and
consists of 977 liquid filled ionization chambers, 0.003 port verification. Using a back projection algorithm, the
cm3 in size, with a detector spacing of 2.5 mm in the measured fluence on the EPID can be converted to dose
central 5.5 cm region, and 5 mm in the outer area. The [64, 65]. High resolution EPID device such as the Varian
total active detector area is 10 × 10 cm2. The chambers aS1200 are now available that can be used for measuring
are filled with liquid ~1000 times denser than air. Thus high dose rate, FFF beams [66]. The Varian aS1200 is an
the liquid ion chambers have very good sensitivity and example of a high resolution EPID device, with an active
have good signal relative to an ion chamber of the same array of 40 × 40 cm2 consisting of 1190 × 1190 pixel array
volume. The energy response to low scattered photons and a pixel pitch of 0.336 mm.
is much more equivalent and the ionization chambers
have been shown to have good stability over time [62]. 18.4.2 3D Calculation on Patient Geometry
The arrays have also shown good agreement within 2% Studies have shown that 2D gamma analysis utilizing a
of an SRS diode for field sizes down to 1 × 1 cm2 [63]. 3%/3 mm passing criteria is too lenient and are unable
to catch systematic errors in plan delivery and during Dose check use a separate dose calculation algorithm
commissioning [67, 68]. Other methods to verify and such as collapsed cone convolution to verify the dose
evaluate the deliverability of the plan is to use the mea- from the primary treatment planning system.
sured delivered dose or the machine log files and recal-
culate the plan using the delivered MLC positions on the 18.4.2.1 Log File Verification
CBCT image of the patient from the day of treatment. Verification of the plan delivery can also be done
Commercial software such as Mobius3D (Mobius through capturing the trajectory log files from the
Medical Systems, Houston, TX, USA) and SunNuclear treatment machine that contains the mechanical
TABLE 18.9 Summary of Periodic (Daily, Monthly, Annual) Dosimetric QA for Radiosurgery Beams
Daily Tolerance Dosimetric Measurement Device User
X-ray output constancy (all ±3% Cross calibrated ion chamber or QMP or individual directly
energies) multi-detector array device supervised by QMP
Monthly Tolerance Dosimetric Measurement Device User
X-ray output constancy 2% Ion chamber
Backup monitor chamber 2% NA QMP or individual directly
constancy supervised by QMP
Typical dose rate output 2% at SRS dose rate, MU Ion chamber
constancy
Photon beam profile constancy 1% Multi-detector array device or film
Annual Tolerance Dosimetric Measurement Device User
X-ray flatness change from 1% Film or multi-detector array device QMP (qualified medical physicist)
baseline
X-ray symmetry change from ±1% Film or multi-detector array device
baseline
SRS arc rotation mode Monitor units set vs. delivered:
(range: 0.5–10 MU/deg) 1.0 MU or 2% (whichever is
greater)
X-ray output calibration (TG 51) ±1% (absolute) NIST traceable ion chamber
Field size dependent output ±2% from baseline Detector used for relative
factors for x-ray (two or more dosimetry, small field
FSs)
X-ray beam quality (PDD10 or ±1% from baseline Detector used for relative
TMR2010) dosimetry
X-ray monitor unit linearity ±5% (2–4 MU), ±2% >=5 MU Ion chamber or detector used for
(output constancy) relative dosimetry
X-ray output constancy vs. dose ±2% from baseline Ion chamber
rate
x-ray output constancy vs. ±1% from baseline Ion chamber
gantry angle
X-ray off-axis factor constancy ±1% from baseline Ion chamber or multi-detector
vs. gantry angle array device
Arc Mode (expected MU/ ±1% from baseline
degrees)
MLC transmission (average of ±0.5% from baseline Ion chamber or film
leaf and interleaf transmission)
all energies
Verification of small-field beam ±2% from baseline for >1.0 cm Detector used for relative small
data – relative output factors apertures, field dosimetry
for cones and or/MLC† ±5% from baseline for <=1.0 cm
apertures
E2E dosimetric evaluation using ±5% measured vs. calculated Ion chamber
SRS frame and or/IGRT system†
† From AAPM-RSS MPPG 9.a.
position of the linac during the treatment delivery. In 2017, the AAPM and RSS published the MPPG 9.a
Using the trajectory logs, the 3D dose can be calcu- for SRS-SBRT [35]. Included in the update are additional
lated to the patient’s CT or CBCT, the day of treatment. annual tests for C-arm linac-based radiosurgery includ-
After calculation of the delivered dose on the CT or ing verification of small field data with different toler-
CBCT, the dose to the patient such as DVH criteria, ances for fields smaller than 1.0 cm as well as dosimetric
3D gamma analysis between the treatment planned tolerances for an end-to-end (E2E) test that should be
dose vs. the delivered dose, and any delivery errors can performed annually.
be analyzed. Examples of different commercial prod-
ucts that use this technology are MobiusFX and Sun ACKNOWLEDGMENT
Nuclear PerFraction. This work was supported by the American Cancer Society
One drawback of this method of using log file verifi- under a Research Scholar Grant RSG-15-137-01-CCE.
cation is that there is no independent MLC or machine
measurement since the log files are obtained from REFERENCES
the machine’s position sensors [69]. Furthermore in 1. O.O. Betti, D. Galmarini, V. Derechinsky, “Radiosurgery
according to ASTRO and ACR guidelines a physical with a linear accelerator. Methodological aspects,”
measurement is often required for reimbursement of Stereotactic and Functional Neurosurgery 57(1–2), 87–98
pre-treatment IMRT QA. (1991).
2. F. Colombo, et al., “Stereotactic radiosurgery utilizing
a linear accelerator,” Applied Neurophysiology 48(1–6),
18.4.2.2 Transmission Measurement 133–145 (1985).
and Back Projection 3. K. R. Winston and W. Lutz, “Linear accelerator as
Other methods for calculation of 3D dose to the patient a neurosurgical tool for stereotactic radiosurgery,”
include using multi-detector arrays such as those Neurosurgery 22(3), 454–464 (1988).
described in Section 18.4.1. The detector array is an inde- 4. J. Kim, et al., “Clinical commissioning and use of the
Novalis Tx linear accelerator for SRS and SBRT,” Journal
pendent measurement device and does not rely on the
of Applied Clinical Medical Physics 13(3), 3729 (2012).
machine secondary positional sensors. The dose is mea- 5. N. Wen, et al., “Characteristics of a novel treatment sys-
sured using the secondary device, and then projected back tem for linear accelerator-based stereotactic radiosur-
on the TPCT to verify the dose delivered to the patient. gery,” Journal of Applied Clinical Medical Physics 16(4),
125–148 (2015).
18.5 RECOMMENDED PERIODICAL 6. Y. Xiao, et al., “Flattening filter-free accelerators: A
QUALITY ASSURANCE report from the AAPM Therapy Emerging Technology
Assessment Work Group,” Journal of Applied Clinical
The AAPM TG-142 [70] gives recommendations for Medical Physics 16(3), 5219 (2015).
periodic QA of medical linacs. It is an update in 2009 7. N. V. Oleg, et al., “Dosimetric properties of photon
to the AAPM TG-40 [71] published in 1994 when the beams from a flattening filter free clinical accelerator,”
quality, accuracy, and precision of modern linacs have Physics in Medicine & Biology 51(7), 1907 (2006).
8. J. Cashmore, “The characterization of unflattened pho-
improved greatly allowing for treatment of small lesions ton beams from a 6 MV linear accelerator,” Physics in
with high levels of dose such as in SRS and SBRT. Medicine & Biology 53(7), 1933–1946 (2008).
The goal of periodic QA is to ensure that the 9. P. F. O’Brien, et al., “Radiosurgery with unflattened
machine characteristics, both dosimetric and mechan- 6-MV photon beams,” Medical Physics 18(3), 519–521
ical, do not deviate significantly from their baseline (1991).
10. M. S. Weinhous and J.A. Meli, “Determining Pion,
values obtained during acceptance and commission-
the correction factor for recombination losses in an
ing. Table 18.9 summarizes the dosimetric tests that ionization chamber,” Medical Physics 11(6), 846–849
should be performed on a periodic basis and the rec- (1984).
ommended tolerances for machines used for SRS/ 11. S. F. Kry, et al., “Ion recombination correction factors
SBRT treatments. Although MLCs are a large part (P(ion)) for Varian TrueBeam high-dose-rate therapy
beams,” Journal of Applied Clinical Medical Physics
of creating conformal dose distributions in SRS, the
13(6), 3803 (2012).
recommended mechanical tests for the MLCs are not 12. M. McEwen, et al., “Addendum to the AAPM’s TG-51
covered in this section; such recommendations were protocol for clinical reference dosimetry of high-energy
provided elsewhere [56, 72]. photon beams,” Medical Physics 41(4), 041501 (2014).
13. P. R. Almond, et al., “AAPM’s TG-51 protocol for factors for microionization chambers,” Physics in
clinical reference dosimetry of high-energy photon Medicine & Biology 54(9), 2951–2969 (2009).
and electron beams,” Medical Physics 26(9), 1847–1870 29. A. J. Scott, et al., “Characterizing the influence of detec-
(1999). tor density on dosimeter response in non-equilibrium
14. G. Xiong and D. W. Rogers, “Relationship between small photon fields,” Physics in Medicine & Biology
%dd(10)x and stopping-power ratios for flattening filter 57(14), 4461–4476 (2012).
free accelerators: A Monte Carlo study,” Medical Physics 30. International Atomic Energy Agency, Dosimetry of Small
35(5), 2104–2109 (2008). Static Fields Used in External Beam Radiotherapy: An
15. P. Papaconstadopoulos, On the Detector Response and IAEA-AAPM International Code of Practice for Reference
the Reconstruction of the Source Intensity Distribution in and Relative Dose determination/International Atomic
Small Photon Fields (McGill theses, McGill University Energy Agency. Technical Report Series no. 483 (2017).
Libraries, 2016). 31. R. Alfonso, et al., “A new formalism for reference dosim-
16. International Electrotechnical Commission, IEC etry of small and nonstandard fields,” Medical Physics
Technical Report 60788: Medical Electrical Equipment – 35(11), 5179–5186 (2008).
Glossary of Defined Terms (2004). 32. G. Cranmer-Sargison, et al., “Monte Carlo modelling of
17. I. J. Das, G. X. Ding, and A. Ahnesjo, “Small fields: diode detectors for small field MV photon dosimetry:
Nonequilibrium radiation dosimetry,” Medical Physics Detector model simplification and the sensitivity of cor-
35(1), 206–215 (2008). rection factors to source parameterization,” Physics in
18. M. B. Sharpe, et al., “Monitor unit settings for inten- Medicine & Biology 57(16), 5141–5153 (2012).
sity modulated beams delivered using a step-and-shoot 33. G. Cranmer-Sargison, et al., “A methodological approach
approach,” Medical Physics 27(12), 2719–2725 (2000). to reporting corrected small field relative outputs,”
19. M. M. Aspradakis, J. P. Byrne, H. Palmans, S. Duane, Radiotherapy and Oncology 109(3), 350–355 (2013).
J. Conway, A. P. Warrington, K. Rosser, IPEM Report 34. E. Pantelis, et al., “On the output factor measurements of
103: Small Field MV Photon Dosimetry (IAEA-CN-182) the CyberKnife iris collimator small fields: Experimental
(International Atomic Energy Agency (IAEA), 2010). determination of the k(Q(clin),Q(msr)) (f(clin),f(msr))
20. Y. Qin, et al., “Deriving detector-specific correction correction factors for microchamber and diode detec-
factors for rectangular small fields using a scintillator tors,” Medical Physics 39(8), 4875–4885 (2012).
detector,” Journal of Applied Clinical Medical Physics 35. P. H. Halvorsen, et al., “AAPM-RSS Medical Physics
17(6), 379–391 (2016). Practice Guideline 9.a. for SRS-SBRT,” Journal of
21. M. S. Huq, et al., “A dosimetric comparison of various Applied Clinical Medical Physics 18(5), 10–21 (2017).
multileaf collimators,” Physics in Medicine & Biology 36. O. A. Sauer and J. Wilbert, “Measurement of output
47(12), N159–N170 (2002). factors for small photon beams,” Medical Physics 34(6),
22. Q. J. Wu, et al., “Impact of collimator leaf width and 1983–1988 (2007).
treatment technique on stereotactic radiosurgery and 37. W. Lechner, et al., “Detector comparison for small field
radiotherapy plans for intra- and extracranial lesions,” output factor measurements in flattening filter free pho-
Radiation Oncology 4(1), 3 (2009). ton beams,” Radiotherapy and Oncology 109(3), 356–360
23. J.-Y. Jin, et al., “Dosimetric study using different leaf- (2013).
width MLCs for treatment planning of dynamic con- 38. M. Westermark, et al., “Comparative dosimetry in nar-
formal arcs and intensity-modulated radiosurgery,” row high-energy photon beams,” Physics in Medicine &
Medical Physics 32(2), 405–411 (2005). Biology 45(3), 685–702 (2000).
24. International Atomic Energy Agency, Absorbed 39. S. Dieterich and G. W. Sherouse, “Experimental com-
Dose Determination in External Beam Radiotherapy. parison of seven commercial dosimetry diodes for mea-
Technical Reports Series (International Atomic Energy surement of stereotactic radiosurgery cone factors,”
Agency, Vienna, 2000). Medical Physics 38(7), 4166–4173 (2011).
25. G. X. Ding and F. Ding, “Beam characteristics and stop- 40. P. Z. Liu, et al., “Plastic scintillation dosimetry:
ping-power ratios of small radiosurgery photon beams,” Comparison of three solutions for the Cerenkov chal-
Physics in Medicine & Biology 57(17), 5509–5521 (2012). lenge,” Physics in Medicine & Biology 56(18), 5805–5821
26. P. Francescon, et al., “Monte Carlo simulated correction (2011).
factors for machine specific reference field dose calibra- 41. A. Darafsheh, et al., “Radiotherapy fiber dosimeter probes
tion and output factor measurement using fixed and based on silver-only coated hollow glass waveguides,”
iris collimators on the CyberKnife system,” Physics in Journal of Biomedical Optics 23(1), 015006 (2018).
Medicine & Biology 57(12), 3741–3758 (2012). 42. P. Carrasco, et al., “Characterization of the Exradin W1
27. P. H. Charles, et al., “Monte Carlo-based diode design scintillator for use in radiotherapy,” Medical Physics
for correction-less small field dosimetry,” Physics in 42(1), 297–304 (2015).
Medicine & Biology 58(13), 4501–4512 (2013). 43. D. A. Jaffray, et al., “X-ray sources of medical linear
28. F. Crop, et al., “The influence of small field sizes, penum- accelerators: Focal and extra-focal radiation,” Medical
bra, spot size and measurement depth on perturbation Physics 20(5), 1417–1427 (1993).
44. D. S. Followill, et al., “The Radiological Physics Center’s 59. L. A. Young, et al., “Rounded leaf end modeling in
standard dataset for small field size output factors,” Pinnacle VMAT treatment planning for fixed jaw lin-
Journal of Applied Clinical Medical Physics 13(5), 3962 acs,” Journal of Applied Clinical Medical Physics 17(6),
(2012). 149–162 (2016).
45. D. B. Wiant, et al., “Commissioning and validation of 60. D. A. Low, et al., “A technique for the quantitative evalu-
BrainLAB cones for 6X FFF and 10X FFF beams on a ation of dose distributions,” Medical Physics 25(5), 656–
Varian TrueBeam STx,” Journal of Applied Clinical 661 (1998).
Medical Physics 14(6), 293–306 (2013). 61. N. Wen, et al., “Precise film dosimetry for stereotactic
46. G. Khelashvili, et al., “Dosimetric characteristics of radiosurgery and stereotactic body radiotherapy quality
the small diameter BrainLab cones used for stereotac- assurance using Gafchromic™ EBT3 films,” Radiation
tic radiosurgery,” Journal of Applied Clinical Medical Oncology 11(1), 132 (2016).
Physics 13(1), 3610 (2012). 62. B.-G. Jalil, et al., “Long-term stability of liquid ioniza-
47. E. Borzov, et al., “Dosimetric characterization of Elekta tion chambers with regard to their qualification as local
stereotactic cones,” Journal of Applied Clinical Medical reference dosimeters for low dose-rate absorbed dose
Physics 19(1), 194–203 (2018). measurements in water,” Physics in Medicine & Biology
48. G. X. Ding, D. M. Duggan, C. W. Coffey, “Commissioning 46(3), 729 (2001).
stereotactic radiosurgery beams using both experimen- 63. M. Markovic, et al., “Characterization of a two-dimen-
tal and theoretical methods,” Physics in Medicine & sional liquid-filled ion chamber detector array used for
Biology 51(10), 2549–2566 (2006). verification of the treatments in radiotherapy,” Medical
49. I. J. Das, et al., “Accelerator beam data commissioning Physics 41(5), 051704 (2014).
equipment and procedures: Report of the TG-106 of the 64. G. Nicolini, et al., “GLAaS: An absolute dose calibra-
Therapy Physics Committee of the AAPM,” Medical tion algorithm for an amorphous silicon portal imager.
Physics 35(9), 4186–4215 (2008). Applications to IMRT verifications,” Medical Physics
50. Z. Yin, R. P. Hugtenburg, A. H. Beddoe, “Response cor- 33(8), 2839–2851 (2006).
rections for solid-state detectors in megavoltage photon 65. B. W. King, D. Morf, P. B. Greer, “Development and test-
dosimetry,” Physics in Medicine & Biology 49(16), 3691– ing of an improved dosimetry system using a backscat-
3702 (2004). ter shielded electronic portal imaging device,” Medical
51. P. E. Galavis, et al., “Characterization of the plastic scin- Physics 39(5), 2839–2847 (2012).
tillation detector Exradin W2 for small field dosimetry,” 66. N. Miri, et al., “EPID-based dosimetry to verify IMRT
Medical Physics 46(5), 2468–2476 (2019). planar dose distribution for the aS1200 EPID and FFF
52. J. A. Gersh, R. C. Best, R. J. Watts, “The clinical impact of beams,” Journal of Applied Clinical Medical Physics
detector choice for beam scanning,” Journal of Applied 17(6), 292–304 (2016).
Clinical Medical Physics 15(4), 4801 (2014). 67. M. F. Chan, et al., “Using a novel dose QA tool to quan-
53. N. Ganesh, et al., “Commissioning an Elekta Versa HD tify the impact of systematic errors otherwise unde-
linear accelerator,” Journal of Applied Clinical Medical tected by conventional QA methods: Clinical head and
Physics 17(1), 179–191 (2016). neck case studies,” Technology in Cancer Research &
54. C. Glide-Hurst, et al., “Commissioning of the Varian Treatment 13(1), 57–67 (2014).
TrueBeam linear accelerator: A multi-institutional 68. B. E. Nelms, et al., “Evaluating IMRT and VMAT dose
study,” Medical Physics 40(3), 031719 (2013). accuracy: Practical examples of failure to detect system-
55. L. J. van Battum, M. Essers, P. R. Storchi, “Conversion atic errors when applying a commonly used metric and
of measured percentage depth dose to tissue maximum action levels,” Medical Physics 40(11), 111722 (2013).
ratio values in stereotactic radiotherapy,” Physics in 69. B. Neal, et al., “A clinically observed discrepancy
Medicine & Biology 47(18), 3289–3300 (2002). between image-based and log-based MLC positions,”
56. T. LoSasso, C. S. Chui, C. C. Ling, “Physical and dosi- Medical Physics 43(6 Part 1), 2933–2935 (2016).
metric aspects of a multileaf collimation system used in 70. E. E. Klein, et al., “Task Group 142 report: Quality
the dynamic mode for implementing intensity modu- assurance of medical accelerators,” Medical Physics
lated radiotherapy,” Medical Physics 25(10), 1919–1927 36(9), 4197–4212 (2009).
(1998). 71. G. J. Kutcher, et al., “Comprehensive QA for radia-
57. W. Yao and J. B. Farr, “Determining the optimal dosi- tion oncology: Report of AAPM Radiation Therapy
metric leaf gap setting for rounded leaf-end multileaf Committee Task Group 40,” Medical Physics 21(4), 581–
collimator systems by simple test fields,” Journal of 618 (1994).
Applied Clinical Medical Physics 16(4), 65–77 (2015). 72. T. Pawlicki, P. Dunscombe, A. J. Mundt, P. Scalliet,
58. M. Snyder, et al., “Modeling the Agility MLC in the “Quality and safety in radiotherapy,” in Imaging in
Monaco treatment planning system,” Journal of Applied Medical Diagnosis and Therapy, edited by W.R. Hendee
Clinical Medical Physics 17(3), 190–202 (2016). (CRC Press, Boca Raton, FL, 2010), p. 643.
Chapter 19
CyberKnife and ZAP-X Dosimetry

Sonja Dieterich
University of California Davis
Sacramento, California
Georg Weidlich
Zap Surgical
Christoph Fuerweger
European Cyberknife Center Munich
Munich, Germany
CONTENTS
19.2 CyberKnife 306
19.2.1 TG-51 Formalism for CyberKnife 306
19.2.2 Output Factors 307
19.2.2.1 Cones 307
19.2.2.2 IRIS 308
19.2.2.3 MLC 308
19.2.3 Tissue-Phantom Ratios (TPRs) 308
19.2.4 Off-Axis Ratios (OARs) 308
19.2.4.1 Cone 308
19.2.4.2 IRIS 308
19.2.4.3 MLC 309
19.3 Zap Surgical Unit 310
19.3.1 Radiation Detectors 310
19.3.2 Beam Data Acquisition 310
19.3.3 Reference Dose Calibration of the ZAP-X 311
19.3.4 Output Factors (OFs) 311
19.3.5 Percent Depth Dose (PDD) 311
19.3.6 Off-Axis Ratios (OARs) 312
19.3.7 Preparation and Processing of Measured Beam Data before Import
into the Treatment Planning System 312
References 312
305
19.1 INTRODUCTION of an MLC in addition to the cones and IRIS collimator,

The CyberKnife (Accuray Inc., Sunnyvale, CA, USA) and many existing systems still rely on the Alfonso mecha-
the ZAP-X device (Zap Surgical Systems Inc., San Carlos, nism of a machine-specific reference (MSR) field. The
CA, USA) are both FDA-approved dedicated radiosur- 60 mm fixed collimator has de facto become the standard
gery devices with linear accelerators (linacs) mounted on for the MSR field.
robots. While the general small field dosimetry principles The first decision in reference dosimetry for
discussed in Chapter 18 apply, there are some designs spe- CyberKnife is which chamber to choose. As Kawachi
cific to each machine which warrant a separate discussion. et al. discussed, the non-flattened MSR field results in
a change of response depending on the length of the
chamber [6]. For a traditional Farmer chamber, a cor-
19.2 CYBERKNIFE
rection factor of about 1.7% is required to account for
The CyberKnife consist of a 6 MV miniature linac the non-flattened field. An alternative to correcting the
mounted on a 6 degree of freedom industrial robot chamber reading based on chamber length and field
(Kuka, Augsburg, Germany). The beams can be shaped shape is to use a shorter chamber for reference dosim-
through three collimator systems which can be attached etry. This can either be an ADCL-calibrated chamber,
to the linac: fixed cones, an IRIS collimator [1], and the or a chamber cross-calibrated with a Farmer chamber.
INCISE MLC collimator. Image guidance is provided The next step is to relate the standard calibration con-
using orthogonal x-ray cameras mounted at the ceiling dition of CyberKnife (80 cm SAD using the 60 mm fixed
with amorphous silicon detectors embedded in the floor. cone as MSR field) to the standard conditions defined
Real-time respiratory motion monitoring is provided in AAPM TG-51 (100 cm SAD, 10 × 10 cm2 field size).
using the Synchrony system with surface beacons [2]. In addition, the physicist needs to handle the gradient
The IRIS collimator consists of two leaf banks with corrections which are included implicitly in kQ.
six tungsten-alloy segments each, offset at 30 degrees to In the first step, the robot is moved up from the water
each other. This dodecagonal treatment field is approxi- tank to an extended source-to-surface distance (SSD)
mately circular. While the IRIS collimator theoretically (for CyberKnife) of 100 cm. In this position, the equiva-
can have unlimited numbers of apertures, for treatment lent square of the 60 mm cone at 100 cm SSD is 6.75 ×
planning optimization purposes 12 apertures matching 6.75 cm2 [7]. The %dd(10, 6.75, 100) is then measured
the fixed cones are used. and compared to %dd(10, 6.75, 100) with local refer-
The INCISE MLC has been released in two versions ence data from a standard linac or, alternatively, the
with different specifications. Version 1 has already been data published by Jordan in BJR supplement 25 [8]. The
phased out and is no longer in clinical use. Version 2 corresponding %dd(10,10,100) from the reference data-
consists of 26 pairs of flat-sided leaves with a height of set will determine kQ. Based on Figure 2 of Kalach and
90 mm and is capable of full overtravel and interdigi- Rogers’ paper [9] comparing flattening-filter free soft
tation. The leaf thickness is 3.85 mm at the reference beams with flattened beams, kQ from a flattened beam is
source-to-axis distance (SAD) of 80 cm, and the maxi- valid for a softer unflattened beam as well.
mum field size for clinical use is 11.5 × 10.1 cm2. In the In the second step, the SSD is then changed to the
absence of jaws, interleaf leakage is reduced to less than CyberKnife calibration condition (SAD or SSD).
0.5% by a leaf tilt of 0.5°, which causes asymmetric pen- Pgr(CK) is then determined at the measurement point
umbras in in-plane profiles. Leaf positioning is achieved and the dose is calculated by Equations 19.1–19.3:
with a mean accuracy of <0.2 mm and verified by a sec-
ondary camera system. As of 2020, the MLC is operated MN DQ,w
in step-and-shoot mode only. DwQ = , (19.1)
%dd or TPR
19.2.1 TG-51 Formalism for CyberKnife

Alfonso et al. [3] developed a formalism for devices which
M = Mraw ∏ P, (19.2)
do not have the standard field size required for AAPM

Pgr (CK )
Task Group 51 (TG-51) [4] or IAEA TRS-398 [5]. While N DQ,w = N D60,Co
w kQ . (19.3)
the newer generation CyberKnife systems offer the option Pgr (10,10,100)
CyberKnife and ZAP-X Dosimetry ◾ 307
TABLE 19.1 Sample Set of Output Factors Acquired for a CyberKnife M6 FIM
Circular Square/Rectangular
Field diameter (mm) Cones IRIS Field size (mm²) MLC
5 0.683 0.507 7.6 × 7.7 0.808
7.5 0.818 0.769 15.4 × 15.4 0.948
10 0.871 0.857 23.0 × 23.1 0.976
12.5 0.908 0.899 30.8 × 30.8 0.988
15 0.934 0.926 38.4 × 38.5 0.994
20 0.960 0.958 46.2 × 46.2 1.000
25 0.973 0.969 53.8 × 53.9 1.005
30 0.979 0.976 69.2 × 69.3 1.013
35 0.984 0.980 84.6 × 84.7 1.020
40 0.987 0.985 100.0 × 100.1 1.025
50 0.993 0.991 115.0 × 100.1 1.026
60 1.000 1.001
It should be emphasized that a peer-review of the to the 60 mm field size at 80 cm SSD. Accuray rec-
output should be conducted after commissioning and ommends detectors optimized for use in small beam
before the first patient is treated, and annually thereaf- dosimetry (e.g., unshielded diodes and synthetic micro-
ter. External peer review checks are offered by several diamond detectors). In all cases, Accuray recommends
entities, including the IAEA, IROC Houston, and other the evaluation and use of published correction factors
dosimetry labs. If this service is not available, an accept- as appropriate, particularly for output factor determina-
able alternative is to request a peer-review from an out- tion in the smallest beams. Other detectors such as film,
side medical physicist with extensive experience who microchambers, liquid filled chambers, and alanine pel-
would be able to bring in their own chambers. lets have also been used [10, 11]. Table 19.1 lists a sample
For CyberKnife units equipped with an MLC, the set of output factors for all three collimators.
10 × 10 cm2 field size needed to perform AAPM TG-51 or Since the publications of Francescon et al. [12–14],
IAEA TRS-398 reference dosimetry is available. For exist- there has been discussion if an energy correction fac-
ing systems that have used the method outlined above for tor should be used to correct the output factors for the
the 60 mm MSR field, the option of retaining that method smallest collimator. IAEA Technical Report 483 has
for consistency and using a separate reference dosimetry extensive guidance on implementation of correction
method for an MLC which was added later could be dis- factors [15]. For beams and/or detectors which are not
cussed. If the reference dosimetry is switched to a tradi- covered in IAEA TRS 483, best clinical practice is:
tional 10 × 10 cm2 reference dosimetry, new output factors
for both collimators should be compared to the previous 1. Use at least two different suitable detectors, prefer-
output factors, and any changes above 2% be investigated. ably with opposite energy response.
For new CyberKnife systems who have an MLC available
2. Apply correction factors from peer-reviewed
from installation, the reference dosimetry field should
literature.
be the 10 × 10 cm2 field. Regardless, the absolute dose
calibration must be conducted such that the CyberKnife 3. Average the results.
output with the 60 mm fixed cone at 800 mm SAD and
4. Compare the result to peer-reviewed literature or
15 mm depth in water is 1 cGy per 1 MU.
use peer review to verify results.
19.2.2 Output Factors
The outputs factors (OFs), also called total scatter fac- 19.2.2.1 Cones
tors (SCP), are measured relative to the output factor The output factors for the fixed cones are measured
of the 60 mm fixed collimator for three different SSDs with the detector centered on the central axis (CAX) at
(65 cm, 80 cm, and 100 cm) intended to cover the SSD 1.5 cm depth in water. Output factors are measured at
range in clinical use. The output factors are normalized 65 cm, 80 cm, and 100 cm SAD. All output factors are
normalized to the 60 mm cone at 80 cm SAD. Measured critical in comparison to circular or small MLC fields –
output factor values have been published for a variety of small errors in measurement depth cause larger differ-
detectors [16], and energy correction factors are avail- ences in detector readings.
able from Francescon et al. [17, 18].
19.2.3 Tissue-Phantom Ratios (TPRs)
19.2.2.2 IRIS The CyberKnife TPS currently requires the measure-
Output factors for the IRIS collimator are very similar to the ment of tissue-phantom ratios (TPRs) as input. With
fixed cone sized output factors down to about the 12.5 mm a regular size water tank, these measurements can
cone. For the three smallest cones (10 mm, 7.5 mm, and be very time consuming, because the robot has to be
5 mm), differences start to emerge. The reason is the moved manually for each measurement depth. For ear-
mechanical aperture tolerance of 0.2 mm. To get an average lier CyberKnife versions, this was usually achieved by
output factor, the collimator opening is changed in between entering the treatment room. For M6, the Kuka teach
repeat measurements for the three smallest collimators. In pendant can be connected to the robot controller in the
clinical practice, the 7.5 mm and 10 mm IRIS collimator equipment room and routed to the control area, which
size is not heavily used in any treatment plan. Most clinical allows for moving the robot from outside the treatment
sites will not use the 5 mm IRIS collimator. room but requires careful monitoring of the robot using
in-room cameras to avoid collision. There is a third,
19.2.2.3 MLC unsupported option applicable to all CyberKnife ver-
While the smallest clinically used MLC field is 1 × 1 cm2, sions, which uses service tools to move the robot from
commissioning of the CyberKnife treatment planning the delivery console.
system (TPS) includes OF measurement of the 0.76 × The general process can be somewhat abbreviated for
0.77 cm2 field. In analogy to the smallest circular fields, the IRIS and the MLC collimator by measuring TPRs at
diodes can be expected to over-, air-filled microchambers a specific depth for all or at least multiple collimator and
to under-respond for small MLC fields. Recently, Monte- field sizes. To make TPR measurements more efficient,
Carlo-based correction factors kΩfclin , fmsr
clin ,Ωmsr
specific to the smaller water tanks with TMR functionality have been
CyberKnife M6 MLC have become available for a range of developed [20]. PDD-to-TMR conversions have been
detectors and should be used for OF determination [19]. attempted but found to fail for the smaller collimators
Several detectors used to measure output factors, (private communication). Figure 19.1 shows a sample set
diodes especially, have an energy response as a func- of TPR data for a range of collimators and field sizes.
tion of field size. While this response is small over the
field size range for fixed and IRIS collimators, it can 19.2.4 Off-Axis Ratios (OARs)
be considerable over the field sizes achievable with the Accurate radiosurgical planning is highly depended
MLC, depending on the specific detector. In this case, on steep dose gradients toward critical structures.
it is recommended to use the same approach as for lin- Therefore, it is essential that the dose gradients are
acs, i.e., measure the output factors for larger field sizes measured accurately and with the least smearing due
using an appropriate chamber. Given the requirement to detector size. Figure 19.2 shows a sample set of OAR
to measure output factors for a discrete selection of field data for a range of collimators and field sizes.
sizes for commissioning, the field size of 4.62 × 4.62 cm2
is an appropriate choice to daisy-chain chamber-based 19.2.4.1 Cone
output factor measurements for larger fields to output OARs for the cones are measured using small area
factor measurements for smaller fields using a diode or detectors such as diodes or diamond detectors. Inline
other appropriate detector [16]. Alternatively, the use of and crossline profiles are measured at several depths as
MC-based correction factors kΩfclin , fmsr
clin ,Ωmsr
can also be con- required by the TPS. The data processing is performed
sidered for the largest fields [19]. similar to taking OARs for external beams.
For the largest MLC fields, the OF measurement
depth of 1.5 cm is well within the build-up region, as the 19.2.4.2 IRIS
maximum dose point is closer to 1.8 cm. Consequently, Because of the shape of the IRIS collimators, two addi-
exact reproduction of the measurement depth is more tional cross profiles must be measured in addition to
FIGURE 19.1 TPRs acquired for selected field sizes of a CyberKnife M6 FIM.
the inline and crossline profiles. The additional cross- IRIS, measured OARs are averaged and reduced to a sin-
profiles are measured at a 15-degree angle offset to the gle half beam profile for each field size, full in- and cross-
inline and crossline profiles. plane profiles of 10 square and one rectangular field are
used for MLC commissioning. Consequently, water tank
19.2.4.3 MLC scanning direction and signs matter for MLC only – in-
The general procedure of measuring OARs for MLC is and cross-plane penumbrae are different, and due to the
similar to fixed cones and IRIS: Beam data is acquired bank tilt, CyberKnife MLC fields will not be perfectly
at the same depths, either at a fixed SSD or SAD of symmetrical even with an optimally tuned beam [21, 22].
800 mm. However, there are two MLC-specific aspects Second, correctly aligning the linac to the water phan-
that require some attention: First, whereas for cones and tom is somewhat trickier due to the mechanical degrees
FIGURE 19.2 OARs for selected field sizes of a CyberKnife M6 FIM acquired at SSD 700 mm and depth 100 mm. The graph
shows averaged OARs for circular and cross-plane OARs for MLC fields.
of freedom of the robot and the absence of a light field treatment chamber sphere. The patient is supported on a
of the MLC, and has to be achieved by a few iterations of moveable treatment table that extends outside the treat-
measuring off-axis profiles of a narrow field, followed by ment sphere but which itself is also enclosed by addi-
moving the robot to adjust the linac orientation. tional radiation shielding during radiosurgery. This
table shielding consists of a rotary shell and pneumatic
19.3 ZAP SURGICAL UNIT door on a steel frame.
The ZAP-X is a new, dedicated self-contained and self- The ZAP-X accomplishes three-dimensional (3D)
shielded radiosurgery system developed and manu- patient registration by means of an integrated planar
factured by ZAP Surgical Systems, Inc. of San Carlos, kilovolt (kV) imaging system that also rotates around
California [23–25]. A cross section of the device is the patient’s head. Pairs of non-coaxial x-ray images and
shown in Figure 19.3. image-to-image correlation are utilized to determine
This device is intended for stereotactic radiosurgery the location of the patient’s anatomy with respect to the
(SRS) treatment of benign and malignant intracranial machine isocenter, both prior to and during radiosur-
and cervical spine lesions. A 3.0 MV S-band linac is the gical treatment. The treatment beam is shaped through
source of therapeutic radiation. Akin to a large gyro- tungsten collimators.
scope, the linac is mounted within a combination of
yoked gimbals with attached radiation shielding, each 19.3.1 Radiation Detectors
of which accurately rotates around a common isocen- Radiation detectors appropriate for small field ZAP-X
ter. This mechanical construct enables the linac beam stereotactic radiosurgery fields require small detection
crossfire from 2π steradians of solid angle, as is ideally volume to prevent averaging of the rapidly changing
required for cranial SRS. radiation intensities across the small fields and near
Accurate therapeutic beam positioning is accom- the field edges. Acceptable detectors include diodes
plished through the two above-mentioned dual-axes, and ionization chambers. Other detectors with simi-
independent rotations of the accelerator, and precise lar detection volumes response are acceptable as well.
movements of a robotic patient table. Most components
needed to produce the beam, such as the radiofrequency 19.3.2 Beam Data Acquisition
power source, waveguiding system, and beam trigger- The ZAP-X TPS utilizes measured and interpolated
ing electronics, as well as significant radiation shielding, beam data only by employing the raytracing method
are mounted on or integrated into the rotating patient with the equivalent path length correction. No modeling
FIGURE 19.3 Cross-sectional view of ZAP-X system. (Figure provided by ZAP Surgical.)
FIGURE 19.4 Setup of 3D PTW water phantom inside the ZAP-X treatment sphere.
or parametrization of the measured beam data is per- deliver 1.0 cGy at the calibration conditions of SAD =
formed. Required ZAP-X beam data include collima- 45.0 cm, at 7.0 mm depth, for the 25 mm collimator.
tor output factors, TPRs, and off-axis ratios (OARs). The MSR field [3] size is the 25 mm collimator. Because
Percent depth dose (PDD) data shall be acquired and the depth of dmax for the MSR field is 7.0 mm and the ion-
converted to TPR. All beam data should be acquired in ization chamber has an outer radius of 3.5 mm, the refer-
a water phantom with a minimum scanning depth of ence calibration is performed at a depth of 5.0 cm. The
25 cm and a step size for PDDs and OARs of 0.1–1.0 mm. measured dose is then referenced to calibration condi-
The SSD should be set to 45 cm for PDD and OAR mea- tions (SAD = 45.0 cm, d = 7.0 mm, circular collimator size
surements, while output factors shall be acquired at the = 25 mm diameter) using the same methodology as out-
calibration point of dmax = 7.0 mm and SSD = 44.3 cm. lined for the CyberKnife in Section 19.2.1. Most suitable
The unique measurement setup inside the ZAP-X treat- for the calculation of dose at the calibration point is the
ment sphere is shown in Figure 19.4. use of TPR ratios. Alternatively, a custom-made acrylic
spherical phantom can be attached to the collimator
19.3.3 Reference Dose Calibration of the ZAP-X housing via a holder (F-bracket) to provide a known and
The reference dose calibration is performed using reproducible measurement geometry and can be used to
AAPM TG-51 [4] protocol with the adaptation for MSR determine the dose at the calibration point.
fields as described by Alfonso et al. [3]. Due to the very
small field sizes of the ZAP-X, the reference calibration 19.3.4 Output Factors (OFs)
of this system is performed with an ionization cham- Output Factors can be measured with small volume
ber mounted with its chamber axis perpendicular to detectors such as a diamond detector, pinpoint ioniza-
the CAX and a maximum sensitive length of 7 mm. tion chamber or SRS diode. Diamond detectors and SRS
An example of suitable chamber is the PTW 0.125 cm3 diodes should be positioned with their stem axis paral-
semiflex ionization chamber, Model 31010. Accredited lel to the central beam axis (CAX) and centered on it.
Dosimetry Calibration Laboratories can provide cali- The ionization chamber can be oriented with its stem
bration factors Nx and ND,W for this type of chamber. The axis parallel to the central beam axis or perpendicular
ionization chamber used for reference dose calibration to it. The output measurements for each collimator are
shall be cross calibrated on a conventional linac with a normalized to the 25 mm collimator. Output factors are
calibration field size of 10 × 10 cm2 and well-established shown in Figure 19.5.
dosimetry. The final calibration of the ZAP-X should be
independently verified by TLD or OSLD exposures or 19.3.5 Percent Depth Dose (PDD)
independent redundant calibration. The final result of PDD measurements are best performed with an SRS
the reference dose calibration will ensure that 1 MU will diode with its stem axis oriented parallel to the CAX
FIGURE 19.7 OAR for the 25 mm collimator at depths 0.7 cm,

FIGURE 19.5 ZAP-X output factor as a function of field size. 5.0, 10.0, 20.0, and 25.0 cm.
and centered on it. Since the collimator sizes are 4, 5, 7.5, 25 cm to the surface of the water. The PDD graph for the
10, 12.5, 15, 20, and 25 mm in diameter at the machine 25 mm collimator is shown in Figure 19.6.
isocenter (45 cm SAD), the use of a small volume refer-
ence detector positioned in the primary beam is not prac- 19.3.6 Off-Axis Ratios (OARs)
tical due to its interference with the radiation measured OAR measurements shall be performed with the SRS
by the field detector. Instead, it is recommended to use diode and the same phantom and reference detector
a transmission reference ionization chamber of parallel setup as with PDD measurements. OAR profiles shall be
plate structure with a diameter significantly larger than measured in the wheel plane (parallel with the collima-
25 mm. An example of such a reference chamber is the tor wheel) and orthoplane (perpendicular to the wheel
PTW T-Ref chamber. This chamber should be positioned plane). Depth of measurement shall be performed at
at least 15 cm above the water surface of the water phan- dmax = 0.7 cm, 5.0 cm, 10.0 cm, 20.0 cm, and 25.0 cm or
tom and should be centered on the CAX. This transmis- maximum achievable depth. OAR data for the 25 mm
sion chamber is considered to be radio-translucent and collimator are shown in Figure 19.7.
not to impact the beam intensity in a measurable man-
ner. The surface of the water phantom shall be set at SSD 19.3.7 Preparation and Processing of Measured
= 45 cm and the PDD is measured from a depth of at least Beam Data before Import into the
Treatment Planning System
PDD and OAR data will be processed and converted into
a standard text format before import into Zap-X TPS.
PDD and OAR data are recommended to be smoothed
with a low-order polynomial. OAR measurements shall
be smoothed, centered, normalized to its CAX value,
and a half-beam scan is generated. The processed beam
data can now be imported into the TPS.
REFERENCES
1. G. G. Echner, W. Kilby, M. Lee, E. Earnst, S. Sayeh, A.
Schlaefer, B. Rhein, J. R. Dooley, C. Lang, O. Blanck, E.
Lessard, C. R. Maurer, W. Schlegel, “The design, physi-
cal properties and clinical utility of an IRIS collima-
tor for robotic radiosurgery,” Physics in Medicine and
FIGURE 19.6 PDD for the 25 mm collimator size. Biology 54, 5359–5380 (2009).
2. A. Schweikard, H. Shiomi, J. Adler, “Respiration track- specific reference field dose calibration and output fac-
ing in radiosurgery,” Medical Physics 31, 2738–2741 tor measurement using fixed and IRIS collimators on
(2004). the CyberKnife system,” Physics in Medicine and Biology
3. R. Alfonso, P. Andreo, R. Capote, M. S. Huq, W. Kilby, P. 57, 3741 (2012).
Kjall, T. R. Mackie, H. Palmans, K. Rosser, J. Seuntjens, 15. H. Palmans, P. Andreo, K. Christaki, M. Huq, J.
W. Ullrich, S. Vatnitsky, “A new formalism for reference Seuntjens, Dosimetry of Small Static Fields Used
dosimetry of small and nonstandard fields,” Medical in External Beam Radiotherapy: An IAEA-AAPM
Physics 35, 5179–5186 (2008). International Code of Practice for Reference and Relative
4. P. R. Almond, P. J. Biggs, B. M. Coursey, W. F. Hanson, Dose Determination (International Atomic Energy
M. S. Huq, R. Nath, D. W. Rogers, “AAPM’s TG-51 pro- Agency, Vienna, 2017).
tocol for clinical reference dosimetry of high-energy 16. S. Dieterich, G. W. Sherouse, “Experimental compari-
photon and electron beams,” Medical Physics 26, 1847– son of seven commercial dosimetry diodes for measure-
1870 (1999). ment of stereotactic radiosurgery cone factors,” Medical
5. D. T. B. Pedro Andreo, K. Hohlfeld, M. Saiful Huq, T. Kanai, F. Physics 38, 4166–4173 (2011).
Laitano, V. Smyth, S. Vynckier, Absorbed Dose Determination 17. P. Francescon, S. Beddar, N. Satariano, I.J. Das,
in External Beam Radiotherapy: An International Code of “Variation of kQclin, Qmsr fclin, fmsr for the small-field
Practice for Dosimetry based on Standards of Absorbed Dose dosimetric parameters percentage depth dose, tissue-
to Water (International Atomic Energy Agency, Report No. maximum ratio, and off-axis ratio,” Medical Physics 41,
TRS-398, Vienna, 2000). 101708 (2014).
6. T. Kawachi, H. Saitoh, M. Inoue, T. Katayose, A. 18. P. Francescon, S. Cora, C. Cavedon, “Total scatter fac-
Myojoyama, K. Hatano, “Reference dosimetry condi- tors of small beams: a multidetector and Monte Carlo
tion and beam quality correction factor for CyberKnife study,” Medical Physics 35, 504–513 (2008).
beam,” Medical Physics 35, 4591–4598 (2008). 19. P. Francescon, W. Kilby, J. Noll, L. Masi, N. Satariano,
7. M. Day and E. Aird, “Central axis depth dose data for S. Russo, “Monte Carlo simulated corrections for beam
use in radiotherapy: 1996,” British Journal of Radiology. commissioning measurements with circular and MLC
Supplement 25, 138–151 (1996). shaped fields on the CyberKnife M6 System: a study
8. T. Jordan, “Central axis depth dose data for use in radio- including diode, microchamber, point scintillator, and
therapy,” British Journal of Radiology 5, 62–109 (1996). synthetic microdiamond detectors,” Physics in Medicine
9. N. Kalach and D. Rogers, “Which accelerator photon & Biology 62, 1076 (2017).
beams are “clinic-like” for reference dosimetry pur- 20. N. C. Knutson, M. C. Schmidt, M. D. Belley, N. Nguyen,
poses?,” Medical Physics 30, 1546–1555 (2003). M. Price, S. Mutic, E. Sajo, H. H. Li, “Equivalency of
10. C. Antypas and E. Pantelis, “Performance evaluation beam scan data collection using a 1D tank and auto-
of a CyberKnife G4 image-guided robotic stereotactic mated couch movements to traditional 3D tank mea-
radiosurgery system,” Physics in Medicine and Biology surements,” Journal of Applied Clinical Medical Physics
53, 4697–4718 (2008). 19, 60–67 (2018).
11. P. Francescon, S. Cora, C. Cavedon, P. Scalchi, S. 21. G. Asmerom, D. Bourne, J. Chappelow, L. Goggin, R.
Reccanello, F. Colombo, “Use of a new type of radio- Heitz, P. Jordan, W. Kilby, T. Laing, C. Maurer Jr, J. Noll,
chromic film, a new parallel-plate micro-chamber, “The design and physical characterization of a multileaf
MOSFETs, and TLD 800 microcubes in the dosimetry collimator for robotic radiosurgery,” Biomedical Physics
of small beams,” Medical Physics 25, 503–511 (1998). & Engineering Express 2, 017003 (2016).
12. P. Francescon, S. Cora, C. Cavedon, P. Scalchi, J. 22. C. Fürweger, P. Prins, H. Coskan, B. J. Heijmen,
Stancanello, “Cyberknife dosimetric beam characteris- “Characteristics and performance of the first commer-
tics: comparison between experimental results & Monte cial multileaf collimator for a robotic radiosurgery sys-
Carlo simulation,” in Robotic Radiosurgery, Vol. 1, tem,” Medical Physics 43, 2063–2071 (2016).
edited by R.F. Mould, pp. 71–80 (CyberKnife Society 23. G. A. Weidlich, M. B. Schneider, J. R. Adler, “Self-
Press, Sunnyvale, CA, 2005). shielding analysis of the Zap-X System,” Cureus 9(12),
13. P. Francescon, W. Kilby, N. Satariano, “Monte Carlo e1917 (2017).
simulated correction factors for output factor mea- 24. G. A. Weidlich, M. B. Schneider, J. R. Adler,
surement with the CyberKnife system—results for new “Characterization of a novel revolving radiation colli-
detectors and correction factor dependence on mea- mator,” Cureus 10(2), e2146 (2018).
surement distance and detector orientation,” Physics in 25. G. A. Weidlich, M. Bodduluri, Y. Achkire, C. Lee, J. R.
Medicine and Biology 59, N11 (2014). Adler Jr, “Characterization of a novel 3 megavolt linear
14. P. Francescon, W. Kilby, N. Satariano, S. Cora, “Monte accelerator for dedicated intracranial stereotactic radio-
Carlo simulated correction factors for machine surgery,” Cureus 11(3), e4275 (2019).
Chapter 20
Dosimetry in the Presence

of Magnetic Fields
Carri Glide-Hurst
University of Wisconsin-Madison
Madison, Wisconsin
Hermann Fuchs and Dietmar Georg

Wien, Austria
Dongsu Du
City of Hope Medical Center
Duarte, California
CONTENTS
20.2 Physics Interaction Considerations 316
20.2.1 Physics of Interactions 316
20.2.2 Considerations for Photons 316
20.2.3 Considerations for Protons 316
20.2.4 Electron Return Effect Dependencies 316
20.3 Dosimetry Equipment Considerations and Performance 318
20.4 Treatment Planning Considerations 319
20.4.1 Dose Calculation/Optimization Considerations 319
20.4.2 Quantifying the Electron Return Effect 321
20.5 Potential Clinical Impact 321
20.5.1 Beam Orientation and Cavity Size 321
20.5.2 Optimization of Intensity-Modulated Radiation Therapy 322
20.6 Unmet Needs and New Directions 322
References 323
20.1 INTRODUCTION of a magnetic field. This chapter will describe the fun-
The advent of magnetic resonance-guided radiation damental physics interactions, quantify the impact on
therapy (MRgRT) has introduced new considerations for measurement devices, and describe clinical manage-
the management of radiation dosimetry in the presence ment scenarios to highlight the major considerations.
315
20.2 PHYSICS INTERACTION 20.2.2 Considerations for Photons

CONSIDERATIONS In MR-guided photon therapy, the primary beam itself
20.2.1 Physics of Interactions is not affected by the magnetic field. However, the sec-
The external magnetic field surrounding the target ondary electrons which are responsible for dose deposi-
area including neighboring organs at risk needs to be tion are affected. This may lead to distorted dose maps
considered for MRgRT. Although physics processes, compared to conventional photon therapy and may
such as particle creation and dose deposition, remain require being accounted for during treatment planning.
unaffected, the external magnetic field leads to not The most notable effects are observed in proximity to
obvious effects. In general, a moving charged particle interfaces such as a material with low atomic number Z
in a magnetic field is affected by the Lorentz force. If we and the corresponding density gradient that may cause
 charge q mov-
disregard electric fields, a particle with significant changes in the remaining particle range. For

ing with speed v in a magnetic field B experiences the example, an electron exiting a dense material into a less
Lorentz force F dense material such as air, may have sufficient remaining
range to turn around and re-enter the previous material.
This “electron return effect” [1] (ERE) leads to signifi-
F = qv × B (20.1)
cantly increased or reduced entry and exit doses espe-
cially at tissue surfaces having been reported up to 40%
The Lorentz force introduces a perpendicular deflec-
[2–4]. The ERE also impacts air-filled dosimeters [5–9]
tion of the particle path, generally resulting in a circular
and selection of measurement phantom geometries [10].
trajectory alongside the particles original track, i.e., a
spiralling track. Combining it with the centripetal force 20.2.3 Considerations for Protons
and some rearrangement we can introduce the radius of
MR-guided proton beam therapy is still in its infancy.
this circular motion, the gyroradius r
In contrary to the effects in therapeutic photon beams
described above, the primary proton therapy beam will
mv
r = ¬ (20.2) be affected. The bending of the treatment beam results in
qB¬
a lateral displacement of the Bragg-peak of up to 3 cm for
 a 250 MeV proton beam in a 1 T magnetic field [11–13].
where m is the mass of the particle and v ¬ and B¬ are
Furthermore, this bending of the beam path leads to a
the perpendicular components of the respective vectors.
small reduction in penetration depth. In addition, proton
The relativistic representation is
beams experience an effect also encountered in spectrom-
eters, an energy separation occurs and consequently the
mvγ
r= (20.3) lateral beam shape gets deformed. This holds true also for
qB
initially mono-energetic beams due to the energy/range
straggling occurring inside a patient.
1
where γ is the Lorentz factor, γ = . Rewriting In proton therapy, the energy distribution of the
2
Equation 20.3 leads to 1 − vc2 generated secondary electrons is generally lower com-
pared to photons, except for low energetic protons close
mc γ 2 − 1 to the Bragg-peak as shown in Figure 20.1. This leads
r= (20.3b)
qB to a reduced range and consequently a reduced ERE.
To date, Monte Carlo simulations performed by dif-
Resulting in ferent groups and limited experimental measurements
indicate that electron return specific dose differences at
2 material surfaces are on the order of 2–8% and depend
 E 
mc  1 + kin2  − 1 on the magnetic field strength and orientation [4, 11, 14].
 mc 
r= (20.3c)
qB 20.2.4 Electron Return Effect Dependencies
The magnetic field effect on dose distribution has been
where c is the speed of light, Ekin the kinetic energy. primarily studied through Monte-Carlo based dose
Dosimetry in the Presence of Magnetic Fields ◾ 317
FIGURE 20.1 Electron spectra of a 6 MV photon beam, a 50, FIGURE 20.3 Electron return trajectory radius (mm) in
150, and 250 MeV proton beam in water, corresponding to vacuum, depends on the magnetic field strength and electron
ranges in water of 2.2, 15.8, and 38 cm, respectively. The elec- energy. (Based on data from Raaijmakers et al. [13].)
tron spectra were scored at 2 cm depth.
simulation [15–18]. These studies found that magnetic which will affect magnetic field dose is the field strength.
field induced dose effects are dependent on the field orien- Raaijmakers et al. [13] have investigated the ERE dose
tation, field strengths [1] and beam energy. As described changes with different field strength in four situations
above, the Lorentz force will act on traveling charged par- (1) ERE at distal side of the beam, (2) the lateral ERE, (3)
ticles such as secondary electrons. Figure 20.2 shows the ERE in cylindrical air cavities, and (4) ERE in the lungs.
Lorentz force direction when the electron motion is per- Comparing the ERE for four different field strengths (B =
pendicular to magnetic field. 0.2, 0.75, 1.5, and 3 T), the ERE has minimal impact for
Generally speaking, a higher energy beam has a larger B = 0.2 T while considerable dose increases were observed
return electron trajectory’s radius and as a result, a less- for all the other fields. Figure 20.3 (using data derived
ened magnetic field dose effect. Another important fact from reference [13]) shows the radius of electron return
trajectory depends on the energy of the electron and the
magnetic field strength.
Another major factor in the impact of magnetic
field on dosimetry is orientation, including that of the
magnetic field (inline vs. perpendicular). Here, inline is
defined as the MRI scanner and linac axes of symme-
try coinciding while perpendicular involves the treat-
ment beam being perpendicular to the main magnetic
field [19]. Advantages and disadvantages of each orienta-
tion are described by Keall et al. in detail [20]. The overall
impact of geometry depends on field strength, radius of
gyration, and tissue type. Inline orientations tend to use
dose perturbation in terms of electron contamination
being focused by the MRI field whereas perpendicular
orientation tends to be linked to ERE. Another orien-
FIGURE 20.2 (Left) Typical interactions of photons and sec-
tation consideration is the surface angle of entry (both
ondary electrons with tissue in the absence of a magnetic exit and entry sides) [21]. Monte Carlo simulations of a 6
field. (Right) Electron return effect due to the presence of the MV beam in a transverse 3 T magnetic field showed that
magnetic field perpendicular to the page. entry surface angles ranging from −30° to −60° yielded
comparable entry skin doses to simulations performed An area of emerging interest is gel dosimetry, where
with no magnetic field [21]. the dosimeter is MR-visible, near-water equivalent,
and offers potential for 3D measurement and subse-
20.3 DOSIMETRY EQUIPMENT quent analysis. One such example is the radiochromic
CONSIDERATIONS AND PERFORMANCE plastic PRESAGE® dosimeter (Heuris Pharma, LLC,
When considering dosimetry equipment in magnetic Skillman, NJ) that exhibits optically visible changes
fields, several key features need to be considered: MR occurring with radiation exposure. Another experi-
safe or conditional considerations, orientation with mental dosimeter developed at MD Anderson Cancer
respect to the magnetic field, and the overall impact of Center is the radiochromic gel FOX, similar to a
the magnet on the dosimeter performance. Table 20.1 Fricke dosimeter, that is prepared in-house. Both gel
summarizes these key characteristics for published lit- dosimeters can be prepared to include air cavities and
erature pertaining to dosimeter utility for MRgRT. demonstrated the ERE that agreed well with Monte
Table 20.2 summarizes key detectors that have been Carlo simulations. Commercially available options
tested in MRgRT environments, their chief findings, and are currently under development. One such example
related references. Key takeaways from the summary of is a radiochromic gel dosimeter (ClearView, Modus
the literature is that detector performance depends on Medical Device Inc.), consisting of gellan gum, pro-
several factors and orientation as listed, thus end users pylene glycol, deionized water, and BNC tetrazolium
should be cognizant of the magnetic field dose–response prepared in a transparent vessel (15 cm diameter, 12 cm
characterization for each individual detector in clinical effective length) as shown in Figure 20.4. Optical
use as summarized in Table 20.2. readout was conducted using a Vista 16 cone-beam
FIGURE 20.4 (A) Radiochromic gel container immobilized and marked in CT using lasers, (B) axial slice of a single isocenter
multi-target brain treatment plan generated for a 0.35 T MR-linac, (C) corresponding gel dosimetry plane, (D) measured dose dis-
tribution of an intraprostatic dominant lesion prostate plans with the (E) corresponding dose profile through the dominant lesion
and urethra sparing for the measured (blue) and calculated (orange) dose distribution of the prostate plan showing good agreement
for the complex geometry. Gel experiments performed in conjunction with Modus Medical Devices, Inc., London, ON.
TABLE 20.1 Performance Consideration for Main Dosimeters Evaluated in the Presence of a Magnetic Field
Dosimeter Conditions Tested Impact of magnet on performance Reference
Ion chamber Array, IC Study using 1.5 T MR-Linac with The IC PROFILER linearity, Smit et al. [22]
PROFILER parallel plate chambers within reproducibility, pulse rate frequency
profiler. dependence, are unaffected by the
Measurement results are magnetic field.
normalized to the center The ionization chamber measurement in
reference chamber. magnetic field has some dependence on
the ion chamber shape and orientation.
Therefore, the setup orientation require
attention when normalize the reading
to central ion chamber.
Cylindrical diode array, Study using 1.5 T MR-Linac Short-term reproducibility, dose Houweling et al. [23]
ArcCHECK-MR (transverse magnetic field) and a linearity, dose rate dependence, field
regular linac. size dependence, dose per pulse
dependence, and inter-diode
variation have very little difference
with and without magnetic field.
Thermoluminescent Inserted in lung and head & neck – Magnetic field results vs. 0 T for 1.5 Steinmann et al. [24]
dosimeters (TLDs) phantom T MR-linac and MR-Co60 were 0.5%
TLD measurements and 0.6%, respectively.
Gafchromic EBT2 (Ashland, Irradiation conducted for a Co-60 – Presence of magnetic field may affect Reynoso et al. [25]
Inc., Covington, KY) MR-IGRT system (B = 0.35 T) crystal orientation and
with/without magnetic field polymerization within active layer of
under same conditions. radiochromic film.
– NetOD decreased by 8.7%, 8.0%, and
4.3% in red, green, and blue channels,
respectively. Underresponse was
dose-dependent and differed by up to
15% at 17.6 Gy.
Gafchromic EBT3 (Ashland, Photons: Compared 0 and 0.35 T Negligible absolute differences in DTA Steinmann et al. [24]
Inc., Covington, KY) beam profiles generated via were compared at 80%, 50%, and 20%
Monte Carlo to differences in profiles (maximum average difference
distance to agreement between between 0 and 0.35 T was 0.6 mm for
0 and 0.35 T beam profiles film and 0.4 mm for Monte Carlo), no
measured on EBT3 film. shift/shape change in beam profiles.
Protons: Transversal film Film performance not impacted Padilla-Cabal et al. [26]
Irradiation in PMMA with significantly, dose impact <1% at all
clinical proton beams; B = 0–1 T. positions.
Electronic Portal Imaging aS500 EPID modified with 0.1 cm – Comparable flood field, dark field, Rathee et al. [27]
Device (EPID) copper plate to function in 0.5 T linearity, uncertainty, radiation
parallel magnetic field at 6 MV. profiles, and lag found with and
without magnetic field.
– Modified EPID responds to 6 MV
photons similarly with and without a
magnetic field in parallel orientation
to the radiation beam.
optical computed tomography scanner a 3D image 20.4 TREATMENT PLANNING

of the dose was reconstructed using an iterative pro- CONSIDERATIONS
cess from 500 projections taken by step-wise rotation 20.4.1 Dose Calculation/Optimization
around the central axis of the jar a single isocenter Considerations
multi-target (SIMT) cranial stereotactic radiosur- For photon-based treatments, the primary beam is not
gery plan with 5 metastatic lesions (0.1–0.7 cm3), each affected by the magnetic field. However, dose is depos-
receiving 18 Gy. ited by secondary particles. This leads to asymmetric
TABLE 20.2 Commonly Used Detectors in Magnetic Fields, Testing Conditions, and Correction Factor Considerations Based on Current
Literature
Correction Factor
Detectors Conditions Tested Correction factor Dependence Reference
PTW 30013, PTW Measurement in different Only calculated M1.5T/M0 Depends on the field Agnew et al. [28]
31006, PTW magnetic field strength. (not correction factor): strength and air gap
31010, Extradin (0.75 T, 1.0 T, 1.5 T, 1.75 T 1.05–1.09 position around the
A1SL transverse). chamber.
PTW 30016 Simulation study in both Correction factor: Correction factor is Pojtinger et al. [29]
PTW 30015 0.35 T and 1.5 T magnetic 0.968–0.977 (0.35 T) within 0.99 for parallel
NE 2571 field (transverse and 0.954–0.98 (1.5 T) magnetic field.
PTW 30013 parallel).
PTW 30013 and Measurement in water tank. PTW 30013 Depends on chamber Spindeldreier et al. [30]
5 custom build Magnetic field ranging ~0.97 (0.3 T) radius and magnetic
farmer type from 0 T to 1.1 T. And ~0.955 (1.5 T) field orientation.
chamber Monte Carlo simulation in transverse field.
study from 0 T to 3 T.
NE2571 and Monte Carlo simulation Transverse magnetic field Dose–response and Reynolds et al. [31]
PR06C ionization dose–response in air and (perpendicular to beam) correction factor
chambers water tank with different had maximum of 11% depends on detector
magnetic field and detector response change near 1.0 T. type, energy, magnetic
orientations. Longitudinal magnetic field field strength and
(parallel to the beam) orientation.
produced slight increase in
dose–response (2% at 1.5 T).
PTW 60003 Monte Carlo simulation on Significant dose–response in Correction factors higher Reynolds et al. [32]
diamond detector, PTW 60003 diamond transverse field geometries in transverse magnetic
IBA PFD diode detector and IBA PFD (up to 20% at 1.5 T) and field for both PTW
detector diode detector. Simulated dependence on orientation 60003 and IBA PFD
the dose–response in air in magnetic field, energy, diode detector.
and in water tank with and detector. Longitudinal
different magnetic field and magnetic fields show little
detector orientations. dose–response, rising
slowly with magnetic field,
and reaching 0.5–1% at 1.5 T
regardless of orientation.
dose deposition perpendicular to the magnetic field pencil beam algorithms, remains the clinical standard,
or for larger field sizes to a small increase in the beam although first Monte Carlo based solutions are enter-
penumbra. In addition, a shift of the depth dose curve ing the market. The magnetic field related effects require
occurs. The changes were found to be dependent on new dose calculation algorithms. Such semi-analytical
magnetic field strengths and orientation, showing dose calculation algorithms taking magnetic field related
notable deviations from symmetric profiles [1, 33–35]. effects into account have already been developed, show-
Consequently, employing existing semi-analytic dose ing good agreement with full Monte Carlo simulations
calculation algorithms is not possible, and thus MRgRT [39–41]. In addition to the dose calculation itself, the bent
requires dedicated dose calculation systems as well as proton beams potentially require new optimization strat-
beam modeling that takes the presence of the magnetic egies for optimal efficiency. The Bragg-peak displacement
field into account. The currently available MR-guided is energy dependent, requiring adapted spot positions for
photon and Cobalt 60 therapy units employ Monte Carlo each energy layer. Consequently, simple linear ray cast-
based dose calculation engines for treatment planning ing to determine the initial spot positions needs to be
and optimization [36–38]. adapted. Furthermore, due to the bending of the primary
In proton beam therapy, treatment planning based on beam, the commonly employed methods to determine
semi-analytical dose calculation algorithms, so-called beam angles are not well-suited [41, 42].
FIGURE 20.5 (a) Film sandwiched in a water slab in an MR-guided Co-60 system and (b) planned and delivered dose dis-
tributions. The dashed lines represent the film measurement and the solid line represent dose calculation with magnetic field
confirming agreement with the treatment planning system.
20.4.2 Quantifying the Electron Return Effect many groups have conducted simulation and validation
An example experimental setup to measure the ERE for studies in the presence of a magnetic field. The chief
a Co-60 MRgRT machine (ViewRay, Oakwood Village, concerns are dose enhancement occurring at transitions
OH) and validate it against the planning system is shown from high to low density regions (such as soft tissue to
in Figure 20.5a. A 10 × 10 cm posterior–anterior beam was lung) and the reduction in dose in the transition between
delivered to a Gafchromic™ film sandwiched by a water low to higher density tissue. However, the impact has
slab phantom. The dose was also calculated in the ViewRay been shown to vary based on the beam arrangement,
Treatment Planning System (TPS) using a Monte Carlo number of fields, and orientation between the magnetic
based dose calculation algorithm that models the ERE. field and beam, which will be covered in the following
The resultant isodose lines (Figure 20.5b) from the calcu- sections.
lated dose distribution (solid line) and film measurement
(dashed line) are shown. The measured dose distribution 20.5.1 Beam Orientation and Cavity Size
agreed well with the calculated dose with a gamma index The MRIdian Linac system (Viewray, Ohio, US) con-
passing rate of 99.87% using passing criteria of 3%/3 mm, sists of a double-donut superconducting wide (80 cm)
confirming that the ERE at the distal end of the phantom split-bore magnet with 0.35 T field strength and a 6 MV
was accurately modeled in ViewRay TPS. flattening filter-free (FFF) Linac. The Linac components
Wooten et al. performed measurements in a Cobalt-60 are spread among five cylindrical ferromagnetic com-
with a low field 0.35 T magnet in a custom-designed het- partments around a magnetically shielded ring located
erogeneous phantom [43]. Open field and IMRT plans between the double donuts to avoid magnetic field
were generated on a CT scan of the phantom, plans were interference while the shielding also contains carbon
delivered, and radiographic film and ionization chamber fiber to absorb radiofrequency energy. To evaluate the
measurements were obtained. Ionization chamber and impact of the 0.35 T magnet on treatment planning, a
film measurements made with the phantom suggested that series of anthropomorphic phantom calculations were
the dosimetric effect was minimal, and the TPS predicts conducted using a novel pelvic end-to-end phantom
dose reasonably well for complex heterogeneous scenarios. (PETE) [44] with rectal air volumes ranging from 30
to 120 cm3. The ViewRay TPS uses a convex, nonlinear
20.5 POTENTIAL CLINICAL IMPACT programming model for IMRT optimization and fast
While few clinics have had the ability to measure the Monte Carlo dose calculation (source model simulating
impact of the magnetic field on dose due to the magnet particle transport in the linear accelerator head (exclud-
remaining on at all times once the program is clinical, ing the MLC) and patient model which performs the
FIGURE 20.6 In this dose calculation experiment, the main 0.35 T superconducting split magnet provides the static B0 field
in the horizontal field direction. (Left) Axial cross section of a pelvis phantom with simulated rectal air volumes for a single
anterior–posterior (AP) field (top) and 4-field box (bottom) highlighting the reduction in dose difference with added beams.
(Right) Dose profile of a single anterior 6 MV FFF beam traversing simulated rectal gas cavities ranging from 30 to 120 cm3
(r30–r120). (Figure credit: Laza Rakotondravohitra, Ph.D.)
transport toward the MLC opening and patient geom- the beam axis perpendicular to the magnet’s 1.5 T main
etry) based on VMC++ [45]. The end user can determine field [47]. With the exception of 1 lung case, all planning
whether the 0.35 T magnetic field is implemented in the target volume differences with the transverse magnetic
optimization or dose calculation. For this experiment, field were <4% different from no magnetic field calcu-
it was determined that increasing the rectal air volume lations. Their results also suggested that incorporated
yielded a local increase of dose due to the ERE based on many beam angles as used in IMRT, and particularly
cavity size with subsequent decrease after the cavity as for VMAT, reduce the impact of the magnetic field on
shown in Figure 20.6. However, increasing the number clinical plan quality. Nevertheless, the authors found
of treatment beams decreased the overall impact of the that by incorporating the transverse magnetic field into
presence of the air and related dose perturbation. the optimization, the maximum doses and dose to 1 cm3
were reduced, particularly for skin dose.
20.5.2 Optimization of Intensity-Modulated Overall, the current literature suggests that account-
Radiation Therapy ing for the presence of the magnetic field in the opti-
Chen et al. performed step-and-shoot intensity-modulated mization and subsequent dose calculation is appropriate
radiation therapy (IMRT) and volumetric-modulated for MRgRT. Furthermore, the impact can be lessened by
arc therapy (VMAT) plan optimizations in a 1.5 T trans- adding beams and counteracting local effects.
verse magnetic field for several disease sites with air/tis-
sue interfaces including pancreas, lung, breast, and head 20.6 UNMET NEEDS AND NEW DIRECTIONS
and neck cancer cases [46]. Treatment plans were gener- While clinical treatments with MRgRT have been
ated for the following scenarios: (1) no magnetic field, underway in the past decade, several unmet needs still
(2) with recalculation (no optimization using the mag- exist. MR-guided adaptive radiation therapy, particu-
netic field), and (3) full reoptimization including the larly for the online setting where the treatment plan is
impact of the magnetic field. Standard clinical dosimet- changed on daily basis using updated anatomy of the
ric endpoints were considered and all calculations were day, is an emerging area of clinical trial development.
performed with an Elekta Agility linac head 6 MV flat This requires significant needs for benchmarking and
beam model with fast graphic processing unit (GPU)- workflow development to ensure high confidence in the
based Monte Carlo dose calculation (1% statistical delivered plan. Dosimetric calibration protocols in the
uncertainty/1 mm grid size). Of note is that the Elekta presence of magnetic fields are currently under devel-
Agility linac implemented in this work is very similar opment at standardized laboratories but have yet to be
to the 160-leaf multi-leaf collimator used in the Elekta implemented into standard clinical practice. Recently,
Unity MRI-linac (Elekta AB, Stockholm, Sweden) with real-time MRI guidance in proton or particle therapy
(MRIgPT) has been introduced [48] and a proof of con- 12. S. M. Schellhammer, S. Gantz, A. Lühr, B. M. Oborn, M.
cept system has been used to acquire MR images of a Bussmann, A. L. Hoffmann, “Experimental verification
of magnetic field-induced beam deflection and Bragg
phantom during proton-beam delivery [49], thus this
peak displacement for MR-integrated proton therapy,”
modality appears to be the next generation of MRgRT. Medical Physics 45(7), 3429–3434 (2018).
Future work will likely encompass dosimetric and plan- 13. J. Hartman, C. Kontaxis, G. Bol, et al., “Dosimetric
ning studies of particle and proton therapy in the pres- feasibility of intensity modulated proton therapy in a
ence of magnetic fields. transverse magnetic field of 1.5 T,” Physics in Medicine
& Biology 60(15), 5955 (2015).
REFERENCES 14. A. Lühr, L. N. Burigo, S. Gantz, S. M. Schellhammer,
A. L. Hoffmann, “Proton beam electron return effect:
1. A. J. Raaijmakers, B. W. Raaymakers, J. J. Lagendijk, Monte Carlo simulations and experimental verifica-
“Magnetic-field-induced dose effects in MR-guided tion,” Physics in Medicine and Biology 64(3), 035012
radiotherapy systems: Dependence on the magnetic (2019).
field strength,” Physics in Medicine and Biology 53(4), 15. Y. M. Yang, M. Geurts, J. B. Smilowitz, E. Sterpin, B.
909–923 (2008). P. Bednarz, “Monte Carlo simulations of patient dose
2. A. E. Rubinstein, Z. Liao, A. D. Melancon, et al., “A perturbations in rotational-type radiotherapy due to
Monte Carlo study of magnetic-field-induced radiation a transverse magnetic field: A tomotherapy investiga-
dose effects in mice,” Medical Physics 42(9), 5510–5516 tion,” Medical Physics 42(2), 715–725 (2015).
(2015). 16. B. M. Oborn, P. E. Metcalfe, M. J. Butson, A. B. Rosenfeld,
3. B. Oborn, P. E. Metcalfe, M. Butson, A. B. Rosenfeld, P. J. Keall, “Electron contamination modeling and skin
“High resolution entry and exit Monte Carlo dose cal- dose in 6 MV longitudinal field MRIgRT: Impact of the
culations from a linear accelerator 6 MV beam under MRI and MRI fringe field,” Medical Physics 39(2), 874–
the influence of transverse magnetic fields,” Medical 890 (2012).
Physics 36(8), 3549–3559 (2009). 17. B. M. Oborn, Y. Ge, N. Hardcastle, P. E. Metcalfe, P. J.
4. B. W. Raaymakers, A. J. Raaijmakers, J. J. Lagendijk, Keall, “Dose enhancement in radiotherapy of small lung
“Feasibility of MRI guided proton therapy: Magnetic tumors using inline magnetic fields: A Monte Carlo
field dose effects,” Physics in Medicine & Biology 53(20), based planning study,” Medical Physics 43(1), 368 (2016).
5615 (2008). 18. J. M. Park, S. Y. Park, H. J. Kim, H. G. Wu, J. Carlson,
5. D. J. O’Brien, J. Dolan, S. Pencea, N. Schupp, G. O. J. I. Kim, “A comparative planning study for lung SABR
Sawakuchi, “Relative dosimetry with an MR-linac: between tri-Co-60 magnetic resonance image guided
Response of ion chambers, diamond, and diode detec- radiation therapy system and volumetric modulated arc
tors for off-axis, depth dose, and output factor measure- therapy [published online ahead of print 2016/07/13],”
ments,” Medical Physics 45(2), 884–897 (2018). Radiotherapy & Oncology 120(2), 279–285 (2016).
6. V. N. Malkov and D. Rogers, “Monte Carlo study of ion- 19. D. E. Constantin, R. Fahrig, P. J. Keall, “A study of the
ization chamber magnetic field correction factors as a effect of in-line and perpendicular magnetic fields on
function of angle and beam quality,” Medical Physics beam characteristics of electron guns in medical linear
45(2), 908–925 (2018). accelerators,” Medical Physics 38(7), 4174–4185 (2011).
7. D. O’Brien, D. Roberts, G. Ibbott, G. Sawakuchi, 20. P. J. Keall, M. Barton, S. Crozier, “The Australian
“Reference dosimetry in magnetic fields: Formalism magnetic resonance imaging-linac program [pub-
and ionization chamber correction factors,” Medical lished online ahead of print 2014/06/17],” Seminars in
Physics 43(8 Part 1), 4915–4927 (2016). Radiation Oncology 24(3), 203–206 (2014).
8. I. Meijsing, B. W. Raaymakers, A. Raaijmakers, et al., 21. B. Oborn, P. E. Metcalfe, M. Butson, A. B. Rosenfeld,
“Dosimetry for the MRI accelerator: The impact of a “Monte Carlo characterization of skin doses in 6 MV
magnetic field on the response of a Farmer NE2571 ion- transverse field MRI-linac systems: Effect of field size,
ization chamber,” Physics in Medicine & Biology 54(10), surface orientation, magnetic field strength, and exit
2993 (2009). bolus,” Medical Physics 37(10), 5208–5217 (2010).
9. M. Reynolds, S. Rathee, B. G. Fallone, “Ion cham- 22. K. Smit, J. Kok, J. Lagendijk, B. Raaymakers, “Performance
ber angular dependence in a magnetic field,” Medical of a multi-axis ionization chamber array in a 1.5 T mag-
Physics 44(8), 4322–4328 (2017). netic field,” Physics in Medicine & Biology 59(7), 1845
10. D. J. O’Brien and G. O. Sawakuchi, “Monte Carlo study (2014).
of the chamber-phantom air gap effect in a magnetic 23. A. Houweling, J. De Vries, J. Wolthaus, et al.,
field,” Medical Physics 44(7), 3830–3838 (2017). “Performance of a cylindrical diode array for use in a 1.5
11. H. Fuchs, P. Moser, M. Gröschl, D. Georg, “Magnetic T MR-linac,” Physics in Medicine & Biology 61(3), N80
field effects on particle beams and their implications (2016).
for dose calculation in MR-guided particle therapy,” 24. A. Steinmann, D. O’Brien, R. Stafford, et al.,
Medical Physics 44(3), 1149–1156 (2017). “Investigation of TLD and EBT3 performance under
the presence of 1.5 T, 0.35 T, and 0 T magnetic field 37. S. B. Ahmad, A. Sarfehnia, M. R. Paudel, et al.,
strengths in MR/CT visible materials,” Medical Physics “Evaluation of a commercial MRI Linac based Monte
46(7), 3217–3226 (2019). Carlo dose calculation algorithm with geant 4,” Medical
25. F. J. Reynoso, A. Curcuru, O. Green, S. Mutic, I. J. Das, L. Physics 43(2), 894–907 (2016).
Santanam, “Magnetic field effects on Gafchromic-film 38. M. R. Paudel, A. Kim, A. Sarfehnia, et al., “Experimental
response in MR-IGRT,”Medical Physics 43(12), 6552–6556 evaluation of a GPU-based Monte Carlo dose calcula-
(2016). tion algorithm in the Monaco treatment planning sys-
26. F. Padilla-Cabal, P. Kuess, D. Georg, H. Palmans, L. tem,” Journal of Applied Clinical Medical Physics 17(6),
Fetty, H. Fuchs, “Characterization of EBT3 radiochro- 230–241 (2016).
mic films for dosimetry of proton beams in the presence 39. R. Wolf and T. Bortfeld, “An analytical solution to pro-
of magnetic fields,” Medical Physics 46(7), 3278–3284 ton Bragg peak deflection in a magnetic field,” Physics in
(2019). Medicine & Biology 57(17), N329 (2012).
27. S. Rathee, B. G. Fallone, S. Steciw, “EPID’s response 40. F. Padilla-Cabal, D. Georg, H. Fuchs, “A pencil beam
to 6 MV photons in a strong, parallel magnetic field,” algorithm for magnetic resonance image-guided proton
Medical Physics 46(1), 340–344 (2019). therapy,” Medical Physics 45(5), 2195–2204 (2018).
28. J. Agnew, F. O’Grady, R. Young, S. Duane, G. J. Budgell, 41. S. M. Schellhammer and A. L. Hoffmann, “Prediction
“Quantification of static magnetic field effects on radio- and compensation of magnetic beam deflection in
therapy ionization chambers,” Physics in Medicine & MR-integrated proton therapy: A method optimized
Biology 62(5), 1731 (2017). regarding accuracy, versatility and speed,” Physics in
29. S. Pojtinger, O. S. Dohm, R.-P. Kapsch, D. Thorwarth, Medicine & Biology 62(4), 1548 (2017).
“Ionization chamber correction factors for MR-linacs,” 42. O. Jäkel and J. Debus, “Selection of beam angles for
Physics in Medicine & Biology 63(11), 11NT03 (2018). radiotherapy of skull base tumours using charged par-
30. C. Spindeldreier, O. Schrenk, A. Bakenecker, et al., ticles,” Physics in Medicine & Biology 45(5), 1229 (2000).
“Radiation dosimetry in magnetic fields with Farmer- 43. H. Wooten, O. Green, H. Li, V. Rodriguez, S. Mutic,
type ionization chambers: Determination of magnetic “Measurements of the electron-return-effect in a com-
field correction factors for different magnetic field mercial magnetic resonance image guided radiation
strengths and field orientations,” Physics in Medicine & therapy system. WE-G-17A-4. Paper presented at:
Biology 62(16), 6708 (2017). AAPM Annual Meeting (Austin, TX, 2014).
31. M. Reynolds, B. Fallone, S. Rathee, “Dose response of 44. J. M. Cunningham, E. A. Barberi, J. Miller, J. P. Kim, C.
selected ion chambers in applied homogeneous trans- K. Glide-Hurst, “Development and evaluation of a novel
verse and longitudinal magnetic fields,” Medical Physics MR-compatible pelvic end-to-end phantom,” Journal of
40(4), 042102 (2013). Applied Clinical Medical Physics 20(1), 265–275 (2019).
32. M. Reynolds, B. Fallone, S. Rathee, “Dose response of 45. F. Hasenbalg, M. K. Fix, E. J. Born, R. Mini, I. Kawrakow,
selected solid state detectors in applied homogeneous “VMC++ versus BEAMnrc: A comparison of simulated
transverse and longitudinal magnetic fields,” Medical linear accelerator heads for photon beams,” Medical
Physics 41(9), 092103 (2014). Physics 35(4), 1521–1531 (2008).
33. A. Raaijmakers, B. W. Raaymakers, J. J. Lagendijk, 46. X. Chen, P. Prior, G. P. Chen, C. J. Schultz, X. A. Li,
“Experimental verification of magnetic field dose effects “Dose effects of 1.5 T transverse magnetic field on tis-
for the MRI-accelerator,” Physics in Medicine & Biology sue interfaces in MRI-guided radiotherapy,” Medical
52(14), 4283 (2007). Physics 43(8 Part 1), 4797–4802 (2016).
34. H. K. Looe, B. Delfs, D. Poppinga, D. Harder, B. Poppe, 47. M. Glitzner, P. Woodhead, P. Borman, J. Lagendijk,
“Magnetic field influences on the lateral dose response B. Raaymakers, “MLC-tracking performance on the
functions of photon-beam detectors: MC study of Elekta unity MRI-linac,” Physics in Medicine & Biology
wall-less water-filled detectors with various densities,” 64(15), 15NT02 (2019).
Physics in Medicine & Biology 62(12), 5131 (2017). 48. G. P. Liney, B. Whelan, B. Oborn, M. Barton, P. Keall,
35. S. J. Woodings, J. Bluemink, J. de Vries, et al., “Beam “MRI-linear accelerator radiotherapy systems [pub-
characterisation of the 1.5 T MRI-linac,” Physics in lished online ahead of print 2018/09/10],” Clinical
Medicine & Biology 63(8), 085015 (2018). Oncology (Royal College of Radiologists (Great Britain))
36. B. Raaymakers, I. Jürgenliemk-Schulz, G. Bol, et al., 30(11), 686–691 (2018).
“First patients treated with a 1.5 T MRI-Linac: Clinical 49. S. Schellhammer, L. Karsch, J. Smeets, et al., “OC-0605:
proof of concept of a high-precision, high-field MRI First in-beam MR scanner for image-guided proton
guided radiotherapy treatment,” Physics in Medicine & therapy: Beam alignment and magnetic field effects,”
Biology 62(23), L41 (2017). Radiotherapy and Oncology 127, S318–S319 (2018).
Chapter 21
Helical Tomotherapy
Treatment and Dosimetry
Reza Taleei
Sidney Kimmel Medical College at Thomas Jefferson University
Sarah Boswell
Accuray Incorporated
Madison, Wisconsin
CONTENTS
21.2 System Details 327
21.2.1 Beam Production and Monitoring 327
21.2.2 Beam Collimation 329
21.2.3 MVCT Imaging 332
21.3 Patient Positioning and Motion Synchronization 333
21.3.1 Patient Positioning 334
21.3.2 Motion Synchronization 334
21.4 Optimization and Dose Calculation for IMRT Plans 335
21.4.1 User-Defined Parameters 335
21.4.2 Beamlets: The Basic Unit of Optimization 336
21.4.3 Optimization Process 336
21.4.4 Dose Calculation by Convolution/Superposition 337
21.5 Quality Assurance 338
21.5.1 Machine QA 338
21.5.2 Absolute Dose Measurement and Calibration 338
21.5.3 Patient Quality Assurance 339
21.6 Summary 339
Acknowledgments 340
Conflict of Interest 340
Notes 340
References 340
325
21.1 INTRODUCTION
The concept of helical tomotherapy was formed in the late
1980s with a series of projects defined by Mackie’s research
group at the University of Wisconsin [1]. The idea started
by investigating the delivery of non-uniform optimized
beams introduced by Brahme [1, 2]. The project evolved
into designing a linac delivery system on a CT ring gan-
try with helical delivery [3]. The helical TomoTherapy®
System (Accuray Incorporated) is an intensity-modulated
radiation therapy (IMRT) delivery system capable of
treating targets with thousands of coplanar beamlets as
the patient continuously translates through the beam in
the longitudinal direction (see Figures 21.1 and 21.2).
The following list is summarized and freely adapted
from the Physics Essentials Guide with permission by
Accuray Incorporated [4]. The helical tomotherapy sys-
tem features:
FIGURE 21.2 Major components of the helical tomotherapy
system: (1) linac; (2) jaws; (3) MLC; (4) beam; (5) CT detector;
• A patient couch that travels at a constant velocity (6) beam stop. (Image reprinted from [4]; used by permission
(per treatment plan) in the longitudinal direction. of Accuray Incorporated.)
The target, which can be less than 1 cm to more
than 1 m in length, is treated from the superior end • A standing wave linear accelerator (6 MV),
to the inferior end as the couch translates through mounted on the rotating gantry with source-to-
the beam. axis distance 85 cm. See Figure 21.5. The same lin-
• A gantry that rotates at a constant velocity (per ear accelerator can be tuned to a lower fluence rate
treatment plan) in the clockwise direction.1 The and energy to produce the beam for megavoltage
combined effect of the couch translation and computed tomography (MVCT) imaging.
gantry rotation is a helical delivery pattern on • Transmission monitor chambers, to provide feed-
the patient, with 3264 coplanar beamlets avail- back for achieving a stable dose rate during the
able for optimization at each gantry rotation. delivery. See Figure 21.6.
See Figures 21.3 and 21.4 for sample plan dose
distributions. • Fixed tungsten collimator that produces a fan
beam. The beam size is 40 cm in the transverse
direction at isocenter.
• A single pair of moveable jaws that collimate the
beam to 1 cm, 2.5 cm, or 5 cm in the longitudi-
nal direction at isocenter. To sharpen the supe-
rior/inferior dose fall-off, the jaws gradually open
as the superior end of the target enters the beam,
and gradually close as the inferior end of the target
exits the beam. See Figure 21.7.
• A 64-leaf, binary multi-leaf collimator (MLC) for
intensity modulation in the transverse direction.
See Figure 21.8.
FIGURE 21.1 Helical tomotherapy system. (Image used by • A single-slice MVCT detector mounted opposite
permission of Accuray Incorporated.) the linac.
Helical Tomotherapy Treatment and Dosimetry ◾ 327
FIGURE 21.3 Head and neck tomotherapy plan. The prescription to PTV1 is 70 Gy in 35 fractions. Fraction delivery time
is 285 s.
• Lead beam stop, opposite the linac and beyond the a uniform dose distribution. For simple cases that do
MVCT detector, to attenuate the primary beam not require sophisticated IMRT optimization, a 3D
to the extent that room shielding calculations are Conformal planning option is available. The delivery
dominated by head leakage and patient scatter. geometry of a 3D Conformal plan is identical to its
IMRT counterpart (helical or fixed angle), but the MLC
• The tomotherapy gantry is enclosed by gantry cov-
delivery instructions are determined differently for a
ers. No accessories can be attached to the linear
3D Conformal plan, without user-provided optimiza-
accelerator. There is no light field or front pointer.
tion inputs [4].
In addition, some systems include motion tracking
equipment: 21.2 SYSTEM DETAILS
21.2.1 Beam Production and Monitoring
• A kV imaging tube and flat panel detector for The high voltage power supply generates direct current
acquiring radiographs of the patient at intervals to charge a solid-state modulator. The solid-state modu-
during treatment. See Figure 21.9. lator acts as a switch and provides pulsed high voltage
• An optical camera for continuously monitoring direct current to the magnetron (300 Hz for treatment,
the respiratory waveform, by tracking light-emit- 134 Hz for imaging). The magnetron generates the radio-
ting markers attached to the patient surface. See frequency power used by the linac to accelerate electrons.
Figures 21.10. The standing wave linac accelerates electrons from the
electron gun to the transmission-style tungsten target,
The helical tomotherapy system is an FFF (Flattening where photons are produced via the bremsstrahlung
Filter Free) system; see Figure 21.11. All tomotherapy effect. When operated in treatment mode, the linac pro-
treatments are delivered with moving leaves to achieve duces a 6 MV energy spectrum, as shown in Figure 21.12.
FIGURE 21.4 Above: Helical tomotherapy IMRT breast plan. The prescription is 50 Gy in 25 fractions. Fraction delivery time
is 650 s. Below: 3D breast plan with two fixed gantry angles. The prescription is Canadian Fractionation of 42.56 Gy in 16 frac-
tions. Fraction delivery time is 325 s.
Dose monitor chambers are located downstream of the treatment, such as high dose rate (HDR) brachytherapy
linac and upstream of the jaws and MLC; see Figure 21.6. or cobalt external beam therapy. (By contrast, in conven-
The monitor chambers are composed of two layers of tional linac-based systems, the treatment plan specifies a
independently sealed, ceramic, parallel plate chambers. number of monitor units, and the beam turns off when
The chambers are transmission chambers: the beam the exact number of monitor units has been delivered).
passes through the chambers before entering the colli- The time-based planning and delivery approach is well-
mation system. The monitor chambers provide feedback suited to tomotherapy treatments, for which the couch
to the Dose Control System, which is used to adjust the translates at a constant velocity, and targets are treated
magnitude of the magnetron pulse that establishes the from the superior end to the inferior end. Each voxel
acceleration field of the linac. The Dose Control System within the target passes through the beam plane during
stabilizes the dose rate so that it does not fluctuate cycli- a limited amount of time, and is affected only by the dose
cally with gantry rotation. An additional feedback loop delivered while that voxel is in the beam plane (or within
on the gun current helps to stabilize the injector current, scatter’s reach of the beam plane).
which affects the beam energy. Although tomotherapy delivery is not controlled by
Tomotherapy treatments are planned and delivered monitor units, monitor units are displayed for infor-
based on time, given a known, fixed dose rate. In this mational purposes. The conversion from raw monitor
aspect, a tomotherapy treatment is like radioactive source counts to monitor units can be calibrated by the physicist
FIGURE 21.7 Jaws. The jaw setting determines the field size
in the longitudinal direction. (Image reprinted from [4]; used
by permission of Accuray Incorporated.)
During treatment, the raw counts on the moni-

FIGURE 21.5 A side view of the beamline, showing (1) linac;
(2) primary collimator and transmission monitor chambers; tor chambers are continually monitored. If the count
(3) jaws; (4) MLC. (Image reprinted from [4]; used by permis- rate falls outside system tolerances, an interlock will
sion of Accuray Incorporated.) stop the treatment. A make-up procedure can then be
delivered.
so that the display is meaningful, i.e., the display reads
1 cGy/MU under some reference conditions chosen by 21.2.2 Beam Collimation
the user. The monitor unit display calibration does not The linac is supported by a tungsten fixture that con-
control the beam delivery. tains a rectangular aperture to collimate the radiation
FIGURE 21.8 MLC. There are 66 leaves, but the outermost

two leaves always remain closed. MLC leaves move along the
longitudinal direction. The amount of time that the leaves
FIGURE 21.6 Transmission-style dose monitor chambers. stay open controls the intensity distribution in the transverse
(Image reprinted from [4]; used by permission of Accuray direction. (Image reprinted from [4]; used by permission of
Incorporated.) Accuray Incorporated.)
FIGURE 21.9 kV detector (1), tube (2), and generator (3) mounted on gantry for motion tracking. (Image reprinted from [4];
used by permission of Accuray Incorporated.)
FIGURE 21.10 Motion synchronization. To monitor the patient’s breathing, an optical camera mounted to the ceiling tracks
the position of light-emitting markers attached to the patient surface. To continuously monitor the target position, the posi-
tions of the markers are correlated with intermittent kV radiographs collected during treatment delivery. (Figure by Sam
Durst of Accuray Incorporated. Image reprinted from [27]; used by permission of Accuray Incorporated.)
MLC; see Figure 21.7. The jaws control the width of the
radiation beam in the longitudinal direction; there is no
collimator rotation. The field width is selected during
treatment planning. Available field widths include: 5.0 cm,
2.5 cm, and 1.0 cm; these are nominal full-width at half-
maximum (FWHM) values projected to isocenter.
Treatments can be delivered with either fixed or mov-
ing (dynamic) jaws. When using fixed jaws, the 5-cm
field width has the advantage of the shortest treatment
time, but the superior/inferior penumbra is sharpest for
the 1-cm beam. Moving the jaws dynamically during
treatment allows the planner to achieve treatment times
similar to the 5-cm beam, without compromising the
superior/inferior penumbra.
Figure 21.13 illustrates a dynamic jaws treatment
FIGURE 21.11 Sample transverse profile of the tomotherapy
with the 5-cm field width. The optimizer opens beam-
beam. The unflattened fluence distribution of photons pro- lets superior and inferior to the target, where necessary
duced by bremsstrahlung interactions in the target is roughly to achieve a uniform target dose. The first beamlets
twice as high in the center as it is on the edges. This profile available for optimization occur when the superior end
was measured in a water tank at 85 cm SSD, with the 5 cm of the target crosses the inferior edge of the beam, and
jaw size and all leaves open. (Image used by permission of the last beamlets available for optimization occur when
Accuray Incorporated.) the inferior edge of the target crosses the superior edge
of the beam. By narrowing the jaw opening at the supe-
cone beam produced by the linac into a fan beam. This rior and inferior ends of the target, the penumbra is
rectangular aperture cannot be adjusted. improved. By using a larger field width internal to the
The tomotherapy system has only one set of jaws, target, the treatment time is improved.
which are made of a tungsten alloy, and located down- The MLC is located downstream of the jaws and pro-
stream of the rectangular aperture and upstream of the vides intensity modulation along the transverse direction
during treatment (see Figure 21.8). There are 64 binary,
moveable leaves (tongue and groove design), arranged
in banks of 32 leaves each. The two banks of leaves are
interlaced and do not meet in the center of the field; when
closed, each leaf covers the full longitudinal extent of the
beam. The leaves are 10 cm high, and made of tungsten.
The leaves are arranged on a curve with a focus near the
radiation source. Leaves are equiangular with an average
width of 6.25 mm when projected to a line perpendicular
to the beam at 85 cm distance. The maximum transverse
width of the beam, when all leaves are open, is 6.25 mm
per leaf × 64 leaves = 400 mm at isocenter.
Each leaf is controlled independently and pneumati-
cally (unlike other MLC’s, the tomotherapy MLC has no
motors to replace). The leaves are binary: they are always
programmed to be opened or closed. When opened,
the leaves are fully removed from the beam, and the
FIGURE 21.12 Example energy spectrum for the tomother- fan beam shape is determined by the primary collima-
apy treatment beam. (Image reprinted from [4]; used by per- tor. When closed, each leaf covers the full longitudinal
mission of Accuray Incorporated.) extent of the beam. MLC leaf-open times are stored as
FIGURE 21.13 Stages of a dynamic jaws treatment. The couch travels at a constant velocity in the +IEC Y (longitudinal)
direction, into the beam. The intersection of the thicker perpendicular dashed lines marks the radiation isocenter. The thin-
ner dashed line represents the 5-cm field width. The lighter and darker solid lines represent the front and back jaw positions,
respectively (the back jaw is always located on the +IEC Y side of the front jaw). As the target enters the beam, the back jaw
moves just ahead of the target (beginning with a minimum field width of about 1 cm), continuously expanding until the 5-cm
field width is reached. The jaws are fully open to the 5-cm field width while the central portion of the target is being treated,
and the leaves modulate during the entire treatment. As the inferior portion of the target begins to move past the beam, the
front jaw position narrows to about 1 cm, following the inferior target edge. Discrete jaw positions are illustrated here, but the
actual jaw motion is continuous. (Image reprinted from [4]; used by permission of Accuray Incorporated.)
floating-point values, resulting in a very large number of data is used by the planning system to adjust leaf-open
discrete intensity levels (higher precision than deliver- time commands to achieve the planned leaf-open times.
able). The number of possible intensity levels for a single MLC leakage can be measured by comparing a “leak-
beamlet depends on the linac pulse rate and the projec- age” film exposed with the jaws open and leaves closed,
tion time (see Section 21.4.2 for a discussion of projec- to a “reference” film exposed with the jaws open and
tion time). leaves open. The maximum rate in the MLC leakage film
To ensure that the leaves open and close at the is less than 2% of the rate in the center of the reference
planned times, each leaf is continuously monitored by film, and the average rate is less than 0.5% of the average
two optical position sensors. If the leaves fail to meet rate in the reference film.
their positions, the treatment is interrupted. Newer
tomotherapy systems use continuous feedback from the 21.2.3 MVCT Imaging
optical sensors to correct the timing of commands to MVCT images are acquired using the treatment linac.
each leaf during treatment. On older systems, the field The linac parameters are adjusted for imaging to achieve
service engineer uses the CT detector to measure the a lower energy (approximately 3 MV spectrum) and
average response of the MLC to timing commands; this lower dose rate than the treatment beam.
The on-board detector is an array of parallel plate are 1.6 mm in diameter, with center-to-center spacing
ionization chambers (pressurized Xenon-filled cavities of 3.2 mm. The ability to resolve objects in the longitu-
separated by stainless steel septa) housed in an alumi- dinal direction depends on the collimation size, couch
num case. The detector consists of about 640 channels, motion, and the spacing of the reconstructed slices. Slice
520 of which are within reach of the imaging beam and sensitivity profiles can be measured using a method
used for image reconstruction. The size of the detector described by Polacin et al. [6] and implemented for
channels is approximately 2.5 cm in the longitudinal tomotherapy by Chao et al. [7, 8]: a thin disk is scanned,
direction, and approximately 1.24 mm in the transverse and images are reconstructed at 0.5 mm intervals (using
direction (since tomotherapy uses a single-slice detector, non-commercial software outside of the treatment sys-
tomotherapy MVCT imaging is fan beam CT, not cone tem). The average pixel values for the high contrast object
beam CT). are plotted as a function of slice position. The FWHM of
When a photon interacts in the detector, secondary the profiles is 3.9 mm ± 1.0 mm for Fine pitch, 5.3 mm
electrons liberate additional electrons in the gas volume ± 1.0 mm for Normal pitch, and 7.2 mm ± 1.0 mm for
through ionization. Under the influence of an applied Coarse pitch.
electric field (−200 V bias on every other plate), charges To help maintain the stability of CT numbers pro-
flow to the collecting electrode. duced by the system, the physicist can run a weekly
Data is acquired helically with a six-second gan- CT number calibration procedure, which scales the CT
try period and 134 Hz pulse rate, providing 804 linac numbers by comparing measurements in air and in a
pulses and detector measurements per gantry rotation. phantom against their expected values.
The jaw width for imaging projects to approximately The primary purpose of the MVCT imaging system
4 mm at isocenter. The couch velocity is determined is to acquire daily pre-treatment images to verify the
by the user-selectable acquisition pitch. Couch speeds patient setup. However, MVCT images can also be used
are 4 mm/rotation (Fine), 8 mm/rotation (Normal), or for treatment planning, provided the 39-cm field of view
12 mm/rotation (Coarse). Thus, a Coarse scan requires is adequate for the patient. A convenient feature of using
one-third of the time and delivers one-third of the a megavoltage beam is the lower attenuation through
patient imaging dose as compared to a Fine scan, but metal, leading to a significant reduction in streak arti-
the Fine scan has the best longitudinal resolution. The facts for patients with high-density features such as
patient imaging dose for a Fine scan is less than 3 cGy, prostheses and dental fillings [9, 10] (see Figure 21.14).
when measured with an ion chamber near the center of In addition, daily MVCT images can be used for
a 30-cm diameter phantom, over a scan distance that adaptive radiation therapy, to monitor changes in the
spans the entire chamber collection volume. patient anatomy. The daily dose can be calculated on
The imaging system uses two different types of algo- the daily MVCT images, deformed to the planning
rithms for image reconstruction: filtered backprojection image and summed across fractions to see the cumu-
and iterative reconstruction [5]. While analytical algo- lative effect in the patient. This information is helpful
rithms such as filtered backprojection are based on a single to determine if the patient needs to be re-planned due
reconstruction, an iterative algorithm uses multiple recon- to weight loss, tumor shrinkage, or other anatomical
structions and statistical weighting to reduce the image changes.
noise while preserving spatial resolution and image fidelity. Finally, MVCT detector data has significant utility as
Helical data is interpolated onto discrete slices a quality assurance (QA) tool, allowing the physicist to
along the longitudinal axis. Two or four images are monitor the consistency of the transverse beam profile for
reconstructed per gantry rotation, resulting in image various automated machine QA tasks within the TQA®
reconstruction intervals of 2 mm or 1 mm (Fine), software (Total Quality Assurance, Accuray Incorporated).
4 mm or 2 mm (Normal), and 6 mm or 3 mm (Coarse).
Reconstruction is performed on a 512 × 512 pixel 21.3 PATIENT POSITIONING AND
matrix. The reconstruction field of view is approxi- MOTION SYNCHRONIZATION
mately 39 cm. Helical tomotherapy calculations on the planning CT
The spatial resolution in the transverse plane is suf- image exhibit uniform target coverage and a high degree
ficient to resolve pin holes in a CT resolution plug that of critical structure sparing. To successfully deliver the
FIGURE 21.14 MVCT images produced on the Radixact® Treatment Delivery System (a form of helical tomotherapy). This
patient has a bone prosthesis, but due to the MV energy, the image is not obscured by streak artifacts. The field of view of the
MVCT is limited to 39 cm diameter. (Images courtesy of the Radiation Oncology Department at the Froedert and Medical
College of Wisconsin.)
plan to the patient, the daily treatment position must be significant target motion is an unavoidable reality, and
consistent with the planned position. the physician may choose to expand target volumes to
account for motion.
21.3.1 Patient Positioning
At simulation, the patient receives small tattoo dots or 21.3.2 Motion Synchronization
semi-permanent marks.2 Radiopaque markers are tem- To synchronize the beam delivery with involuntary
porarily affixed to the patient surface at the same posi- patient motion such as motion caused by respiration or
tion of these tattoos so that they will be visible in the CT digestion, some helical tomotherapy systems have a kV
image. During planning, patient-specific laser positions imaging source and a flat-panel detector mounted on
are assigned so that they correspond to the patient’s the rotating gantry, offset 90 degrees from the linac and
radiopaque markers on the planning CT. CT detector; see Schnarr et al. [11]. After initially align-
In preparation for daily treatment, the patient is aligned ing the patient based on a pre-treatment MVCT image,
so that their tattoos correspond to the patient-specific laser motion detection is accomplished by collecting 2D kV
positions in the treatment room. Next, a helical MVCT radiograph images at selected gantry angles while the
image is used to visualize the patient’s internal anatomy treatment is in progress (two to six images per gan-
in 3D, to ensure that the target is in its planned position. try rotation). The position of the target or a surgically
(MVCT images are acquired with all leaves open. Due implanted fiducial set is determined from the radio-
to MLC leaf modulation, the system cannot reconstruct graphs. Motion synchronization is accomplished by
MVCT images during treatment delivery.) continuously updating jaw and leaf positions to address
MVCT images are reconstructed and rigidly reg- target motion.
istered to the planning kVCT image by automatic Target motion can follow a cyclical pattern (e.g.,
or manual methods. Translational adjustments are breathing motion), or the motion can be irregular (e.g.,
applied to the couch, prior to delivering the daily digestive motion). The time between sequential radio-
treatment. Rotations about the patient’s longitudinal graphs can range from 1 s to 55 s, depending on the gan-
axis (roll offsets) are implemented by adjusting the try period and the number and placement of imaging
gantry start angle of the daily treatment. The tomo- angles. The motion tracking system must estimate the
therapy couch does not have rotational degrees of target position in between the radiographs:
freedom, so if pitch or yaw adjustments are required,
the patient is manually adjusted in the treatment • To track respiratory motion in between the times
room and another MVCT image is acquired to verify of radiograph acquisition, light emitting markers
the adjustment. are placed on the patient surface, and their posi-
Accurate treatment requires that the target retain its tion is monitored by an optical camera system (see
registered position until the daily treatment is complete. Figure 21.10).3 The system builds a statistical model
Target motion can be minimized through patient educa- to correlate the continuously monitored marker
tion and restraint devices. However, in some treatments, positions with the target structure or fiducials in
the intermittent radiographs. The jaws and leaves laterally beyond the field of view, directional block-
lock onto the target position as determined by this ing can be used to prevent beam angles from entering
Correlation Model. through unspecified amounts of tissue.
The planner also specifies the prescription and opti-
• For non-respiratory motion, the target is assumed
mization constraints, and basic information about
to maintain the last measured position until the
the plan geometry, including the treatment modality
next radiograph is acquired. A statistical Position
(helical or fixed-angle), jaw size, jaw mode (fixed or
Model predicts three-dimensional (3D) fiducial
dynamic), calculation resolution, pitch, and modulation
positions from sequential, two-dimensional (2D)
factor. Pitch and modulation factor are discussed below.
radiographs. Each time a radiograph is acquired,
For helical plans, the pitch defines the relationship
the collimation system locks onto the updated tar-
between couch and gantry positions: pitch = d/W, where
get position.
d is the couch travel distance per gantry rotation, and
W is the nominal field width projected to isocenter
A new model is built for each treatment fraction, using
(1.0 cm, 2.5 cm, or 5.0 cm). The maximum allowed pitch
only tracking data collected during that session. The
is 0.5. The helical pitch may be conceptualized as the
beam-on time for treatment with motion tracking is
“tightness,” or amount of overlap, in the helical delivery
the same as for delivery without motion tracking, since
pattern. A smaller number for the pitch implies more
motion correction is accomplished in real-time with-
overlap in the rotations. For a smaller pitch number,
out gating. Additional time is required for acquiring
more rotations are required to cover the same target
kV images to build the model prior to treatments with
length.
motion tracking, as well as pausing the delivery if adjust-
For a helical plan, as the gantry rotates around an off-
ments to the motion tracking parameters are needed.
axis target, the source-to-target distance varies, impact-
ing the dose rate (inverse square effect), beam size
21.4 OPTIMIZATION AND DOSE (divergence effect), and fluence rate (transverse profile
CALCULATION FOR IMRT PLANS shape). The overlapping rotations can result in a pattern
21.4.1 User-Defined Parameters of ripples in the dose distribution along the longitudi-
Planning CT images are usually acquired on a kV simu- nal axis, which are sometimes referred to as the helical
lator, although an MVCT image from the tomotherapy thread effect. The ripples are predicted by the dose cal-
system may also be used. To account for the attenua- culation and are a consequence of the helical delivery
tion of the treatment couch, the planning CT couch is pattern. By selecting the pitch value appropriately for the
replaced with an image of the tomotherapy couch. field width and off-axis position of the target, the thread
Mass density is required as an input to the tomo- effect can be minimized. See Table II in Mingli Chen
therapy dose calculator. For each CT scanner that sup- et al. [12] for recommended pitch values. In addition
plies images for tomotherapy planning, the customer to considering the helical thread effect, the appropriate
physicist scans a phantom that contains inserts of vari- pitch depends on the prescription dose: higher fraction
ous known mass densities, measures the corresponding doses require smaller pitch values to provide enough
Hounsfield Unit (HU) values, and creates a table that gantry rotations to deliver the dose without exceeding
converts HU values to mass density. the longest allowed gantry period of 60 s. Conversely,
The system calculates dose on all the voxels in the if the selected pitch value is too small, the gantry
patient CT image (including the patient, restraint reaches its shortest allowed gantry period of 11.8 s,
devices, couch, air, etc.), so it is important that all the and the delivery becomes inefficient. The tomotherapy
voxels in the CT image accurately represent the density planning system presents a histogram of leaf open times
of the patient at the time of treatment. The CT scan is that can be used to evaluate a plan’s selected pitch value
expected to extend a sufficient distance superior and for delivery efficiency and reasonable leaf-open times
inferior to the target to account for the primary beam as that contribute to better patient QA pass rates [13].
well as scatter. Density override can be used to correct For fixed-angle plans, the pitch is defined as the couch
any imaging artifacts, and to remove contrast that will travel in centimeters per sinogram projection. A projec-
not be present for daily treatments. If the patient extends tion represents an opportunity for each leaf to open and
close not more than one time. Thus, the fixed-angle pitch called source super-sampling. For helical plans, the dose
specifies the resolution of the fixed-angle delivery pat- calculation approximates the dose to be delivered from
tern along the longitudinal axis. A smaller number for three discrete samples per projection, each correspond-
the pitch implies finer resolution in the delivery pattern. ing to a 2.35-degree arc. For fixed-angle plans, at least
Although the pitch has a different definition in the con- two discrete samples per slice of the planning image are
text of helical or fixed-angle deliveries, similar numerical generated. In this context, a “sample” is a distinct source
values are typically selected for planning. position from which rays are traced, and corresponds to
The modulation factor is a user-specified parameter a discrete gantry and couch position.
that controls the range of leaf-open times in the plan by A beamlet represents the leaf-open time or fluence
putting a cap on the ratio between the maximum and through a given leaf at a given projection position.
average leaf-open times, for the non-zero beamlets. If the Beamlets represent the basic units of optimization. For
modulation factor is exceeded during optimization, the a helical treatment plan, there are 64 × 51 = 3264 beam-
leaf-open times are re-distributed to improve the treat- lets per gantry rotation. The number of gantry rotations
ment time. A modulation factor of 1.0 indicates that all depends on the target length, field size, and pitch.
leaves, if used, are equally weighted. A modulation factor A sinogram is a two-dimensional data array of MLC
of 2.0 indicates that the maximum leaf-open time is twice delivery instructions, consisting of rows represent-
the average leaf-open time; this plan takes twice as long to ing projections and 64 columns representing the MLC
deliver as a plan with a modulation factor of 1.0. leaves. See Figure 21.15. Each element in the sinogram
The physician or planner draws contours on the represents a beamlet. The gray-level of each beamlet
patient CT to represent targets and avoidance struc- indicates the percentage of fluence or leaf-open time
tures. A single plan may include one or more targets, for a particular leaf at a particular range of gantry and
but one target is identified as the prescription target. The couch positions corresponding to that projection.
entire plan is scaled to meet the prescription, typically The vertical axis of the sinogram is a discretized
specified as a DVH point. Importance values specify the representation of time and couch position. For helical
relative priority of achieving the goals for the various treatments, the vertical axis of the sinogram is also a
contoured structures, and DVH goal points are set with discretized and cyclical representation of gantry posi-
their relative penalty weightings. tion. Stepping through the rows (projections) of the
sinogram in time, the leaves are configured as indicated
21.4.2 Beamlets: The Basic Unit of Optimization in each row, while the couch, gantry, and jaws (where
The MLC delivery pattern is the only aspect of the deliv- applicable) move as specified in the plan. Sinograms
ery that is optimized by the planning system. get their name from the sinusoidal pattern seen in the
The continuous gantry rotation and couch translation beamlet intensity for a helical plan with a target that is
are divided into discrete units called projections. For a offset from the machine isocenter.
helical plan, there are 51 projections per 360-degree gan-
try rotation. Each projection corresponds to a 7.06-degree 21.4.3 Optimization Process
arc of continuous gantry rotation. For fixed-angle plans, Optimization is the iterative process of adjusting the
each gantry angle is divided into a number of projections MLC sinogram to achieve a dose distribution that
as determined by the user-defined pitch value. meets the prescription DVH point and attempts to sat-
A projection represents an opportunity for each MLC isfy other plan objectives indicated by DVH points and
leaf to open and close up to one time. A leaf may be open maximum/minimum doses.
for the entire projection, closed for the entire projection, The optimizer starts by building an initial sinogram
or open for some percentage of the projection. If a leaf is that identifies those beamlets that are available for opti-
open for less than 100% of the projection time, that leaf mization. Beamlets are considered available for optimi-
is opened when the gantry and couch are positioned in zation if any part of the beamlet passes through a target,
the central portion of the projection. and if no part of the beamlet passes through a contour
For improved dose calculation accuracy, the plan flu- that has been designated by the user as blocked.
ence may be sampled more than once per projection to Beamlets are projected through the patient to deter-
account for slightly different source positions – a process mine if they intersect a target or blocked structure. The
FIGURE 21.15 A sinogram for a prostate case with rectal blocking. The columns represent the 64 MLC leaves and the rows
represent the projection angles. White pixels indicate that leaves are open for 100% of the projection, black pixels indicate that
leaves are closed, and gray pixels represent open leaves with lower intensity values. The plot on the right illustrates the leaf
open times for the projection marked by the vertical line in the sinogram. (Images are from the Delivery Analysis™ software
(Accuray Incorporated). Image used by permission of Accuray Incorporated.)
initial sinogram thus determined is an array of zeros optimization is finished, the optimized sinogram is
and ones. Beamlets with zero intensity are not available converted into a delivery sinogram to be used by the
for optimization; those beamlets will remain closed. system. This delivery sinogram discards undeliver-
Beamlets that are available for optimization are referred able leaf-open times shorter than 18 ms, and generates
to as “used” beamlets. In the initial sinogram, all the machine instructions that meet the requirements for the
used beamlets are assigned 100% intensity (i.e., leaves allowed ranges of gantry rotation period, couch speed,
open for 100% of their corresponding projections). The jaw speed, etc. A final, full dose calculation is performed
intensity weightings for the used beamlets are adjusted on the delivery sinogram and displayed for the planner’s
during the optimization process. evaluation.
At each iteration, the dose distribution in the patient
is calculated, either by performing a full dose calculation 21.4.4 Dose Calculation by Convolution/
or by using a simple ray-tracing method to predict small Superposition
changes from the most recent full dose calculation. Then, The treatment planning system uses a Convolution/
beamlet weights are adjusted according to plan objectives. Superposition dose calculation algorithm similar to
Any beamlet weights that exceed the maximum allowed the TomoTherapy one described originally in Ahnesjö
by the modulation factor are truncated. Dose is calculated [14]. A more recent description of the TomoTherapy
using the new beamlet weights, the entire plan is scaled algorithm, as implemented on GPU hardware, was pub-
to meet the primary prescription point, and the iteration lished in Chen et al. [15].
DVH and isodose display is updated. Dose calculation is a three-step process:
If the optimization has not been paused by the user
and the designated number of iterations has not been 1. At each projection, the fluence incident on the
exceeded, the beamlet weights are adjusted again. When patient is calculated in a 2D plane.
2. Each 2D fluence plane is projected into the patient density representation is a critical requirement for accu-
volume. Density is interpolated along fluence rays rate plan dose calculations. Patient QA measurements,
to determine radiological path lengths in water which are performed not in the patient but in a phan-
or a bone-like material. Radiological path lengths tom, do not verify that the patient density is properly
are used to determine the mass attenuation coef- represented in the treatment plan.
ficients from a look-up table. The mass attenuation
coefficients are multiplied by the fluence to get 21.5.2 Absolute Dose Measurement and Calibration
the TERMA, which is then projected into the 3D In contrast to conventional C-arm gantry-based linacs
patient dose calculation grid. that deliver output in terms of monitor units, tomother-
apy output is delivered based on time. The tomotherapy
3. To determine the dose, TERMA is convolved with
treatment planning system assumes a constant dose
a stored scatter kernel, which was originally com-
rate which is specified in the beam model as an energy
puted using Monte Carlo.
fluence rate. The absolute dose calibration of the tomo-
therapy system is performed differently from the C-arm
21.5 QUALITY ASSURANCE gantry-based linacs that follow the TG-51 dosimetry
21.5.1 Machine QA protocol. TG-51 specifies that dose is measured at 10 cm
TG-148 [16] is the AAPM Task Group report specifi- depth for a 10 × 10 cm2 static open field in a water tank
cally for quality assurance of helical tomotherapy. [17]. However, tomotherapy physical limitations only
This report outlines requirements for daily, monthly, permit a maximum field size of 5 cm in the longitudinal
quarterly, annual, and post-service QA. It also pro- (y) direction at 85 cm SSD (source to surface distance).
vides dose calculation guidance for the non-standard The closest field size to the TG-51 protocol is 5 × 10 cm2
field size (maximum 5 × 40 cm 2) and 85 cm SAD of at 85 cm SSD. Additionally, the lowest available couch
the tomotherapy system. Most clinics choose to position in the tomotherapy bore with 85 cm diameter
adopt the TG-148 recommendations for routine QA does not provide sufficient space for measurements at
procedures. 100 cm SSD and 10 cm depth with appropriate backscat-
A second task group, TG-306, has been formed to ter. In order to overcome the obstacles to implementing
update the TG-148 recommendations, and to account for the TG-51 protocol for reference dosimetry, a machine-
features that were not available when TG-148 was pub- specific reference dosimetry has been suggested for
lished in 2010, such as fixed-angle treatments, dynamic tomotherapy (IAEA-AAPM Joint Committee) [18]. The
jaws, and the dose control system. TG-51 extension for tomotherapy is performed at 85 cm
As described in Ref. [4], the helical tomotherapy sys- SSD for a field size of 5 × 10 cm2. The main component of
tem is provided by the manufacturer with all beam data this protocol is the machine-specific beam quality factor
installed. Thus, the site physicist only needs to check the calculated from the beam quality factor for the calibra-
system routinely to ensure that the machine is perform- tion beam:
ing consistently with the installed beam model. Most
system parameters are not user-adjustable, so the site Dwfmsr fmsr fmsr
, Qmsr = M Qmsr .N D ,w ,Q0 .k Qmsr ,Q (21.1)
physicist would ask the field service engineer to adjust
f ,f
the machine as needed. k fQmsrmsr ,Q = kQ ,Q0 . k Qmsrmsr ,Qref (21.2)
Many system checks can be self-reported, by running
the provided procedures and using vendor-provided where, fmsr is the machine-specific reference (msr) field
software to analyze data from the on-board monitor size (5 × 10 cm2), Qmsr is the machine-specific beam
chambers and CT detector. However, since the dose quality for 10 cm × 5 cm and 85 cm SSD, Q is the TG-51
monitor chamber is used to control the beam through beam quality for a conventional reference 10 × 10 cm2
the dose control system, it is important to verify the field size and 100 cm SSD, MQfmsr msr
is the corrected read-
dose with an external ion chamber on a regular basis. ing (for temperature, pressure, polarity, ion recombi-
Finally, it is essential to give careful attention to the nation, and electrometer) for fmsr field size, N D ,w ,Q0 is
density models used to convert Hounsfield Units in the absorbed dose to water calibration factor based on
the planning CT images to mass density. Correct mass TG-51, and k fQmsrmsr ,Q is the beam quality correction factor
that takes into account the machine-specific beam qual- (gantry, couch, jaws, and leaves) are delivered to a phan-
ity and field size, kQ ,Q0 is the beam quality correction fac- tom CT image set.
f ,f
tor for beam quality Q, k Qmsrmsr ,Qref corrects the beam quality The cylindrical “cheese” phantom has cavities for ion
correction factor for machine-specific beam quality and chambers and a place to insert a film. To avoid the incon-
field size. TG-148 lists the beam quality correction fac- venience of film, most users have transitioned to detec-
tor for multiple ion chambers including the Exradin tor array QA devices such as MatriXX [21], ArcCHECK
A1SL ion chamber that is used by most clinics for tomo- [22], and MapCHECK [23]. In order to set up the phan-
therapy beam profile scanning and absolute dosimetry tom for measurement, MVCT of the phantom is nor-
[16, 19, 20]. mally performed.
The machine-specific reference dosimetry is pro- The University of Virginia has developed a detector-
posed for a fixed beam geometry while tomotherapy based tool for patient specific QA [24, 25]. In addition to
treatments are mainly delivered helically. IAEA-AAPM patient specific QA the toolbox calculates a Monte Carlo
joint committee has proposed a plan-class specific ref- (MC) based secondary dose to verify treatment planning
erence (pcsr) dosimetry for rotational output measure- dose calculations. The MC calculation uses a tomother-
ments [16, 18]. In this method, a homogenous dose is apy simulation model with PENELOPE MC code. The
delivered to an extended and simple target volume. phase-space files from tomotherapy were created from
TG-148 suggests a 2 Gy uniform dose to an 8 cm the MC simulation and analytical transfer function for
diameter and 10 cm long target in a water-equivalent modeling the binary MLC leaves. Patient specific QA is
phantom for pcsr absolute dosimetry. The setup of the performed through the MLC/log file analysis. The infor-
water-equivalent phantom is performed with MVCT. mation recorded for the treatment delivery are linac ion
The output measurement in the phantom is com- chamber measurements (MU1 and MU2), position of
pared with the treatment planning dose calculation at gantry and couch, and MVCT exit detector measure-
the position of the ion chamber volume from the CT ments. Tomotherapy log files record the gantry angle
images. The dose in the ion chamber can be calculated and couch position for each projection angle. The MLC
according to the following equation: positions are derived from the MVCT exit detector mea-
surements. This technique permits 3D dose verification
f f f
,Q pcsr = M Q pcsr .N D ,w ,Q0 .k Q pcsr ,Q
Dwpcsr (21.3)
pcsr pcsr
of the treatment plan, and patient specific QA by com-
paring the dose delivery plan measured with an MVCT
f f ,f f ,f
k Qpcsr
pcsr ,Q
= kQ ,Q0 . k Qmsrmsr ,Qref .k Qpcsr msr
pcsr ,Qmsr
(21.4) exit detector against the treatment plan.
Accuray Incorporated offers a tool called Delivery
where, f pcsr is the plan-class specific field size, Q pcsr is the Analysis™ that analyzes data from the on-board leaf
f
plan-class specific beam quality, MQpcsr pcsr
is the corrected sensors for pre-treatment and in-treatment verifica-
reading (for temperature, pressure, polarity, ion recom- tion of MLC leaf-open times. Dose is recalculated on
bination, and electrometer) for f pcsr field size, and N D ,w ,Q0 the planning CT for comparison with the planned dose.
is the absorbed dose to water calibration factor based on Displayed items include leaf sinograms and histograms,
f
TG-51, k Qpcsr
pcsr ,Q
is the beam quality correction factor that planning CTs with isodose lines, MVCTs, DVHs, and
takes into account the plan-class specific beam quality trending data. Linked viewing allows the user to corre-
and field size. TG-148 lists the beam quality factors for late the various data representations for visual analysis.
the pcsr dosimetry [16]. Delivery Analysis verifies the leaf-open times but is not
sensitive to other aspects of the delivery; the dose is
21.5.3 Patient Quality Assurance re-calculated using the beam energy, profiles, and flu-
Patient specific QA is recommended for all tomotherapy ence rate from the model data, and the gantry, couch,
treatment plans including 3D plans. Delivery quality and jaw positions from the plan.
assurance (DQA) for tomotherapy, similar to C-arm
linacs, is traditionally performed by comparing phan- 21.6 SUMMARY
tom measurements to dose calculations in a phantom. Tomotherapy was invented in the late 1980s to deliver
Vendor-provided software calculates the dose distribu- helical IMRT treatments on a CT type rotational plat-
tion when the patient-specific plan delivery instructions form. Since that time, the system has continued to
evolve, most recently to incorporate motion manage- REFERENCES

ment capabilities. The unique design of tomotherapy
1. T. R. Mackie, “History of tomotherapy,” Physics in
enables the delivery of thousands of coplanar beamlets Medicine & Biology 51(13), R427 (2006).
to achieve highly conformal dose distributions, as well 2. A. Brahme, “Optimization of stationary and moving
as MVCT imaging for patient setup. One of the main beam radiation therapy techniques,” Radiotherapy and
advantages of tomotherapy over conventional C-arm Oncology 12(2), 129–140 (1988).
treatment linacs is the ability to efficiently treat elon- 3. T. R. Mackie, et al., “Tomotherapy: A new concept
for the delivery of dynamic conformal radiotherapy,”
gated targets such as total bone marrow irradiation. Medical Physics 20(6), 1709–1719 (1993).
This chapter reviews the technology of beam delivery, 4. Accuray Incorporated, Radixact® Physics Essentials
treatment planning, and dosimetry measurements for Guide 1062470 (Accuray Incorporated, Madison, 2019).
tomotherapy treatments. 5. M. Beister, D. Kolditz, W. A. Kalender, “Iterative recon-
struction methods in X-ray CT,” Physica Medica 28(2),
94–108 (2012).
ACKNOWLEDGMENTS
6. A. Polacin, et al., “Measurement of slice sensitivity pro-
Some text and images were reproduced from the Physics files in spiral CT,” Medical Physics 21(1), 133–140 (1994).
Essentials Guide by Accuray Incorporated with their per- 7. E. Chao, et al., “SU-GG-J-120: Longitudinal resolu-
mission. Section 21.2 is summarized and freely adapted tion of the TomoTherapy® MVCT image and potential
from [4] with permission by Accuray Incorporated. improvements,” Medical Physics 35(6 Part 7), 2706–2706
(2008).
Section 21.3.1 is adapted from [26] with permission 8. E. Chao, et al., “SU-GG-J-32: Evaluation of a thinner-
by Accuray Incorporated. Section 21.3.2 is copied slice MVCT scan mode across several systems,” Medical
and adapted from [27] with permission by Accuracy Physics 37(6 Part 9), 3152–3152 (2010).
Incorporated. Section 21.4 is copied and summarized 9. K. Ruchala, et al., “Megavoltage CT on a tomotherapy
from [4] with permission by Accuray Incorporated. The system,” Physics in Medicine & Biology 44(10), 2597
(1999).
authors thank the following people for their input to 10. A. Beavis, “Is tomotherapy the future of IMRT?, ” The
the Physics Essentials Guide and helpful suggestions on British Journal of Radiology 77(916), 285–295 (2004).
this chapter: Eric Schnarr, Andrea Cox, Ed Chao, Dylan 11. E. Schnarr, et al., “Feasibility of real-time motion man-
Casey, Jacob Shea, Hillary Hausler, Chris Shaw, Michael agement with helical tomotherapy,” Medical Physics
J. Taylor, and Todd Weston. 45(4), 1329–1337 (2018).
12. M. Chen, et al., “Theoretical analysis of the thread effect
in helical TomoTherapy,” Medical Physics 38(11), 5945–
CONFLICT OF INTEREST 5960 (2011).
Sarah Boswell works at Accuray Incorporated. 13. D. C. Westerly, et al., “Treatment planning to improve
delivery accuracy and patient throughput in heli-
NOTES cal tomotherapy,” International Journal of Radiation
Oncology, Biology, Physics 74(4), 1290–1297 (2009).
1. For patients that do not require 360 degrees of gantry 14. A. Ahnesjö, “Collapsed cone convolution of radiant
angles, such as some breast cases for which a pair of energy for photon dose calculation in heterogeneous
tangent beams may be suitable, the tomotherapy sys- media,” Medical Physics 16(4), 577–592 (1989).
tem can be used to deliver fixed-angle treatments. For 15. Q. Chen, M. Chen, W. Lu, “Ultrafast convolution/super-
fixed-angle treatments, the gantry is stationary at the position using tabulated and exponential kernels on
user-selected gantry angles, while the couch translates GPU,” Medical Physics 38(3), 1150–1161 (2011).
in the longitudinal direction through the beam for each 16. K. M. Langen, et al., “QA for helical tomotherapy:
gantry angle [4]. Report of the AAPM Task Group 148,” Medical Physics
2. Patient external marks are placed in a stable location 37(9), 4817–4853 (2010).
near the target, but the exact location is arbitrary. The 17. P. R. Almond, et al., “AAPM’s TG-51 protocol for clini-
concept of a tumor isocenter is not relevant to tomo- cal reference dosimetry of high-energy photon and elec-
therapy treatments, because the couch moves continu- tron beams,” Medical Physics 26(9), 1847–1870 (1999).
ously into the bore during treatment. 18. R. Alfonso, et al., “A new formalism for reference dosim-
3. This method of motion synchronization is called etry of small and nonstandard fields,” Medical Physics
Synchrony® Respiratory Tracking™ and is unique 35(11), 5179–5186 (2008).
to systems produced by Accuray Incorporated: the 19. S. Thomas, et al., “A Monte Carlo derived TG-51 equiv-
Radixact® System (a form of helical tomotherapy) and alent calibration for helical tomotherapy,” Medical
the CyberKnife® System. Physics 32(5), 1346–1353 (2005).
20. M. McEwen, “SU-FF-T-218: Evaluation of the Exradin# 24. L. L. Handsfield, et al., “Phantomless patient-specific
a19 ion chamber for reference dosimetry in megavolt- TomoTherapy QA via delivery performance monitor-
age photon beams,” Medical Physics 34(6 Part 10), 2451– ing and a secondary Monte Carlo dose calculation,”
2451 (2007). Medical Physics 41(10), 101703 (2014).
21. M. Rao, et al., “Comparison of Elekta VMAT with heli- 25. Q. Chen, et al., “TomoTherapy MLC verification using exit
cal tomotherapy and fixed field IMRT: Plan quality, detector data,” Medical Physics 39(1), 143–152 (2012).
delivery efficiency and accuracy,” Medical Physics 37(3), 26. Accuray Incorporated, 1g Patient Positioning Workflows:
1350–1359 (2010). Simulation, Planning, and Daily Treatment 1059569.B.
22. C. Neilson, et al., “Delivery quality assurance with In the Radixact® Physics Essentials Course (Accuray
ArcCHECK,” Medical Dosimetry 38(1), 77–80 (2013). Incorporated, Madison, 2017).
23. P. A. Jursinic, R. Sharma, J. Reuter, “MapCHECK used 27. Accuray Incorporated, Synchrony® Real-Time Motion
for rotational IMRT measurements: Step-and-shoot, Tracking and Correction on the Radixact® Treatment
TomoTherapy, RapidArc,” Medical Physics 37(6 Part 1), Delivery System 1064193.E (Accuray Incorporated,
2837–2846 (2010). Madison, 2019).
Chapter 22
Gamma Knife Dosimetry

Nels C. Knutson
Washington University in St. Louis School of Medicine
St Louis, Missouri
CONTENTS
22.2 Safety First 344
22.3 Confirming The Measurement Geometry 345
22.4 Absolute Dosimetry 345
22.5 Relative Dosimetry: Output Factors 347
22.6 Relative Dosimetry: Off-Axis Distributions 349
22.7 Putting It All Together: End-to-End Testing 350
References 352
22.1 INTRODUCTION (QA) of gamma stereotactic units and will hopefully be

The following chapter will review important dosimet- available to the public soon.
ric measurement techniques used to characterize the This chapter assumes the reader will have a signifi-
Gamma Knife radiosurgery system (Elekta, Stockholm, cant knowledge of general operation and design of the
Sweden). As of writing, the current generation model Gamma Knife. We will not include a discussion of pre-
of the Gamma Knife is the Icon which replaced the vious Gamma Knife models nor their differences com-
Perfexion model which replaced the model 4C. This pared to the new models. We will only briefly go over the
work will focus on the latest two models as the 4C and basics of the new design here. The readers are referred
previous models are being phased out. This chapter will to several of the following excellent references for a
focus in particular on the dosimetric methods used to more in-depth description of the Gamma Knife and dif-
characterize the Gamma Knife. This includes absolute ferences between models [1, 2]. The Gamma Knife is a
dosimetry at the radiation focus and relative dosimetry specialized radio surgery system that focuses nearly 200
techniques to further characterize the system. This chap- Co-60 sources to a single point. The sources are divided
ter will focus on techniques used by the author person- evenly among eight sectors which can be driven inde-
ally as well as some techniques from the literature. Even pendently from a blocked position to any of the three
though attempts to be inclusive will be made, there is collimator sizes (4 mm, 8 mm, and 16 mm). The loca-
always different and new methods being developed, and tion of this focal point is correlated to stereotactic space
the reader is encouraged to keep reviewing the literature via either a rigid frame attached to the couch or via a
for new publications. For instance, at the time of writing calibrated stereotactic cone beam computed tomogra-
this chapter, the American Association of Physicists in phy (CBCT) with the Icon model. The patient is immo-
Medicine (AAPM) task group 178 is preparing for publi- bilized via either a rigid frame or mask system and can
cation of a guidance document on the quality assurance be translated to the radiation focus for treatment. The
343
size of a single radiation focus is controlled by the col- cobalt-60 sources that are always emitting gamma rays.
limation used per sector. Each sector can use the same They are simply moved from a shielded position to an
collimation or consist of different combinations of the unshielded position for treatment. The Perfexion and
collimation used in composite. A single location within Icon models have 5 potential positions starting from
the patient of the focal point is often referred to as a the full retracted position or the home position, then
shot. Treatment plans can consist of multiple shots with the 8 mm collimator exposed position, then a blocked
different times to make the desired radiosurgery dose position, then the 4 mm collimator position, and finally
distributions. These distributions are commonly char- the 16 mm collimator position. The blocked position
acterized by a high level of conformity and sharp gradi- is used when transitioning the patient between differ-
ent in dose from the prescribed dose level covering the ent radiation focus positions. The user should survey
target to the lower dose levels in the surrounding nor- the machine in both conditions; with the sources at the
mal tissues. home position and fully exposed using the largest col-
This chapter will highlight the techniques used limator size. The user should be sure to conduct surveys
during the commissioning of the Gamma Knife. This using a fast-responding detector such as Geiger coun-
process though complex and semi-challenging can be ter or scintillator detector to examine shielding barriers
done in a relatively short period of time. With proper for any potential hot spots that could be present due to
planning, this commissioning process can typically imperfections in the shielding construction, etc. For the
be completed by a physicist in a single week. Note this final determination of the dose rate at a survey location
chapter focuses on the dosimetry aspects of this pro- a large volume calibrated ionization chamber should be
cess. There are numerous other tests involved in the used. For the details on these measurement techniques
commissioning of the Gamma Knife that are outside the user is referred to the following references [8–10].
this scope. The readers are referred to other works in Also, outside of the scope of this work, before ever
the literature for the non-dosimetry commissioning using the CBCT on a real patient, it must be commis-
tests used on the Gamma Knife [3–7]. In all sections sioned. This includes safety, mechanical, image quality,
of this chapter, the author will try to relate their own and positional accuracy tests. This also includes the dosi-
personal experience to try to help guide the reader. metric characterization of the CBCT. Even though this is
However, the reader should realize the end user is dosimetry, this work will focus on the dosimetry of the
ultimately responsible to investigate all measurement treatment beam. The readers are referred to the literature
techniques, measurement equipment, and processes (e.g., Chapter 30) to learn about CBCT dosimetry.
themselves, regardless of anyone else’s personal expe- A wipe test is performed to ensure the sealed sources
riences. The users are ultimately responsible for the are indeed sealed. In the United States this is done in
integrity of their own data. accordance with title 10 of the code of federal regula-
tions (CFR) part 39.35. Sources with a wipe test greater
22.2 SAFETY FIRST than 5 nCi must be removed from service. This wipe test
As stated, the techniques discussed in this chapter are can be completed with the sources in the home position
implemented as part of the commissioning and accep- and opening the shielding doors in the front of the unit.
tance process after an installation or source exchange The service engineer or user can remove the collimator
of a Gamma Knife. It is critical that before performing cover with a long suction cup tool. Once removed the
measurements on any machine for the first time, a full user can wipe outside of this cover that was facing the
radiation safety assessment is completed. This includes collimator and replace the cover. Instructions for this
a full radiation barrier survey, wipe test, confirmation procedure are provided by the manufacturer and should
and accounting of all sources, confirmation of the func- be followed with care as the cover has pressure sensors
tion of all controls and emergency off switches. Once that can break if care is not taken in removing or replac-
completed, this will help ensure the machine is safe to ing the cover.
operate for further testing. Accounting for all sources is important for any
Though outside of the scope of this chapter, radiation Gamma Knife user. A panoramic image using radio-
barrier surveys are extremely important for Gamma chromic film is an easy method for accomplishing
Knife. Unlike linear accelerators, the Gamma Knife uses this. As described by Cho et al. [11], one can wrap a
Gamma Knife Dosimetry ◾ 345
cylindrical phantom with film and irradiate it to capture tool provided by the manufacturer to be within 0.3 mm
an image of the source geometry. This allows for direct of the baseline set at the time of installation by the ser-
accounting for all sources as it may be hard to detect a vice engineer.
single missing source out of 192 with other dosimetric
methods. This can be done using all three collimator 22.4 ABSOLUTE DOSIMETRY
sizes. In Gamma Knife, absolute dosimetry is used to deter-
Finally, the user should test all controls including all mine the dose rate at the radiation focus for the 16 mm
emergency off buttons and door interlock, to ensure they collimator at the center of a 16 cm diameter sphere. This
are functional and operate as intended. This ensures in calibration is used for all patients, therefore an error in
the case of an emergency the user is confident on how this measurement would propagate across all patients
to react and the machine will function as intended. This treated. Therefore, it is critical that this measurement be
should be tested before each use of the Gamma Knife double, and triple checked. Looking in the literature, a
and is particularly important the very first time the round robin study was completed by Drzymala et al. at
machine is used. nine different institutions, using two models of gamma
knife, four different phantoms, three dosimetry proto-
22.3 CONFIRMING THE cols, and two different ionization chambers [12]. The
MEASUREMENT GEOMETRY readers can see how the level of complexity in this type
For Gamma Knife measurements one cannot visually of study can increase rapidly. The reader is referred to
check that the measurement geometry is sound like one this work as it highlights many of the factors for accu-
can on a linear accelerator with a field light vs. radia- rate absolute dosimetry in Gamma Knife. This section
tion test. Instead, the user must rely on the correlation will look at the individual components of the absolute
of the radiation focus position to the couch positions. dosimetry process and will often refer to this round
With the Gamma Knife Icon one can use the on-board robin study.
CBCT imaging to confirm this. One can also test this There are two different 16 cm diameter spherical
correlation between the couch coordinates and radia- phantoms provided by the manufacturer for calibra-
tion focus position using a film holder with a pin at a tion and will be the focus of the work. The readers are
fixed position relative to the frame adapter. This is often referred to the literature if they are interested in using
referred to as the pin prick test. Knowing physical posi- a third party phantom such as the in air method [13]
tion of the pin relative to the frame, one can irradiate the or water phantom method [14] mentioned in the round
film and ensure this correlates with center of the radia- robin. Previously the manufacturer provided a gray
tion focus. This needs to be with 0.3 mm at the center of plastic spherical phantom. Currently this has been
the frame adapter and within 0.5 mm at the outer posi- replaced with a red colored water equivalent plastic
tions of the frame adapter. This should be completed spherical phantom. There are several benefits of the
in all planes for each collimator. This QA tool can be new generation phantom. The new phantom utilizes the
scanned via CBCT to ensure the CBCT coordinates cor- frame adapter that is used for the patient treatments.
relate with the physical positions and can be cross cor- The previous model used a separate adapter to hold the
related with the radiation focus. This is critical because phantom in place which was known to cause just under
the detector positioning is critical for the measurement 1% attenuation of the beam during the measurement
of small radiation fields used in Gamma Knife. If the [15]. Typically, ionization chambers ranging in volume
setup is incorrect this can lead to large differences in from 0.007 cm3 up to 0.125 cm3 are used for calibration.
the measurement. Anytime a new phantom or measure- A list of common ionization chamber models used are
ment technique is being used, the user should confirm shown in Table 22.1, note this is not an all-inclusive list
the geometry via imaging or physical correlation to the and the reader is encouraged to review other chambers
frame/frame adapter via imaging. This can be com- that may be described in the literature. Either phantom
pleted with the frame and fiducial box using CT and has inserts that can be drilled to fit any of these different
for the Icon the phantom can be directly imaged in the detectors.
measurement position with CBCT. Note that CBCT iso- In the United States, the requirements for absolute
center location is confirmed on a daily basis via a QA dosimetry measurement equipment is highlighted in title
TABLE 22.1 Common Ionization Chambers Used for Gamma have been reported for the Exradin A16 chamber in
Knife Absolute Dosimetry
particular [12]. The selection of the dosimetry proto-
Chamber Model Manufacturer Volume (cm3) col also impacts the selection of the ionization cham-
Exradin A16 Standard Imaging 0.007 ber. Within the limits of what is reasonably achievable
Exradin A14SL Standard Imaging 0.015
to the user, it is advisable for a user to select at least
TN31010 PTW 0.125
TN31016 PTW 0.016
two well-characterized ionization chambers for their
PR-05P Capintec 0.07 calibration. Using a well-established chamber is much
PR-05 Capintec 0.14 easier than trying to fully characterize a new chamber,
CC13 IBA 0.13 come up with the appropriate correction factors for the
CC04 IBA 0.04 protocol of use, and validate these corrections, which
CC01 IBA 0.01 is often beyond the skill set of clinical physicists. At
our institution we use the average output as measured
by our two relatively large volume chambers (PR-05P
10 of the code of federal regulations part 35, section 630 and TN31010) as our primary calibration and we have
(10CFR35.630). Specifically, “(1) The system must have an A16 that can be cross-calibrated for end-to-end
been calibrated using a system or source traceable to the to dosimetry testing, but it is not used as a primary
National Institute of Standards and Technology (NIST) calibration due to the stability concerns raised above.
and published protocols accepted by nationally recog- Note this is just our experience with one ion chamber
nized bodies; or by a calibration laboratory accredited and certainly does not represent every A16 chamber,
by the American Association of Physicists in Medicine it just highlights the need for using multiple detectors
(AAPM). The calibration must have been performed over time to reduce the uncertainties and check for
within the previous 2 years and after any servicing consistency.
that may have affected system calibration; or (2) The This work will discuss two of the primary formalisms
system must have been calibrated within the previous used for Gamma Knife absolute dosimetry. The first and
4 years. Eighteen to thirty months after that calibration, longest used is the AAPM TG-21 formalism [16], includ-
the system must have been intercompared with another ing revisions made in 1986 [17]. This formalism has the
dosimetry system that was calibrated within the past benefit of being very generalized with very few assump-
24 months by NIST or by a calibration laboratory accred- tions being made in the measurement geometry. The
ited by the AAPM. The results of the intercomparison user must know the material they are measuring in, the
must indicate that the calibration factor of the licens- construction of the ion chamber, and one can measure
ee’s system had not changed by more than 2 percent. dose fairly easily. The first time setting up this calcula-
The licensee may not use the inter comparison result tion requires substantial work as the user will have to
to change the calibration factor. When intercompar- look up or calculate approximately fifteen different val-
ing dosimetry systems to be used for calibrating sealed ues found in TG-21 to calculate the output. Once this is
sources for therapeutic units, the licensee shall use a completed, the calculation of the output is straightfor-
comparable unit with beam attenuators or collimators, ward. This process must be completed for each phantom
as applicable, and sources of the same radionuclide as and detector combination used. This level of complexity
the source used at the licensee’s facility.” It is the author’s has led many users to seek alternative protocols to deter-
opinion, option 1 is preferred. mine the output. The formulae below show the different
Selection of the desired ionization chamber can components used in determining the measured dose.
seem daunting. Further compounding this is the choice Note that user measured inputs are M and PDD; val-
ues that must be looked up or calculated from TG-21
of phantom to use in combination with the detector
( ) ,( ) ,( ) ,( ) ,( ) ,( )
wall air cap air wall med
µen µen
of choice and getting the appropriate plugs drilled for are Awall, α , L
ρ gas
µen
ρ
L
ρ gas ρ
L
ρ air ρ
,
wall cap wall
the detector. It is however important to note that the
() ( )
med water
L , Pion , Prepl, ESC , and µen ; values from the
small volume ionization chambers seem attractive ρ air ρ
med
due to relatively small fields being measured, however ADCL calibration are N x and Aion ; and k ( We ) is a
caution should be used as short term stability issues physical constant.
N Gas = and kQfmsr ⋅ fref

is a single looked up value from Table 14 of
msr ⋅Qo
TRS 483. Note f indicates field size, Q indicates beam
W
k   × Aion × Awall × βwall quality, M is charge, k is the correction, N is the cali-
 e 
Nx × wall air cap air bration factor from the ADCL, Qo indicates the calibra-
 L µ   L µ  tion beam quality, ref indicates the reference field for the
α ×   ×  en  + (1 − α ) ×   ×  en 
ρ
  gas  ρ  wall   gas  ρ  cap
ρ calibration, msr indicates a machine specific reference.
(22.1) f f ⋅f
fmsr fmsr
Dwater ,Qmsr = MQmsr × N D ,W ,Qo × kQmsr ⋅Qo (22.5)
ref msr ref
wall med med

  L µ   L 
α ×   ×  en  + (1 − α ) ×    At our institution during the commissioning of our
  ρ  air  ρ  wall  ρ  air  first Icon unit which was a simple upgrade and reload
Pwall = med (22.2) of an existing Perfexion. We elected to do both formal-
 L
 ρ  isms in the new and old phantoms with two chambers.
air Therefore, 8 measurements of the output were com-
 L
med pleted and averaged. The maximum difference between
Dmed = M × N gas ×   × Pwall × Pion × Prepl (22.3) two methods was 0.6% and the maximum difference
 ρ  air
from the average was 0.3%, giving us confidence in our
water absolute calibration. However, no matter how confident
µ 
Dmed × ESC ×  en  the user is, it is highly recommended the user do an
 ρ  med independent audit of the output. This can be completed
Dwater ( at dmax ) = (22.4)
PDD via the Imaging and Radiation Oncology Core (IROC)
100 at MD Anderson in Houston Texas. Independent audit
stated our output was 1.01 × our institution stated out-
In general, the simpler TG-51 [18] formalism has all put. Post the initial install at our institution the output
but replaced TG-21 for typical linear accelerators. This was measured using the TRS 483 formalism with two
simplicity comes at a cost of less flexibility in the TG-51 chambers in the solid water phantom, in addition to the
formalism. Machines such as Gamma Knife do not meet third party audit on a yearly basis at our institution and
the criteria of beam quality specification or the reference compared to the TPS output corrected for decay entered
measurement to use this formalism. The corrections at the time of commissioning.
used in the protocol can lead to errors in the calibra-
tion if applied blindly. To address these issues Alfonso et 22.5 RELATIVE DOSIMETRY:
al. have proposed a new formalism to extend the TG-51 OUTPUT FACTORS
style easier formalism to non-standard beam geometries After determining the output of the 16 mm collimator,
such as Gamma Knife [19]. This is the formalism used the output of the 4 mm and 8 mm collimators can be
in IAEA Technical Report Series (TRS) 483 [20]. In this determined relative the 16 mm reference. The Gamma
report, the output is tied directly to a corrected read- Knife treatment planning systems comes with the default
ing, the calibration factor from a standards lab, and a values of 1.000, 0.9005, and 0.8140 for the 16 mm, 8 mm,
conversion from the reference geometry in water, to the and 4mm collimators, respectively. These values are pro-
measurement geometry in the user’s measurement for vided by the manufacturer and are from Monte Carlo
a chamber. This reduces the number of values the user simulation. It is the author’s opinion that it is best to use
has to look up to one. However, these values are tabu- measurements to confirm these values instead of using
lated for very specific detectors, phantoms, geometries, the measurements directly. These measurements are
and machines. Therefore, new machines, phantoms, and completed in the same reference conditions to the output
chambers still pose an issue to future users as these val- measurement. This can be accomplished in either of the
ues will yet to be determined. Again, simplicity comes at calibration phantoms discussed previously. Typically,
a tradeoff for the loss of flexibility. This general equation this is done using film. Film has excellent spatial resolu-
is shown in Equation 22.5, in which MQfmsr msr
is the user tion and can easily be placed in either of the manufac-
fref
measured input, N D ,W ,Qo is obtained from the ADCL, turer provided phantoms. Radiochromic film has been
widely used in Gamma Knife and reported on in the collimator for the fixed times to calculate the output
literature. The convenience of not needing a developer, factors. This is generally done by using a small region
being nearly tissue equivalent, being largely energy inde- of interest on the scanned film. This technique requires
pendent, and having a large dosimetric range, as com- the user to construct a calibration curve for the film
pared to radiographic film has led to radiochromic film first. This is time consuming and any errors (random
to become very popular for Gamma Knife users. As with or systematic) in this process will propagate through to
any detector, radiochromic film is not perfect and has its the output factor measurement. To reduce the random
own drawbacks. There is a time dependency for readout, error, one can take several measurements for each point
the user must have a high-quality scanner, there have on the calibration curve as well as multiple output factor
been reported nonuniformities both per batch and in an measurements and take the average for each. To reduce
individual sheet of film. There are dependencies on the the systematic errors, one should follow the best prac-
color channel being used. With care and skill however, tices for film dosimetry and confirm the output of the
film can be used to measure Gamma Knife dose dis- Gamma Knife as described in the previous section. An
tributions with excellent spatial resolution and precise example calibration curve is shown in Figure 22.1.
measurements of dose. Consistency is also crucial for Another school of thought is to take the ratio of net
film dosimetry so one should take care to be consistent optical densities (i.e., correcting for the inherent optical
in the time of readout, orientation, scanning protocol, density of unirradiated film) for films irradiated with a
and handle and store all film with care to ensure quality fixed time for all collimators. This assumes that dose is
measurement results. These challenges and techniques proportional to net optical density and therefore a sim-
to help minimize their impacts have been described in ple ratio will cancel this out. This is true of course for
the literature [21, 22]. small ranges of dose as a user can see from the above-
In order to use radiochromic film for relative output mentioned calibration curve technique (Figure 22.1). It is
factors, the user must irradiate films with different col- recommended the user does not use the ratio of net opti-
limator settings for fixed time. There are several schools cal densities unless they have well characterized the film
of thought for this measurement. One school of thought they are using and know with a high certainty the dose
is to convert film optical density to dose via a calibra- response is linear in the dose range they are using for
tion curve of known doses from no dose to beyond the the irradiation. The net OD can be read out with a scan-
maximum dose delivered to the output factor films and ner or with a densitometer. Note that Ma et al. proposed
then directly take ratio of maximum doses for each a similar method to eliminate the use of a calibration
FIGURE 22.1 Example calibration curve obtained with the multiple dose level irradiations in the Gamma Knife for EBT3
Gafchromic film. Note the non-linearity in the curve fit near each end of the calibration curve using a piecewise cubic Hermite
interpolating polynomial. This curve would be more useful for measuring profiles with a max dose of 6 Gy.
curve by irradiating a single film with the so-called dou- Just as volume averaging effects are critical in this
ble shot method [23]. They concluded the uncertainty measurement, geometric accuracy is as well. If the
associated with this assumption to be approximately 3% detector holder being used is off even by 1–2 mm, an
assuming a 1–2% random uncertainty in the film. They error in the 4 mm output factor measurement could be
feel this uncertainty is an acceptable trade-off for elimi- well over 10%. Therefore it is highly recommended to
nating the sheet-to-sheet variation and eliminating the confirm this measurement geometry via imaging. This
need for a calibration curve. The author would point out can be done via CT with the fiducial markers or CBCT
traditional output factor measurements can typically be on the machine. This was described previously but the
done on a single sheet of film anyway, but the double author would like to re-emphasize this point due to its
shot technique could be of interest to the reader. importance. At our institution we used radiochromic
Outside of using film, active detectors have also been film and a synthetic diamond detector to confirm the
proposed. These are typically small diodes, diamond or output factors at time of installation and during the
micro ionization chambers that will give the user charge annual QA. Example CBCT images of measurement
(M). TRS 483 defines a measured output factor accord- geometry is shown in Figure 22.2.
ing to Equation 22.6. In which MQfclin clin
and MQfmsr
msr
are the
fclin ⋅ fmsr
user measured inputs and kQclin ⋅Qmsr is obtained from TRS 22.6 RELATIVE DOSIMETRY: OFF-AXIS
483. Note f indicates field size, Q indicates beam qual- DISTRIBUTIONS
ity, M is charge, k is the correction, clin indicated the Along with characterizing the output of the Gamma
clinical field being measured, msr indicates a machine Knife and the relative outputs of the different collima-
specific reference: tor sizes, the characterization of the radiation profiles is
critical. This can and should be done in all planes of the
clin ⋅ fmsr
MQfclin ⋅ fmsr Gamma Knife (Axial, Coronal, Sagittal). The profile data
OFQfclin ⋅Qmsr =
clin
× kQfclin
clin ⋅Qmsr
(22.6)
MQfmsr
msr
in the TPS come from Monte Carlo simulations and can
easily be extracted for comparison to measurements.
Detector selection is critical for this measurement due Looking at the profiles for different collimator settings
to the small field sizes used in Gamma Knife. It becomes one will note the very fast fall off. Typically, Gamma
immediately clear that detectors that are used for the 16 mm Knife plans are prescribed to the 50% isodose to take
collimator are not appropriate for use with the 4 mm advantage of this fast fall off. This is of course highly
and 8 mm collimators due to their relatively large size desirable but demands precision to avoid underdosing
compared to the radiation field size. This is highlighted the target or potentially over dosing OARs.
in Table 25 of TRS 483 where a list of detector correc- Much attention has been given to the output and
(
tion factors kQfclin )
⋅ fmsr
clin ⋅Qmsr
for the different clinical field sizes the relative output factors in this chapter, however in
are given based off of the work by Benmakhlouf et al. radiosurgery the minimal peripheral dose to the target
[24]. Note in TRS 483 there are only five detectors listed is often just as or more important than the maximum
as appropriate for measuring all three collimator sizes dose delivered to the tumor. This is directly governed by
used in the Gamma Knife. One should also note the cor- the off-axis distributions of the shots used. Imagine, the
rection factors listed become large, on the order of 5%, for positioning errors are zero, however the profile width as
the shielded diode. Corrections larger than this should defined by the full-width at half-maximum (FWHM)
indicate to the user that detector is most likely not suitable is incorrect by 2 mm. This could mean your prescrip-
for this measurement and should not be used as the sole tion dose size is incorrect by 2 mm. In the radiosurgery
measurement device. Commonly, the A16 is proposed for this difference could lead to underdosing the target if
use of measurement of the relative output factors. However, the FHWM is smaller than modeled. Conversely if the
due to its relatively large size (1.7 mm length and 2.4 mm FWHM is larger than modeled, one could overdose an
diameter) it will have corrections for not only the cham- OAR. Therefore, it is critical to confirm this FHWM is
ber components (wall, electrode, etc.) but also partial vol- within 1 mm of the TPS value. The FWHM is not configu-
ume averaging. These potential corrections are not given rable in the Gamma Knife TPS given the fixed geometry
for Gamma Knife in the report, however, for a CyberKnife of the Gamma Knife. If the user’s measurement (assum-
6 mm cone the corrections are on the order of 5%. ing there was no measurement error) doesn’t match the
FIGURE 22.2 Examples of confirming the output factor measurement utilizing CBCT. Note position of the shot was found
via CBCT. Notice the relative size of the microchamber (left) and synthetic diamond (right) compared to the 4 mm shot. The
microchamber is very large compared to the 4 mm shot and the synthetic diamond detector is more appropriate as the amount
of volume averaging is reduced due to its size.
expected values this would indicate a manufacturing the desired collimation and orientations. Again, it can be
or installation error which would need to be corrected. tempting to just take a profile of optical density from the
Since the Gamma Knife prescription isodose dose does film and used this as your off-axis ratio. However, cau-
not have to be 50%, and it can be varied significantly, tion should be used as this assumes the net optical density
the user should also confirm the 80–20% penumbra of is directly proportional to dose. This can be true over a
the profiles. In fact the best practice is just to confirm the range of doses for radiochromic film; however, it is cer-
entire off axis distribution via either dose difference and tainly not true over all ranges. Therefore, one should be
distance-to-agreement metrics, or using a 1D Gamma sure of the characterization of the film they are using in
test. The user should keep in mind the appropriate met- the dose ranges they are trying to measure.
rics for these tests. The user’s measurement should be Film measurements can be noisy. It is tempting to
within 3% dose difference or 1 mm distance to agree- smooth these images using various smoothing kernels.
ment. Note that the 1D Gamma test is slightly less strict The user should remember that this will inherently
as it allows for searching the minimum combination of limit the spatial resolution of the film, which is needed
dose difference and distance to agreement for all points to characterize the small radiosurgery fields used in
on the curve. This does not mean it shouldn’t be used, Gamma Knife. Therefore, to improve and reduce this
this is simply a word of caution to the reader. random noise in the measurement, the author recom-
In order to measure the off-axis distribution, the user mends limiting smoothing techniques and instead
is mostly limited to using film. The use of 3D gel dosim- using repeated measurements to average out this statis-
etry systems have been described in the literature, how- tical noise. Example profile measurements compared to
ever, this has yet to see wide spread adoption clinically. the TPS profiles are shown in Figure 22.3.
Currently, there are no other ways to measure off axis
profiles in Gamma Knife. Since film dosimetry is by far 22.7 PUTTING IT ALL TOGETHER:
the most common method used at the time of writing this END-TO-END TESTING
chapter, it will be the main focus here. Measuring off axis The final portion of this chapter will discuss putting
distributions in Gamma Knife with film is done by con- all these principles together in an end-to-end test. This
structing a calibration curve of net optical density and involves taking a phantom and putting it through the
dose, as described in the relative output factors section entire Gamma Knife treatment process from simulation
of this chapter and applying it to films irradiated with through to treatment. There are several phantoms that are
FIGURE 22.3 X, Y, and Z profiles taken with film for 4 mm, 8 mm, and 16 mm collimator compared to the TPS. Note every
4th film data point is plotted for clarity. Note the statistical noise seen in the film measurements.
commercially available that allow the user to end-to-end the delivery of a complex treatment plan. At our institu-
test their process. This allows the user to simulate, plan, tion, it was not immediately clear why our end-to-end test
treat and measure the dose delivered via film, gel, or with ion chamber measurements were systemically lower than
an active detector. The phantom should be treated just as expected for the complex plans and not for a simple single
a patient and go through the entire clinical workflow. The shot plan. After close inspection it was clear that between
resultant measurements should be completed to confirm shots during couch motions the total charge reading was
both the absolute and relative dosimetry delivered. This decreasing due to leakage causing a systematic error in
should be done over a range of case types that are seen
in the clinic, ranging from very simple single shot dose
distributions to complex plans consisting of multiple
composite shots in an anthropomorphic phantom. Note
that these phantoms can make excellent training tools
for all parties involved in the Gamma Knife process. This
allows the users to work through the processes before try-
ing to treat a patient. An example phantom with different
inserts are shown in Figure 22.4.
At the author’s institution an end-to-end test is done
on an annual basis using both an active detector (using an
A16 micro chamber or synthetic diamond detector) and
radiochromic film for both a simple spherical target and
a complex organic shaped target. For this test we use a
5% dose difference 1 mm distance-to-agreement criteria.
Typically, the simpler plans can be measured within 3% or
1 mm distance-to-agreement. Note that during complex
target delivery the user should be aware of potential leak-
age currents in the detector. This can be particularly trou- FIGURE 22.4 An example end-to-end phantom in a frame
blesome if the plan contains numerous couch motions in used with multiple detector and target inserts.
the measurement. This was confirmed with the measure- 3. N. C. Knutson, et al., “Characterization and validation
ment using a second chamber and the original chamber of an intra-fraction motion management system for
masked-based radiosurgery,” Journal of Applied Clinical
was sent out for repair.
Medical Physics 20(5), 21–26 (2019).
Note this same process can be repeated using an 4. I. Aldahlawi, D. Prasad, M. B. Podgorsak, “Evaluation of
independent third-party phantom (IROC in the United stability of stereotactic space defined by cone-beam CT
States) that is planned and irradiated by the user but for the Leksell Gamma Knife Icon,” Journal of Applied
the dosimetry is read out by the third party. This ser- Clinical Medical Physics 18(3), 67–72 (2017).
vice is an excellent independent test for the user and 5. W. Li, et al., “Preliminary evaluation of a novel ther-
moplastic mask system with intra-fraction motion
is highly recommended. This is also required to enroll monitoring for future use with image-guided Gamma
patients on to many clinical trials. Their phantom uses Knife,” Curēus 8(3), e531 (2016).
a combination of film and thermoluminescent dosim- 6. A. Sarfehnia, et al., “Performance characterization of an
eters (TLDs) to report the measured dose in comparison integrated cone-beam CT system for dedicated gamma
to the planned dose to the user. They use a 1D Gamma radiosurgery,” Medical Physics 45(9), 4179–4190 (2018).
7. M. Zeverino, et al., “Commissioning of the Leksell Gamma
comparison on the film profiles and a simple dose dif-
Knife® Icon™,” Medical Physics 44(2), 355–363 (2017).
ference for the TLD measurements. The criteria to pass 8. National Council on Radiation Protection and
this test is currently set to five percent for the point mea- Measurement, NCRP Report no. 151 Structural Shielding
surements and greater than or equal to 85% of points Design and Evaluation for Megavoltage X- and Gamma-
having a gamma value less than one with a five percent Ray Radiotherapy Facilities (2005).
9. P. H. McGinley, Shielding Techniques for Radiation Oncology
dose difference and three millimeter distance to agree-
Facilities (Medical Physics, Madison, WI, 2002).
ment criteria. In the author’s opinion, this is a relatively 10. J. K. Shultis, R. E. Faw, K. R. Kase, Radiation Shielding
lax criteria and represents a very minimal requirement (Prentice Hall PTR, Upper Saddle River, NJ, 1996).
a user should be able to meet as Molineu et al. reported 11. Y. Bin Cho, M. Van Prooijen, D. A. Jaffray, M. K. Islam,
only a single failure to meet this criterion in a relatively “Verification of source and collimator configuration for
small subset of 23 irradiations using Gamma Knife [25]. Gamma Knife® Perfexion™ using panoramic imaging,”
Medical Physics 37(3), 1325–1331 (2010).
Interestingly the pass rates across all modalities was 12. R. E. Drzymala, et al., “A round-robin gamma stereo-
lower at 81%. At the authors institution this was com- tactic radiosurgery dosimetry interinstitution compari-
pleted for both frame and frameless SRS modalities. son of calibration protocols,” Medical Physics 42(11),
6745–6756 (2015).
22.8 CONCLUSIONS 13. S. G. Meltsner and L. A. DeWerd, “Air kerma based
Gamma Knife radiosurgery is an extremely accurate dosimetry calibration for the Leksell Gamma Knife,”
Medical Physics 36(2), 339–350 (2009).
treatment modality for delivering very focused radiation 14. R. E. Drzymala, R. C. Wood, J. Levy, “Calibration of
fields. Like any system, it must be validated prior to ever the Gamma Knife using a new phantom following the
treating a patient. These measurements can be challeng- AAPM TG51 and TG21 protocols,” Medical Physics
ing due to the small radiation fields used, but with care 35(2), 514–521 (2008).
the system can be validated in relatively short amount 15. J. P. Bhatnagar, J. Novotny, M. A. Quader, G. Bednarz,
M. S. Huq, “Unintended attenuation in the Leksell
of time to a high degree of accuracy. If the reader is to
Gamma Knife Perfexion calibration-phantom adaptor
take anything away from this chapter, please remember, and its effect on dose calibration,” Medical Physics 36(4),
proper planning prevents poor performance. Never rely 1208–1211 (2009).
on a single measurement and always get an independent 16. Task Group 21, Radiation Therapy Committee, AAPM,
confirmation of your work! “A protocol for the determination of absorbed dose
from high-energy photon and electron beams,” Medical
REFERENCES Physics 10(6), 741–771 (1983).
17. R. J. Schulz, et al., “Clarification of the AAPM Task Group
1. J. P. Bhatnagar, J. Novotny, M. Saiful Huq, “Dosimetric 21 protocol,” Medical Physics 13(5), 755–759 (1986).
characteristics and quality control tests for the collima- 18. P. R. Almond, et al., “AAPM’s TG-51 protocol for clini-
tor sectors of the Leksell Gamma Knife® Perfexion™,” cal reference dosimetry of high-energy photon and elec-
Medical Physics 39(1), 231–236 (2012). tron beams,” Medical Physics 26(9), 1847–1870 (1999).
2. C. Lindquist and I. Paddick, “THE Leksell Gamma 19. R. Alfonso, et al., “A new formalism for reference dosim-
Knife Perfexion and comparisons with it’s predeces- etry of small and nonstandard fields,” Medical Physics
sors,” Neurosurgery 61(September), 130–141 (2007). 35(11), 5179–5186 (2008).
20. IAEA, Dosimetry of Small Static Fields Used in External Leksell Gamma Knife Perfexion™,” Physics in Medicine
Beam Radiotherapy (2017). and Biology 54(12), 3897–3907 (2009).
21. S. Pai, et al., “TG-69: Radiographic film for megavolt- 24. H. Benmakhlouf, J. Johansson, I. Paddick, P. Andreo,
age beam dosimetry,” Medical Physics 34(6), 2228–2258 “Monte Carlo calculated and experimentally deter-
(2007). mined output correction factors for small field detectors
22. B. M. Coursey, et al., Radiochromic Film Dosimetry in Leksell Gamma Knife Perfexion beams,” Physics in
Radiochromic film dosimetry: Recommendations of AAPM Medicine and Biology 60(10), 3959–3973 (2015).
Radiation Therapy Committee Task Group 55 (1998). 25. A. Molineu, S. Kry, P. Alvarez, N. Hernandez, T.
23. L. Ma, P. Kjäll, J. Novotny, H. Nordström, J. Johansson, Nguyen, D. Followill, “SU-G-TeP2-12: IROCHouston
L. Verhey, “A simple and effective method for valida- and MDAPL SRS anthropomorphic phantom results,”
tion and measurement of collimator output factors for Medical Physics 43(6 Part 27), 3665 (2016).
Chapter 23
Kilovoltage X-Ray Beam Dosimetry

C. M. Charlie Ma
Fox Chase Cancer Center
CONTENTS
23.1 Review of Kilovoltage X-Ray Dosimetry 355
23.2 Kilovoltage X-Ray Dosimetry Protocol 357
23.2.1 Dosimeter Calibration 357
23.2.2 Formalisms for Kilovoltage X-ray Dosimetry 358
23.3 Dosimetric Considerations for Kilovoltage X-Rays 360
23.3.1 Detectors and Equipment 360
23.3.2 Kilovoltage X-Ray Beam Specification 362
23.3.3 Relative Dosimetry Measurement 362
23.3.4 Dose to Medium Conversion 363
23.3.5 Dosimetry Uncertainty 365
23.4 Clinical Implementation of Kilovoltage X-Rays 366
23.4.1 Acceptance Testing 367
23.4.2 Clinical Commissioning 370
23.4.3 Routine Quality Assurance 370
23.5 Summary 371
References 371
23.1 REVIEW OF KILOVOLTAGE and for patient setup and target localization in radio-
X-RAY DOSIMETRY therapy treatment [7, 8]. On the other hand, kilovoltage
Kilovoltage x-ray radiation has been used for medical x-rays are widely used in radiobiology studies for in vitro
applications including imaging and cancer therapy ever cell irradiation and in vivo animal experiment [9–11].
since Roentgen discovered x-rays in 1895 [1, 2]. X-rays The dosimetric and biological effects of kilovoltage
were first used by Emil Herman Grubbe to treat a patient x-ray beams were investigated in many ways from the
with breast cancer, only one year after the discovery of early simple visual observations of skin reddening, to
x-rays [3]. Although the use of kilovoltage x-ray beams photographic or fluorescence methods to measure x-ray
for external beam radiation therapy decreased drasti- intensities, and to primitive chemical and thermal mea-
cally with the development of 60Co teletherapy systems surements of x-ray absorption. The introduction of the
and medical linear accelerators that generate megavolt- roentgen in 1928 at Stockholm Congress of Radiology
age electron and photon beams, kilovoltage x-ray ther- marked the beginning of precise physical measurement
apy is still a viable option for treating superficial targets of the radiation exposure, which was later re-defined
such as skin cancers and for intraoperative radiation at the Chicago Congress of radiology in 1937 for both
therapy [4–6]. Kilovoltage x-ray imaging has received x- and γ-rays. The device used to measure exposure
widespread applications both for diagnostic purposes in accordance with the definition of roentgen was a
355
“free-air” chamber, which has been used in almost all chamber related perturbation correction factor for
the standards laboratories to establish exposure stan- the medium-energy formalism, which was up to 10%
dards. A free-air chamber must be large enough so that at the low-energy end, created some controversies and
the secondary electrons that produce the ionization was later revised. This triggered many studies in the
can originate and complete their tracks in the air of the late 1980s and early 1990s on the various aspects of
chamber. Initial measurements between the standards kilovoltage x-ray dosimetry (c.f., Ma and Seuntjens [5])
chambers of several national laboratories showed good and several new dosimetry protocols were published
agreement (±1%) in the energy range of 100–180 kV as [20–22].
reported by Taylor [12] but were later demonstrated to The AAPM TG61 was formed in 1994 to evaluate
be untrue; Errors of the order of 1.5% and 2.5% were the situation and to produce a dosimetry guide for the
reported 20 years later [13]. American physicists espe- users in North America. The TG61 report “AAPM pro-
cially those at the NIST made major contributions to the tocol for 40-300 kV x-ray beam dosimetry in radiation
standardization of x-ray measurements using free-air therapy and radiobiology” is the only AAPM protocol
chambers [14]. on kilovoltage x-ray dosimetry [23]. It provided detailed
Dosimetry protocols play an essential role in the recommendations for reference as well as relative dosim-
accurate delivery of radiation therapy dose clinically. etry. It provided more complete date for dose conversion
A survey [15] carried out by the AAPM Task Group coefficients, backscatter factors and chamber correc-
61 (TG61) showed that in North America a variety of tion factors, and it also provided guidelines for clinical
dosimetry procedures were used in practice with a com- dosimetry and quality assurance (QA) measurements.
bination of conversion and correction factors measured Different from previous dosimetry protocols, TG61 rec-
and/or taken from different protocols. The National ommended the backscatter method for the low-energy
Council on Radiation Protection and Measurements (up to 150 kV) range and both the backscatter method
(NCRP) Report No. 69 [16] provided a formula to calcu- and the in-phantom method for the medium-energy
late the dose to a phantom material at a point in air (with (100–300 kV) range. Investigators have compared the
a proper buildup) for tube potential 10 kV through the dosimetry consistency between the two dosimetry
medium-energy range. Based on NCRP 69 a backscatter methods for the overlapping x-ray energies and the
factor will be needed to calculate the dose on the phan- results showed good agreement for both methods using
tom surface for clinical dose determination but NCRP the TG61 formalism and dosimetric data [24]. No major
69 did not provide the values of this factor. Therefore, errors have been reported since the publication of the
some hospitals adopted the ICRU Report No. 23 [17], TG61 report in 2001 [8].
which recommended a backscatter method for the New dose determination methods have been inves-
low-energy (40–150 kV) range with the backscatter fac- tigated since then to determine the dose to water in the
tors taken from BJR suppl. 11 [18], and an in-phantom NIST-matched x-ray beams at the University of Wisconsin
method for medium-energy (150–300 kV) x-rays. The Accredited Dosimetry Calibration Laboratory (ADCL),
in-phantom method was for the first time introduced to e.g., the Monte Carlo-based method and the 60Co-based
dose measurement for medium-energy x-rays, which at method [25]. The 60Co method used the absorbed dose-
that time was still widely used for treating deep-seated to-water calibration of an ionization chamber acquired
tumors (therefore the name orthovoltage radiotherapy). from an ADCL and accounted for differences in beam
The IAEA published a technical report series No. 277 quality with a beam quality correction factor (kQ), which
[19] “Absorbed Dose Determination in Photon and is analogous to the procedure that is outlined in AAPM
Electron Beams; An International Code of Practice,” TG51 [26]. Both methods showed excellent agreement
which also recommended the backscatter method for with the TG61 method although their estimated uncer-
the low-energy range and the in-phantom method for tainty has been much reduced compared to TG61. For
the medium-energy range although the energy ranges example, the Monte Carlo method and the TG61 in-
were defined slightly different (low-energy: 10–100 kV; phantom method agreed within 1.8% for all energies,
medium-energy: 100–300 kV). The backscatter factors and it also agreed with the 60Co method within 1.8% for
for the low-energy formalism were calculated using all of the beams. The TG61 in-phantom method and the
60Co method showed a maximum discrepancy of 1.0%.
the Monte Carlo method while the introduction of a
Kilovoltage X-Ray Beam Dosimetry ◾ 357
Their results also compare favorably with calorimetry- protocol and the terminology used here is consistent
based standards established at other institutions [27]. with that defined in the TG61 protocol [23].
Based on a thorough review of the dosimetry protocols
currently available, the Kilovoltage Dosimetry Working 23.2.1 Dosimeter Calibration
Group of the Australasian College of Physical Scientists The TG61 dosimetry formalism is based on the air
and Engineers in Medicine (ACPSEM) Radiation kerma concept [23]. Air kerma, Kair, is defined as the
Oncology Specialty Group (ROSG) recommended the ratio of the kinetic energy, ΔE, transferred from x-rays
use of the AAPM TG61 as the reference dosimetry code to charged particles to a given air mass, Δm (i.e., Kair =
of practice [28]. ΔE/Δm). If Kair is the air kerma at the reference point in
This chapter suggests the use of AAPM TG 61 for air for a given beam quality and M the reading (cor-
kilovoltage x-ray reference dosimetry for now because rected for temperature, pressure, recombination and
it is based on air-kerma standards, which are readily polarity effect) of an ionization chamber to be calibrated
available in most countries and regions, and because with its center at the same point, the air kerma calibra-
it currently contains the widest range of tabulated tion factor, NK, for this chamber at the specified beam
data with respect to beam characteristics and mea- quality is defined as:
surement detectors. The in-air and in-phantom
methods, as recommended by TG61, are easy to use K air
NK = (23.1)
and produce accurate results compared with other M
recent methods that are only available to standards
laboratories and a few advanced centers [25, 27, 28]. Many earlier dosimetry protocols for kilovoltage x-rays
The TG61 protocol also has a comprehensive discus- were based on the exposure, X, which is defined as the
sion on issues related to relative dosimetry, which are total electric charge, ΔQ, of all ions of one sign (e.g., elec-
useful to the users in the clinical implementation of trons), produced in air by x-rays in a unit mass of air,
an x-ray unit. Older reference dosimetry protocols or Δm (i.e., X = ΔQ/Δm). A similar equation can be used to
codes of practice that are based on exposure measure- derive the exposure calibration factor, NX:
ments should not be used because they may contain
erroneous data [8]. The measurement methods and X
NX = (23.2)
dosimetric procedures described in this chapter are M
primarily based on the recommendations of the TG61
report [23]. We can derive the air kerma calibration factor from the
exposure calibration factor from
23.2 KILOVOLTAGE X-RAY
DOSIMETRY PROTOCOL W
N K = N X   (1 − g )−1 (23.3)
Kilovoltage x-rays have some unique properties that  e  air
require special attention in their applications in radia-
tion therapy and radiobiological applications. For where ( We )air has the value 33.97 J/C (or 0.876 cGy per
example, clinically available 50–300 kV x-rays have a roentgen) for dry air, (1 − g) corrects for the effect of
very small electron range (<0.5 mm of water). Their dose radiative losses by charged particles to bremsstrahlung
distributions include a significant (up to 30%) scatter photons (g is the fractional energy lost to bremsstrah-
component, which is energy and field size dependent. lung photons in air, which is practically zero for x-ray
Because of the negligible radiative energy loss for this beams below 300 kV).
energy range the collision kerma is generally consid- Both NK and NX factors can be obtained from stan-
ered to be the same as kerma and the absorbed dose. dards laboratories for a number of available x-ray beam
Commonly used ionization chambers have been gener- qualities to match the actual x-ray beams for an x-ray
ally calibrated as “exposure meters” and used as “pho- unit. In the calculation of dose to a patient for a kilovolt-
ton detectors” since the well-known Bragg–Gray cavity age x-ray beam, the accuracy of the chamber calibration
theory no longer applies to this energy range [29]. The determines the final accuracy (see uncertainty analysis
following sections describe the AAPM TG61 dosimetry in the next section). Each chamber involved in reference
dosimetry of the clinical kilovoltage setup must be cali-

brated, since there can be up to 8% differences in the cal-
ibration factor for different chambers of the same type.
The chamber must be calibrated for all the beam quali-
ties available for the x-ray unit to minimize the effect of
the chamber energy dependence. Well-designed cham-
bers usually exhibit less than 6% energy dependence
between 40 kV and 300 kV.
23.2.2 Formalisms for Kilovoltage X-ray Dosimetry

Two major methods have been investigated extensively
for kilovoltage x-ray dosimetry determination, i.e., the
in-air method and the in-phantom method. The AAPM
TG61 recommends the use of the in-air method for
low-energy (50–100 kV) x-ray dosimetry for superfi-
cial radiotherapy applications. For medium-energy
(100–300 kV) x-ray dosimetry determination the TG61
recommends to use either the in-air or the in-phantom
methods depending on the application [23]. This recom-
mendation was made based on the fact that the determi-
nation of the surface dose can be more accurate using
the backscatter method if the point of interest is the skin
surface. On the other hand, if the point of interest is at
a deeper depth it is more accurate to use the in-phan-
tom method to determine the dose. The uncertainty
would be much greater if one derived the dose at a depth
based on a surface dose calibration (i.e., using the in-air
method) because of the large uncertainties in the depth
dose values near the phantom surface, and vice versa.
The In-Air Method: This method can be used for both
superficial and orthovoltage x-ray dose determination.
In this method the ion chamber is calibrated free in FIGURE 23.1 Schematic diagrams to illustrate (a) the in-air
air in a standards laboratory, and used to measure air method and (b) the in-phantom method.
kerma free in air in a user’s beam without the presence
of a water phantom. A backscatter factor is needed to of a phantom, Bw the backscatter factor that accounts
account for the effect of photon backscatter from the for the effect of the phantom (water) scatter, and Pstem,air
phantom (Figure 23.1a). The dose to water at the phan- the chamber stem correction factor that accounts for
tom surface can be calculated by the change in photon scatter from the chamber stem
between the calibration and measurement due to the
( )
w w
µ  difference in field size. Here µρen is independent of
Dw = MN K  en  Bw Pstem,air (23.4) air
 ρ  air the field size used since it is evaluated over the primary
beam only [30]. The TG61 recommended values of
where M is the corrected ionization chamber reading
free in air, NK the chamber air kerma calibration factor
( )
µen w and B for some typical beam qualities are given
ρ air
w
in Tables 23.1 and 23.2 [23]. Pstem,air is taken as unity for a

( )
w
at the user’s beam quality, µρen the ratio of mass energy
air calibrated chamber if, for a given beam quality, the same
absorption coefficients for water to air, evaluated over field size is used in the calibration and the measure-
the x-ray energy spectrum free-in-air, in the absence ment. Otherwise, the Pstem,air factor can be measured
TABLE 23.1 Ratios of Average Mass Energy Absorption by intercomparing the chamber with unknown Pstem,air
Coefficients for Water to Air to Convert Air Kerma to Water Kerma, with a reference chamber for which Pstem,air is known. A
Free-In-Air for Both Superficial and Orthovoltage X-Rays
Farmer type cylindrical chamber with flat response can
( )
w
HVL µ en
ρ
air
(free-in-air) be used as a reference chamber for the measurement of
(mm Al) (mm Cu) Pstem,air of another chamber since its stem effect varies
0.03 1.047 little with field size (<1%).
0.05 1.046
The In-Phantom Method: For orthovoltage x-rays,
0.08 1.044
TG61 recommends the in-phantom method if the
0.10 1.044
0.3 1.035
point of interest is at a depth in the patient [23]. In this
0.5 1.028 method, the chamber is calibrated free in air in a stan-
0.8 1.022 dards laboratory and used to measure the air kerma at
1.0 1.020 the reference depth in water (Figure 23.1b). Assuming
3.0 1.021 the reference depth is 2 cm in water the dose to water for
5.0 1.029 orthovoltage x-rays can be calculated using
8.0 1.045
0.1 1.020 w
µ 
0.3 1.035 Dw = MN K PQ,cham Psheath  en  (23.5)
0.5 1.050  ρ  air
0.8 1.068
1.0 1.076 where M is the corrected ionization chamber reading
3.0 1.100 in water, NK the chamber air kerma calibration factor
( )
w
5.0 1.109 at the user’s beam quality, and µρen the ratio of mass
air
TABLE 23.2A Water Kerma-Based Backscatter Factors, Bw, for Different Field Diameters (d) and SSD for Superficial X-Rays
SSD d HVL (mm Al)
(cm) (cm) 0.03 0.05 0.08 0.1 0.3 0.5 0.8 1.0 3.0 5.0 8.0
20 1 1.005 1.007 1.011 1.014 1.028 1.036 1.043 1.046 1.061 1.058 1.053
3 1.005 1.008 1.014 1.019 1.049 1.069 1.092 1.105 1.158 1.165 1.158
5 1.005 1.008 1.014 1.019 1.054 1.080 1.112 1.131 1.215 1.234 1.236
10 1.005 1.008 1.014 1.019 1.057 1.088 1.129 1.155 1.291 1.334 1.354
15 1.006 1.008 1.014 1.019 1.058 1.090 1.133 1.162 1.321 1.380 1.414
20 1.006 1.008 1.014 1.019 1.058 1.091 1.136 1.165 1.334 1.402 1.444
30 1 1.005 1.007 1.011 1.015 1.027 1.035 1.043 1.047 1.063 1.059 1.053
3 1.005 1.008 1.014 1.019 1.048 1.069 1.093 1.107 1.164 1.168 1.158
5 1.005 1.008 1.014 1.019 1.053 1.079 1.111 1.130 1.221 1.242 1.237
10 1.006 1.008 1.014 1.019 1.057 1.088 1.130 1.157 1.298 1.350 1.367
15 1.006 1.008 1.014 1.019 1.058 1.091 1.136 1.165 1.332 1.403 1.434
20 1.006 1.008 1.014 1.019 1.058 1.091 1.138 1.169 1.350 1.428 1.472
50 1 1.005 1.007 1.011 1.014 1.027 1.035 1.042 1.045 1.065 1.059 1.052
3 1.005 1.007 1.013 1.018 1.049 1.070 1.093 1.106 1.163 1.169 1.160
5 1.005 1.007 1.013 1.018 1.054 1.081 1.113 1.132 1.226 1.241 1.242
10 1.006 1.007 1.013 1.018 1.057 1.091 1.134 1.159 1.309 1.352 1.375
15 1.006 1.007 1.013 1.018 1.058 1.093 1.140 1.169 1.346 1.411 1.448
20 1.006 1.007 1.013 1.018 1.058 1.094 1.142 1.173 1.363 1.443 1.493
100 1 1.005 1.007 1.011 1.014 1.028 1.036 1.042 1.044 1.062 1.059 1.053
3 1.005 1.008 1.014 1.019 1.050 1.070 1.092 1.104 1.163 1.169 1.162
5 1.006 1.008 1.014 1.019 1.055 1.082 1.113 1.131 1.225 1.240 1.243
10 1.006 1.008 1.014 1.019 1.058 1.091 1.134 1.158 1.311 1.351 1.381
15 1.006 1.008 1.014 1.019 1.059 1.094 1.140 1.169 1.354 1.417 1.460
20 1.006 1.008 1.014 1.019 1.059 1.095 1.143 1.172 1.375 1.451 1.508
TABLE 23.2B (continued). Water Kerma-Based Backscatter Factors, Bw, for Different Field Diameters (d) and SSD for Orthovoltage X-Rays
SSD d HVL (mm Cu)
(cm) (cm) 0.1 0.3 0.5 0.8 1.0 3.0 5.0
20 1 1.061 1.055 1.053 1.048 1.045 1.024 1.018
3 1.158 1.168 1.155 1.147 1.140 1.082 1.057
5 1.214 1.242 1.233 1.219 1.209 1.127 1.088
10 1.290 1.352 1.353 1.339 1.326 1.204 1.141
15 1.320 1.407 1.415 1.403 1.389 1.251 1.174
20 1.333 1.434 1.447 1.436 1.421 1.278 1.194
30 1 1.063 1.056 1.052 1.047 1.044 1.024 1.018
3 1.164 1.169 1.155 1.146 1.139 1.084 1.055
5 1.220 1.242 1.235 1.221 1.211 1.130 1.087
10 1.297 1.363 1.367 1.347 1.332 1.214 1.147
15 1.330 1.417 1.438 1.422 1.405 1.270 1.189
20 1.348 1.446 1.478 1.464 1.446 1.302 1.213
50 1 1.065 1.054 1.052 1.047 1.045 1.025 1.018
3 1.163 1.170 1.157 1.148 1.140 1.084 1.057
5 1.225 1.247 1.240 1.226 1.214 1.131 1.089
10 1.308 1.367 1.376 1.360 1.344 1.222 1.152
15 1.345 1.433 1.452 1.446 1.428 1.285 1.195
20 1.361 1.471 1.499 1.495 1.478 1.325 1.226
100 1 1.062 1.055 1.052 1.047 1.045 1.025 1.018
3 1.163 1.170 1.160 1.150 1.142 1.085 1.057
5 1.224 1.245 1.241 1.227 1.217 1.132 1.090
10 1.310 1.370 1.383 1.369 1.353 1.226 1.155
15 1.353 1.447 1.463 1.458 1.441 1.291 1.204
20 1.373 1.490 1.513 1.516 1.499 1.334 1.237
energy absorption coefficients for water to air, evaluated primary concern) or at a 2 cm depth in water (dose at
over the x-ray energy fluence at 2 cm depth in water, in greater depths is the primary concern). Cylindrical
the absence of the chamber. The overall chamber correc- chambers that have a calibration factor varying with the
tion factor PQ,cham accounts for the effect of the change beam quality by less than 3% in this energy range are
in chamber response due to photon energy and angular recommended for the beam output measurement. For
variation between chamber calibration in air and mea- superficial x-rays the in-air method is recommended.
surement in water, the effect of chamber stem between Cylindrical chambers that have a calibration factor
calibration in air and measurement in water and the
effect of displacement of water by the chamber in the TABLE 23.3 Ratio of Average Mass Energy Absorption
measurement in water. The sheath correction factor Coefficients for Water to Air at 2 cm Depth in Water for a 100 cm2
Psheath is needed when a waterproofing sheath is used, Field Defined at 50 cm SSD for Orthovoltage X-Rays
which is not directly related to the individual chamber
( )
µ en w
( )
w ρ air
type. The values of µρen , PQ,cham, and Psheath for some HVL
air (mm Cu) (mm Al)
typical beam qualities are given in Tables 23.3–23.5 [23].
0.1 2.9 1.026
0.3 6.3 1.037
23.3 DOSIMETRIC CONSIDERATIONS
0.5 8.5 1.046
FOR KILOVOLTAGE X-RAYS
0.8 10.8 1.055
23.3.1 Detectors and Equipment 1.0 12.0 1.060
Dosimeters: TG61 recommends the use of air-filled 2.0 15.8 1.081
ionization chambers for kilovoltage x-ray beam ref- 3.0 17.9 1.094
erence dosimetry [23]. Measurements are performed 4.0 19.3 1.101
5.0 20.3 1.105
either free in air (in cases where the surface dose is the
TABLE 23.4 Overall Chamber Correction Factors PQ,cham for Commonly Used Cylindrical Chambers in
Orthovoltage X-Ray Beams
Chamber Type NE2571 Capintec PR06C PTW N30001 NE2611 or NE2561
HVL (mm Cu)
0.10 1.008 0.992 1.004 0.995
0.30 1.023 1.008 1.021 1.017
0.50 1.025 1.010 1.023 1.019
0.80 1.024 1.010 1.022 1.018
1.0 1.023 1.010 1.021 1.017
2.0 1.016 1.007 1.015 1.011
3.0 1.009 1.005 1.010 1.006
4.0 1.004 1.003 1.006 1.003
5.0 1.002 1.001 1.002 1.001
The data applies to the in-phantom method for a chamber at 2 cm depth in water and a 100 cm2 field.
varying with the beam quality by less than 5% between ion recombination and polarity effect before applying
50 and 150 kV are recommended. Extensive studies have Equations 23.1–23.5.
been carried out on the correction factors for the com- Electrometers: The TG61 protocol [23] recommends
monly used Farmer-type chambers for the in-phantom that electrometers for kilovoltage x-ray therapy mea-
measurement [5, 31, 32–35]. Other cylindrical cham- surements should be capable of reading currents on
bers may also be used. However, correction factors must the order of 0.1 nA, with an accumulated charge of
then be determined by comparing these chambers with 50–100 nC. They are calibrated by a standards labora-
a chamber with known correction factors. If measure- tory with proper correction factors applied in the mea-
ments are performed in water, a thin waterproofing surement. These correction factors are generally close to
sleeve should be used and appropriate correction factors 1.000; but can occasionally be 5% different from 1.000.
should be applied depending on the material and thick- Electrometers and ionization chambers can be cali-
ness used [47]. If measurements are performed in air, brated either together or sometimes separately and if so
the thimbles of cylindrical chambers are thick enough, their calibration factors must be combined.
so no build-up cap is required. For x-ray energies below Phantoms: The TG61 protocol recommends the use
70 kV, calibrated parallel-plate chambers with a thin of a water phantom when the in-phantom method is
entrance window are recommended. Thin plastic build- used in orthovoltage x-ray reference dosimetry [23].
up foils should be added to the entrance window, if nec- Conversion factors are used to convert from the air
essary, to provide full electron buildup and to eliminate kerma in the sensitive chamber cavity to the dose to
electron contamination (see Table 23.6). All chamber water. Plastic phantoms are not recommended for in-
readings should be corrected for temperature, pressure, phantom reference dosimetry for orthovoltage x-rays as
TABLE 23.5 Correction Factors for PMMA Sheaths when Using Cylindrical Chambers for In-water
Measurements in Orthovoltage X-Ray Beams
HVL PMMA (Lucite)
(mm Cu) (mm Al) t = 0.5 mm t = 1 mm t = 2 mm t = 3 mm
0.1 3.0 0.998 0.995 0.991 0.986
0.3 6.1 0.998 0.997 0.994 0.991
0.5 8.5 0.999 0.998 0.996 0.994
0.8 11.0 0.999 0.998 0.997 0.996
1.0 12.1 1.000 0.999 0.998 0.997
2.0 15.2 1.000 1.000 1.000 0.999
3.0 17.6 1.000 1.000 1.000 1.000
4.0 19.4 1.000 1.000 1.000 1.000
5.0 20.9 1.000 1.000 1.000 1.000
The data applies to 2 cm depth in water and a 100 cm2 field.
TABLE 23.6 Thickness of Build-Up Material for Thin-Window

Parallel-Plate Chambers Used for In-Air Calibrations with X-Ray
Energy below 100 kV
Generating Potential Thickness of Foil
(kV) (mg cm−2)
50 1.5
60 3.0
70 4.7
80 6.6
90 8.7
100 10.9 FIGURE 23.2 Recommended experimental setup for HVL
measurement including the source (target), the HVL filter
the chamber correction factors and conversion factors (absorber), the diaphragm and the measurement ion chamber.
to derive dose at a depth in water for these phantoms
are poorly known. Some “water equivalent” plastics are and the measurement chamber. The field size is reduced
commercially available for routine consistency measure- to just enough to encompass the chamber. The absorb-
ments. Their properties must be investigated before they ing materials (Al or Cu) are placed at least 50 cm from
can be used as water substitutes for reference dosimetry. the measurement chamber and close to the collimating
aperture. The absorber should be made of high-purity
23.3.2 Kilovoltage X-Ray Beam Specification (99.9%) material and the thickness of the absorber should
X-ray beams with different energy spectra exhibit be measured accurately. The thickness that reduces the
penetrating powers in biological media. The quality air kerma value to one half is obtained by interpolation.
of an x-ray beam refers to its penetrating power, with A monitor detector is used to correct fluctuations of air
high beam quality indicating deeper penetration in a kerma rate (Figure 23.2). The measurement chamber
medium or less attenuation by the medium. The exact must have limited beam quality dependence (within
knowledge of beam quality plays an important role in 5% between 40 and 300 kV) for accurate HVL mea-
clinical radiotherapy to ensure proper treatment target surements. Thin-walled chambers are used for lightly
coverage and adequate skin sparing. TG61 recommends filtered beams to measure the low-energy x-ray compo-
the use of more than one beam quality parameter to nents accurately.
specify an x-ray beam [23]. The usual quantities used are It is well known that the peak voltage across an x-ray
the half-value layer (HVL) and the accelerating poten- tube determines the maximum photon energy in an
tial. A literature review shows that most dosimetric data x-ray spectrum. Traditionally, the accelerating poten-
for kilovoltage x-rays have been given using one beam tial is expressed in kV for a kilovoltage x-ray beam to
quality parameter, usually HVL. indicate that the x-ray beam actually has an energy
The AAPM TG61 defines the HVL as the thickness of spectrum with its maximum energy up to its acceler-
an absorber that reduces the air kerma rate of a narrow ating potential. It should be noted that most often the
unidirectional x-ray beam at a point distant from the accelerating potential of a kilovoltage x-ray beam is
absorber to 50% of that of the non-attenuated beam [23]. nominal rather than actual. Higher accelerating poten-
For kilovoltage x-rays, aluminum (Al) and copper (Cu) tials produce more penetrating x-rays and, therefore
have been used as the absorbing material for the deter- have a direct effect on the beam quality. A review of
mination of HVL. The beam quality is expressed in mm x-ray equipment for low and medium energies can be
Al or mm Cu. For superficial and contact x-ray beams, found in Ma [6].
only mm Al is used to express HVL. For orthovoltage
x-rays, both Al and Cu can be used to specify the beam 23.3.3 Relative Dosimetry Measurement
quality although Cu is more frequently used in practice. Output Factors: For kilovoltage x-ray dosimetry, the out-
The AAPM TG61 protocol recommends the use of put factor is defined as the surface dose value for a given
an ionization chamber to measure the HVL [23]. A nar- source-to-surface distance (SSD) and field size relative
row beam setup should be used, in which a collimat- to that under the reference conditions. Clinically, out-
ing aperture is placed half way between the x-ray target put factors are required for all combinations of SSD and
field size used for kilovoltage x-ray treatment. Since the 20–40% for kilovoltage x-rays. Significant corrections
scatter contribution from the inside of a cone applica- with depth may be required for the PDD measurement
tor may be significant, it is not sufficiently accurate to with these chambers. Specifically designed parallel-
estimate output factors for different applicators using plate chambers for low-energy x-rays usually have a flat
the ratio of the backscatter factors corresponding to the energy response in air but not at a depth in a phantom.
respective field sizes. The output factor for each individ- For instance, more than 10% variations in chamber
ual applicator must be measured at each beam quality. response have been observed for the Capintec PS-033
If the in-phantom calibration method has been used for chambers. Thus, a depth-related correction factor may
orthovoltage x-rays it is necessary to obtain the depth be required for these chambers to be used in accurate
dose values in order to determine the dose at the surface PDD measurements.
(see Section 23.3.4). This may result in large uncertain- A water tank with a small-volume scanning ioniza-
ties in the output factors since the uncertainties are gen- tion chamber is recommended for the PDD and profile
erally high in the depth dose values near the surface. measurement. A monitor chamber should be used to
Percentage Depth Dose and Lateral Dose Profiles: measure any erratic fluctuations in dose rate. If a thin
Percentage depth dose (PDD) and lateral dose profiles window parallel plate chamber is used, the chamber
are important clinical quantities for treatment planning. must have sufficient buildup material placed over its
The measurement uncertainties are high for kilovolt- entrance window (see Table 23.6). Because of the finite
age x-rays because of the significant variation of x-ray size of the ionization chamber in the beam direction, it
energy (especially for lightly filtered beams) and angu- is necessary to extrapolate the dose to the surface. Since
lar distribution with depth and field size. Care must be the dose distribution may be nonlinear near the phan-
exercised in the measurement of PDD and dose profiles tom surface, caution should be exercised in extrapolat-
for kilovoltage x-ray beams because of their significant ing over the last few millimeters [24, 36].
depth and field size dependence. The PDD can also be measured by placing thin sheets
The suitability of x-ray detectors for relative dosim- of water-equivalent material over the chamber in a phan-
etry measurement has been evaluated extensively [5, 24, tom and moving the chamber back by the same amount
36]. Detectors for the PDD and dose profile measure- to maintain a constant SSD. The water-equivalence of the
ment should have high spatial resolution and constant material in the energy range of interest must be verified.
energy and angular response. As a general requirement PMMA is not suitable for this purpose. Strictly speak-
for the evaluation of a specific detector, the relative ing, an ion chamber measures the “depth-ionization”
in-air chamber response and the relative in-phantom distribution rather than the depth-dose distribution.
response should be compared with a well-behaved cylin- However, the difference between them is small [5] except
drical chamber at depths where reasonable measure- for the surface dose measurement. If a suitable detector
ments with the cylindrical chamber can be performed. for relative dosimetry cannot be identified in the clinic it
Diamond detectors and the NACP parallel plate cham- is recommended to use the data from the British Journal
ber (type) have been found to require relatively small of Radiology Supplement 25 [37] or published data that
depth-dependent correction factors for orthovoltage match the kV and HVL values of the user’s beams (e.g.,
x-ray beams [5]. Other solid detectors, attractive because Podgorsak et al. [38], Gerig et al. [39], Butson et al. [40],
of the small size of their sensitive volume (diode detec- Aukett et al. [41], Kurup and Glasgow [42]).
tor, TLD, film), usually show significant beam quality
dependence or large experimental uncertainties. Well- 23.3.4 Dose to Medium Conversion
designed cylindrical chambers have nearly constant Dose on the Surface: The AAPM TG61 protocol provides
energy response in this beam quality range and are suit- detailed guidelines to determine dose to other biological
able for in-phantom measurements. However, the mea- tissues on the surface or at a depth of a human body for
surement depth is limited to no less than the outer radius clinical radiotherapy and radiobiology [23].
of the chamber. Parallel-plate chambers have been used The surface dose for other media (med) can be calcu-
for measurements at smaller depths. Those parallel- lated from:
plate chambers designed for electron beams usually
have a calibration factor varying with beam quality by Dmed,z = 0 = Cwmed Dw ,z = 0 (23.6)
TABLE 23.7 Free-in-Air Ratios of Mass Energy Absorption Coefficients of Biological Tissue to Water for Application in Conjunction with
the In-air Method (The Data Are for SSD = 50 cm)
HVL “Free-in-Air” Mass Energy Absorption Coefficient Ratio of the Specified Tissues to Water
(mm Al) (mm Cu) Tissue Muscle Lung Skin Bone
0.3 0.917 1.016 1.031 0.890 4.200
0.4 0.918 1.020 1.035 0.893 4.289
0.5 0.919 1.022 1.037 0.895 4.335
0.6 0.920 1.024 1.039 0.897 4.382
0.8 0.921 1.028 1.043 0.901 4.475
1.0 0.923 1.031 1.046 0.904 4.494
1.2 0.925 1.031 1.046 0.907 4.469
1.5 0.927 1.032 1.047 0.910 4.427
2.0 0.930 1.032 1.047 0.915 4.350
3.0 0.934 1.032 1.045 0.922 4.179
4.0 0.939 1.030 1.042 0.929 3.975
5.0 0.943 1.028 1.039 0.935 3.769
6.0 0.947 1.026 1.036 0.940 3.557
8.0 0.955 1.021 1.030 0.950 3.133
0.1 0.934 1.032 1.045 0.921 4.209
0.2 0.942 1.029 1.040 0.934 3.808
0.3 0.947 1.026 1.036 0.940 3.561
0.4 0.952 1.023 1.032 0.946 3.314
0.5 0.956 1.020 1.029 0.952 3.068
0.6 0.960 1.018 1.026 0.957 2.859
0.8 0.964 1.015 1.022 0.961 2.657
1.0 0.967 1.012 1.018 0.965 2.456
1.5 0.975 1.006 1.009 0.975 1.952
2.0 0.981 1.001 1.003 0.980 1.637
3.0 0.986 0.996 0.997 0.985 1.280
4.0 0.988 0.994 0.994 0.987 1.128
5.0 0.990 0.992 0.992 0.989 1.026
with the conversion factor from dose-to-water to dose- Table 23.8 shows the ratios of the backscatter factors,
to-medium given by: bone to water, for photon beams 50–300 kV (0.875–20.8
mm Al) with different field sizes at various SSD.
B  µ  med  Dose at a Depth: The absorbed dose at a depth in a
C med
w = med  en   (23.7) non-water phantom can be calculated by
Bw  ρ  w air
( )
w
where  µρen
med
 represents the ratio of mass energy µ 
 w  air Dmed = MN K PQ,cham  en  Cwmed (23.8)
 ρ  air
absorption coefficients medium to water averaged over
the primary photon spectrum free-in-air, and BBmed the
w
where
ratio of kerma based backscatter factors medium to
water at the surface of the medium versus air. med
The AAPM TG61 report gives the ratios of mass P med  µen 
C med = Q,cham . (23.9)
 ρ  w
w
energy absorption coefficients averaged over the photon PQ,cham
w
fluence spectrum free-in-air of media of clinical inter-

est relative to water (Table 23.7). The backscatter factor is the conversion factor. The chamber correction fac-
ratios relative to water for these biological tissues, except tor for non-water phantoms is not well known but it is
bone is not differing from unity by more than 1.2% for expected to be similar to that for water for low-Z mate-
a large field size and can therefore be ignored. For bone, rials if their density does not differ significantly from
TABLE 23.8 Ratios of the Kerma-Based Backscatter Factors, Bone to Water, for Photon Beams 50–300 kV (0.05–5 mm Cu) with Different
Field Sizes and SSD
Bbone
SSD HVL Bw
(cm) (mm Cu) (mm Al) 1 × 1cm2 2 × 2cm2 4 × 4cm2 10 × 10cm2 20 × 20cm2
10 0.05 1.6 0.958 0.929 0.897 0.861 0.854
0.1 2.9 0.976 0.945 0.905 0.853 0.838
0.5 8.5 1.019 1.011 0.974 0.910 0.875
1 12.0 1.031 1.041 1.026 0.974 0.943
2 15.8 1.038 1.065 1.077 1.047 1.023
3 17.9 1.037 1.066 1.092 1.086 1.070
4 19.3 1.028 1.053 1.082 1.087 1.075
5 20.3 1.022 1.043 1.074 1.087 1.078
30 0.05 1.6 0.958 0.926 0.889 0.850 0.833
0.1 2.9 0.976 0.940 0.894 0.837 0.809
0.5 8.5 1.019 1.011 0.981 0.887 0.833
1 12.0 1.031 1.042 1.033 0.959 0.902
2 15.8 1.038 1.067 1.083 1.043 0.989
3 17.9 1.037 1.067 1.101 1.090 1.047
4 19.3 1.029 1.055 1.088 1.091 1.065
5 20.3 1.023 1.045 1.077 1.091 1.079
50 0.05 1.6 0.958 0.927 0.891 0.847 0.827
0.1 2.9 0.975 0.942 0.897 0.832 0.800
0.5 8.5 1.018 1.009 0.977 0.881 0.825
1 12.0 1.031 1.040 1.032 0.958 0.894
2 15.8 1.038 1.066 1.085 1.047 0.983
3 17.9 1.036 1.069 1.100 1.095 1.048
4 19.3 1.028 1.057 1.084 1.094 1.066
5 20.3 1.022 1.047 1.072 1.094 1.080
unity. Thus, the conversion factor can be given as the depth and SSD deviate significantly from 100 cm2, 2 cm
ratio of the mass energy-absorption coefficients for the and 50 cm, respectively.
non-water phantom to water averaged over the photon
23.3.5 Dosimetry Uncertainty
( )
med
spectrum at the depth z, µρen . The waterproofing
w The International Commission on Radiation Units and
sleeve correction factor is not needed for solid phan-
Measurements (ICRU) report 24 recommended that the
toms. In general, ( )
µen med
ρ w is depth, SSD and field-size uncertainty on the absorbed dose that can be delivered
dependent. to the treatment target in a clinical situation be prefer-
The AAPM TG61 provides the ratios of mass-energy ably better than ±5% [43]. The overall uncertainty on the
absorption coefficient ratios of tissue to water at depth dose delivered to the target consists of several compo-
in water (Table 23.9). It was found that for all the tis- nents, the first part of which occur as a result of uncer-
sues listed in Table 23.9, except for bone, depth depen- tainties in the calibration chain linking the calibration
dence of the conversion factors is at most 0.9%, field-size of the clinical beam to the standards laboratory (NK-
dependence at most 0.8%, and SSD dependence at most factor, uncertainties on conversion and correction fac-
0.1% [48]. Therefore, this table can be, except for bone, tors, beam quality specification uncertainties, etc.) and
applied to any depth, field size and SSD. For bone, the a second part associated with clinical errors in patient
differences were found to be up to 16% (depth depen- setup, immobilization, target localization and dose
dence), 27% (field size dependence), and 1.5% (SSD delivery. Table 23.10 lists several components contrib-
dependence), as shown in Figures 23.3, if the field size, uting to the uncertainty (including type A and type B
TABLE 23.9 Ratios of the Mass Energy Absorption Coefficients of Tissue to Water for
Converting Dose to Water to Dose to Tissue at an Equivalent Depth
“In-Phantom” Ratio of Mass Energy Absorption
HVL Coefficients of the Specified Tissue to Water
(mm Al) (mm Cu) Tissue Muscle Lung Bone
0.3 0.921 1.030 1.046 4.540

0.4 0.922 1.033 1.049 4.611
0.5 0.922 1.033 1.049 4.615
0.6 0.923 1.034 1.049 4.619
0.8 0.924 1.035 1.050 4.627
1.0 0.925 1.035 1.050 4.585
1.2 0.927 1.035 1.049 4.514
1.5 0.930 1.035 1.049 4.433
2.0 0.932 1.034 1.048 4.345
3.0 0.936 1.033 1.046 4.177
4.0 0.939 1.031 1.044 4.025
5.0 0.942 1.030 1.041 3.873
6.0 0.945 1.028 1.039 3.719
8.0 0.951 1.024 1.034 3.410
0.1 0.936 1.033 1.047 4.200
0.2 0.941 1.030 1.042 3.901
0.3 0.945 1.028 1.039 3.722
0.4 0.948 1.026 1.036 3.542
0.5 0.951 1.024 1.033 3.363
0.6 0.954 1.022 1.031 3.205
0.8 0.958 1.019 1.027 3.014
1.0 0.961 1.017 1.024 2.823
1.5 0.969 1.011 1.016 2.346
2.0 0.974 1.006 1.011 2.015
3.0 0.981 1.000 1.002 1.565
4.0 0.985 0.997 0.997 1.330
5.0 0.988 0.994 0.994 1.136
Note: The data are for SSD = 50 cm, 100 cm2 field size and 2 cm depth.
uncertainties) in the dose to medium, as described in the TG61 recommendations. This was confirmed by recent
TG61 report [23] without considering the clinical errors. studies using different methods with high-precision cal-
These uncertainties are fairly large due to both the lim- culations and measurements at standards laboratories
ited amount of experimental data and the limited com- [25, 27].
putational resources available when it was published.
For example, the uncertainties on the dose conversion 23.4 CLINICAL IMPLEMENTATION
and detector correction factors can be much reduced OF KILOVOLTAGE X-RAYS
[25] and the depth dose measurement can be performed The clinical implementation of a therapeutic x-ray unit
more accurately with well-designed chambers. The requires acceptance testing, clinical commissioning
estimated uncertainty for kilovoltage x-ray dosimetry and a routine QA program to ensure its safe operation.
following TG61 is likely to be better than 2% at refer- Acceptance testing is required to ensure that the x-ray
ence points, i.e., at the phantom surface for the in-air unit meets the technical specifications set by the manu-
method, and at 2 cm depth for the in-phantom method. facturer as well as the local department. Clinical com-
For dose determination at depths other than the refer- missioning involves the collection of dosimetry data
ence points and for medium other than water, the esti- for beam characterization, treatment planning and QA
mated uncertainty is likely to be about 3% following the testing for the routine QA program.
FIGURE 23.3 Depth and field size dependence of the ratio of mass energy absorption coefficients, bone to water. Half value
layers for the indicated radiation qualities are: 0.88 mm Al (50 kV), 2.65 mm Al (100 kV), 0.57 mm Al (150 kV), 1.6 mm Cu
(200 kV), and 4.3 mm Cu (300 kV).
23.4.1 Acceptance Testing checks of the x-ray unit, radiation tests, radiation sur-
It is recommended that the user of a kilovoltage x-ray unit veys and checking the general operation and functional-
should work with the vendor to establish an acceptance ity of the unit as well as the treatment room.
testing protocol. The acceptance tests that are required The following tests are examples for the acceptance
for typical clinical applications include mechanical testing of a kilovoltage treatment unit as recommended
TABLE 23.10 Components and Typical Values of the 1σ Uncertainty on Determining the Dose to Medium in the Users
Beam in the Clinic
Type of Quantity or Procedure Uncertainty (%)
In-air method (low and medium energies)
NK from calibration lab. 0.7
Effect of beam quality difference between calibration and measurement 2.0
Backscatter factor B 1.5
Pstem,air 1.0
1.5
( )
µen water
ρ
air
Free air measurement in clinic 1.5

Overall uncertainty for Dw,z=0 3.5
Transfer to other points in water 3.0
Conversion to dose to tissue 1.0
Overall uncertainty for Dmed,z 4.7
In-phantom method (medium energies)

NK from calibration lab. 0.7
Effect of beam quality difference between calibration and measurement 2.0
Chamber correction factors 1.5
Psheath 0.5
1.5
( )
µen water
ρ
air
In-phantom measurement in clinic 2.0

Overall for Dw,z=2 3.6
Transfer to other points in water 3.0
Conversion to dose to tissue 1.0
Overall uncertainty for Dmed,z 4.8
by the ACPSEM kilovoltage dosimetry working group • Reproducibility of positioning of the treatment
[28]. These tests can be performed along with any tests applicators to ensure correct position and treat-
that are specified by the manufacturer. ment distances in the clinical setting.
Safety tests
• Validation of any light field with the mechanical
• The safety switches and emergency buttons func- readout for a range of field sizes at an appropriate
tion properly. SSD.
• The radiation signs at the door and at the console • Confirmation of travel range of the x-ray unit in
activate when the beam is on. any direction available for the treatment should be
consistent with any indicators display on the stand
• The interlock for the room entrance or door works or on the x-ray tube.
correctly and disables the beam from being able to
be switched unless the door is closed or interlock • If the x-ray tube has the ability to be angled
is armed. in different directions, testing of the accuracy
and mechanical rigidity of the system holding
• The patient remote monitoring cameras are func- the treatment applicator in place is required. In
tioning properly. addition, a comparison of the radiation field
Mechanical tests defined by a standard treatment applicator in
a non-rotated position to that in an extremely
• Verification of the alignment of the mounting rotated position should be performed.
device that holds the beam applicator to ensure Ensure the locking mechanism on the unit is
that it is centered about the x-ray source. functional.
• Testing filter interlocks to verify that the selected beam profile, particularly for the largest fields, giv-
filter and kV combination can produce the correct ing rise to the “heel effect” produced by the elec-
x-ray beam and that if an inappropriate filter is tron interaction onto the angled target.
inserted, x-ray will not be produced.
• Beam quality is usually determined by HVL mea-
• Check the integrity of each treatment applicator surements using a setup consistent with the recom-
for cracks or dents. Measure lengths and diameter mendations in the AAPM TG61 code of practice
of each treatment applicator to ensure that it meets (see Section 23.3).
specifications.
Radiation dosimetry Tests
Radiation tests
• Focal spot size and position should be determined
• Light field versus x-ray field: the congruence to ensure that the treatment distances indicated are
between light field and x-ray field should be correct and can be assessed using a pinhole tech-
checked for a number of field sizes at the standard nique to ensure that the treatment applicators are
SSD. The comparison should be performed at the well aligned with the focal spot of the x-ray tube.
FWHM of the x-ray field and the edge of the light • Timer error should be investigated if the x-ray unit
field. The two edges should agree to within ±3 does not have a monitor chamber. The accuracy
mm for field sizes greater than 10 cm and should and linearity of the timer should be evaluated. The
agree to better than ±2 mm for smaller field sizes. system should also have a backup timer, in case
The accuracy of the light field for small field sizes the primary timer fails. Calculation of the timer
should be carefully evaluated before clinical use. error is important due to existence of the “ramp
• X-ray leakage: all sources of leakage should be up region” in kilovoltage x-ray beams machines.
assessed (head or tube leakage, leakage transmit- During the ramp-up, the dose rises from zero to
ted through the applicator and the applicator sup- the steady state dose rate over a period of a few sec-
porting device) and should meet both regulations onds. If the steady state dose rate is used for the
and specifications. The leakage should be deter- calculation of dose, it can lead to a dose deficit.
mined using both film and ionization chamber • Dose output linearity as a function of monitor units
measurements and should be conducted with the (MUs) or with treatment time (if using timer mode)
highest beam kV and mA available. should be determined. The stability of the kV and
the mA over operating time can be checked with a
X‑ray beam performance checks series of short and long radiation exposure times.
• Dose output reproducibility should be checked by
• Field flatness should be evaluated by measurement
performing a number of dose measurements with
of beam profiles in water for the largest treatment
the same monitor units (or time).
field available of each beam to be used clinically.
Profiles of the beam should be measured in a scan- • Rotation of the x-ray tube should not produce any
ning water tank both perpendicular to and in variation in the exposure rate, beam output, field
the cathode–anode axis. If it is assumed that the flatness or symmetry. An ionization chamber may
beam energy does not change across the field, then be attached to the applicator, parallel to the anode–
radiochromic film in a solid phantom may be used cathode axis, and measurements then be per-
for flatness measurements. For kilovoltage x-ray formed for a variety of tube angles. Alternatively,
beams, the uniformity of the field is usually ±5% film can be attached to the applicator for monitor-
over the field area. ing beam uniformity at different tube angles.
• Field symmetry may be evaluated using the same • Short/long term reproducibility of the dose rate
beam profile measurement as for flatness. However, for the clinical treatment techniques should be
differences may be seen in the cathode–anode monitored.
23.4.2 Clinical Commissioning Planning data

Clinical commissioning involves the collection of
dosimetry data that is used for beam characterization. • Relative output factors for standard lead cutouts
This is required for both treatment planning and refer- • Transmission measurements for lead thickness
ence dosimetry. Relative dosimetry measurements are
also performed for beam characterization and use in Backscatter factor verification
treatment planning data. Reference dosimetry is per-
formed for each of the x-ray beams that will be made • Check the Back Scatter Factors (BSFs) via direct
available on the treatment unit. Finally the QA baselines measurement or comparison with measured out-
are established and benchmarks are set which will be put factors
used to monitor the performance of the x-ray unit by
routine QA tests. Reference dosimetry
It is recommended that the user of a kilovoltage
• Perform reference dosimetry measurements for
x-ray unit work with the vendor to establish a compre-
each of the clinical x-ray beams
hensive commissioning protocol based on the speci-
fications of the x-ray unit and its intended clinical
Independent audit
applications. The following commissioning tests are
examples that were recommended by the ACPSEM • Independent audit of the reference dosimetry of
kilovoltage dosimetry working group [28], which can each clinical x-ray beam as well as a review of the
be included in the commissioning protocol together relative dosimetry data
with other commissioning tests recommended by the
vendor.
HVL measurements 23.4.3 Routine Quality Assurance
A QA program is required for the safe and accurate
• Measurement of the first and second HVL for delivery of radiation therapy treatments using kilo-
each of the radiation beams using narrow beam voltage x-rays. The safe delivery of radiation therapy is
geometry ensured by machine interlocks and strategically placed
emergency-off buttons. Accurate delivery of radiation
treatment is ensured by maintaining the mechanical
Relative dosimetry for each x-ray beam and treatment
accuracy within the specification of treatment unit,
applicator combination
maintaining the beam quality in its original condition,
and maintaining the accuracy of dosimetry calculation
• Relative output factor and applicator factor
and measurements for treatment planning and plan
• Percentage depth dose curve verification. All dosimetry data and mechanical limits
are established during the initial machine commission-
• Lateral profiles in both axes perpendicular
ing period before its clinical use and thereafter annu-
to beam direction for at least the largest
ally, or after any change which may significantly alter
applicator
the dosimetry data and the mechanical limits. Once the
• Isodose measurements as required for treatment baseline is established, maintaining the baseline data
planning becomes the mission of the dosimetry QA program.
The test frequency for a kilovoltage x ray therapy
Validity of inverse square law (ISL) with increasing system is determined by the significance of the tests
SSD that indicate beam dosimetry changes and mechanical
tolerance changes within the limited amount of time.
• Assess the validity of the ISL correction for The number of patients that are treated using kilovolt-
increasing distances of stand-off from the end of age x-ray beams are much less than for megavoltage
each applicator; this should be performed for every radiotherapy. Therefore, the check of the beam dosim-
x-ray beam energy etry should depend on the frequency of the beam usage.
Documentation of weekly and monthly spot check mea- imaging, radiation therapy and radiobiology applica-
surements is maintained for 2 years. Documentation for tions at time of writing of this chapter.
each full yearly calibration is maintained for 5 years after The TG61 report recommends two measurement
the completion of calibration. The TG61 dosimetry QA methods based on ionization chambers calibrated
items and their frequencies [23] are summarized below: free-in-air in terms of air-kerma. If the point of inter-
Daily Checks: est is clinically at or close to the surface, the code rec-
ommends one unified approach over the entire energy
• Beam output constancy for energy and filter com- range to determine absorbed dose to water at the surface
binations in use of a water phantom based on in-air measurements (the
“in-air” method). If the point of interest is at depth, it is
• Functionality of the audio-visual monitor
recommended to perform an in-phantom ion chamber
• Door and energy interlock circuits and emergency measurement at 2 cm for kVp values ≥100 kV (the “in-
stops phantom” method) whereas the in-air method is recom-
mended for kVp ≤ 100 kV. The TG61 report provides a
Monthly Checks: comprehensive data set and produces consistent dose
values regardless of the method used and presents esti-
• Items included in the daily checks, and mates of uncertainties on the final dose values. Whereas
the code provides guidelines for dose measurements
• Beam flatness and symmetry
using water, it is supplemented with data to calculate
• Timer operation dose to other materials of clinical interest.
This chapter does not deal with very low energy
• Light-radiation congruence
x-rays (e.g., 50 kV or lower) such as those used in sur-
face or contact therapy or interstitial radiosurgery
Annual Checks:
[6, 8, 44–46]. For x-ray beams with accelerating poten-
tials less than 40 kV, the AAPM TG61 protocol may be
• Items included in the monthly checks, and
used provided the required data values are present in the
• Dose rate for all energy and filter combinations associated tables. For values of beam potential down to
10 kV, the in-air method may be used provided the med-
• Output factors for each of the applicator (cone)
ical physicist can locate appropriate correction factors
• Timer accuracy (verification of the timer error) and associated uncertainties in the literature. It should be
noted that for x-ray beams with energies less than 40 kV,
• Accuracy of the light localizer system
the attenuation of the x-rays in air becomes significant.
• Accuracy of distance measuring devices Further dosimetric studies on detector response and
dose conversion factors are needed to establish reliable
• Beam quality
and practical reference and relative dosimetry methods
• Accuracy of depth dose data and isodose charts and procedures for the widespread clinical application
of these techniques at very low x-ray energies.
• Accuracy of field size dependence data
• Agreement of dose rate with distance from target REFERENCES
• Attenuation in lead for patient block thickness 1. O. Glasser (ed.), The Science of Radiology (Bailliere,
London, 1933).
2. R. Paterson, The Treatment of Malignant Disease by
23.5 SUMMARY Radium and X-Rays (Williams and Wilkins, Baltimore,
This chapter provides a brief historical review of the 1949).
kilovoltage x-ray dosimetry and describes the AAPM 3. E. H. Grubbe, “Priority in the therapeutic use of
X-rays,” Radiology 21, 156–162 (1933). https://doi.
TG61 code of practice for low- and medium-energy x-rays org/10.1148/21.2.156.
(40 kV ≤ kVp ≤ 300 kV). TG61 still represents the 4. P. Biggs, C.-M. Ma, K. Doppke, A. Niroomand-Rad, J.
best kilovoltage x-ray dosimetry protocol for clinical Beatty, “Chapter 9: Kilovoltage x-rays,” in The Modern
Technology of Radiation Therapy, edited by J. Van Dyk 20. S. C. Klevenhagen, R. J. Aukett, R. M. Harrison, C.
(Medical Physics Publishing, Madison, WI, 1999). Moretti, A. E. Nahum, K. E. Rosser, “The IPEMB code
5. C.-M. Ma and J. P. Seuntjens (eds.), Kilovoltage X-ray of practice for the determination of absorbed dose for
Beam Dosimetry for Radiotherapy (Medical Physics x-rays below 300 kV generating potential (0.035 mm
Publishing, Madison, WI, 1999). Al-4mm Cu HVL; 10–300 kV generating potential),”
6. C.-M. Ma, “X-ray therapy equipment, low and medium Physics in Medicine and Biology 41, 2605–2625 (1996).
energy,” in Encyclopedia of Medical Devices and 21. Netherlands Commission on Radiation Dosimetry (NCS),
Instrumentation (John Wiley & Sons, Inc., New York, Dosimetry of Low and Medium Energy X-rays, a Code of
NY, 2006). Practice for Use in Radiotherapy and Radiobiology, NCS
7. C.-M. Ma and K. Paskalev, “In-room CT techniques for Report 10 (NCS, Delft, The Netherlands, 1997).
image-guided radiation therapy,” Medical Dosimetry 31, 22. International Atomic Energy Agency (IAEA), Absorbed
30–39 (2006). Dose Determination in External Beam Radiotherapy: An
8. R. Hill, B. Healy, L. Holloway, Z. Kuncic, D. Thwaites, C. International Code of Practice for Dosimetry Based on
Baldock, “Advances in kilovoltage X-ray beam dosime- Standards of Absorbed Dose to Water, ISSN 1011-4289
try,” Physics in Medicine and Biology 59(6), R183 (2014). (IAEA, Vienna, 2000).
9. C. McKerracher and D. I. Thwaites, “Calibration of 23. C.-M. Ma, C. Liu, C. Coffey, R. Nath, S. Seltzer, J.
an x-ray cabinet unit for radiobiology use,” Physics in Seuntjens, “AAPM protocol for 40–300 kV x-ray beam
Medicine and Biology 51, 3315 (2006). dosimetry for radiotherapy and radiobiology,” Medical
10. F. Verhaegen, P. Granton, E. Tryggestad, “Small animal Physics 28, 868–893 (2001).
radiotherapy research platforms,” Physics in Medicine 24. C.-M. Ma, X. A. Li, J. Seuntjens, “Consistency study on
and Biology 56, R55–R83 (2011). kilovoltage x-ray beam dosimetry for radiotherapy,”
11. R. Pidikiti, S. Stojadinovic, M. Speiser, K. H. Song, F. Medical Physics 25, 2376–2384 (1998).
Hager, D. Saha, T. D. Solberg, “Dosimetric characteriza- 25. L. Dimaso, B. Palmer, J. Micka, et al., “Monte Carlo
tion of an image-guided stereotactic small animal irra- and 60Co-based kilovoltage x-ray dosimetry methods,”
diator,” Physics in Medicine and Biology 56, 2585–2599 Medical Physics 45, 5564–5576 (2018).
(2011). 26. P. Almond, P. J. Biggs, B. M. Coursey, et al., “AAPM’s
12. L. S. Taylor, Radiology 18(99), 627 (1932). TG-51 protocol for clinical reference dosimetry of high-
13. H. O. Wyckoff, G. H. Aston, E. E. Smith, “A comparison energy photon and electron beams,” Medical Physics 26,
of x-ray standards,” British Institute of Radiology 27, 325 1847–1870 (1999).
(1954). 27. L. Buermann, A. S. Guerra, M. Pimpinella, et al., “First
14. G. Ibbott, C.-M. Ma, D. W. Rogers, S. M. Seltzer, J. F. international comparison of primary absorbed dose to
Williamson, “Anniversary paper: Fifty years of AAPM water standards in the medium energy X-ray range,”
involvement in radiation dosimetry,” Medical Physics Metrologia 53, 06007 (2016).
35, 1418–1427 (2008). 28. R. Hill, Healy, D. Butler, D. Odgers, S. Gill, J. Lye, T.
15. C.-M. Ma, C. W. Coffey, L. A. Dewerd, C. Liu, R. Nath, Gorjiara, D. Pope, B. Hill, “Australasian recommenda-
S. M. Seltzer, J. P. Seuntjens, “Status of kilovoltage x-ray tions for quality assurance in kilovoltage radiation therapy
beam dosimetry in radiotherapy,” in Kilovoltage X-ray from the Kilovoltage Dosimetry Working Group of the
Beam Dosimetry for Radiotherapy and Radiobiology, Australasian College of Physical Scientists and Engineers in
edited by C.-M. Ma and J. P. Seuntjens (Medical Physics Medicine,” Australasian Physical & Engineering Sciences
Publishing, Madison, WI, 1999). in Medicine 41, 781–808 (2018).
16. NCRP, NCRP Report 69, Dosimetry of X-ray and gamma 29. C.-M. Ma and A. E. Nahum, “Bragg–Gray theory and
ray beams for Radiation Therapy in the Energy Range ion chamber dosimetry in photon beams,” Physics in
10 keV to 50 MeV (NCRP Publications, Washington, Medicine and Biology 36, 413–428 (1991).
DC, 1981). 30. A. E. Nahum and R. T. Knight, “Consistent formal-
17. International Commission on Radiation Units and ism for kV x-ray dosimetry,” in Proceedings of the IAEA
Measurements (ICRU) Report 23, Radiation Dosimetry: International Symposium on Measurement Assurance in
Measurement of Absorbed Dose in a Phantom Irradiated Dosimetry (IAEA, Vienna, Austria, 1994), pp. 451–459.
by a Single Beam of X- or Gamma Rays (International 31. C.-M. Ma and A. E. Nahum, “Monte Carlo calcu-
Commission on Radiation Units and Measurements, lated stem effect corrections for NE2561 and NE2571
Bethesda, MD, 1973). chambers in medium-energy x-ray beams,” Physics in
18. M. Cohen, D. E. A. Jones, D. Greene (Eds.), “Central Medicine and Biology 40, 63–72 (1995).
axis depth dose data for use in radiotherapy,” British 32. C.-M. Ma and A. E. Nahum, “Calculations of ion cham-
Journal of Radiology Supplement 11 (1972). ber displacement effect corrections for medium-energy
19. International Atomic Energy Agency (IAEA), Absorbed x-ray dosimetry,” Physics in Medicine and Biology 40,
Dose Determination in Photon and Electron Beams; An 45–62 (1995).
International Code of Practice, Technical Report Series 33. J. P. Seuntjens and F. Verhaegen, “Dependence of overall
(IAEA, Vienna, Austria, 1987), vol. 277. correction factor of a cylindrical ionization chamber on
field size and depth in medium energy X-ray beams,” 42. R. G. Kurup and G. P. Glasgow, “Dosimetry of a kilovolt-
Medical Physics 23, 1789–1796 (1996). age radiotherapy x-ray machine,” Medical Dosimetry 18,
34. J. P. Seuntjens, H. Thierens, A. Van der Plaetsen, O. 179–86 (1993).
Segaert, “Determination of absorbed dose to water with 43. International Commission on Radiation Units and
ionisation chambers calibrated in free air for medium Measurements (ICRU) Report 24, Determination of
energy X-rays,” Physics in Medicine and Biology 33, Absorbed Dose in a Patient Irradiated by Beams of X or
1171–1185 (1988). Gamma Rays in Radiotherapy Procedures (International
35. J. P. Seuntjens, H. Thierens, U. Schneider, “Correction Commission on Radiation Units and Measurements,
factors for a cylindrical chamber used in medium Bethesda, MD, 1976).
energy x-ray beams,” Physics in Medicine and Biology 44. M. Dinsmore, K. J. Harte, A. P. Sliski, D. O. Smith, P.
38, 805–832 (1993). M. Nomikos, M. J. Dalterio, A. J. Boom, W. F. Leonard,
36. X. L. Li, C.-M. Ma, D. Salhani, “Measurement of per- P. E. Oettinger, J. C. Yanch, “A new miniature x-ray
centage depth dose and lateral beam profile for kilo- source for interstitial radiosurgery: Device description,”
voltage x-ray therapy beams.” Physics in Medicine and Medical Physics 23, 45–52 (1996).
Biology 42, 2561–2568 (1997). 45. Y. Rong and J. S. Welsh, “Surface applicator calibration
37. British Institute of Radiology (BIR), “Central axis depth and commissioning of an electronic brachytherapy sys-
dose data for use in radiotherapy,” British Journal of tem for nonmelanoma skin cancer treatment,” Medical
Radiology Supplement 25 (1996). Physics 37, 5509–5517 (2010).
38. E. B. Podgorsak, M. Gosselin, M. D. C. Evans, “Superficial 46. J. P. Gerard, A. S. Myint, O. Croce, J. Lindegaard, A.
and orthovotage x-ray beam dosimetry,” Medical Physics Jensen, R. Myerson, J. M. Hannoun-Levi, S. Marcie,
25, 1206–1211 (1998). “Renaissance of contact x-ray therapy for treating rec-
39. L. Gerig, M. Soubra, D. Salhani, “Beam characteristics tal cancer,” Expert Review of Medical Devices 8, 483–92
of the Therapax DXT-300 orthovoltage therapy unit,” (2011)
Physics in Medicine and Biology 39, 1377–1392 (1994). 47. C.-M. Ma and J. P. Seuntjens, “Correction factors for
40. M. J. Butson, J. Mathur, P. Metcalfe, “Dose characteristics water-proofing sleeves in kilovoltage x-ray beams,”
of a new 300 kVp orthovoltage machine,” Australasian Medical Physics 24, 1507–1513 (1997).
Physical & Engineering Sciences in Medicine 18, 133–138 48. C.-M. Ma and J. P. Seuntjens, “Mass energy-absorption
(1995). coefficient and backscatter factor ratios for kilovoltage
41. R. J. Aukett, D. W. Thomas, A. W. Seaby, J. T. Gittins, x-ray beams,” Physics in Medicine and Biology 44, 131–
“Performance characteristics of the Pantak DXT-300 143 (1999).
kilovoltage X-ray treatment machine,” British Journal of
Radiology 69, 726–734 (1996).
Chapter 24
Electron Dosimetry
John A. Antolak
Mayo Clinic
Rochester, Minnesota
CONTENTS
24.2 Properties of Electron Beams 375
24.3 Water-Phantom Dosimetry 376
24.3.1 Reference Dosimetry 376
24.3.2 Reference Dosimetry Verification 378
24.3.3 Reference Dosimetry for Non-Standard Electron Modalities 378
24.3.4 Commissioning and QA Measurements 380
24.3.5 Detector Choice 380
24.3.6 Percent Depth Dose 381
24.3.7 Lateral Profiles 384
24.3.8 Output Factors 385
24.4 Solid Phantom Measurements 387
24.4.1 General Considerations 387
24.4.2 Quality Assurance Measurements 387
24.4.3 Solid Phantom Film Measurements 387
24.4.4 Other Solid Phantom Measurements 388
24.5 In Vivo Dosimetry 388
24.6 Summary 388
Notes 388
References 388
24.1 INTRODUCTION There are many very good references that are broadly
The dose deposited by an x-ray beam is primarily due to available, and nothing in this chapter is meant to replace
secondary charged particles (mostly electrons), which the information in those references. The author assumes
implies that any detector that is suitable for measuring that the reader has access to the relevant reference materi-
dose in an x-ray beam is generally suitable for measuring als, avoiding the possibility that transcription errors are
dose in an electron beam. The goals of this chapter are to introduced that could lead the reader to make mistakes
provide the reader with (1) a basic understanding of the having downstream clinical effects.
physics of electron beams that can affect electron dosime-
try, (2) recommendations for water-phantom dosimetry of 24.2 PROPERTIES OF ELECTRON BEAMS
electron beams, (3) guidance for solid phantom measure- Electrons are fundamental particles with a small mass,
ments using electron beams, and (4) an overview of the approximately 2000 times lighter than a nucleon (pro-
use of in vivo dosimeters for electron beam radiotherapy. ton or neutron). Because they are also charged particles,
375
they readily interact with all of the charged particles a problem of calculating the fluence distribution of the
that make up normal matter. If the electron comes close electrons in the material.
enough to the nucleus of an atom, the electric field of the This is a very simplified description of the electron
nucleus can deflect the trajectory of the electron, with the beam transport, but it is good enough to be able to
amount of deflection depending on the strength of the qualitatively understand some of the basic properties
interaction or how close the electron gets to the nucleus. of electron beams used in radiation therapy and how
This interaction is called Coulomb scattering. The num- dosimetry of the electron beams might be influenced.
ber of scattering interactions is approximately 750 per
mm of water for >5-MeV electrons [1] and it is not 24.3 WATER-PHANTOM DOSIMETRY
practical to follow each interaction on a microscopic The primary purpose of this section is to discuss consid-
scale. On a macroscopic scale, the deflection of the erations for measurements commonly carried out dur-
electron will be the end result of those many interac- ing commissioning or annual quality assurance (QA),
tions and this is called multiple Coulomb scattering. primarily using water phantoms with automated scan-
The scattering power of a material is a measure of ning systems. The American Association of Physicists
how much a given material changes the angular dis- in Medicine (AAPM) Report 106 (TG-106 [2]) discusses
tribution of the electrons passing through the mate- beam commissioning in detail for both photons and
rial. In some cases, the electron comes close enough electrons, including the use of water phantom systems
to the nucleus that it interacts more strongly with and AAPM Report 99 (TG-70 [3, 4]) reiterates those
the nucleus and in addition to changing direction, recommendation in the context of performing the
a bremsstrahlung x-ray is emitted as a result of the measurements for electron beams. Most of the recom-
interaction. This is the basic mechanism for produc- mendations that follow are from those reports, but also
ing x-rays in the anode of an x-ray tube or the target includes some preferences of the author based on prior
of a linear accelerator. These bremsstrahlung events experience.
remove energy from the electron beam, but this
energy is not deposited locally and hence contributes 24.3.1 Reference Dosimetry
very little to the deposited dose. Electron beams must be calibrated before they are used
The most common interaction is with orbital elec- clinically. Calibration is the process by which you estab-
trons. Even if the incident electron doesn’t come very lish the relationship between the response of the accel-
close to the orbital electron, the interaction can lead to erator’s monitor chamber and the delivered dose under
a small amount of energy loss and scattering because a specified calibration condition. The most common
the masses of the incident and orbital electrons are the current calibration protocols are TG-51 (AAPM Report
same. This is also where most of the energy loss occurs, 67 [5, 6]) in the United States and Canada and TRS398
leading to heating of the matter and ionization of the elsewhere [7]. Tailor and Hanson [8] showed that the
atoms, the latter of which can eventually lead to bio- newer calibration protocols result in 1–2% higher dose
logical effects. The mass collisional stopping power is in electron beams, depending on the beam quality and
a measure of the energy lost to collisional interactions ionization chamber used for calibration. The funda-
per unit length divided by the density of the material mental changes in the latest protocols were that the ion
and the deposited dose is the product of the electron chamber calibration factor is dose to water for a specified
fluence (number of electrons per unit area) and the beam quality (60Co) and all calibrations must be done
mass collisional stopping power. The mass collisional in a water phantom. Acceptable dosimeters for these
stopping power is relatively constant for biological protocols include cylindrical ionization chambers and
materials. Therefore, the energy deposition is deter- well-guarded plane-parallel ionization chambers. The
mined primarily by the total electron track length in dosimeter and electrometer should be calibrated by an
a given volume, which is equivalent to the electron Accredited Dosimetry Calibration Laboratory (ADCL)
fluence. The electrons in the beam also redistribute in the United States, or equivalent in other countries.
themselves depending on the scattering power of the The calibration of the ionization chamber should gener-
materials that the beam passes through and calculating ally be done every two years or more often if required by
the dose deposition due to electron beams is primarily regulations. While calibration laboratories will provide
Electron Dosimetry ◾ 377
a calibration factor for plane-parallel ionization cham- 2 ≤ I50 ≤ 10 cm [11, 12]. The advantage of the new beam
bers, a process for cross-calibrating such ionization quality specifier is that dref (as defined above) is very close
chambers is also described in the reports [5, 7]. The to R100 for lower energies and is monotonically increas-
basic process is to perform a calibration measurement ing as a function of energy, which is not true of R100. In
in a high-energy electron beam using both ionization the previous TG-21 calibration protocol, the measure-
chambers and solve for the unknown calibration factor ment depth for electron beam calibration was at R100 or
by assuming that the dose at the reference depth is the dmax [13], which can be quite variable at higher energies
same for both measurements. (depending on the design of the beamline).
The following general description of the calibration When determining the beam quality using a cylin-
process is based on TG-51, but also generally applies drical ionization chamber, the effective point of mea-
to TRS398 since the protocols are very similar and surement must be used. For electron beams, the effective
based on the same underlying research. With respect point of measurement is located upstream of the center of
to electron-beam calibration, the primary difference the ionization chamber by a distance of 0.5 rcav , where rcav
between the protocols is that while TRS398 requires is the radius of the ionization chamber [14, 15]. The soft-
the use of plane-parallel ionization chambers below ware for computer-controlled water phantom systems
10 MeV (R50 ≤ 4.3 cm ), TG-51 only recommends their allows users to specify the coordinate system relative to
use for those energies and requires plane-parallel the effective point of measurement. However, if you do
ionization chambers for energies less than 6 MeV so, you must be careful to reset the coordinate system
(R50 ≤ 2.6 cm ). The reader should refer to the original prior to calibration. Failing to do this would mean that
protocol documents to ensure their process is com- the point of measurement would be deeper than dref and
pliant with the protocol and an independent check the calibration would be incorrect. Therefore, TG-51
of the calibration, such as having another person do recommends that measurement of the beam quality be
an independent measurement with different instru- done using center of ionization chamber coordinates.
mentation, is good practice. In addition, it would be The entire curve can then be shifted upstream by 0.5 rcav
useful to review the article describing common issues to get the depth ionization curve. Equations (24.2) and
found by the Radiological Physics Center (now IROC) (24.1) can then be used to determine R50 and dref .
as they were helping many centers with their conver- The TG-51 protocol specifies that the calibration field
sion to the TG-51 protocol [9]. size must be greater than or equal to 10 × 10 cm 2 for
The TG-51 protocol for electron beam calibration R50 ≤ 8.5 cm , or 20 × 20 cm 2 for R50 > 8.5 cm . The source
specifies the reference depth, dref , for an electron beam as to surface distance (SSD) for the calibration may be any
SSD from 90 to 110 cm. Because most typical accelerators
dref = 0.6 R50 − 0.1 ( cm ) (24.1) use an applicator system with final collimation at 90–95 cm
from the nominal source position and have a maximum
where R50, the beam quality specifier, is the depth at energy of 20 MeV or less, the most commonly used cali-
which the dose drops to 50% of the maximum central- bration standard condition is a 10 × 10 cm 2 field at an
axis dose, D ( dmax ) [10]. The reference depth is where the SSD of 100 cm, although 15 × 15 cm 2 is also quite com-
ion chamber’s point of measurement must be placed mon. The measurement phantom for calibration must
when doing the calibration measurement. The point of be liquid water and measure at least 30 × 30 × 30 cm3.
measurement is the center of a cylindrical ion chamber, If the calibration beam is directed through a solid wall
or on the inside of the front electrode of a plane-parallel that is greater than 0.2-cm in thickness, all depths need
ion chamber. If you are determining the beam quality to be scaled to water-equivalent depths.
by measuring ionization instead of dose, the beam qual- Assuming that dref has been appropriately deter-
ity parameter can be determined using mined, the point of measurement of the ionization
chamber is placed at dref and the ionization is measured
R50 = 1.029 I50 − 0.06 ( cm ) (24.2) for a set number of monitor units. The fully-corrected
ionization, M, is calculated using
where I50 is the depth at which the ionization drops
to 50% of the maximum central-axis ionization and M = Pion PTP Pelec Ppol M raw (24.3)
where M raw is the measured ionization, Pion corrects for good practice would dictate that a verification (second
incomplete collection of ionization, PTP corrects the check) of this crucial step would be prudent and there are
reading to standard temperature (22°C) and pressure several ways to accomplish this task. In a larger depart-
(1 atm), Pelec accounts for any innacuracies in the elec- ment, there are likely multiple “calibrated” ionization
trometer reading, and Ppol corrects the reading for polar- chambers and electrometers, so this could be as simple
ity effects. The dose in water for an electron beam of as having a colleague repeat the setup and measurement
specific quality, DwQ, is then given by the following equa- using a different ionization chamber and electrometer.
tion from the TG-51 report [5]. In a smaller department with just a single calibration
system, you could have a second person repeat the mea-
DwQ = M PgrQ kR′ 50 kecal N D60,wCo (24.4) surement with the same equipment independently; how-
ever, if there is a problem with the dosimetry equipment,
From left to right, M is the fully-corrected ionization this process will not catch that problem. It is possible to
reading, PgrQ is a gradient correction to account for the hire a medical physicist specifically to verify the calibra-
difference between the dose at dref and the dose at the tion using their own equipment, or a dosimetry service
ion chamber’s effective point of measurement, kR′ 50 is a (e.g., IROC Houston Quality Assurance Center1) can be
factor that converts the calibration factor for a standard used to help verify the calibration [16, 17]. The latter is
electron beam to the calibration factor for the current not as accurate but can catch gross errors in calibration.
electron beam quality, kecal is a factor that converts the Depending on any other dosimeters that are available,
calibration factor in a 60Co beam to the calibration fac- it is also possible to cross-calibrate another dosimeter
tor for a standard electron beam, and N D60,wCo is the cali- and then use that to verify the calibrations. An example
bration factor (dose to water) in a 60Co beam. The kecal of this might be to use an energy-independent dosim-
factor depends on the ionization chamber chosen for eter (e.g., p-type electron diode) and cross-calibrate this
the calibration and was introduced to prepare for a dosimeter for one beam energy and then check all of the
future where primary standards for electron beams may other energies using that calibration factor. The cross-
be available. The dose gradient at the reference depth calibration could also be done on another machine, if
depends on beamline details for electron beams and is available. In my experience, medical physicists often
not just a function of the beam quality; hence PgrQ must check the electron beam calibration using multiple
be explicitly included in the dose equation. methods.
To summarize, the general process for calibrating
an electron beam is to first determine the beam quality 24.3.3 Reference Dosimetry for Non-Standard
from I50 or R50, then move the point of measurement of Electron Modalities
the ionization chamber to the dref depth and measure For a normal linear accelerator, the calibration proto-
the ionization, then measure the ionization correction col, as described above is relatively straightforward and
factors to get the fully-corrected ionization reading, and easy to apply. There are a couple of scenarios where it
then move the ionization chamber deeper by 0.5 rcav to is not so obvious that the protocols can be applied; the
determine the gradient correction factor. The dose to first is intraoperative electron beam therapy (IOEBT or
water at the reference depth, dref , can then be calculated IOERT or IORT) and the second is total skin electron
using Equation (24.4). For the vast majority of institu- therapy (TSET or TSEI). For the former, the applica-
tions that prefer to calibrate a dose rate at the R100 depth, tor or cone system is quite different from the standard
the measured dose to water should then be divided by linear accelerator and for the latter, there is usually no
the clinical percent depth dose at the reference depth to applicator and the distance to the treatment position is
get the dose to water at R100. much greater than standard calibration protocols allow.
IORT can be carried out using a standard medical
24.3.2 Reference Dosimetry Verification linear accelerator installed in the operating room [18],
The initial calibration is arguably the most important and to increase the safety of the procedure, the linear
step in commissioning a device for radiation treatments accelerator is often modified to disable photon (or x-ray)
and the electron modality should be handled in a simi- mode. The standard applicator system is modified to
lar manner to any other treatment modalities. As such, meet the needs of the IORT procedure, but standard
electron applicators are not precluded for quality assur- incident on the wall of the water phantom. However, by
ance purposes. Therefore, the calibration of each electron the time the beam has exited the linear accelerator and
beam energy can be done using the standard calibration made its way to the patient plane, the energy is very low
protocols as described above. Each IORT cone is then and the reference depth for calibration will be within
characterized by an output factor that relates the dose the wall unless you have a thin window in the phan-
output with the IORT cone to the standard electron cali- tom wall. Therefore, if calibration is done at the patient
bration geometry. plane, solid phantoms are preferable. Since the current
Because of the disadvantages of installing a linear standard dose calibration protocols do not allow solid
accelerator in an operating room, a more common phantoms and very large distances, any dose measure-
approach to IORT is to use a mobile linear accelera- ment at the patient plane should really be considered an
tor. These devices have only been available for about output verification and not a dose calibration.
the last 20 years and have their own challenges. The It is possible to measure the dose rate using the TSEI
AAPM Report 92 (TG-72 [19]) provides a nice sum- mode at a standard SSD (e.g., 100 cm) in a water phan-
mary of the considerations involved in using such a sys- tom, which would be compliant with the calibration
tem and Wootton et al. [20] describes commissioning protocols. The energy of the beam at 100-cm SSD will
and use of a newer model device. It recommends that be higher than at the patient plane is likely to be high
the TG-51 calibration protocol be used for calibration enough that a cylindrical ionization chamber would be
of the beams with a 10-cm diameter circular applicator. allowed for the TG-51 calibration protocol. However, the
These devices do not have normal electron applicators, higher dose rate used for the TSEI mode generally leads
but Monte Carlo calculations of the in-phantom water to more ion recombination, similar to the case for IORT.
to air stopping power ratio for these devices shows that If Pion is greater than 1.05, a different ionization cham-
they are very close to those recommended in the cali- ber should be used to be compliant with the calibra-
bration protocols, as might be expected [21]. One sig- tion protocols. AAPM Report 92 [19] gives an example
nificant issue for calibrating these devices comes from where a cylindrical ionization chamber was not suitable
their higher dose per pulse (relative to normal medical at a dose rate of 10 Gy min −1 for IORT, but the Pion for
linear accelerators), which can cause the ion recombina- an advanced Markus (plane-parallel) ionization cham-
tion correction, Pion, to be more significant than normal ber was no greater than 1.03 and was therefore suitable
medical linear accelerator electron beams. Depending for dose calibration. The dose rate for the TrueBeam™
on the ionization chamber chosen for the measurement, (Varian Medical Systems, Palo Alto, CA) TSEI mode is
Pion may be larger than 1.05, and AAPM Report 92 rec- 25 Gy min −1. Since plane-parallel ionization chambers
ommends switching to a different ionization chamber are also recommended for lower electron energies, using
if this is the case [19]. Moretti et al. [22] described a such a chamber would be recommended for the TSEI
method for calibrating a mobile linear accelerator using dose calibration.
radiochromic film, which has little dependence on the To determine the relationship between the dose out-
dose per pulse. However, most physicists would likely be put at 100-cm SSD and at the treatment plane, it is not
more comfortable using this technique as a verification nearly as straightforward as it is for normal electron
of the in-water calibration according to current calibra- beams or IORT. The energy of the beam at the treatment
tion protocols [23]. plane for TSEI is much lower than at 100-cm SSD, and
TSEI represents a different set of problems for ref- this change in energy needs to be considered. Using a
erence dosimetry. For normal electron beams, the dosimeter with less energy and dose rate dependence,
calibration condition is not very far removed from such as radiochromic film or a semiconductor diode, can
the irradiation condition. For TSEI, the applicator help. For example, one could calibrate at 100-cm SSD
is removed, and the patient is treated very far away. using a plane-parallel ionization chamber, then make
Traditionally, TSEI calibrations are done at a location a measurement of the output using another dosimeter
where the patient is treated [24]. Because of the lower in a Solid Water® (Gammex RMI, Middleton, WI) or
energy of the beams, plane-parallel ionization chambers equivalent phantom. The same phantom and dosimetry
are strongly recommended. The approximately hori- can then be put at the patient plane to determine the
zontal beam geometry implies that the beam has to be dose output at that point. It would also be prudent to
have multiple ways to check the output. For example, planning system. The initial commissioning also estab-
Antolak et al. [25] used thermoluminescent dosimeters lishes a baseline for the overall performance of the beam
placed on the abdomen of TSEI patients and normalized delivery system. In most clinics, the majority of the
the readings to the dose output of a standard 9-MeV commissioning measurements are carried out in a scan-
electron beam (not the TSEI beam). While the spread of ning water phantom and AAPM Report 106 describes in
the readings was significant, which was indicative of the detail the best practice for carrying out these measure-
patient setup uncertainty, the average over all patients ments [2]. On an annual basis, the current performance
was within 1% of the desired dose. level of the linear accelerator is checked using the same
The polarity effect is also significant when using scanning water phantom. If required, beam parameters
smaller ionization chambers at larger distances. Das are adjusted to bring the performance back to the base-
et al. [26] reported on measurements of bremsstrahlung line. QA is also done on a daily and monthly basis, as
dose for a TSEI beam and found that the length of cable prescribed by regulations and professionally-defined
in the beam significantly affected the results. Therefore, good practice, for example, AAPM Report 46 (TG-40
they recommended using dosimeters without cables [27]). Daily and monthly QA does not generally require
when making measurements under those conditions. It a large water phantom and methods for doing these
is interesting to note that they did not mention anything dosimetric measurements will be described later.
about investigating the polarity dependence of the read-
ing. When calibrating a TSEI beam using a small ion- 24.3.5 Detector Choice
ization chamber, Antolak and Hogstrom [24] found that The two most common types of detectors used for com-
correcting the raw readings for polarity was necessary. missioning and QA measurements are ionization cham-
Anecdotally, the magnitude of the polarity effect does bers and semiconductor diodes. Ionization chambers are
depend on the length of cable in the beam, but the polar- probably the most commonly used detectors, with cylin-
ity corrected reading is not very strongly dependent on drical chambers being the most popular choice in North
the length of the cable in the beam and steps can be America due to AAPM calibration protocols allow-
taken to minimize this (as recommended by Das et al. ing their use for energies down to 6 MeV [5]. In other
[26]). Therefore, I would not go as far as recommending parts of the world, well-guarded plane-parallel ioniza-
that ionization chambers and other cabled dosimeters tion chambers are preferred for electron beam measure-
be avoided, but that cable lengths within the beams be ments [7, 28]. For cylindrical ionization chambers, the
minimized. A non-cabled dosimeter can be useful for a user needs to be aware that the effective point of mea-
second check, however. surement is located at a point that is 0.5 rcav upstream of
Assuming that the calibration is done at 100-cm SSD, the center of the ionization chamber, while the point of
the fact that the treatment condition is far removed from measurement in the calibration protocols is the center
the calibration condition complicates the quality assur- of the ionization chamber [3, 14, 29]. For plane-parallel
ance of the procedure. The relative beam output and ionization chambers, the effective point of measure-
profile at the patient plane are strongly dependent on ment is on the inside of the front electrode, which is
the energy of the beam [24]; therefore changes in beam the same as the calibration point of measurement. For a
output at the patient plane should not lead to immedi- typical cylindrical ionization chamber used for general
ate adjustment of the linear accelerator without further purpose water phantom scanning, the active volume is
investigation. approximately 0.1 cm3, with a radius of approximately
3 mm and a length of approximately 6 mm. Plane-
24.3.4 Commissioning and QA Measurements parallel ionization chambers used for water phantom
While calibration of the electron beams is arguably the scanning generally have a diameter of 5 mm or greater.
most important aspect of electron beam dosimetry, com- Semiconductor diodes, more specifically p-type, have
missioning of the electron beams is also very important. several advantages for relative electron beam dosimetry.
Commissioning the electron beams prepares them for The sensitive volume of the diode is fairly small with a
clinical use, which in a modern clinic means measuring radius of 1–2.5 mm and a thickness of less than 0.1 mm.
the data required for input into the treatment planning Despite the small volume, the nominal sensitivity (cur-
system and to validate the operation of the treatment rent per unit dose) is generally as large or larger than
typical ionization chambers used for beam scanning. The microscopic nature of the semiconductor and
Similar to plane-parallel ionization chambers, the effec- diode devices is also a potential issue. The measurement
tive point of measurement is very well defined and is the device is encapsulated and not visible to the naked eye.
position of the semiconductor die in the detector. The Examination with x-rays can be useful for confirming
stopping power ratio (silicon to water) is not very energy the location of the active volume but will not tell you
dependent, and it is generally accepted that ionization if the semiconductor is p-type (preferred) or n-type. Over
and dose are interchangeable [3, 29, 30]. However, it is the years, there have been many different variations in
also considered good practice to routinely verify this construction of the devices. For example, there are pho-
by comparing to percent depth dose determined using ton-specific diodes with added filtration to provide bet-
ionization chambers. The response of diode detectors is ter energy response when measuring x-ray beams. These
reduced as a result of being irradiated (< 4%/kGy), so devices are not suitable for electron beam measurements
monitoring the response is warranted for extended mea- and it can be difficult to tell the difference between pho-
surement sessions. The temperature dependence of the ton and electron diodes. Semiconductor diode detec-
response is approximately 0.3–0.4% K −1, which is similar tors are also subject to radiation damage (p-type diodes
to ionization chambers. are better in this regard) and if the diode is being used
Diamond detectors can also be considered for elec- extensively, the perceived risk of the detector not per-
tron beam commissioning and QA. The Institute of forming as expected increases. In the author’s opinion,
Physics and Engineering in Medicine (IPEM) code of many of these concerns are not warranted and it is very
practice and the International Atomic Energy Agency simple to ascertain that a particular diode is perform-
(IAEA) calibration protocols specifically mention their ing well prior to use. It would also be good practice to
use for directly measuring percent depth dose curves, carefully track and label any semiconductor detectors in
similar to semiconductor diodes [29, 31]. However, your possession to ensure that the appropriate detector
AAPM has no specific guidance on the use of these is used.
detectors and the author has no practical experience
with their use for electron beam dosimetry. As with 24.3.6 Percent Depth Dose
semiconductor diode detectors, it is recommended that When measuring percent depth dose in water, PDDw,
before a diamond detector be used for extensive dosi- using a scanning water phantom, it is recommended to
metric measurements, it is thoroughly commissioned, scan the detector starting at depth and moving toward
including comparisons with trusted ionization chamber the surface [2, 3]. If scanning is done in the reverse direc-
measurements [32]. tion, surface tension effects can pull the water surface
Despite the significant advantage of not having to down with the detector for a few millimeters, leading to
convert the ionization signal to dose, semiconductor incorrect data at shallow depths. While it is easy to reduce
diode and diamond detectors are often not trusted for surface tension effects using something like a few drops
commissioning measurements. Ionization chambers are of liquid soap, it is still good practice to scan from depth
essentially mechanical devices with demonstrated long- toward the surface. As mentioned above, semiconduc-
term stability and very little variation from one device tor diodes or diamond detectors can be used to measure
to another (of the same model). Because of this macro- percentage depth dose directly. However, it is also recom-
scopic sameness and the fact that ionization chambers mended that results are compared to ionization chamber
are universally used for calibration, it is very easy to measurements, so it is important to understand how to
be confident in their use for commissioning measure- convert percent ionization to percent dose.
ments. Anecdotally, some physicists can cite instances According to AAPM Report 99 [3], PDDw is calcu-
where semiconductor diodes have led to bad measure- lated from the percent depth ionization in water, PDI w ,
ments and they are therefore hesitant to trust them for using the following equation.
measurements where ionization chambers can be used,
even if they are not ideal. Ionization chambers can also
occasionally fail (e.g., excessive leakage), but perhaps PDDw ( d ) = PDI w ( d ) ×
( L /ρ )air ( R50 , d ) ⋅ Pfl ( Ed )
w
less often and physicists tend to still trust their use for ( L /ρ )wair ( R50 , dmax ) ⋅ Pfl ( Ed )
max
most measurements. (24.5)

The two corrections are for the restricted stopping power resulting curve upstream by 0.5 rcav . The second method
ratio from water to air and the fluence correction. Prior is to adjust the origin of the scanning coordinate system
to the implementation of the TG-51 calibration protocol, so that the center of the ionization chamber is at a depth
AAPM Report 32 (TG-25 [14]) recommended the use of of 0.5 rcav in the water phantom when the scanning depth
the restricted stopping powers from the TG-21 calibra- coordinate reads zero. Either method is acceptable, but
tion protocol [13], which were computed using monoen- the author has a slight preference for shifting the curve
ergetic electron beams. Burns et al. [10] improved upon after measurement. Modern scanning systems have the
this approach by calculating restricted stopping power capability to store user preferences for each measure-
ratios for realistic electron beams and parameterizing ment device that is used, so scanning coordinates can
the restricted stopping power as a function of the depth be adjusted as simply as specifying a certain detector
of 50% dose, R50, and the depth, d. for the measurement. If you choose to have the system
implicitly apply the gradient correction, then you risk
( L /ρ )air ( R50 , d ) = having the point of measurement set incorrectly when
w
calibrating the beam. No matter which method is cho-

1.0752 − 0.50867 ( ln R50 ) + 0.08867 ( ln R50 ) − 0.08402 ( d /R50 )
2
sen, having a standard process for doing this in a repro-
1 − 0.42806 ( ln R50 ) + 0.064627 ( ln R50 ) + 0.003085 ( ln R50 ) − 0.1246 ( d /R50 ) ducible manner is most important.
2 3
(24.6) The fluence correction factor accounts for the change

in the primary electron fluence spectrum caused by the
For the TG-51 calibration protocol, Equation (24.2) is presence of the air cavity. For a well-guarded plane-
used for calculating R50 from the measured value of I50 parallel ionization chamber, the fluence correction fac-
and Rogers [33] showed that using Equation (24.6) for the tor is unity, which essentially means that the restricted
stopping power ratio produces dose estimates that are stopping power ratio is the only correction needed to
accurate at all depths to within 1% of the maximum dose convert ionization to dose. For other plane-parallel
on central axis. This equation is recommended by AAPM ionization chambers, AAPM Report 48 (TG-39 [34])
Report 99 [3] and is also consistent with recommenda- should be used to get the fluence correction factor. For
tions in TRS398 [7]. The TG-51 calibration protocol also a cylindrical ionization chamber, the fluence correc-
mentions that to be entirely consistent, the R50 value tion factor is given in Table I of AAPM Report 99 [3].
determined from the converted ionization curve should Note that AAPM Report 99 and AAPM Report 32 [14]
be compared with the R50 value previously determined both indicate that the table is showing the replacement
using Equation (24.2). If they are significantly different, it correction, even though it is really the fluence correc-
is likely that an error was made along the way. tion. For every ionization chamber diameter listed in
The second correction factor for conversion of ioniza- the table, a second order polynomial fit reproduces the
tion to dose is the fluence correction factor. In AAPM fluence correction factor to within 0.1% and the polyno-
Report 99, the authors say to determine the replacement mial coefficients are listed in Table 24.1. Because Pfl is
correction factor, Prepl, it can be divided into a gradient a slowly varying function of both the ionization cham-
correction factor, Pgr , and a fluence correction factor, Pfl . ber diameter and the mean energy at depth, Ed , and the
In practice, Pgr is accounted for by using the effective polynomial coefficients vary smoothly as a function of
point of measurement of the ionization chamber, which ionization chamber diameter, interpolation of the poly-
is why you don’t really need to determine Prepl. For a nomial coefficients is almost the same as direct inter-
plane-parallel ionization chamber, the effective point of polation in the original table. However, the polynomial
measurement is the same as the point of measurement coefficients are easier to use in a spreadsheet or com-
(inside surface of the front electrode), so Pgr is unity. For puter program, if needed (assuming that the correction
a cylindrical ionization chamber, commonly used in is not already handled by the scanning software). The
North America, the effective point of measurement is mean energy at depth is given by
0.5 rcav upstream of the center of the ionization cham-
ber. There are two practical ways to handle this correc-
 d 
tion. The first method is to measure using the point of Ed = E0  1 −  (24.7)
measurement (center of chamber) and then shift the  Rp 
TABLE 24.1 Polynomial Fit Coefficients for Cylindrical Ionization Chamber Fluence
Correction Factor, Pfl
Chamber Diameter (mm) A B C
3 0.9720 2.351 54.8
5 0.9547 3.841 89.0
6 0.9470 4.301 94.0
7 0.9385 4.957 105
Note: The cylindrical ionization chamber fluence correction factor is approximately given
by Pfl = A + B × 10−3 ⋅ Ed + C × 10−6 ⋅ Ed2 , where Ed is the mean energy at depth d. The
above coefficients reproduce Table I in AAPM Report 99 [3] to within 0.1% for all
values in the table. Since Pfl is a slowly varying function of both energy and depth and
the coefficients also vary slowly as a function of diameter, the coefficients can be
interpolated to get the fluence correction factor for ionization chamber diameters not
included in the table.
where E0 is the mean energy at the surface of the water hence R50 using Equation (24.2) and then all of the cor-
phantom and R p is the practical range of the electron rections can be made as described above.
beam [35]. The mean energy at the surface can be deter- Figure 24.1 shows how these correction factors work
mined using the relationships given in the IPEMB code for a lower energy beam, 9 MeV, and a higher energy
of practice [28] for R50 or I50 (in cm). beam, 20 MeV. For the unshifted percent depth ioniza-
tion curves, you’ll notice that there is a slight upward
E0 = 0.656 + 2.059 R50 + 0.022 R50
2 (24.8) curve near the surface. This is due to the fact that the
ionization chamber is transitioning from being entirely
E0 = 0.818 + 1.935 I50 + 0.040 I50
2 (24.9) submerged to only partially submerged. In this case, the
radius of the ionization chamber is 3 mm, so the data for
While it is recognized that there are more accurate
depths up to 3 mm is affected. Shifting the percent depth
data in the literature, differences between the simple
ionization curve upstream by 1.5 mm makes the curve
approach described above and other approaches are
look a little better, but the first 1.5 mm is still not reliable.
clinically insignificant [3, 36].
The dose near the surface should have a steeper delta-ray
Equation (24.5) implies that the depth dmax is known.
buildup, but using just a cylindrical ionization cham-
However, if we are starting with an unknown beam, we
ber will not be able to resolve this feature. For the lower
will know the depth of maximum ionization, I100 , but
energy (9-MeV) beam, the stopping power and fluence
the depth of maximum dose, dmax or R100, is not neces-
correction factors have a fairly subtle effect, with the stop-
sarily the same. Therefore, what most users will really do
ping power correction reducing the percent dose in the
is calculate the percent depth dose using the following
buildup region (relative to percent depth ionization) and
equation.
the fluence correction factor slightly increasing the dose
in the buildup region. For the higher energy (20-MeV)
PDI w ( d ) ⋅ ( L /ρ )air ( R50 , d ) ⋅ Pfl ( Ed )
w
PDDw ( d ) = beam, you can see similar effects. However, because the
{
max PDI w ( d ) ⋅ ( L /ρ )air ( R50 , d ) ⋅ Pfl ( Ed )
w
} change of energy is larger, the stopping power and flu-
(24.10) ence corrections have a greater effect, especially when it
comes to the perceived therapeutic depth. The depth to
In other words, the PDI (or ionization) values are mul- 90% of maximum ionization is 5.5 cm, while the depth
tiplied by the restricted stopping power and fluence to 90% of maximum dose is approximately 6.0 cm. The
corrections, then the entire curve is normalized to the depth of maximum dose is also about 1 cm deeper than
maximum of the corrected values. For most scanning the depth of maximum ionization. In addition, there is a
systems, the computer analysis software will have these larger difference between the curves with just the stop-
corrections built in. Once the correct operation of the ping power correction and the fully-corrected percent
software has been verified, converting ionization to dose depth dose curve; hence the fluence correction factor is
is quite easy because the software can determine I50 and also more significant for the higher energy.
FIGURE 24.1 Percent depth ionization and percent depth dose for 15 × 15 cm 2 field size (a) 9-MeV, (b) 9-MeV zoomed in,
(c) 20-MeV, and (d) 20-MeV zoomed in. The purple line is the percent depth ionization, the blue line is the same curve shifted
upstream by 0.5r, the red curve includes the stopping power correction, and the thicker green curve is fully corrected for stop-
ping power and the fluence correction factors.
24.3.7 Lateral Profiles point on the profile are constant, so no corrections are
When measuring lateral profiles in a simple water phan- needed as a function of off-axis position. The true pro-
tom, the situation is quite a bit simpler than what was file at any depth is a convolution of a rectangular func-
described above for percent depth dose. As long as the tion (the beam aperture) and a dose-spread kernel that
depth is constant and there are no added heterogene- can be characterized by a standard deviation of lateral
ities, the energy and fluence correction factor at each position. Because of the finite size of the detector, the
FIGURE 24.2 Calculated electron beam penumbra for a 10 × 10 cm 2 field size at a depth of 2.5 cm. The “True Profile” is a
convolution of a rectangular function with a scattering kernel with a standard deviation (sigma) of 0.3 cm. The “Diode” profile
is a moving average of the true profile with a width of 0.2 cm, which is a representative size for a semiconductor diode detector.
The “Ion Chamber” profile is a moving average of the true profile with a width of 0.6 cm, which is a representative size for a
cylindrical ionization chamber detector.
measured lateral profile is a convolution of a detector functions. If we assume that the response function of
response function and the true profile and is therefore the detector is a rectangular function with the same
a smoothed version of the true profile. The dose-spread width as the sensitive volume of the detector, w, the
kernel is due to the angular variance of the electrons at standard deviation of the response function is w/2 3.
the position of the electron aperture and the multiple For a detector width of 2 mm, which is typical for a
Coulomb scattering in the water phantom [37]. Using diode detector, the standard deviation is 0.58 mm. If
the measured penumbra width, Antolak et al. [38] the true scattering kernel standard deviation is 3.0 mm,
extracted the standard deviation of the scattering kernel the effective standard deviation is 3.06 and, as shown in
for 12-MeV and 16-MeV beams. For both energies, the Figure 24.2, the measured penumbra is almost identical
standard deviation is approximately 2 mm at the surface to the true penumbra. If the detector width is increased
and 3–3.5 mm at the depth of maximum dose, increas- to 6 mm (e.g., typical cylindrical ionization chamber),
ing to a maximum of 11.5–14 mm at a depth of 0.8 R p. the detector response standard deviation is now 1.73 mm
To illustrate the influence of detector size on the and the effective standard deviation is now 3.46 cm. In
beam profile measurement, we will assume a uniform Figure 24.2, we see that the measured penumbra with
beam fluence and a dose-spread kernel standard devia- the larger detector is visibly different from the true pen-
tion of 3 mm at a depth of 2.5 cm (approximate depth umbra. Note that the detector profiles in Figure 24.2
of maximum dose for 12 MeV electrons). The normal- were calculated using a moving average filter as an illus-
ized dose profile is calculated using the first few terms tration of what the measured profiles might look like.
of Equation (24.13) from Hogstrom et al. [37] and At deeper depths, the detector size has less influence
Figure 24.2 shows the penumbra of this profile for a because the penumbra width increases, but it should be
10 × 10 cm 2 field size. When we add a detector to the sys- fairly clear that using a smaller detector (e.g., semicon-
tem, the measured profile is a convolution of the true ductor diode) is desirable.
profile and the detector response function. The effec-
tive scattering kernel standard deviation is the square 24.3.8 Output Factors
root of the quadrature sum of the standard deviations The output factor is the ratio of the dose per monitor
of the true scattering kernel and the detector response unit for a give field size to the dose per monitor unit for
a reference condition. The reference condition is usually the depth of maximum dose. For the standard field size
the same field size and SSD that was used for the calibra- and SSD, this depth dose should be readily available. For
tion of the linear accelerator (e.g., 10 × 10 cm 2 field size the measured field size, the true depth dose may not be
at 100-cm SSD). While the reference depth, dref , is the readily available (e.g., an irregular patient cutout), but
specified depth to use for calibrating the linear accelera- there may be data for similar field sizes that may allow
tor, it is much more common to use the clinical percent the user to closely approximate the true depth dose. If
depth dose to nominally calibrate the linear accelerator that is not good enough, it may be necessary to perform
(i.e., set 1 cGy per MU) at dmax or R100. The AAPM Report a depth dose scan.
99 [3] defines the output factor, Se using the following If the output factors are needed as input data for the
equation treatment planning system, the user needs to carefully
check the physics manual to see what the treatment
D ( dmax ( r ) , r , SSD ) planning expects. The treatment planning system may
Se ( dmax ( r ) , r , SSD ) = (24.11)
(
D dmax ( r0 ) , r0 , SSD0) only require output factors at a nominal depth of maxi-
mum dose to make data entry easier for the user since
where r is the field size for the output factor, r0 is the it would be easy for the computer algorithm to calculate
reference field size, SSD is the SSD for the output fac- the dose elsewhere.
tor measurement, SSD0 is the reference SSD, and D is If the user decides to use the actual depth of maxi-
the dose rate (or dose per monitor unit). In this equa- mum dose, using a semiconductor diode or diamond
tion, the depth of maximum dose is determined for each detector that does not require depth-dependent cor-
irradiation condition. Having to determine the depth of rection factors would probably be more convenient.
maximum dose for each field size makes this measure- However, comparison with the treatment planning sys-
ment more complicated than a simple ratio of doses at tem becomes more complicated since you need to con-
a single depth. In a water phantom, the user could per- sider that the calculated depth dose may not be the same
form a depth dose scan (or convert ionization to dose) as the measured depth dose and how that can affect the
around dmax to determine its location and dose rate to interpretation of the results. Fortunately, this added
compare to the dose rate at dmax for the reference field. complication is rarely necessary and should be avoided
However, for practical purposes, this process is gener- if possible.
ally more complicated than necessary. The user might could also consider using the reference
The most common choice for output factor mea- depth from the calibration, dref , instead of the nominal
surements is to use a nominal dmax depth for both mea- depth of maximum dose. The reference depth is usually
surements, with a reasonable choice being the depth close to dmax, but it can be deeper than dmax for higher
of maximum dose of the reference field size. Once the energies. For semiconductor diodes, diamond detectors,
nominal depth is determined, a simpler phantom (e.g., and plane-parallel ionization chambers, where a gradi-
small water phantom with manual dosimeter position- ent correction is not needed, placing the detector at dref
ing or solid phantom material) could be used to repro- is unambiguous. However, if a cylindrical ionization
ducibly place the effective measurement point of the chamber is used, the user should ensure that the effec-
detector at the desired depth. If an ionization chamber is tive point of measurement of the ionization chamber is
used, doing the measurements at the same depth means placed at dref . Since the reference calibration is done with
that the stopping power and replacement corrections are the center of the ionization chamber at that depth, this
also the same and the nominal output factor is simply a could cause some confusion.
ratio of measured ionization values. Given the above considerations, it is the author’s rec-
If the intent of the measurement is to check the dose ommendation that output factors be measured using
output in the treatment planning system, using a nomi- a nominal depth of maximum dose, preferably with a
nal depth of maximum dose means that you can use the semiconductor diode or diamond detector. If output
point dose tool to easily compare measurement and cal- factors at the true depth of maximum dose are neces-
culation. If it was desired to determine the output factor sary, clinical depth dose curves can generally be used to
at the actual depths of maximum dose, clinical depth correct the measured value with sufficient accuracy for
doses could be used to correct the measured values to clinical use.
24.4 SOLID PHANTOM MEASUREMENTS assurance such as daily and monthly constancy checks,
24.4.1 General Considerations but some level of quality control should be exercised in
For any type of phantom measurement, there are two their implementation.
primary considerations for choosing an appropriate Currently, there are several more advanced solid
dosimeter and phantom. The first is whether the pres- material water substitutes that are suitable for radia-
ence of the dosimeter significantly alters the radiation tion dosimetry. Examples of such materials are Solid
transport in the phantom material. For example, does Water® (Gammex RMI, Middleton, WI, USA) and
placing a film between 2 slabs of phantom material alter Virtual Water™ (Med-Cal, Verona, WI, USA), and there
how the electron beam traverses the material? The sec- are several other similar materials on the market. They
ond consideration is how the energy response of the are generally sold in the form of solid slabs in a variety
detector compares to the energy response of the phan- of thicknesses and thicker slabs can usually be ordered
tom material. A good example of this for ionization with machined cavities for insertion of standard ioniza-
chambers is the conversion of percent ionization to per- tion chambers. Their density and atomic composition
cent dose in an electron beam using depth-dependent are designed such that interactions with x-rays and elec-
stopping power and fluence correction factors. For that trons are much the same as in liquid water. Before put-
conversion, the assumption is that the beam is perpen- ting any of these materials into routine use, they should
dicular to the water phantom surface so that the energy be examining radiographically to verify that no unex-
is primarily a function of depth. For any other geome- pected voids exist in the material. One should also be
try, the relatively simple formula for the stopping power careful to examine prior literature regarding the partic-
correction in TG-51 may not be applicable. This is one ular brand and model that is being used to ensure that it
reason why ionization chambers may not be the dosime- is suitable for the intended purpose. Some early formu-
ter of choice for a particular measurement. For example, lations were known to be reasonably good substitutes in
semiconductor diode detectors were necessary to mea- x-ray beams, but not so good for electron beams [42, 43].
sure dose distributions behind heterogeneities such as
24.4.3 Solid Phantom Film Measurements
rods and discs, used for evaluating electron beam dose
algorithms [39, 40] and for the standard electron dose When making measurements in solid phantoms using
distribution data set measured by Boyd et al. [41]. film, the phantom and film measurement system should
fit together without any air gaps and the film edge
should be flush with the edge of the phantom if the film
24.4.2 Quality Assurance Measurements is being irradiated edge-on [44, 45]. While Cherenkov
For periodic quality assurance, solid phantom materials emission is not a major energy loss mechanism for elec-
can be more convenient to use in place of setting up a tron beams, the film may be sensitive to the resulting
water phantom. However, care must be taken to ensure photons, which means that opaque phantom materials
that the chosen phantom material is appropriate for the are preferred for electron beams. For radiographic films
measurement purpose. When AAPM Report 32 [14] was in paper jackets, the film may be left in the jacket, but it
published, the most common solid phantom materials is recommended to put pinpricks in the jacket to release
were plastics such as polystyrene and acrylic (PMMA) air inside the jacket and to allow for good compression
and the TG-21 calibration protocol allowed for reference of the film-phantom combination [45].
dosimetry to be done in plastic phantoms [13]. When To assist with standard planar dosimetry measure-
doing these calibrations, the absolute density of the plas- ments, Bova [46] designed a hinged film phantom using
tic might not be the same as water, which meant that the solid water slabs and pins for reproducible alignment.
depth of dmax might be different in the plastic as com- To allow for slight variations in film size from batch
pared to dmax in water. Differences in scattering power to batch, shims were available to get a better fit of the
meant that the actual dose at dmax would be different film within the phantom and to ensure coincidence
than simple density scaling would imply. The manufac- of the edge of the film and the edge of the phantom.
turing of these plastic materials can also introduce addi- Unfortunately, this phantom is not commercially avail-
tional uncertainties that complicate their use even more able to be the best of the author’s knowledge. A differ-
[3]. These phantoms can still be used for periodic quality ent phantom-cassette design is commercially available
(Radiation Products Design, Albertville, MN 55301), more difficult to get a good measurement and for larger
but the author has no experience with the device. dosimeters, it may affect the dose distribution inside the
If the film orientation is perpendicular (or at least patient as well. Before doing any sort of in vivo dosim-
not close to parallel) to the beam direction, the actual etry with electron beams, it may be useful to do some
edge of the film relative to the edge of the phantom is phantom measurements with the dosimetry system to
not as important. However, the slabs used for the phan- ensure that you can obtain reasonable results under
tom must be flat, which is usually true for phantom controlled conditions. To estimate the degree of dose
slabs manufactured expressly for radiation dosimetry. perturbation inside the patient, it would be useful to
However, off-the-shelf plastic slab materials will gener- place the dosimeter on a film phantom where the film
ally require machining to create a uniform-thickness is parallel to the beam direction and close to the phan-
slab with flat surfaces. tom surface. This will allow you to make an informed
decision regarding the utility of your particular system.
24.4.4 Other Solid Phantom Measurements The in vivo dosimeter should ideally be thin in the beam
In addition to solid materials that mimic water, there are direction to minimize the perturbation. It should also
solid materials available that mimic muscle tissue, adi- attach firmly to the skin surface and it may be necessary
pose, lung, bone, and other tissues. There are commercial to remove body hair so that air gaps between the dosim-
phantoms available that have included heterogeneities, eter and patient do not affect the measurements [25].
and these can be modified to allow for dosimeters to be
placed inside the phantom. Hogstrom et al. [47] created 24.6 SUMMARY
such phantoms using patient CT data to help evaluate This chapter was limited to aspects of measuring dose
the accuracy of an electron dose calculation algorithm. distributions for therapeutic electron beams and there is
In a later study, the same phantoms were used to evalu- much more that can be learned about their use in radia-
ate customized bolus electron treatment plans [48]. The tion therapy. The AAPM, IPEM, and IAEA documents
phantom had holes drilled out to accept small TLD cap- mentioned in this chapter provide much more detail than
sules to measure the dose and because the capsules were what was provided here. Hogstrom and Almond [50]
relatively small, the dose perturbation of the dosimeters provide a reasonably comprehensive review of the phys-
was reasonably small. As mentioned above, a key con- ics of electron beam radiotherapy and AAPM Report
sideration for any type of phantom measurement for 99 [3] supplements AAPM Report 32 [14] by providing
electron beams is minimizing the effect of the dosimeter several clinical examples of electron beam radiotherapy.
on the dose distribution. Treatment planning aspects of electron beam radiother-
apy have also been described elsewhere by the author [51].
24.5 IN VIVO DOSIMETRY
There are many different kinds of dosimeters that are NOTES
suitable for in vivo dosimetry for electron beams and
1. http://rpc.mdanderson.org/RPC/home.htm, accessed
there is a good chance that the reader already has access
Oct 7, 2020.
to at least one in vivo dosimetry system. For those that
do not yet have access to such a system, Chapter 11 of REFERENCES
this book [49] describes various systems that are avail-
able. For the most part, any in vivo dosimetry system 1. H. Huizenga and P. R. M. Storchi, “Numerical calcula-
tion of energy deposition by broad high-energy electron
that is suitable for use with x-ray beams can be used
beams,” Physics in Medicine and Biology 34(10), 1371–1396
for measurements in electron beams, provided that (1989). https://doi.org/10.1088/0031-9155/34/10/003.
you consider how the dosimeter may affect the irradia- 2. Indra J. Das, Chee-Wai W. Cheng, Ronald J. Watts,
tion itself. Some in vivo dosimeters may have inherent Anders Ahnesjo, John P. Gibbons, X. Allen Li, Jessica
buildup that is useful for x-ray beams and it is generally Lowenstein, Raj K. Mitra, William E. Simon, Timothy
C. Zhu, “Accelerator beam data commissioning
not advised to use such dosimeters for electron beams.
equipment and procedures: Report of the TG-106
When placing a dosimeter on the patient surface, it of the Therapy Physics Committee of the AAPM,”
is important to remember that the dosimeter can per- Medical Physics 35(9), 4186–4215 (2008). https://doi.
turb the electron beam. This perturbation may make it org/10.1118/1.2969070.
3. Bruce J. Gerbi, John A. Antolak, F. Christopher Deibel, Medical Physics 10(6), 741–771 (1983). https://doi.
David S. Followill, Michael G. Herman, Patrick D. org/10.1118/1.595446.
Higgins, M. Saiful Huq, et al., “Recommendations for 14. Faiz M. Khan, Karen P. Doppke, Kenneth R. Hogstrom,
clinical electron beam dosimetry: Supplement to the Gerald J. Kutcher, Ravinder Nath, Satish C. Prasad, James
recommendations of Task Group 25,” Medical Physics A. Purdy, Martin Rozenfeld, Barry L. Werner, “Clinical
36(7), 3239–3279 (2009). https://doi.org/10.1118/ electron-beam dosimetry: Report of AAPM Radiation
1.3125820. Therapy Committee Task Group No. 25,” Medical Physics
4. Bruce J. Gerbi, John A. Antolak, F. Christopher Deibel, 18(1), 73–109 (1991). https://doi.org/10.1118/1.596695.
David S. Followill, Michael G. Herman, P. D. Higgins, 15. IAEA, “Absorbed dose determination in photon and
M. Saiful Huq, et al., “Erratum: ‘Recommendations electron beams: An international code of practice,”
for clinical electron beam dosimetry: Supplement to Technical Report Series (IAEA, Vienna, 1987), vol. 277.
the recommendations of Task Group 25’ [Med. Phys. 16. J. Homnick, G. Ibbott, A. Springer, J. Aguirre, “TH-
36, 3239–3279 (2009)],” Medical Physics 38(1), 548–548 D-352-05: Optically stimulated luminescence (OSL)
(2010). https://doi.org/10.1118/1.3532916. dosimeters can be used for remote dosimetry services,”
5. Peter R. Almond, Peter J. Biggs, Bert M. Coursey, Medical Physics 35(6 Part 27), 2994–2995 (2008). https://
William F. Hanson, M. Saiful Huq, Ravinder Nath, doi.org/10.1118/1.2962948.
David W. O. Rogers, “AAPM’s TG-51 protocol for clini- 17. J. Aguirre, P. Alvarez, D. Followill, G. Ibbott, C.
cal reference dosimetry of high-energy photon and elec- Amador, A. Tailor, “SU-FF-T-306: Optically stimulated
tron beams,” Medical Physics 26(9), 1847–1870 (1999). light dosimetry: commissioning of an optically stimu-
https://doi.org/10.1118/1.598691. lated luminescence (OSL) system for remote dosim-
6. Malcolm R. McEwen, Larry A. DeWerd, Geoffrey S. etry audits, the radiological physics center experience,”
Ibbott, David S. Followill, David W. O. Rogers, Stephen Medical Physics 36(6 Part 13), 2591–2592 (2009). https://
Seltzer, Jan P. Seuntjens, “Addendum to the AAPM’s doi.org/10.1118/1.3181785.
TG-51 protocol for clinical reference dosimetry of high- 18. Jatinder R. Palta, Peter J. Biggs, John D. Hazle, M. Saiful Huq,
energy photon beams,” Medical Physics 41(4), 041501 Robert A. Dahl, Timothy G. Ochran, Jerry Soen, Ralph
(2014). https://doi.org/10.1118/1.4866223. R. Dobelbower, Edwin C. McCullough, “Intraoperative
7. IAEA, “Absorbed dose determination in external beam electron beam radiation therapy: Technique, dosimetry,
radiotherapy.” Technical Report Series (IAEA, Vienna, and dose specification: Report of Task Force 48 of the
Austria, 2000). https://doi.org/10.1097/00004032- Radiation Therapy Committee, American Association of
200111000-00017. Physicists in Medicine,” International Journal of Radiation
8. R. C. Tailor and W. F. Hanson, “Calculated absorbed- Oncology, Biology, Physics 33(3), 725–746 (1995). https://
dose ratios, TG51/TG21, for most widely used cylin- doi.org/10.1016/0360-3016(95)00280-C.
drical and parallel-plate ion chambers over a range of 19. A. Sam Beddar, Peter J. Biggs, Sha Chang, Gary A. Ezzell,
photon and electron energies,” Medical Physics 29(7), Bruce A. Faddegon, Frank W. Hensley, Michael D. Mills,
1464–1472 (2002). https://doi.org/10.1118/1.1487857. “Intraoperative radiation therapy using mobile electron
9. R. C. Tailor, W. F. Hanson, G. S. Ibbott, “TG-51: linear accelerators: Report of AAPM Radiation Therapy
Experience from 150 institutions, common errors, Committee Task Group No. 72.” Medical Physics 33(5),
and helpful hints,” Journal of Applied Clinical Medical 1476–89 (2006). https://doi.org/10.1118/1.2194447.
Physics 4(2), 102–111 (2003). https://doi.org/10.1120/ 20. Landon S. Wootton, Juergen Meyer, Edward Kim, Mark
jacmp.v4i2.2524. Phillips, “Commissioning, clinical implementation, and
10. D. T. Burns, George X. Ding, David W. O. Rogers, performance of the Mobetron 2000 for intraoperative
“R50 as a beam quality specifier for selecting stopping- radiation therapy,” Journal of Applied Clinical Medical
power ratios and reference depths for electron dosim- Physics 18(1), 230–242 (2017). https://doi.org/10.1002/
etry,” Medical Physics 23(3), 383–388 (1996). https://doi. acm2.12027.
org/10.1118/1.597893. 21. Sergio Righi, Evis Karaj, Giuseppe Felici, Fabio Di
11. George X. Ding, David W. O. Rogers, Thomas R. Martino, “Dosimetric characteristics of electron beams
Mackie, “Calculation of stopping-power ratios using produced by two mobile accelerators, Novac7 and Liac,
realistic clinical electron beams,” Medical Physics 22(5), for intraoperative radiation therapy through Monte
489–501 (1995). https://doi.org/10.1118/1.597581. Carlo simulation,” Journal of Applied Clinical Medical
12. M. Saiful Huq, Ning Yue, N. Suntharalingam, Physics 14(1), 6–18 (2013). https://doi.org/10.1120/jacmp.
“Experimental determination of fluence correction v14i1.3678.
factors at depths beyond dmax for a farmer type cylin- 22. E. Moretti, M. R. Malisan, A. Trianni, R. Padovani,
drical ionization chamber in clinical electron beams,” “Use of radiochromic films to calibrate a high dose-per-
Medical Physics 24(10), 1609–1613 (1997). https://doi. pulse electron accelerator for intraoperative radiation
org/10.1118/1.597978. therapy (IORT),” in International Conference on Quality
13. AAPM, “A protocol for the determination of absorbed Assurance and New Techniques in Radiation Medicine
dose from high-energy photon and electron beams,” (IAEA, Vienna, Austria, 2006), pp. 488–489.
23. Evis Karaj, Sergio Righi, Fabio Di Martino, “Absolute 34. Peter R. Almond, Frank H. Attix, Leroy J. Humphries,
dose measurements by means of a small cylindrical Hideo Kubo, Ravinder Nath, Steve Goetsch, David W.
ionization chamber for very high dose per pulse high O. Rogers, “The calibration and use of plane-parallel
energy electron beams,” Medical Physics 34(3), 952–958 ionization chambers for dosimetry of electron beams:
(2007). https://doi.org/10.1118/1.2436979. An extension of the 1983 AAPM Protocol Report of
24. John A. Antolak and Kenneth R. Hogstrom, “Multiple AAPM Radiation Therapy Committee Task Group No.
scattering theory for total skin electron beam design,” 39,” Medical Physics 21(8), 1251–1260 (1994). https://doi.
Medical Physics 25(6), 851–859 (1998). https://doi. org/10.1118/1.597359.
org/10.1118/1.598295. 35. Dietrich Harder, “Einfluß Der Vielfachstreuung
25. John A. Antolak, Jackson H. Cundiff, Chul S. Ha, von Elektronen Auf Die Ionisation in Gasgefüllten
“Utilization of thermoluminescent dosimetry in total Hohlräumen,” Biophysik 5(2), 157–164 (1968). https://
skin electron beam radiotherapy of mycosis fungoides,” doi.org/10.1007/BF01202901.
International Journal of Radiation Oncology, Biology, 36. John A. Antolak, “Ionization to dose for electron
Physics 40(1), 101–108 (1998). https://doi.org/10.1016/ beams: A comparison of three approaches (Abstract
S0360-3016(97)00585-3. SU-FF-T-205),” Medical Physics 32(6), 1997 (2005).
26. Indra J. Das, John F. Copeland, Harry S. Bushe, “Spatial https://doi.org/10.1118/1.1997933.
distribution of bremsstrahlung in a dual electron beam 37. Kenneth R. Hogstrom, M. D. Mills, Peter R. Almond,
used in total skin electron treatments: Errors due to ion- “Electron beam dose calculations.” Physics in Medicine
ization chamber cable irradiation,” Medical Physics 21 and Biology 26(3), 445–459 (1981).
(11), 1733–1738 (1994). https://doi.org/10.1118/1.597215. 38. John A. Antolak, Ernest Mah, John W. Scrimger,
27. Gerald J. Kutcher, Lawrence Coia, Michael T. Gillin, “Optimization of pencil beam widths for electron-
William F. Hanson, Steven Leibel, Robert J. Morton, beam dose calculations,” Medical Physics 22(4), 411–419
Jatinder R. Palta, et al., “Comprehensive QA for radia- (1995). https://doi.org/10.1118/1.597606.
tion oncology: Report of AAPM Radiation Therapy 39. J. E. Cygler, Jerry J. Battista, John W. Scrimger, Ernest
Committee Task Group 40,” Medical Physics 21(4), 581– Mah, John A. Antolak, “Electron dose distribu-
618 (1994). https://doi.org/10.1118/1.597316. tions in experimental phantoms: A comparison with
28. D. I. Thwaites, D. T. Burns, S. C. Klevenhagen, A. E. 2D pencil beam calculations,” Physics in Medicine
Nahum, W. G. Pitchford, “The IPEMB code of practice and Biology 32(9), 1073–1086 (1987). https://doi.
for electron dosimetry for radiotherapy beams of initial org/10.1088/0031-9155/32/9/001.
energy from 2 to 50 MeV based on an air kerma calibra- 40. Ernest Mah, John A. Antolak, John W. Scrimger, Jerry
tion,” Physics in Medicine and Biology 41(12), 2557–2603 J. Battista, “Experimental evaluation of a 2D and 3D
(1996). https://doi.org/10.1088/0031-9155/41/12/001. electron pencil beam algorithm,” Physics in Medicine
29. D. I. Thwaites, A. R. DuSautoy, T. Jordan, Malcolm R. and Biology 34(9), 1179–1194 (1989). https://doi.
McEwen, A. Nisbet, Alan E. Nahum, W. G. Pitchford, org/10.1088/0031-9155/34/9/004.
“The IPEM code of practice for electron dosimetry for 41. Robert A. Boyd, Kenneth R. Hogstrom, John A. Antolak,
radiotherapy beams of initial energy from 4 to 25 MeV Almon S. Shiu, “A measured data set for evaluating
based on an absorbed dose to water calibration,” Physics electron-beam dose algorithms,” Medical Physics 28(6),
in Medicine and Biology 48(18), 2929–2970 (2003). 950–958 (2001). https://doi.org/10.1118/1.1374245.
https://doi.org/10.1088/0031-9155/48/18/301. 42. D. I. Thwaites, “Measurements of ionisation in water,
30. G. Rikner, “Characteristics of a P-Si detector in high energy polystyrene and a ‘solid water’ phantom material for
electron fields.” Acta Radiologica: Oncology 24(1), 71–74 electron beams.” Physics in Medicine and Biology
(1985). https://doi.org/10.3109/02841868509134368. 30(1), 41–53 (1985). https://doi.org/10.1088/0031-
31. IAEA, “Review of data and methods recommended in 9155/30/1/005.
the international code of practice for dosimetry IAEA 43. Bruce Thomadsen, Chris Constantinou, Anthony Ho,
Technical Reports Series No. 381. The use of plane par- “Evaluation of water-equivalent plastics as phantom
allel ionization chambers in high energy electron and material for electron-beam dosimetry,” Medical Physics
photon beams.” IAEA-TECDOC (IAEA, Vienna, 2000), 22(3), 291–296 (1995). https://doi.org/10.1118/1.597453.
vol. 1173. 44. J. Dutreix and A. Dutreix, “Film dosimetry of
32. Wolfram U. Laub and Richard Crilly, “Clinical radia- high-energy electrons.” Annals of the New York
tion therapy measurements with a new commercial Academy of Sciences 161(1), 33–43 (1969). https://doi.
synthetic single crystal diamond detector,” Journal of org/10.1111/j.1749-6632.1969.tb34039.x.
Applied Clinical Medical Physics 15(6), 92–102 (2014). 45. Sujatha Pai, Indra J. Das, James F. Dempsey, Kwok L.
https://doi.org/10.1120/jacmp.v15i6.4890. Lam, Thomas J. Losasso, Arthur J. Olch, Jatinder R.
33. David W. O. Rogers, “Accuracy of the burns equation Palta, Lawrence E. Reinstein, Dan Ritt, Ellen E. Wilcox,
for stopping-power ratio as a function of depth and “TG-69: Radiographic film for megavoltage beam
R50,” Medical Physics 31(11), 2961–2963 (2004). https:// dosimetry,” Medical Physics 34(6), 2228–2258 (2007).
doi.org/10.1118/1.1803811. https://doi.org/10.1118/1.2736779.
46. Francis J. Bova, “A film phantom for routine conformal therapy,” Medical Physics 40(7), 71720 (2013).
film dosimetry in the clinical environment.” https://doi.org/10.1118/1.4811104.
Medical Dosimetry 15(2), 83–85 (1990). https://doi. 49. D. Bollinger and A. Darafsheh, “Clinical considerations
org/10.1016/0958-3947(90)90040-O. and dosimeters for in vivo dosimetry,” in Radiation
47. Kenneth R. Hogstrom, Michael D. Mills, Jeff A. Eyer, Therapy Dosimetry: A Practical Handbook , edited by A.
Jatinder R. Palta, David E. Mellenberg, Raul T. Meoz, Darafsheh (CRC Press, Boca Raton, FL, 2021) , Chapter
Robert S. Fields, “Dosimetric evaluation of a pencil- 11, pp. 151–171.
beam algorithm for electrons employing a two-dimen- 50. Kenneth R. Hogstrom and Peter R. Almond, “Review
sional heterogeneity correction,” International Journal of electron beam therapy physics,” Physics in Medicine
of Radiation Oncology, Biology, Physics 10(4), 561–569 and Biology 51(13), R455–R489 (2006). https://doi.
(1984). https://doi.org/10.1016/0360-3016(84)90036-1. org/10.1088/0031-9155/51/13/R25.
48. Robert L. Carver, Kenneth R. Hogstrom, Connel Chu, 51. John A. Antolak, “Electron beam treatment planning,” in
Robert S. Fields, Conrad P. Sprunger, “Accuracy of Khan’s Treatment Planning in Radiation Oncology, edited
pencil-beam redefinition algorithm dose calculations by F. M. Khan, J. P. Gibbons, P. W. Sperduto (Wolters
in patient-like cylindrical phantoms for bolus electron Kluwer, Philadelphia, PA, 2016), 4th ed., pp. 288–312.
Chapter 25
Proton Therapy Dosimetry

Michele M. Kim and Eric S. Diffenderfer
CONTENTS
25.2 Fundamentals of Proton Physics 394
25.3 Detectors 396
25.3.1 Absolute Detectors 396
25.3.1.1 Ionization Chambers 396
25.3.1.2 Calorimetry 397
25.3.1.3 Faraday Cup 398
25.3.2 Relative Detectors 398
25.3.2.1 Radiochromic Film 399
25.3.2.2 Thermoluminescent Detectors 399
25.3.2.3 Optically Stimulated Luminescence Detectors 399
25.3.2.4 Solid State Diode Detectors 399
25.3.3 Multidimensional Detectors 400
25.3.3.1 Multielement Detectors 400
25.3.3.2 Scintillating Detectors 400
25.4 Proton Beam Delivery Techniques 401
25.4.1 Scattered Beams 401
25.4.2 Scanned Beams 402
25.5 Measurement Techniques 403
25.6 Dosimetry Protocols 404
25.7 Quality Assurance for Proton Therapy 404
25.8 Uncertainties 405
25.9 Summary 406
References 408
25.1 INTRODUCTION specific quality assurance (QA). Many aspects of radia-

Radiation dosimetry encompasses the measurement tion dosimetry are equivalent between different radia-
techniques and procedures that allow quantification and tion sources and delivery modalities; however, a few
verification of the radiation dose delivered during radia- aspects are particular to particle therapy with protons.
tion therapy. These techniques are applied throughout The fundamental advantage introduced by proton
the processes of accepting an accelerator system, com- beam therapy over conventional x-ray and electron
missioning a treatment planning system, perform- therapies is that the protons penetrate tissue and travel
ing daily, monthly, annual, and patient treatment plan mostly with a straight trajectory before coming to rest
393
and generating a high density of ionization in a small but still small, angles. Protons go through many of these
region. Exit dose beyond the Bragg peak is negligible, interactions as it passes through matter, and statisti-
with a small contribution due to secondary particles cally all of these interactions add up in a net angular
and gamma rays produced in nuclear interactions. The and radial deviation, which has been called “multiple
highly localized deposition of proton dose enhances the Coulomb scattering.”
normal tissue sparing capabilities of external beam ther- There are two main types of nuclear interactions as
apy while maintaining a high dose to the tumor target protons travel through a medium, elastic and nonelastic
and surrounding margin. However, the scattering and collisions. In elastic collisions between the proton and a
energy loss characteristics of protons lead to dosimetry nucleus, the nucleus is left intact while the proton loses
and measurement techniques that differ from conven- a significant fraction of its energy and is deflected. In
tional electron radiation. nonelastic collisions, the nucleus is broken apart as the
Calibration of proton beams requires measure- proton loses a significant fraction of its energy and is
ment of absolute physical dose which typically differs deflected. Typically, the light fragment of the nucleus is
from prescribed dose by a biological factor intended to knocked out with considerable speed as it leaves behind
account for the difference in biological effect [1, 2] such a heavy fragment that stays close to the place of interac-
that the physical dose is multiplied by a relative biologi- tion and is very ionizing.
cal effectiveness (RBE) factor. The product is called the Bremsstrahlung photons are generated when charged
RBE weighted dose and written using the International particles are passing in the field of an atomic nucleus.
Commission on Radiation Units and Measurements However, the likelihood of bremsstrahlung is roughly
suggested notation Gy(RBE) to indicate that the SI unit proportional to the inverse square of the particle mass,
of Gy is modified by an RBE factor [3]. and therefore do not enact any clinical significance in
proton therapy.
25.2 FUNDAMENTALS OF PROTON PHYSICS As protons travel through media, the dose depos-
Protons interact in matter with an atom or nucleus via ited rises sharply near the end of their range, creating a
several mechanisms. There are Coulombic interactions Bragg peak. The shape of the Bragg peak is due to mul-
with atomic electrons, Coulombic interactions with the tiple effects. The first of which is the slow loss of energy
atomic nucleus, nuclear interactions with atomic nuclei, that protons undergo due to Coulomb interactions with
and Bremsstrahlung. Protons will lose kinetic energy atomic electrons. The linear energy transfer (LET) or
continuously with multiple inelastic Coulombic interac- stopping power is the proton’s linear rate of energy loss,
tions with the atomic electrons, hence depositing dose, measured in units of MeV/g/cm2. This can be described
as they penetrate matter. For the most part, protons by the Bethe–Block formula [5, 6]
travel in a straight line since the rest mass of protons
are 1832 times that of an electron [4]. Protons and elec- S dE
=−
trons having opposite charges causes attraction and the ρ ρdx
electrons can be released from the atoms. This results in
Z z2  2me c 2γ 2 β 2 δ C
ionization of atoms, and loose electrons go on to ionize = 4π N Are2me c 2  ln − β2 − − 
A β2 I 2 Z
further atoms in the vicinity of the atom from which it
came. Protons on average lose relatively little energy in (25.1)
individual ionizations and do not get deflected as much.
They will undergo hundreds of thousands of interac- where S is the stopping power, ρ is the mass density of
tions per centimeter of material before eventually losing the absorbing material, (−dE/dx) describes the energy
all of their energy and coming to rest. loss rate, NA is Avogadro’s number, re and me are the
Since protons are so much heavier than electrons, electron radius and mass, respectively, z is the charge
they are barely deflected by Coulomb interactions with of the projectile, Z is the atomic number of the mate-
atomic electrons. Passing through material, however, rial, A is the atomic weight of the absorbing material, c is
protons do experience a repulsive force when passing the speed of light, β = v/c where v is the projectile veloc-
positively charged atomic nuclei. Nuclei are larger than ity, γ = (1 − β 2)(−1/2), I is the mean excitation potential of
electrons, resulting in deflections of protons at larger, the absorbing material, δ is the density correction from
Proton Therapy Dosimetry ◾ 395
shielding of remote electrons by close electrons, and C is path, so a narrow peak is higher for the total energy in
the shell correction term that is only important for low the peak to be constant. Proton range is often character-
energies. Approximately, the energy loss is proportional ized by water-equivalent thickness (WET). The general
to the inverse square of the proton’s mean speed and equation is
(Z/A)z2. The dose rises quickly as the protons slow down.
The Bragg peak is blurred due to two main effects. ρm Sm
t w = tm (25.2)
The first is called “range straggling,” which are statisti- ρw Sw
cal fluctuations in the ionization process and causes a
smearing out of the depth of penetration of the stopping where tw and tm are thickness in water and material, ρw
protons. The magnitude of this is roughly 1% of the pro- and ρm are the mass densities of water and material, and
ton’s range. The second effect is from the energy spread Sw and Sm are the mean proton mass stopping powers for
due to the specifics in producing high energy proton water and material [7].
beams. There is never a pristine monoenergetic beam of Monoenergetic proton beams will penetrate matter
protons, resulting in an energy spread of around 1% in of a given density up to a depth which is defined by the
magnitude. beam energy. The relation between depth of penetration
Nuclear interactions of protons occur at a rate of and energy is beneficial in the practical use of protons
around 1%/g/cm2 until the last few millimeters of their for radiation therapy.
end of range. Nuclear interactions reduce the number of To make clinical use of the very narrow proton Bragg
primary protons in the beam. They also produce a halo peak, multiple monoenergetic Bragg peaks can be deliv-
of scattered primary protons and knocked-out secondary ered to create a flat distal region of near-constant high
protons that travel long distances and create a tail to the dose, called the spread-out Bragg peak (SOBP). The
lateral dose profile of a proton beam. Nuclear interactions monoenergetic peaks are not equally weighted, but the
also increase the RBE in the vicinity of interaction by more proximal peaks have less weight and contribute
creating heavily ionizing fragments with very high stop- the least to the dose delivered. While an SOBP does not
ping power which then deposit dose. The nuclear interac- deliver dose beyond the most distal high dose region,
tions also create a halo of neutrons that mostly escape the the entrance dose does increase due to the stacking of
patient without further interaction but are responsible for multiple Bragg peaks, and reduces the peak-to-plateau
a small contribution to the dose inside and outside the ratio that is seen in monoenergetic Bragg peaks.
primary radiation field. As protons travel in depth in a Similar to photons, protons experience the inverse
medium, the proton fluence decreases. Roughly, the 80% square effect. Compared to an initially parallel beam,
dose on the falling edge of the Bragg peak is very close to proton fluence and dose at a point that is a distance r
the same depth as the 50% fluence of the falling edge of from the source will be reduced by a factor of 1/r2.
the fluence distribution. The flat entrance plateau region This can be compensated by increasing the weights of
of a proton depth dose curve is due to the combination the upstream Bragg peaks, but then will cause a much
between the rising energy deposition and the diminish- higher entrance dose. For this reason, proton beam
ing primary proton fluence. source distances are at least 2 m from the isocenter so
Proton range is defined as the depth where half of that the proximal tissue sparing can be maximized with
the protons traveling in the medium have come to rest. the use of an SOBP.
It can be best described by the distal 80% point on the Lateral dose distribution of a proton beam is due to
Bragg peak. Range is greater with higher energy proton four main effects that cause broadening of a narrow
beams. Lower energy proton beams tend to have nar- beam: multiple Coulomb scattering – near-Gaussian
rower Bragg peaks since the range straggling effect is core, multiple Coulomb scattering – long tail, proton
relative to its own proton range, namely ~1.5%, result- nuclear interactions, and neutron nuclear interactions.
ing in a smaller absolute broadening of the Bragg peak. Multiple Coulomb scattering was worked on by Molière
The ratio of the Bragg peak to the plateau (or entrance) in 1947 [8]. This is the main cause of the spreading out
dose is higher for lower energy beams because of the of an initially infinitesimal pencil beam. The princi-
narrower Bragg peak. The same energy is delivered in pal component is a nearly Gaussian distribution in the
the Bragg peak in the last couple of g/cm2 in a proton’s angle of deviation and in the lateral spread of a pencil
beam. This lateral blurring is a function of depth within available for absolute measurement of dose which oper-
the beam. In addition to this Gaussian shape, there is ate under one of three principles, ionometric measure-
a long tail that is due to large angle scattering in only ment with an IC, chemical dosimetry such as Fricke
one or a few collisions. This long tail portion of lateral dosimetry [12], and calorimetry. Additionally electron
dose spread has a relatively small amplitude and can be paramagnetic resonance (EPR) dosimetry with alanine
approximated by a second, broader Gaussian [9]. High [13, 14] and optically stimulated luminescence detectors
energy protons that are scattered from the elastic and (OSLDs) [15] are each becoming widely available for use
nonelastic nuclear collisions contribute to the tails of a in proton therapy. However, for the purpose of proton
proton pencil beam. As the depth increases, there is a dosimetry the accepted international standards [3, 16,
larger halo of dose around the beam that can also be 17] for reference dosimetry use IC measurements with
approximated by a Gaussian distribution. Dose can be national standard traceable calibration certificates.
underestimated if this halo dose is not considered while
performing absolute dosimetry. Finally, there is a halo 25.3.1.1 Ionization Chambers
of neutrons that is generated but will usually escape the ICs are the standard dose measurement device employed in
patient without further interaction. However, the neu- the reference dosimetry protocols discussed in Section 25.6
trons can be responsible for low doses being deposited and are generally recommended when access to a calorim-
in and outside of the proton beam path. Since the lateral eter or Faraday cup (FC) is not available. They can be con-
spread of a pencil beam is unavoidable, it is best to min- structed using tissue equivalent materials with a wide variety
imize the air gap between the patient surface and the of shapes and sizes for convenience of use, and perform reli-
final beam modifying material, whether that is a range ably with well-studied response variables, refer to Chapter 2
shifter or an aperture. for detailed information on the construction and use of ICs
for radiation dosimetry.
25.3 DETECTORS For protons, the largest uncertainty with measuring
Many of the same detectors that are used in conven- absorbed dose to water is with the beam quality correc-
tional x-ray radiation therapy are also useful in proton tion factor, kQ which depends on the ratios of stopping
radiotherapy. Film, diodes, ionization chambers (ICs), power and the energy required to produce an ion pair
scintillation, and array detectors are all widely used in in air with the reference beam quality [17]. These values
proton therapy. There are many sources of information vary as a function of beam energy, LET, and the residual
on commonly used ionizing radiation detectors includ- range of the proton beam. There are no reference pro-
ing the textbooks by Attix [10] and Knoll [11]. However, ton beams available at calibration laboratories, so in
there are certain characteristics of a detector that make most cases 60Co is used as the primary reference quality.
it suitable for use in proton dosimetry. Detectors must Stopping power calculations depend on the mean exci-
be available for relative or absolute dosimetry with tation energy value, I, for water and the chamber mate-
traceable calibration and exhibit response that is mini- rial (Section 25.2). Gomà et al. [18] used Monte Carlo
mally dependent on beam energy. Namely the variation techniques to calculate kQ with I values from ICRU
of detector response with proton stopping power ratio, reports [19, 20] and more recent values from Andreo
or LET variation should be minimized for the particle et al. [21] and Burns et al. [22]. They found agreement
energies relevant to proton therapy. For this reason, ICs with tabulated values in IAEA TRS Report 398 within
constructed of tissue or air equivalent materials are typi- 2.3% and agreement with experimentally determined
cally used. Other detector types such as film, scintilla- values within 1.1%.
tors, solid state detectors, etc. can often exhibit a strong Additionally, Vatnitsky et al. [23] used water calo-
LET dependence which will alter their response com- rimetry to compare kQ for ICs in 60Co reference proton
pared to ICs as the beam LET changes in the Bragg peak. beams. They found agreement with calculation within
1.2% for the entrance region of a monoenergetic 250 MeV
25.3.1 Absolute Detectors proton beam and within 1.8% in the SOBP of a 155 MeV
Absolute dosimetry involves measurement of the funda- beam. Sorriaux et al. [24] used Monte Carlo methods
mental deposition of energy per unit mass of the active to evaluated beam quality correction factor dependence
volume of the detector. There are a few technologies on scanned proton beams compared to proton dose
delivered as a broad beam. They found no additional 3.5%, dependent on energy and measurement depth,
uncertainty when using kQ factors measured or com- for unscanned pencil beams by increasing the chamber
puted for a broad beam with a scanned proton beam. diameter from 8.2 cm to 12 cm [33].
Vatnitsky et al. [25] performed an international dosim-
etry intercomparison between 13 institutions and found 25.3.1.2 Calorimetry
that when using ICs with 60Co calibration factors trace- Calorimetry is the gold standard for absolute dosimetry
able to a standards laboratory, the variations between and the fundamental measurement technique for physi-
centers were within 3%. cal absorption of energy in matter and has been recom-
Often it is preferable to use a parallel plate type IC mended for primary proton dosimetry standards [16, 34].
for proton beam dosimetry over a cylindrical or Farmer Based on measurement of heat energy deposited in a
type chamber. IAEA TRS Report 398 recommends the material when subjected to radiation, calorimetry is the
use of cylindrical chambers for reference dosimetry due most direct measurement of absorbed dose available
to lower uncertainty of kQ. They also recommend taking and does not rely on detection of the products of radia-
the central axis as the measurement point rather than tion interaction such as ionization. Very small changes
the effective point of measurement at 0.75 times the in temperature must be measured and often for this rea-
inner radius. However Gomà et al. [26, 27] have since son graphite calorimeters are used because compared to
shown discrepancy between the tabulated kQ values with water it has a lower heat capacity but with similar radio-
experiment. The discrepancy can be traced to the recom- logical properties. Even so, there are a number of fac-
mendation to use the central axis as the measurement tors contributing to uncertainty in these measurements,
point with better agreement when the effective point which include the heat defect (deposited energy not
is used. Bichsel [28] showed that a thin parallel plate resulting in a temperature change) and electronic stop-
detector filled with air is most appropriate to preserve ping power ratio of water to the calorimeter material.
the shape of the particle Bragg curve. Palmans et al. [29] The reader is advised to refer to Chapter 3 for additional
found that the perturbation correction factor for plane details on the construction and use of calorimeters for
parallel ICs was unity within experimental uncertain- absorbed dose measurement.
ties. There are significant uncertainties associated with However, there are additional drawbacks when con-
using cylindrical ICs in the sharp dose gradients around sidering this dosimetry technology for use with proton
proton Bragg peaks. This leads to a large uncertainty in beams and calibration services for direct calorimetry-
gradient correction factor and, with the finding of negli- based measurement of absorbed dose do not currently
gible contribution from perturbation corrections, there- exist [16]. Corrections were calculated by Lourenço
fore favors the use of a plane parallel chamber. However, et al. using measurement and Monte Carlo techniques to
in their Monte Carlo study Gomà et al. [18] deduced that correct for fluence differences between water and graph-
there may be significant variation from unity of pertur- ite at the same radiological depth [35]. The fluence cor-
bation factors dependent on chamber type. rections were found to be dependent on beam energy,
Large area parallel plate ICs have become available measurement depth, and chamber geometry ranging
specifically to support proton depth profile measure- from 0.99 for a 60 MeV beam at all depths, 0.99–1.04
ment of pristine Bragg peaks for pencil beam scanning for a 180 MeV beam, to 0.99–1.01 for a larger area detec-
(PBS) delivery (See Section 25.4.2). Commissioning of tor when moving from the entrance to greater depths.
PBS requires measurement of the unscanned integrated Additionally, the heat defect is found to depend on the
depth dose (IDD) profile collecting the full lateral pro- LET which varies with depth in ion beams [36–40].
file of the beam which includes a small non-Gaussian Still there have been a number of studies performing
contribution from secondary events known as the absorbed dose measurements with calorimeters in pro-
nuclear halo [9, 30]. Clasie et al. [31] modeled the nuclear ton beams and this work has gone a long way towards
halo contribution as a secondary Gaussian distribution validating IC calibration and beam quality corrections.
that is nearly invariant with depth in phantom. Lin Medin et al. [38] used a sealed water calorimeter to mea-
et al. [32] measured the nuclear halo dependency on sure kQ for 10 MV photon and a 175 MeV proton fields.
proton nozzle design by measuring PBS field size effects The comparison confirmed the photon beam quality
on output. Collection efficiency can be increased by values tabulated in TRS-398 but yielded a difference of
1.8% from the proton values. Medin [39] also measured not available. Cambria et al. [44] compared two differ-
absorbed dose in a high energy scanned proton beam ent types of FC and an IC. They found good agreement
with a water calorimeter and Farmer type ICs calibrated between FC (better than 3.6%), but a larger difference
in a 60Co beam to obtain kQ factors for proton beams. between FC and IC up to 8.2%. Lorin et al. [45] used an
The results agreed well with those tabulated in TRS-398 FC to calibrate an IC and correct the deficiency of ICs due
[17] and within 1.1% of prior measurements in a pas- to ion recombination in the high ionization density cre-
sively scattered beam. The discrepancy can be attributed ated by scanned pulsed proton beams.
to differences in ion recombination when using ICs in FCs will completely stop and collect the proton charge
scanned proton beams. Sassowsky et al. [36] studied the in a thick electrode leading to an accurate measure of
use of water calorimetry in a scanned proton beam for the total charge on the FC. Secondary charged particles
evaluation of beam quality correction factors and the are created with recoil velocity sufficient to escape the
variation of heat effect due to LET as well as calorim- FC, thereby affecting the charge measured in the device.
eter response changes due to spatial-temporal variations Therefore, FCs are typically designed with a guard elec-
in spot scanned proton beam delivery. Brede et al. [37] trode to suppress escape electrons and ensure an accu-
measured absorbed dose with a water calorimeter in rate charge reading. Though Cascio and Gottschalk [46]
a proton beam with uncertainty better than 1.8% and showed that a simplified FC can be constructed to agree
cross compared with IC measurements, themselves hav- with a traditional FC within 1–5%, but without an elec-
ing uncertainty of 2.0%, for an agreement between calo- tron suppression guard ring.
rimeter and ion chamber of 0.7%. Delacroix et al. [41] The elementary charge e is known with great precision
ran a comparison between calorimeter constructed of and the total number of protons incident on the FC can
tissue equivalent material, various IC and FCs at four be derived. Given the cross-sectional distribution of the
proton therapy centers and found an agreement of bet- beam, the total dose can be calculated from the energy
ter than 1% between calorimeter and the other detectors. and stopping power of the protons. A constant energy at
Vatnitsky et al. [23] developed a formalism for proton the FC is assumed by Vynckier et al. [43] which provides
dosimetry with ICs based on absorbed dose to water the following formula for dose calculation from a FC,
calibration factors and beam quality correction factors.
They used water calorimetry to measure kQ γ in a 60Co D = N /a × ( S /ρ ) × (1.602 × 10−10 ) (25.3)
beam and a proton beam to measure kQp. Sarfehnia et al.
[40] used water calorimetry to measure absorbed dose Here D is dose [Gy] to the medium, N is the number of
to water in double scattered and scanned proton beams. protons, a is the area of the beam at the measurement
The water calorimeter had been previously verified in position, (S/ρ) [MeV cm2 g−1] is mass stopping power
photon and electron beams, and the overall uncertainty in the medium for the protons at the known energy.
was 0.4% for double scattered and 0.6% for scanned pro- Care must be taken in the application of this protocol
ton beams, comparing very favorably with the overall to avoid contribution from low energy scattered pro-
uncertainty of 1.9% for the TRS-398 reference absorbed tons which will deposit energy with a different stopping
dose protocol. power. Demonstrating the sensitivity to measurement
conditions, including the effects of scattered protons,
25.3.1.3 Faraday Cup Delacroix et al. [41] found deviations up to 17% between
FCs are designed to completely collect the total charge of FC measurements at different proton centers. Grusell
a particle beam and can be used for indirect measurement et al. [47] developed a method to use a FC to calibrate
of proton beam dosimetry. Verhey et al. [42] described an an IC in an uncollimated beam, thereby decreasing the
absorbed dose measurement method using a FC at the contribution from scattered protons and the uncertainty
160 MeV Harvard cyclotron and compared with mea- of the stopping power of the protons in the gas filling the
surements using a parallel plate IC, thimble type IC, and cavity of the IC.
tissue equivalent calorimeter and found agreement at a
range of depths within 2.6%. Vynckier et al. [43] devel- 25.3.2 Relative Detectors
oped a proton dosimetry protocol that recommends the Often the purpose of dosimetry measurements is to
use of a FC for calibrating an IC when a calorimeter is measure the dose relative to a reference measurement
without regard for the absolute physical dose. This can response. Spielberger et al. [61] evaluated the response
take the form of point measurements for compari- of silver emulsion film with different particle LET and
son with other measurements or 1D, 2D, and 3D rela- found a clear suppression of response with higher LET.
tive dose distributions. In proton dosimetry and QA A linear dependence for radiochromic film quenching
it is frequently useful to compare measured against in proton radiation was found by Anderson et al. [62]
benchmarked distributions, but it is impractical using who note that under response of >15% on the distal edge
a point dosimeter to measure with high spatial resolu- of the Bragg peak is possible and that LET calibration is
tion. Therefore, when benchmarking constancy of spa- necessary for accurate film calibration.
tial variation in delivered dose, it is more often useful to
employ relative detectors to normalize out fluctuations 25.3.2.2 Thermoluminescent Detectors
between repeated measurements or to use high spatial Thermoluminescent detectors (TLDs) are discussed in
resolution relative detectors for the comparisons. Chapter 6, and they have had a limited role in proton
dosimetry. TLDs have been studied for use in proton
25.3.2.1 Radiochromic Film therapy and have shown good agreement with ICs and
Film is very useful for relative proton dosimetry due to diamond solid state detectors [63–66]. While they can
its convenience, high spatial resolution, availability in be made to small dimensions, they have a relatively
large dimensions that are easily custom fit to applica- large WET. However, the dose record for TLDs is tem-
tions, and long-term storage capability. The use of self- porally stable, and the primary use in proton therapy is
developing radiochromic film has all but superseded for remote auditing of proton therapy sites by irradia-
silver emulsion film due to the clear advantage of not tion and shipment of TLDs embedded in standardized
requiring the film developing equipment and chemicals. phantoms [67].
Recent formulations exhibit increased sensitivity, and
removing the uncertainties introduced by the chemi- 25.3.2.3 Optically Stimulated Luminescence Detectors
cal developing process opens the possibility of absolute OSLDs are discussed in Chapter 7 and their operation
dosimetry with radiochromic films. Several groups have is similar to TLDs, but light is used to stimulate lumi-
developed detailed handling, irradiation, scanning, nescence instead of heat as in the case of TLDs. With
and processing protocols for accurate film dosimetry their temporal stability and compact size, OSLDs are
[48–53]. attractive for use in health physics settings and for sur-
The main issue faced with using radiochromic film for face measurements in proton therapy. OSLDs have been
proton dosimetry is the quenching effect that increases studied for use in proton therapy with minimal observed
with LET and leads to an under response in the Bragg LET dependence for relatively low proton radiation [15,
peak [54–57]. MD-55 Gafchromic film (International 68]. However, LET dependence has been observed for
Specialty Products (ISP), Wayne, New Jersey) is a radio- higher LET particles, and may be of interest for response
chromic film that has been used for relative proton due to secondary neutron radiation produced by the pri-
dosimetry of depth dose and lateral profiles for both mary proton beam [69–71].
modulated and unmodulated proton beams in the
energy range 100–250 MeV [58]. The film exhibits good 25.3.2.4 Solid State Diode Detectors
agreement with IC measurements, but under-responds Solid state detectors in use for proton therapy include
in the distal region of the Bragg peak due to a quench- diamond, metal oxide field effect transistors (MOSFET),
ing effect. Still they found agreement in the distal falloff and doped semiconductor (p-type or n-type) detectors
depth (80–50% distance) was within 0.1–0.2 mm of IC [42, 64, 72, 73]. These detectors can be constructed with
measurements. Martisíková and Jäkel [59] found that very small proportions which lend them to high spatial
Gafchromic EBT film response was comparable between resolution scanning measurements or small field dosim-
Co-60 and protons. Another study by Mumot et al. [60] etry [73]. Additionally, diamond detectors have an active
comparing MD-55 radiochromic film and a silicon diode volume constructed with carbon and can be designed
found a 4% difference in a modulated SOBP and a 12% with near tissue equivalent composition. Fidanzio et al.
difference in an unmodulated Bragg peak, demonstrat- [74] measured the LET dependence of a diamond detec-
ing the effect of LET variations on radiochromic film tor in a 62 MeV proton beam. Grusell and Medin [75]
have observed an increase in sensitivity with LET of but offer better than 0.4 mm range constancy measure-
n-type silicon diodes after pre-irradiation. On the other ment which can be used for QA. However, Takayanagi
hand, in this study the p-type diode did not exhibit et al. [80] developed a dual ring MLIC and a correction
LET correlated increased sensitivity. Onori et al. [76] technique that suppresses the difference between IDDs
measured significant variations in response from both measured with the MLIC and in a water phantom.
silicon diode and diamond detector. They also observed Multilayer FCs have also been designed for measure-
a small dose rate effect with silicon diodes, while there ments of proton fluence change with depth [81–84]. The
was no observed effect from the diamond detector up dose can be derived from the fluence measurements in
to 5 Gy/min. Pacilio et al. [77] studied the response a multilayer FC through the relative stopping powers.
of a diamond detector and a p-type silicon diode in a However, the FC electrodes are typically constructed
10–59 MeV proton beam. The diamond detector showed of copper which has a WET dependent on energy.
large quenching of response in the Bragg peak, where Corrections must be made to obtain the range of a pro-
the p-type diode did not. However, the diode exhibited ton beam, but comparison against baseline measure-
signs of radiation damage with a decrease in response ments validated by a depth dose measurement with a
after a high accumulated dose. The variable reports of water equivalent device can achieve good precision for
LET dependence and susceptibility to radiation damage range constancy measurements [84].
have probably limited the routine use of silicon diode Multi-pad or arrays of ICs have been developed to
detectors in proton dosimetry. combine good spatial resolution with the well-studied
and reliable dose response of ICs [79]. Arjomandy et al.
25.3.3 Multidimensional Detectors [85] characterized a commercial 2D array of ICs. The
Multidimensional detectors have seen a recent rise in MatriXX detector (IBA Dosimetry, Schwarzenbruck,
use in proton therapy primarily due to a rapid increase Germany) consists of 32 × 32 array of 4.5 mm diameter
in highly conformal and intensity modulated treatment ICs with integrated multichannel electrometer and data
techniques such as PBS and intensity-modulated proton acquisition system. The device gave excellent agreement
therapy (IMPT). These techniques with the capability of with film and ion chamber measurements in water when
generating high dose gradients with complex treatment used in a passively scattered proton beam [85, 86]. Lin et al.
volumes covered by multiple isocentric beam arrange- [87] developed a method to characterize the position and
ments require enhanced dosimetry and QA. width of PBS pencil beams to submillimeter precision.
Lin et al. [88] tested a newer version of the device adapted
25.3.3.1 Multielement Detectors to the increased local proton dose rate that is inherent
For ease of use and to potentially save considerable mea- in actively scanned PBS systems. The small pencil beam
surement time, several multichannel depth dose mea- size is delivered with a higher local intensity compared
surement devices have been developed. These devices to passive delivery because of the increased delivery time
typically consist of a stack of air-filled parallel plate ICs scanning from spot to spot and layer to layer over the
with intervening material designed to place the mea- 3D treatment volume. The higher intensity is associated
surement depth of each IC at the equal geometrical and with increased ion recombination effects in the original
water equivalent depth. An early example developed at MatriXX which is operated at a fixed voltage. The newer
the GSI facilities in Darmstadt, Germany consisted of a device, dubbed MatriXX PT, was designed with a smaller
multi-wire proportional chamber for position resolution IC plate gap which minimizes the ion recombination and
which was coupled to a stack of large area parallel plate provides a significant improvement in absolute dose mea-
ICs for a 3D dose measurement system [78]. A multi- surement in PBS beams.
layer IC (MLIC) was developed along with a multi-pad
IC (MPIC) to measure depth and lateral profiles of pro- 25.3.3.2 Scintillating Detectors
ton beams [79]. Bäumer et al. [33] evaluated a commer- Light emitting scintillation detectors can also be used for
cially available MLIC and note a small distortion of the proton dosimetry to offer rapid measurement and high
Bragg peak when using an MLIC with large area paral- spatial resolution. The detectors consist of a scintillating
lel plate ICs. The authors note that the MLICs are likely screen plate coated with a scintillating material which
not suitable for baseline commissioning data acquisition is typically a gadolinium-based material (Gd2O2S:Tb).
The screen is coupled to a charge coupled device (CCD) of the planning target volume (PTV). Protons entering
camera in a light tight box [89]. The scintillation light water or tissue in the same direction will start going in
output is dependent on ionization density and there- slightly different directions by a degree or so by a few
fore LET, the scintillator is also known to experience centimeters of depth due to multiple Coulomb scat-
quenching at high density regions such as in the Bragg tering. Scattering in a material increases as the proton
peak [90]. The rapid turnaround time for image acquisi- slows down. Net scattering is roughly proportional to
tion, high spatial resolution, and sensitivity make these the square-root thickness of a thin sheet of material.
detectors useful for 2D dosimetry at fixed depth [91]. A The net energy loss is directly proportional to the thick-
QA program for modulated PBS can be designed based ness of the material. High Z (atomic number) materials
around commercial versions of these detectors [92] and scatter protons more strongly while low Z materials are
there have also been studies to use the detectors to char- more effective in slowing protons down. Combinations
acterize the nuclear halo effect that can alter PBS field of high and low Z materials are used to control both
size factors especially for small field sizes [9, 30]. energy loss and scattering.
Additionally, cone shaped scintillating screen detec- The simplest passive scattering scheme utilizes a
tors have been developed for isocentric QA and dosim- single scatterer such as lead to maximize scattering
etry. The devices consist of a CCD camera with a light and minimize energy loss. However, this system is not
tight coupling directed into a cone with the interior very efficient, since the fraction of protons within the
coated with a scintillating material. When a narrow useful region of being within ±2.5% the desired dose is
pencil beam intersects with the cone, the light pattern only around 5%, following the Gaussian spread of the
that is produced in the interior scintillator is imaged by scattered proton beam. The transverse dose distribu-
the CCD and used to reconstruct the pencil beam shape tion would also not be flat over this region. In addition,
and path through the cone [93]. Cai et al. [94] developed the protons would lose penetration energy in passing
a QA procedure for image guided proton therapy that through the lead. This would also result in depth with
employs a cone shaped detector that proved useful for a narrow Bragg peak with limited usefulness in provid-
verifying gantry angle, patient couch positioning and ing coverage of larger targets. These limitations can be
imager-radiation isocenter coincidence. overcome by a double scattered system.
To obtain the depth-dose distribution that is useful
25.4 PROTON BEAM DELIVERY TECHNIQUES for providing dose coverage of a target in the direction
There are several beam delivery techniques that are used of the proton beam, a range modulator can be used to
in proton radiotherapy and it is important to identify create an SOBP. The range modulator is typically a rotat-
the influence that each will have on the dosimetry tech- ing object with steps that puts successively thicker lay-
niques we employ. In general, it is necessary for particle ers of material into the beam path [95]. The thickness of
therapy with projectiles much heavier than electrons each layer and the time the beam spends on each step
(e.g. Protons, Helium, Carbon, etc.) to modulate the are adjusted so that the resulting SOBP is flat. In addi-
beam energy in order to spread the Bragg peak width tion to this device, a range shifter can be added in the
over a usable area. Without sufficient energy modula- beam path to provide coverage to more shallow targets
tion the Bragg peak is typically too narrow for effective so that the entire dose distribution is shifted to shallower
treatment of a tumor larger than a few millimeters in depths. Range shifters are typically placed upstream
diameter. This energy modulation is accomplished by (further from the patient) so that it can be smaller.
either varying the beam energy transported into the Double scattering systems (Figure 25.1) were devel-
treatment room in the case of a scanned modality, or in oped to reduce energy loss and improve efficiency in
the case of a scattered modality by rapidly varying the proton delivery. This made the use of larger proton fields
energy in the treatment room through the use of a spin- practical. A double scattering system is comprised of a
ning modulator wheel. uniform first scatterer that produces a Gaussian beam
profile. The nonuniform second scatterer then modifies
25.4.1 Scattered Beams the Gaussian beam profile to produce a flat dose distri-
Passive beam scattering is a technique that was first bution at the target location [95]. One way to design the
developed to cover the cross section and extent in depth second scatterer is to have a flat portion blocked by a
FIGURE 25.1 Double scattered beam delivery system schematic.
cylindrical plug or ring. The more preferred method has worse the further downstream it is located A single scat-
a contoured second scatterer so that the protons from terer with upstream modulator produces the sharpest
the center of the profile are more strongly scattered. The dose, and if a second scatterer is used to generate a large
contoured scatterer causes the center protons to lose field, it should be as far upstream as possible. There is
more energy than the protons along the outer radius so more scattering in the patient, therefore the dose gradient
an appropriately shaped energy compensator needs to will be poor for deep targets no matter the nozzle design.
be added. One drawback of double scattered systems is Large air gaps between the last beam modifying
the sensitivity to beam steering. If the beam is off center device and the patient will result in suboptimal lateral
by even a millimeter on the second scatterer, the dose dose fall off due to scattering. In addition, large air gaps
distribution at the target will be tilted. compromise the accuracy with which the range com-
Double scattered systems require patient-specific pensator deals with tissue inhomogeneities [99]. Distal
hardware for conformation of the dose to the target vol- dose falloff is determined by the Bragg peak. Distal fall
ume. The first piece of hardware is an aperture, or a brass off does not increase with depth in the patient; however,
beam-stop with a hole that is shaped to the outer projec- it does depend on range straggling and will thus increase
tion of the target in the beam’s eye view. This can also be with energy. A better distal fall off will be achieved if a
achieved with a multileaf collimator [96]. With the use beam of sufficient energy is used rather than a higher
of this hardware, there can be two areas of unwanted energy beam that has been degraded.
high dose proximal to the target towards the edges of
the target (thus the edges of the range compensated por- 25.4.2 Scanned Beams
tion) due to the constant range modulation of the field. Recent proton systems have scanned beams. Proton
This is not seen in scanned beams. The treatment plan- charge enables magnetic scanning of narrow pencil
ning system can be used to prepare files for automatic beams. PBS involves a narrow beam of protons enter-
fabrication of patient-specific, field-specific hardware ing the patient at different locations by using magnetic
such as range compensators [97]. These can be designed fields to deflect the beam (Figure 25.2). PBS delivers
to have different goals, such as guaranteeing target cov- dose inside the patient through sequential superposition
erage with respect to alignment errors or patient inter- of many physical pencil beams of small size. Usually
nal organ motion [98]. there is no patient specific hardware or beam modifying
Sharp dose distributions are the benefit of pro- devices involved with scanned beams.
ton beams, but with poorly designed beam modifying Conformal 3D dose distribution is achieved by
devices, resulting dose distributions around the PTV beam scanning. The width of an SOBP can be set along
boundary can be suboptimal. The final aperture closest each pencil beam to the length of the target along that
to the patient results in sharp shadowing or dose fall off. beam’s axis. With this technique, the region of full pre-
The addition of any material after the aperture, such as scription dose is confined to the target since the mod-
a second scatterer, modulator, or range shifter, near the ulation width of the SOBP is varied along the target
patient causes the beam to have less sharp dose fall off. length. SOBP is varied by energy switching. The first
The perturbing effect of any beam-modifying devices is layer of pencil beams delivered has the highest energy,
FIGURE 25.2 Pencil beam scanning system schematic.
covering the most distal edge of the target. Subsequent 25.5 MEASUREMENT TECHNIQUES
layers have lower energy with pencil beam weights Many of the same techniques used for electron beam
that are optimized so that the addition of the entrance dosimetry can also be used for proton beam therapy.
region of the earlier layers of pencil beams in addition Differences arise from the great differential in mass
to the Bragg peaks of lower energy layers add up to the between the two types of particle. The proton is approxi-
prescription dose. mately 2000 times heavier than the electron, and there-
Scanned beams have multiple advantages, making fore is deflected by a much smaller amount from its path
them the preferred proton beam delivery system in the during interaction with the elementary and nuclear par-
modern era. Nearly every proton in the delivery system ticles that make up the medium.
enters the patient, so it is more efficient. This reduces There are many ways to measure proton beams.
activation of material near the patient since the protons Absolute dosimetry can be performed with calorim-
are all deposited within the patient and not stopped by eters, FCs, and ICs, to name a few. Calorimeters are
other materials. Therefore, the neutron dose is mini- devices that measure changes in heat generated per
mized except for those produced in the patient, which unit mass of absorbing material. Corrections for cal-
is an advantage for pediatric patients in reducing the orimeters include a small estimate of the fraction of
induction of secondary cancers [99]. energy that goes into induction of chemical changes
With the advent of PBS, intensity-modulated (IMPT) rather than heat.
fields are achievable. Multiple inhomogeneous dose FCs are measurement devices that are essentially a
fields can be used to deliver a single conformal dose shielded, insulated block that is thick enough to stop the
in the target but spare critical structures [99]. Another protons being measured [46]. They are used to measure
advantage of PBS is the simplification of the mechanical proton fluence. The charge deposited in the block is pro-
design and size of the gantry. Lateral and distal dose fall portional to the number of protons that have stopped
off can be improved with scanning systems. in the block since the proton charge is very accurately
A challenge in beam scanning is the sensitivity to known. With the knowledge of stopping power of the
organ motion during field delivery [100]. Beam spot protons entering the FC and the proton fluence, the dose
scanning is usually applied only once, or a few times, can be computed. Corrections using this device include
over the whole target volume. Therefore, organ motion any change in collected charge due to charged particles
can quickly reduce the homogeneity of the dose in the escaping the stopping material.
target. Well-mobilized tumors, such as those in the head ICs are the most commonly used for absolute dosim-
and neck, spinal cord, and pelvis, were treated first with etry. They are made with a pair of electrodes between
beam scanning. Recently, motion management strate- which is a known amount of gas. As radiation passes
gies and immobilization techniques are utilized in addi- the gas, it is ionized and the applied electric potential
tion to techniques such as dose painting, where the field between the two electrodes will separate the positive
is split and delivered twice or more to ensure that the ions and the electrons. The electrons are collected, and
prescription is not completely dependent on one field. the total charge is measured. For proton pencil beams,
Careful selection of field angles and beam entry points ICs can be large enough to collect the whole beam to
within the patient can be incorporated to minimize the integrate the total dose. Small volume ICs often have
effects of organ motion as well. cylindrical shapes with a central electrode wire or are
made with two small parallel plates, which is useful for diameter of cylindrical chambers should not be larger than
measurements of dose close to the surface. half the width of an SOBP. IAEA TRS-398 recommends
Relative dosimetry is useful for characterizing the dose using a plane-parallel chamber for relative dosimetry.
distribution both laterally and in depth. Furthermore, Beam quality specifiers for proton beams have been
relative dosimetry can be used for beam line monitoring. specified by the effective energy, which is defined as the
Between the accelerator and the beam delivery system, energy of a monoenergetic proton beam having a range
monitors are used to measure the beam intensity and equal to the residual range Rres of the clinical proton
position along the beam line. Typically, these are large- beam. Residual range at a depth z is defined as
area parallel-plate chambers. Machine output can also
be measured with a large parallel-plate IC that is cross- Rres = R p − z (25.4)
calibrated against an absolute dosimeter. Diode arrays,
CCD cameras, films, or scintillation detectors can also where Rp is the practical range and is the depth at which
be used to measure dose distributions. Diode arrays are the absorbed dose beyond the Bragg peak or SOBP falls
particularly useful for lateral dose distributions rather to 10% of its maximum value. Both Rp an Rres have units
than in depth, where they exhibit LET dependencies. of g/cm2. Use of residual range was justified by the small
energy dependence of water/air stopping-power ratios
25.6 DOSIMETRY PROTOCOLS and since the effective energy is close to the maximum
The International Atomic Energy Agency (IAEA) pro- energy in the proton energy spectrum at the reference
vides a series of technical reports (TRS) and TRS-398 depth. Residual range is easily measurable, making it a
provides a code of practice for reference dosimetry good measure of beam quality. This will slightly under-
(beam calibration) for proton beams with energies in estimate the stopping power ratios in the middle of the
the range of 50–250 MeV. In order to do this calibration, SOBP but will not exceed 0.3%.
a factor in terms of absorbed dose to water (ND,w,Q0) for a Beam quality should be measured in water with
dosimeter is needed for a reference beam that has beam either the cylindrical or plane-parallel chamber refer-
quality Q0. For clinical proton beams, cylindrical and ence point at the point of interest and with a clinical
plane-parallel ICs are recommended for use as reference source-to-surface distance (SSD). The field size at the
instruments for calibration. However, the dose to water phantom surface should be 10 × 10 cm2.
uncertainty will be slightly higher for plane-parallel ICs Absorbed dose to water at reference depth (zref ) in a
than for cylindrical chambers due to the higher uncer- proton beam with quality Q and in the absence of the
tainty for pwall (the factor to correct the response of an chamber is given by
IC for the non-medium equivalence of the chamber wall
and any waterproofing material). Cylindrical cham- Dw ,Q = MQ N D ,w ,Q0 kQ ,Q0 (25.5)
bers are however limited to proton beams with quali-
ties at the reference depth Rres ≥ 0.5 g/cm2. Graphite wall where MQ is the reading of the dosimeter with the refer-
chambers are preferred over plastic wall chambers due ence point of the chamber positioned at zref. N D ,w ,Q0 is
to their improved long-term stability and smaller cham- the calibration factor in terms of absorbed dose to water
ber to chamber variations. for the dosimeter at the reference quality Q0 and kQ ,Q0
For cylindrical chambers, the reference point is taken is a chamber-specific factor that corrects for differences
to be on the central axis of the chamber at the center of between the beam quality Q0 and the actual quality
the cavity volume. being used Q. MQ should be corrected for temperature-
Plane-parallel chambers can be used for dosimetry in pressure, electrometer calibration, polarity effects, and
proton beams but must be used for proton beams with ion recombination.
qualities at the reference depth Rres < 0.5 g/cm2. For these,
the reference point is taken to be on the inner surface of 25.7 QUALITY ASSURANCE FOR
the entrance window at the center. The cavity diameter of PROTON THERAPY
a plane-parallel ion chamber or the length of the cavity QA aims to maximize the treatment effect while mini-
of a cylindrical chamber should not be larger than approxi- mizing side effects of the treatment. The QA process
mately half the reference field size. In addition, the outer is continually getting more complex as there are great
advances in technology. QA criteria may be set by regu- Monthly QA procedures will often include measure-
latory agencies at the national or state level. The report ment of system parameters that have a smaller chance
of the American Association of Physicists in Medicine’s of drifting over a short time scale [102]. Typically, these
(AAPM) Task Group number 40, titled “Comprehensive will include output constancy check with a small vol-
QA for Radiation Oncology” is a resource that describes ume parallel plate IC at a benchmarked depth in water
the physical aspects of QA for photon and electron beam or plastic phantom. Additional dosimetry checks for
therapy that can be applied to protons as well [101]. More PBS systems may include imaging and radiation iso-
recently, there has been Task Group report number center alignment checks, spot positioning checks, and
224, entitled “Comprehensive proton therapy machine spot shape verification. A sample test pattern for PBS
quality assurance,” detailing the specific machine QA monthly QA is shown in Figure 25.4. The test pattern
requirements and methods for proton beam delivery was measured in air with a 2D scintillation detector
systems [102]. Computerized treatment planning sys- (Lynx, IBA Dosimetry) in panel (a) and compared to a
tems must undergo routine QA as well. IAEA has pub- baseline measurement in panel (b) through a gamma
lished a technical report on the topic of QA for treatment index evaluation [106].
planning systems, and the AAPM Task Group report
number 53 outlines QA for treatment planning systems 25.8 UNCERTAINTIES
[103, 104]. As in all external beam treatment modalities, dosimetry
Routine QA can be divided by the frequency at for proton beam therapy must take into account a num-
which they should be performed. On a daily basis, ber of uncertainties. Aside from the inherent dosimetry
proton beam output should be checked. Mechanical uncertainties, positioning uncertainties play a signifi-
lasers, snout position alignment, beam delivery safety cant role, and in many cases have a larger risk in proton
interlocks, patient monitoring systems, radiation mon- therapy due to changes in water equivalent path length.
itors, and motion stops should be assessed. Monthly, In most modern uses of proton therapy, the Bragg peak
the beam modulation system should be verified and is used to take advantage of the superior dosimetric
checked. Any beam modifying devices should be properties of protons relative to x-rays, electrons, or neu-
checked as well as imaging devices. Annually, there trons. As discussed in Section 25.2, the positional depth
should be extensive verification of the output under of the Bragg peak depends on both the beam energy
a wide variety of operating conditions, lateral profile and the stopping power of the material through which
flatness and symmetry, the virtual source location, MU the proton beam passes. In the case of a shoot through
linearity, dose per MU, and patient positioning devices technique, the variability of stopping power through the
should be checked for mechanical integrity and coinci- material and the uncertainty in its determination has
dence with the isocenter. much less impact dosimetrically.
Patient-specific QA can be delivered using two This variability in path length is clear when visualiz-
dimensional ion chamber array detectors, such as the ing a proton beam ‘ray’ passing through a complex het-
MatriXX PT (IBA Dosimetry, Germany) [105]. The pla- erogeneous material. When the material, or the patient
nar dose distribution and treatment field profiles can as the case may be, shifts with respect to the ray a dif-
be collected for each field. The measured isodose distri- ferent path length to the terminal end of the Bragg peak
butions (Figure 25.3(a)) can be compared with the pla- will be realized.
nar dose calculated by the treatment planning system Apart from the typical variability that is inherent in
(Figure 25.3(b)). Profiles can be compared along the two external beam therapy, the primary dosimetric uncer-
directions at any point (Figure 25.3(c)). Gamma analy- tainty unique to proton therapy, as opposed to conven-
sis provides information about the measured planar tional treatment modalities (e.g. x-ray and electrons
dose compared to the calculated dose (Figure 25.3(d)). beams), is the uncertainty in the proton stopping power
Different depths can be used to analyze dose; however, as the beam passes through a material.
a typical measurement is at the middle of the SOBP The proton stopping power and subsequently the LET
for each field. This can be achieved by using different rises significantly at the maximum through distal edge
thicknesses of water equivalent plastic on top of the ion of the Bragg peak. Mixing of this high LET in an SOBP
chamber array. which is created through the superposition of multiple
FIGURE 25.3 (a) Measured planar dose distribution with MatriXX 2D ion chamber array. (b) Treatment planning system
calculated planar dose. (c) Profile comparison between measured and planning system calculated dose. (d) Gamma analysis
with 3%/3 mm criterion.
energies causes an elevation of the average LET through- of protons depend strongly on the material where the
out the SOBP relative to the entrance region, Figure 25.5. protons deposit dose and the energy of the protons at
The elevation is most pronounced at the distal edge due the point where they deposit dose. This energy and
to the relative purity of low energy protons in this region, material dependence comes out in the proton stopping
there are fewer contributions from high energy protons powers and is also the primary reason why protons are
as compared to shallower depths. This increased LET, an attractive alternative to high energy x-ray therapy.
increases the biologically effective dose at the distal edge Protons stop in tissue and deposit an increased dose at
of the SOBP and contributes to increased uncertainty in the end of their range. This property can be exploited
the effective proton range [107, 108]. with the newest scanning and range modulation tech-
nologies to precisely sculpt proton dose around a tumor
25.9 SUMMARY with the goal of sparing healthy tissues. A thorough
Dosimetry for proton therapy makes use of many of the understanding of the dosimetric properties of protons
same techniques and equipment that are in widespread and the effects of stopping power variation will aid the
use with other modes of external beam radiotherapy. physicist in choosing appropriate tools and methods
The significant issue to consider when designing a pro- to safely and effectively realize the potential of proton
gram for proton therapy is that the dosimetric properties therapy.
FIGURE 25.4 (a) Measured planar dose planar spot pattern in air with Lynx 2D scintillation detector. (b) Benchmarked 2D
planar measurement. (c) Gamma analysis with 2%/2 mm criterion.
FIGURE 25.5 (a) Proton spread out Bragg peak (SOBP) generated through the superposition of multiple pristine Bragg peaks
with weights chosen to produce a flat topped SOBP. (b) The high LET mixing at points in the distal portion of the SOBP lead
to enhanced relative biological effective dose. (Adapted from [107] (Figure 9).)
REFERENCES 4. W. D. Newhauser and R. Zhang, “The physics of pro-

ton therapy,” Physics in Medicine and Biology 60(8),
1. H. Paganetti, et al., “Relative biological effectiveness R155–R209 (2015). https://doi.org/10.1088/0031-9155/
(RBE) values for proton beam therapy,” International 60/8/R155.
Journal of Radiation Oncology, Biology, Physics 53(2), 5. H. Bethe, “Zur Theorie des Durchgangs schneller
407–421 (2002). Available at: http://www.ncbi.nlm.nih. Korpuskularstrahlen durch Materie,” Annalen der
gov/pubmed/12023146. Physik 397(3), 325–400 (1930). https://doi.org/10.1002/
2. H. Paganetti, “Relative biological effectiveness (RBE) andp.19303970303.
values for proton beam therapy. Variations as a function 6. F. Bloch, “Zur Bremsung rasch bewegter Teilchen
of biological endpoint, dose, and linear energy trans- beim Durchgang durch Materie,” Annalen der Physik
fer,” Physics in Medicine and Biology 59(22), R419–R472 408(3), 285–320 (1933). https://doi.org/10.1002/andp.
(2014). https://doi.org/10.1088/0031-9155/59/22/R419. 19334080303.
3. ICRU, Report 59, Clinical Proton Dosimetry Part I: Beam 7. R. Zhang and W. D. Newhauser, “Calculation of water
Production, Beam Delivery and Measurement of Absorbed equivalent thickness of materials of arbitrary density,
Dose, ICRU (International Commission on Radiation elemental composition and thickness in proton beam
Units and Measurements, Bethesda, MD, 1998). https:// irradiation,” Physics in Medicine and Biology 54(6), 1383–
doi.org/10.1111/j.1440-1673.2001.0944a.pp.x. 1395 (2009). https://doi.org/10.1088/0031-9155/54/6/001.
8. B. Gottschalk, et al., “Multiple coulomb scattering of determine Wair and Ic,” Physics in Medicine and Biology
160 MeV protons,” Nuclear Instruments and Methods in 59(6), 1353–1365 (2014). https://doi.org/10.1088/0031-
Physics Research Section B 74, 467–490 (1993). 9155/59/6/1353.
9. E. Pedroni, et al., “Experimental characterization and 23. S. M. Vatnitsky, J. V Siebers, D. W. Miller, “kQ factors
physical modelling of the dose distribution of scanned for ionization chamber dosimetry in clinical proton
proton pencil beams,” Physics in Medicine and Biology beams,” Medical Physics 23(1), 25–31 (1996).
50(3), 541–561 (2005). https://doi.org/10.1088/0031-9155/ 24. J. Sorriaux, et al., “Consistency in quality correction fac-
50/3/011. tors for ionization chamber dosimetry in scanned pro-
10. F. H. Attix, Introduction to Radiological Physics and ton beam therapy,” Medical Physics 44(9), 4919–4927
Radiation Dosimetry (Wiley, New York, 1986). (2017). https://doi.org/10.1002/mp.12434.
11. G. F. Knoll, Radiation Detection and Measurement, 3rd 25. S. Vatnitsky, et al., “Proton dosimetry intercompari-
edn (Wiley, New York, 2000). Available at: http://he4. son,” Radiotherapy and Oncology 41(2), 169–177 (1996).
phys.tohoku.ac.jp/~booklet/seminar/files/Knoll_con- https://doi.org/10.1016/S0167-8140(96)01800-2.
tents.pdf (Accessed: 14 January 2011). 26. C. Gomà, et al., “Proton beam monitor chamber calibra-
12. H. Fricke and E. Hart, “Chemical dosimetry,” in tion,” Physics in Medicine and Biology 59(17), 4961–4971
Radiation Dosimetry: Instrumentation, edited by F. (2014). https://doi.org/10.1088/0031-9155/59/17/4961.
Attix, W. Roesch, E. Tochilin (Academic Press, New 27. C. Gomà, et al., “Experimental validation of beam
York, 1966), pp. 167–239. quality correction factors for proton beams,” Physics in
13. A. Ableitinger, et al., “Dosimetry auditing procedure Medicine and Biology 60(8), 3207–3216 (2015). https://
with alanine dosimeters for light ion beam therapy,” doi.org/10.1088/0031-9155/60/8/3207.
Radiotherapy and Oncology 108(1), 99–106 (2013). 28. H. Bichsel, “Calculated Bragg curves for ionization
https://doi.org/10.1016/j.radonc.2013.04.029. chambers of different shapes,” Medical Physics 22(11),
14. A. Carlino, et al., “End-to-end tests using alanine dosim- 1721–1726 (1995). https://doi.org/10.1118/1.597535.
etry in scanned proton beams,” Physics in Medicine and 29. H. Palmans, et al., “Dosimetry using plane-parallel ion-
Biology 63(5) (2018). https://doi.org/10.1088/1361-6560/ ization chambers in a 75 MeV clinical proton beam,”
aaac23. Physics in Medicine and Biology 47(16), 2895–2905
15. J. R. Kerns, S. F. Kry, N. Sahoo, “Characteristics of (2002). https://doi.org/10.1088/0031-9155/47/16/305.
optically stimulated luminescence dosimeters in the 30. J. Harms, et al., “Nuclear halo measurements for accu-
spread-out Bragg peak region of clinical proton beams,” rate prediction of field size factor in a Varian ProBeam
Medical Physics 39(4), 1854–1863 (2012). https://doi. proton PBS system,” Journal of Applied Clinical Medical
org/10.1118/1.3693055. Physics 21(1), 197–204 (2020). https://doi.org/10.1002/
16. ICRU, “Report 78, prescribing, recording, and reporting acm2.12783.
proton-beam therapy,” Journal of the ICRU 7(2) (2007). 31. B. Clasie, et al., “Golden beam data for proton pencil-
https://doi.org/10.1093/jicru/ndn001. beam scanning,” Physics in Medicine and Biology 57(5),
17. IAEA, Technical Report Series No. 398, Absorbed Dose 1147–1158 (2012). https://doi.org/10.1088/0031-9155/
Determination in External Beam Radiotherapy, IAEA 57/5/1147.
Technical Report Series (IAEA, Vienna, Austria, 2000). 32. L. Lin, et al., “Experimental characterization of two-
https://doi.org/10.1007/s00259-008-0767-4. dimensional spot profiles for two proton pencil beam
18. C. Gomà, P. Andreo, J. Sempau, “Monte Carlo calcu- scanning nozzles,” Physics in Medicine and Biology
lation of beam quality correction factors in proton 59(2), 493–504 (2014). https://doi.org/10.1088/0031-
beams using detailed simulation of ionization cham- 9155/59/2/493.
bers,” Physics in Medicine and Biology 61(6), 2389–2406 33. C. Bäumer, et al., “Evaluation of detectors for acquisition
(2016). https://doi.org/10.1088/0031-9155/61/6/2389. of pristine depth-dose curves in pencil beam scanning,”
19. ICRU, “ICRU Report 37: Stopping powers for electrons Journal of Applied Clinical Medical Physics 16(6), 151–
and positrons,” Journal of the ICRU (1984). https://doi. 163 (2015). https://doi.org/10.1120/jacmp.v16i6.5577.
org/10.1093/jicru/ndm020. 34. J. T. Lyman, et al., Protocol for Heavy Charged-Particle
20. ICRU, Report 49, Stopping Powers and Ranges for Protons and Therapy Beam Dosimetry: A report of Task Group 20
Alpha Particles (International Commission on Radiation Radiation Therapy Committee American Association
Units and Measurements, Bethesda, MD, 1993). of Physicists in Medicine, New York (1986). Available at:
21. P. Andreo, et al., “Consistency in reference radiotherapy https://mail.aapm.org/pubs/reports/RPT_16.pdf.
dosimetry: Resolution of an apparent conundrum when 35. A. Lourenço, et al., “Experimental and Monte Carlo
60 Co is the reference quality for charged-particle and studies of fluence corrections for graphite calorim-
photon beams,” Physics in Medicine and Biology 58(19), etry in low- and high-energy clinical proton beams,”
6593–6621 (2013). https://doi.org/10.1088/0031-9155/ Medical Physics 43(7), 4122–4132 (2016). https://doi.
58/19/6593. org/10.1118/1.4951733.
22. D. T. Burns, et al., “Use of the BIPM calorimetric and 36. M. Sassowsky and E. Pedroni, “On the feasibility of
ionometric standards in megavoltage photon beams to water calorimetry with scanned proton radiation,”
Physics in Medicine and Biology 50(22), 5381–5400 dosimetry quality assurance,” Physics in Medicine
(2005). https://doi.org/10.1088/0031-9155/50/22/011. and Biology 52(14), 4211–4225 (2007). https://doi.org/
37. H. J. Brede, et al., “Absorbed dose to water determina- 10.1088/0031-9155/52/14/013.
tion with ionization chamber dosimetry and calorime- 50. H. Bouchard, et al., “On the characterization and
try in restricted neutron, photon, proton and heavy-ion uncertainty analysis of radiochromic film dosimetry,”
radiation fields,” Physics in Medicine and Biology 51(15), Medical Physics 36(6), 1931–1946 (2009). https://doi.
3667–3682 (2006). https://doi.org/10.1088/0031-9155/51/ org/10.1118/1.3121488.
15/005. 51. A. Micke, D. F. Lewis, X. Yu, “Multichannel film
38. J. Medin, et al., “Experimental determination of beam dosimetry with nonuniformity correction,” Medical
quality factors, kQ, for two types of Farmer chamber in Physics 38(5), 2523–2534 (2011). https://doi.org/10.1118/
a 10 MV photon and a 175 MeV proton beam,” Physics 1.3576105.
in Medicine and Biology 51(6), 1503–1521 (2006). https:// 52. D. Lewis, et al., “An efficient protocol for radiochromic
doi.org/10.1088/0031-9155/51/6/010. film dosimetry combining calibration and measure-
39. J. Medin, “Implementation of water calorimetry in a ment in a single scan,” Medical Physics 39(10), 6339–
180 MeV scanned pulsed proton beam including an 6350 (2012). https://doi.org/10.1118/1.4754797.
experimental determination of kQ for a Farmer cham- 53. G. Gambarini, et al., “Measurements of 2D distributions
ber,” Physics in Medicine and Biology 55(12), 3287–3298 of absorbed dose in protontherapy with Gafchromic
(2010). https://doi.org/10.1088/0031-9155/55/12/002. EBT3 films,” Applied Radiation and Isotopes 104,
40. A. Sarfehnia, et al., “Direct absorbed dose to water 192–196 (2015). https://doi.org/10.1016/j.apradiso.2015.
determination based on water calorimetry in scanning 06.036.
proton beam delivery,” Medical Physics 37(7), 3541–3550 54. L. Zhao and I. J. Das “Gafchromic EBT film dosim-
(2010). https://doi.org/10.1118/1.3427317. etry in proton beams,” Physics in Medicine and Biology
41. S. Delacroix, et al., “Proton dosimetry comparison involv- 55(10), N291–N301 (2010). https://doi.org/10.1088/0031-
ing ionometry and calorimetry,” International Journal 9155/55/10/N04.
of Radiation Oncology, Biology, Physics 37(3), 711–718 55. B. Arjomandy, et al., “EBT2 film as a depth-dose mea-
(1997). https://doi.org/10.1016/S0360-3016(96)00536-6. surement tool for radiotherapy beams over a wide range
42. L. J. Verhey, et al., “The determination of absorbed dose of energies and modalities,” Medical Physics 39(2), 912–
in a proton beam for purposes of charged-particle radi- 921 (2012). https://doi.org/10.1118/1.3678989.
ation therapy,” Radiation Research 79(1), 34–54 (1979). 56. S. Reinhardt, et al., “Comparison of Gafchromic EBT2
https://doi.org/10.2307/3575020. and EBT3 films for clinical photon,” Medical Physics
43. S. Vynckier, D. E. Bonnett, D. T. L. Jones, “Code of prac- 39(August), 5257–5262 (2012).
tice for clinical proton dosimetry,” Radiotherapy and 57. B. R. Smith, et al., “LET response variability of
Oncology 20(1), 53–63 (1991). https://doi.org/10.1016/ GafchromicTM EBT3 film from a 60Co calibration in
0167-8140(91)90112-T. clinical proton beam qualities,” Medical Physics 46(6),
44. R. Cambria, et al., “Proton beam dosimetry: A compari- 2716–2728 (2019). https://doi.org/10.1002/mp.13442.
son between the Faraday cup and an ionization cham- 58. S. M. Vatnitsky, “Radiochromic film dosimetry for
ber,” Physics in Medicine and Biology 42(6), 1185–1196 clinical proton beams,” Applied Radiation and Isotopes
(1997). https://doi.org/10.1088/0031-9155/42/6/014. 48(5), 643–651 (1997). https://doi.org/10.1016/S0969-
45. S. Lorin, et al., “Reference dosimetry in a scanned pulsed 8043(97)00342-4.
proton beam using ionisation chambers and a Faraday 59. M. Martisíková and O. Jäkel, “Dosimetric properties of
cup,” Physics in Medicine and Biology 53(13), 3519–3529 Gafchromic EBT films in monoenergetic medical ion
(2008). https://doi.org/10.1088/0031-9155/53/13/008. beams,” Physics in Medicine and Biology 55(13), 3741–
46. E. W. Cascio and B. Gottschalk, “A simplified vacuum- 3751 (2010). https://doi.org/10.1088/0031-9155/55/13/011.
less Faraday cup for the experimental beamline at the 60. M. Mumot, et al., “The comparison of doses measured
Francis H. Burr Proton Therapy Center,”IEEE Radiation by radiochromic films and semiconductor detector in a
Effects Data Workshop (2009), pp. 161–165. https://doi. 175 MeV proton beam,” Physica Medica 25(3), 105–110
org/10.1109/REDW.2009.5336294. (2009). https://doi.org/10.1016/j.ejmp.2008.06.001.
47. E. Grusell, et al., “Faraday cup dosimetry in a proton 61. B. Spielberger, et al., “Experimental investigations of the
therapy beam without collimation,” Physics in Medicine response of films to heavy-ion irradiation,” Physics in
and Biology 40(11), 1831–1840 (1995). https://doi. Medicine and Biology 46(11), 2889–2897 (2001). https://
org/10.1088/0031-9155/40/11/005. doi.org/10.1088/0031-9155/46/11/309.
48. A. Niroomand-Rad, et al., “Radiochromic film dosim- 62. S. E. Anderson, et al., “A linear relationship for the LET-
etry: Recommendations of AAPM Radiation Therapy dependence of Gafchromic EBT3 film in spot-scanning
Committee Task Group 55,” Medical Physics 25(11), proton therapy,” Physics in Medicine and Biology 64(5)
2093–2115 (1998). https://doi.org/10.1118/1.598407. (2019). https://doi.org/10.1088/1361-6560/ab0114.
49. M. Fuss, et al., “Dosimetric characterization of 63. C. Carlsson and G. Carlsson, “Proton dosimetry:
GafChromic EBT film and its implication on film Measurement of depth doses from 185-MeV protons by
means of thermoluminescent LiF,” Radiation Research Physics in Medicine and Biology 45(10), 3045–3058
42(2), 207–219 (1970). (2000). https://doi.org/10.1088/0031-9155/45/10/320.
64. S. Vatnitsky, et al., “Application of solid state detectors 77. M. Pacilio, et al., “Characteristics of silicon and dia-
for dosimetry of therapeutic proton beams,” Medical mond detectors in a 60 MeV proton beam,” Physics
Physics 22(4), 469–473 (1995). https://doi.org/10.1118/ in Medicine and Biology 47(8) (2002). https://doi.
1.597608. org/10.1088/0031-9155/47/8/403.
65. M. G. Sabini, et al., “The use of thermoluminescent detec- 78. C. Brusasco, et al., “A dosimetry system for fast measure-
tors for measurements of proton dose distribution,” ment of 3D depth-dose profiles in charged-particle tumor
Radiation Protection Dosimetry 101(1–4), 453–456 (2002). therapy with scanning techniques,” Nuclear Instruments
https://doi.org/10.1093/oxfordjournals.rpd.a006024. and Methods in Physics Research, Section B: Beam
66. J. R. Zullo, et al., “LiF TLD-100 as a dosimeter in high Interactions with Materials and Atoms 168(4), 578–592
energy proton beam therapy – can it yield accurate (2000). https://doi.org/10.1016/S0168-583X(00)00058-6.
results?,” Medical Dosimetry 35(1), 63–66 (2010). https:// 79. D. Nichiporov, et al., “Multichannel detectors for profile
doi.org/10.1016/j.meddos.2009.03.001. measurements in clinical proton fields,” Medical Physics
67. G. S. Ibbott, et al., “Challenges in credentialing institu- 34(7), 2683–2690 (2007). https://doi.org/10.1118/1.2746513.
tions and participants in advanced technology multi- 80. T. Takayanagi, et al., “Dual ring multilayer ionization
institutional clinical trials,” International Journal of chamber and theory-based correction technique for
Radiation Oncology, Biology, Physics 71(1 Suppl.), 71–75 scanning proton therapy,” Medical Physics 43(7), 4150–
(2008). https://doi.org/10.1016/j.ijrobp.2007.08.083. 4162 (2016). https://doi.org/10.1118/1.4953633.
68. C. S. Reft, “The energy dependence and dose response of 81. H. Paganetti and B. Gottschalk, “Test of GEANT3 and
a commercial optically stimulated luminescent detector GEANT4 nuclear models for 160 MeV protons stop-
for kilovoltage photon, megavoltage photon, and elec- ping in CH2,” Medical Physics 30(7), 1926–1931 (2003).
tron, proton, and carbon beams,” Medical Physics 36(5), https://doi.org/10.1118/1.1586454.
1690–1699 (2009). https://doi.org/10.1118/1.3097283. 82. S. Lin, et al., “More than 10 years experience of beam
69. G. O. Sawakuchi, et al., “Determination of average LET monitoring with the Gantry 1 spot scanning proton
of therapeutic proton beams using Al2O3:C optically therapy facility at PSI,” Medical Physics 36(11), 5331–
stimulated luminescence (OSL) detectors,” Physics in 5340 (2009). https://doi.org/10.1118/1.3244034.
Medicine and Biology 55(17), 4963–4976 (2010). https:// 83. I. Rinaldi, et al., “An integral test of FLUKA nuclear mod-
doi.org/10.1088/0031-9155/55/17/006. els with 160 MeV proton beams in multi-layer Faraday
70. D. A. Granville, N. Sahoo, G. O. Sawakuchi, “Calibration cups,” Physics in Medicine and Biology 56(13), 4001–4011
of the Al2O3:C optically stimulated luminescence (OSL) (2011). https://doi.org/10.1088/0031-9155/56/13/016.
signal for linear energy transfer (LET) measurements 84. B. Tesfamicael, et al., “Technical Note: Use of commer-
in therapeutic proton beams,” Physics in Medicine cial multilayer Faraday cup for offline daily beam range
and Biology 59(15), 4295–4310 (2014). https://doi. verification at the McLaren Proton Therapy Center,”
org/10.1088/0031-9155/59/15/4295. Medical Physics 46(2), 1049–1053 (2019). https://doi.
71. E. G. Yukihara, et al., “Time-resolved optically stimulated org/10.1002/mp.13348.
luminescence of Al2O3:C for ion beam therapy dosime- 85. B. Arjomandy, et al., “Use of a two-dimensional ioniza-
try,” Physics in Medicine and Biology 60(17), 6613–6638 tion chamber array for proton therapy beam quality
(2015). https://doi.org/10.1088/0031-9155/60/17/6613. assurance,” Medical Physics 35(9), 3889–3894 (2008).
72. M. R. Raju, “The use of the miniature silicon diode as a https://doi.org/10.1118/1.2963990.
radiation dosemeter,” Physics in Medicine and Biology 86. M. Varasteh Anvar, et al., “Quality assurance of carbon
11(3), 371–376 (1966). https://doi.org/10.1088/0031-9155/ ion and proton beams: A feasibility study for using the
11/3/302. 2D MatriXX detector,” Physica Medica 32(6), 831–837
73. S. M. Vatnitsky, et al., “Dosimetry techniques for nar- (2016). https://doi.org/10.1016/j.ejmp.2016.05.058.
row proton beam radiosurgery,” Physics in Medicine 87. Y. Lin, et al., “Design of a QA method to characterize sub-
and Biology 44(11), 2789–2801 (1999). https://doi. millimeter-sized PBS beam properties using a 2D ion-
org/10.1088/0031-9155/44/11/308. ization chamber array,” Physics in Medicine and Biology
74. A. Fidanzio, et al., “A correction method for diamond 63(10), 0–11 (2018). https://doi.org/10.1088/1361-6560/
detector signal dependence with proton energy,” aabd89.
Medical Physics 29(5), 669–675 (2002). https://doi. 88. L. Lin, et al., “Use of a novel two-dimensional ioniza-
org/10.1118/1.1469634. tion chamber array for pencil beam scanning proton
75. E. Grusell and J. Medin, “General characteristics of the therapy beam quality assurance,” Journal of Applied
use of silicon diode detectors for clinical dosimetry in Clinical Medical Physics 16(3), 270–276 (2015). https://
proton beams,” Physics in Medicine & Biology 45, 2573– doi.org/10.1120/jacmp.v16i3.5323.
2582 (2000). 89. S. N. Boon, et al., “Fast 2D phantom dosimetry for scan-
76. S. Onori, et al., “Dosimetric characterization of silicon ning proton beams,” Medical Physics 25(4), 464–475
and diamond detectors in low-energy proton beams,” (1998). https://doi.org/10.1118/1.598221.
90. S. N. Boon, et al., “Performance of a fluorescent screen compensating bolus and patient Related content A
and CCD camera as a two-dimensional dosimetry sys- pencil beam algorithm for proton dose calculations,”
tem for dynamic treatment techniques,” Medical Physics Physics in Medicine and Biology 34, 1309 (1989).
27(10), 2198–2208 (2000). https://doi.org/10.1118/ 100. M. H. Phillips, et al., “Effects of respiratory motion on dose
1.1289372. uniformity with a charged particle scanning method,”
91. S. Russo, et al., “Characterization of a commercial scin- Physics in Medicine and Biology 37(1), 223 (1992).
tillation detector for 2-D dosimetry in scanned pro- 101. G. J. Kutcher, et al., “Comprehensive QA for radia-
ton and carbon ion beams,” Physica Medica 34, 48–54 tion oncology: Report of AAPM Radiation Therapy
(2017). https://doi.org/10.1016/j.ejmp.2017.01.011. Committee Task Group 40,” Medical Physics 21(4), 581–
92. S. Rana, et al., “Development and long-term stability of 618 (1994). https://doi.org/10.1118/1.597316.
a comprehensive daily QA program for a modern pencil 102. B. Arjomandy, et al., “AAPM task group 224:
beam scanning (PBS) proton therapy delivery system,” Comprehensive proton therapy machine quality assur-
Journal of Applied Clinical Medical Physics 20(4), 29–44 ance,” Medical Physics 46(8), e678–e705 (2019). https://
(2019). https://doi.org/10.1002/acm2.12556. doi.org/10.1002/mp.13622.
93. S. Rana and E. J. J. Samuel, “Feasibility study of utiliz- 103. B. Fraass, et al., “American association of physicists in
ing XRV-124 scintillation detector for quality assur- medicine radiation therapy committee task group 53:
ance of spot profile in pencil beam scanning proton Quality assurance for clinical radiotherapy treatment
therapy,” Physica Medica 66, 15–20 (2019). https://doi. planning,” Medical Physics 25(10), 1773–1829 (1998).
org/10.1016/j.ejmp.2019.09.078. https://doi.org/10.1118/1.598373.
94. W. Cai, et al., “Semi-automated IGRT QA using a 104. International Atomic Energy Agency, “Commissioning
cone-shaped scintillator screen detector for pro- and quality assurance of computerized planning sys-
ton pencil beam scanning treatments,” Physics in tems for radiation treatment of cancer,” Technical
Medicine and Biology 64(8), 0–10 (2019). https://doi. Reports Series 430 (2004).
org/10.1088/1361-6560/ab056d. 105. B. Arjomandy, et al., “Verification of patient-spe-
95. A. M. Koehler, R. J. Schneider, J. M. Sisterson, “Range cific dose distributions in proton therapy using a
modulators for protons and heavy ions,” Nuclear commercial two-dimensional ion chamber array,”
Instruments and Methods 131(3), 437–440 (1975). Medical Physics 37(11), 5831–5837 (2010). https://doi.
https://doi.org/10.1016/0029-554X(75)90430-9. org/10.1118/1.3505011.
96. E. S. Diffenderfer, et al., “Comparison of secondary 106. D. A. Low, et al., “A technique for the quantitative evalu-
neutron dose in proton therapy resulting from the use ation of dose distributions,” Medical Physics 25(5), 656
of a tungsten alloy MLC or a brass collimator system,” (1998). https://doi.org/10.1118/1.598248.
Medical Physics 38(11), 6248–6256 (2011). https://doi. 107. H. Paganetti, “Range uncertainties in proton therapy
org/10.1118/1.3656025. and the role of Monte Carlo simulations,” Physics in
97. M. S. Wagner, “Automated range compensation for pro- Medicine and Biology 57(11), R99–R117 (2012). https://
ton therapy,” Medical Physics 749–752 (1982). https:// doi.org/10.1088/0031-9155/57/11/R99.
doi.org/10.1118/1.595123. 108. A. Carabe, et al., “Range uncertainty in proton ther-
98. P. L. Petti, “New compensator design options for apy due to variable biological effectiveness,” Physics in
charged-particle radiotherapy,” Physics in Medicine and Medicine and Biology 57, 1159–1172 (2012). https://doi.
Biology 42, 1289–1300 (1997). org/10.1088/0031-9155/57/5/1159.
99. L. Hong, M. Goitein, M. Bucciolini, “Distal penetra-
tion of proton beams: The effects of air gaps between
Chapter 26
Ion Range and Dose Monitoring with

Positron Emission Tomography
Katia Parodi
Ludwig-Maximilians-Universität München
Garching b. München, Germany
CONTENTS
26.1 Introduction: In Vivo Dose Verification in Particle Therapy 413
26.2 PET Signal Formation 414
26.3 PET Signal Imaging 415
26.4 PET-Based Range Monitoring 417
26.5 PET-Based Dose Reconstruction 419
26.6 Conclusion And Outlook 421
References 422
26.1 INTRODUCTION: IN VIVO DOSE Bragg peak in the tumor, thus calling for methods of in
VERIFICATION IN PARTICLE THERAPY vivo visualization of the beam range and, ideally, recon-
Light ion beams (i.e., protons and heavier ions up to struction of the applied dose. Such approaches should
charge Z ≈ 10) offer the possibility of a highly precise and be able to provide in vivo confirmation of the successful
(particularly for Z > 1) biologically effective radiation tumor-conformal dose delivery. This would, on the one
therapy. This is especially due to their ability to concen- hand, improve treatment confidence and promote full
trate the dose delivery and, for heavier ions, to reach also clinical exploitation of the Bragg peak and the resulting
at the microscopic level an elevated ionization density dosimetric advantages of ion beams for safe dose esca-
for more effective tumor cell killing in a very localized lation in clinical practice. On the other hand, in vivo
region toward their end of range, so called Bragg peak [1]. monitoring should also provide a prompt feedback on
In combination with recent technological advances such any possible mismatch between the planned and actu-
as pencil beam scanning delivery and intensity modu- ally delivered treatment in order to enable a treatment
lated treatment planning, all these advantages open new adaptation prior to the application of the next treatment
opportunities for dose escalation to the target and/or fraction or, ideally, during the actual beam delivery for
superior sparing of normal tissue and critical organs. an almost real-time intervention (e.g., interruption of
This in turn promises improved clinical outcome for the delivery or, ultimately, on-the-fly treatment adapta-
several tumor indications in comparison to convention). This is deemed especially crucial for the emerging
tional external beam radiotherapy. However, because high-dose hypo-fractionated treatment regimens, where
of the increased physical selectivity, ion beam therapy less or even no subsequent fractions are available for
is also more sensitive than conventional photon radia- compensation of errors. Since the primary ions are to be
tion to uncertainties in treatment planning and deliv- stopped in the tumor, monitoring methods have to rely
ery, which may undermine the correct placement of the on penetrating secondary emissions induced by the ion
413
irradiation, able to emerge from the patient for noninva- typically corresponding to about 1–4 mm residual ion
sive detection. To this end, several monitoring methods range in tissue. In homogenous targets, this results in a
are currently under investigation (see Chapters 27 and 28 characteristic track of β+-activity with an approximately
of this book). This chapter focuses on Positron-Emission- constant or slowly rising slope followed by a rather
Tomography (PET), which still represents the clinically abrupt distal fall-off located few millimeters in front
most extensively investigated approach for a volumetric, of the Bragg-peak (Figure 26.1, left, [2, 3]). For primary
in vivo and noninvasive verification of the actual treat- proton beams, the activation of the traversed tissue is
ment delivery and, in particular, of the ion beam range in mostly due to primary and secondary protons, while
the patient during or shortly after irradiation. for heavier primary ions (1 < Z ≤ 4), also longer ranging
non-positron-emitting projectile fragments contribute a
26.2 PET SIGNAL FORMATION tail of β+-emitting target fragments beyond the Bragg
The unconventional application of the well-established peak (Figure 26.1, middle, [2, 4]).
PET nuclear medicine technique to ion therapy moni- In addition to such a pedestal of β+-activated target
toring exploits the coincident detection of the anni- fragments, heavier ions (Z ≥ 5) can also yield positron-
hilation photon pairs resulting from the β+-decay of emitting projectile fragments through the so called
positron-emitting isotopes formed as a by-product of “auto-activation” mechanism [6]. Being the ion stop-
the therapeutic ion irradiation. This irradiation-induced ping process typically faster than the radioactive decay,
β+-activation provides a surrogate signal of the ion beam these β+-emitting projectile fragments mostly decay
range in tissue, with a varying degree of correlation to after having been slowed down to rest at their end of
the applied dose, due to the different PET signal forma- range. This offers specific advantages for beam range
tion depending on the primary ion species [2]. In fact, the and dose verification, since the β+-active isotopes of
mechanism of β+-activation includes either target frag- the primary ion beam are among the most abundant
mentation only, or formation of both target and projec- reaction products and accumulate shortly before the
tile positron-emitting fragments. Protons and light ions range of the primary stable ions [4], due to the nuclear
up to beryllium can only contribute to the β+-activation reaction kinematics and the different stopping power.
of the target nuclei of the irradiated medium, since they Hence, the auto-activation of heavy ion beams results
either do not fragment at all (protons Z = 1), or cannot in a marked activity maximum located shortly before
produce positron-emitting projectile fragments (Z ≤ 4). the Bragg peak, superimposed onto the minor pedestal
Typical reaction products feature light isotopes such as of β+-active target fragments. This is shown in Figure
11C, 15O, 13N, which exhibit half-lives T
1/2 of about 20, 2 26.1, right, [2, 6] for a primary 12C ion beam, where the
and 10 min, respectively. Target fragmentation occurs activity maximum is originated from the major con-
all along the beam penetration path, as long as the beam tribution of the long-lived 11C and, to a lesser extent, of
energy is above the threshold for nuclear interaction, the short-lived 10C (T1/2 ≈ 20 s).
FIGURE 26.1 Calculated depth dose distribution (dashed line) and corresponding measured β+-activity profile (solid line) for
1H (left), 3He (middle), and 12C (right) irradiation of a homogeneous plastic target of polymethyl methacrylate, PMMA. (Taken
from reference [2], adapted from refs. [3–5].)

Ion Range and Dose Monitoring with Positron Emission Tomography ◾ 415
26.3 PET SIGNAL IMAGING minimizing the signal degradation from uncertainties
After formation, β -active isotopes undergo β -decay
+ + in patient position and washout effects. However, they
at a random time depending on the isotope half-life. are technologically more demanding, requiring dedi-
The emitted positron is slowed down in the surround- cated detector development and fast electronics syn-
ing medium, and most likely annihilates with an atomic chronized with the beam delivery, ideally able to reject
electron (either as a free particle or in a bound state or at least suppress the undesired background of prompt
called positronium), after having (almost) entirely lost radiation, e.g., prompt gamma (see Chapter 27 of this
its kinetic energy, thus resulting in the (quasi) opposite book) during actual beam extraction [15, 16]. Notable
emission of two annihilation gamma quanta of (approx- examples of such customized installations are the tomo-
imately) 511 keV energy each. These energetic photons graphic dual-head PET camera installed at the fixed
have a high probability to escape the patient and can beamline of the Gesellschaft für Schwerionenforschung
be detected by surrounding detector pairs operated in (GSI Darmstadt, Germany [8]) and the planar dual-head
coincidence, according to the standard nuclear medi- camera integrated in the rotating gantry of the National
cine principles of PET imaging [7]. Cancer Center (NCC Kashiwa, Japan [17]). These sys-
As for other monitoring techniques, the low amount tems have been clinically used to image PET activity
of measurable events is among the major challenges induced by actively scanned carbon ions during irra-
for the unconventional application of PET imaging to diation (only in the pauses of the pulsed delivery) and
the verification of ion treatment. In fact, the amount of passively scattered protons immediately after end of
theoretically available signal is limited by the reaction irradiation, respectively. More recent latest-generation
cross sections, resulting in irradiation-induced activity in-beam PET detectors just starting or close to start
concentrations of approximately 0.2–10 kBq/Gy/cm3, clinical operation include the planar dual-head camera
depending on the ion type, time course of irradiation installed at the fixed beamline of the Centro Nazionale
and anatomical site [8–10], for typical fraction doses of di Terapia Oncologica (CNAO, Italy [18]) and the full-
~0.5–2 Gy [2]. In living tissue such activity does not only ring axially-shifted “openPET” design planned for inte-
undergo radioactive decay, but is also subject to a com- gration in one treatment room of the National Institute
plex hot chemistry and can be smeared out by diffusion for Radiological Sciences (NIRS, Japan [19]). In particu-
processes or even dislocated far from the initial produc- lar, the system at CNAO is being evaluated in a first clin-
tion place due to other physiological processes (e.g., per- ical trial for proton and carbon ion beams, and offers
fusion). As a result of this so called biological “washout,” a dedicated data acquisition and image reconstruction
the produced activity reduces more quickly not only due software for visualizing the build-up of activity (cur-
to the physical decay, but also to a so called biological rently with best-quality events only recorded in the
decay, for which half-lives from 2 s up to 10,000 s have pauses of the pulsed delivery) during irradiation at ~10s
been reported, depending on ion species and tissue time resolution [20] (Figure 26.2). However, due to the
type [11–14]. Hence, all these considerations pose severe large development costs and challenges for data acquisi-
requirements to the efficient detection of the irradiation- tion and reliable image reconstruction with unconven-
induced PET signal, ideally at the treatment place. tional detector geometries, such dedicated solutions are
However, different from diagnostic imaging, space still restricted to few specialized institutions and not yet
restrictions from the patient couch, the beam direc- widely available.
tion and additional medical instrumentation constrain Alternative clinically explored solutions relying on
the geometry of the PET scanner to a typical dual-head nuclear medicine instrumentation feature a movable
design for integration at the irradiation place (so called neurological full-ring PET scanner in the treatment
in-beam or on-board). Such on-site installations are in room (in-room, [21]), or fixed installations of full-ring
principle the most appealing solution and the only pos- PET [10, 22] and PET/CT (Compute Tomography) [13, 23]
sibility for an almost real-time monitoring. In fact, they scanners in a separate room in walking distance of the
enable detection of the formed activity directly during treatment room (offline), for imaging from a couple of
irradiation, thereby better preserving its correlation minutes up to ~20 min after irradiation, respectively.
with the delivered dose by recording the major activ- Here, major challenges entail accurate replication
ity contribution from short-lived emitters (e.g., 15O) and and fixation of the treatment position, for which an
FIGURE 26.2 Left panel: The dedicated in-beam PET scanner in the horizontal beamline at CNAO, in measuring position.
Right panel: Treatment plan (with dose colorwash superimposed onto the three orthogonal grayscale CT views) of a patient
treated with scanned proton beams at CNAO (a), and corresponding PET activity (also in colorwash superimposed on the
grayscale planning CT) measured with the dedicated in-beam PET scanner, reconstructed every 60 s and for the entire treat-
ment time with additional 30 s after end of irradiation (from left to right) for two different treatment fractions (top and bottom
row) (b). (Taken from [20].)
additional CT is beneficial to relate treatment and imag- portals for multi-field irradiation. Indeed, conventional
ing anatomy (at the expense of minor additional dose), geometries of full-ring PET scanners can offer improved
as well as loss of counting statistics due to physical and tomographic imaging capabilities compared to custom-
biological decay, requiring an according prolongation ized in-beam geometries, but are challenged by the low
of the measuring time. The latter typically ranges from counting statistics, well below the regimes of standard
5 min for in-room implementations, up to 30 min for tracer imaging for which such scanners and reconstruc-
offline solutions. Besides the reduction of counting tion algorithms are designed [24].
statistics, an additional drawback of the time elapsed According to the typical detector technologies (with
between activity formation and imaging is the already pixelization of ~4–6 mm), along with the non-perfect
mentioned degradation of spatial correlation between co-linearity of the annihilation photons (~0.3° FWHM)
the detectable PET signal and the original positron and the finite positron range (few mm in tissue, depend-
emitter production due to biological washout, which can ing on the endpoint energy of the considered isotope),
directly impair the reliability of the PET signal as a sur- PET imaging typically exhibits spatial resolutions of
rogate of the delivered treatment dose. Moreover, imag- approximately 4–5 mm FWHM in the center of the
ing after completion of a given treatment fraction can scanner field-of-view [2]. However, such coarse resolu-
only provide the PET signal produced by the integral tion does not prevent the potential ability of mm-range
activation of all the applied treatment fields. Unless spe- verification accuracy, as experimentally demonstrated
cial data processing techniques are employed to disen- in controlled phantom studies with the GSI in-beam
tangle the inevitably overlapped signals, this limitation PET installation of moderate spatial resolution, as
has so far restricted the range verification abilities of long as sufficient counting statistics is available [25].
such a post-irradiation method to non-opposing beam Nevertheless, further improvements are expected by
ongoing developments in PET imaging technologies the energy dependence of the underlying cross section
both at the hardware (e.g., smaller crystal size, ultra-fast for positron emitter production.
timing performance) and software (e.g., by incorporat- Regardless of the level of correlation, range informa-
ing the spatially dependent detector response in the tion can be obtained by comparing the activity distal-off
image reconstruction process) level. position of two measurements from different treatment
fractions [17], or between a measurement and an expec-
26.4 PET-BASED RANGE MONITORING tation, which can be obtained from an analytical [28–
Owing to the described different β+-activity formation 30] or stochastic (Monte Carlo) [23, 31, 32] calculation.
processes and the basic physical difference between PET The former approach can only provide a reproducibility
activation (due to nuclear interactions) and energy depo- check, without ensuring that the range inferred from
sition (dominated by Coulomb inelastic interactions), the repeated PET measurements accurately matches the
there is a varying degree of correlation between the irra- one intended at the treatment planning stage. Moreover,
diation-induced PET signal and the beam range, typi- due to the above-mentioned considerations, it may be
cally defined as a certain percentage position of the dose sensitive to the reproducibility of the timing of irra-
fall-off. In particular, light ions limited to β+-activation diation and data acquisition, which can also change
through target fragmentation only exhibit a weak spa- between treatment fractions. On the contrary, the com-
tial correlation between the depth-distributions of dose parison of a measurement to a calculated prediction can
and positron emitters, whereby beam range information enable an accuracy check of the intended delivery, pro-
can be inferred from the position of the distal activity vided that the PET computational engine shares similar
fall-off, typically few millimeters before the beam range physical beam models as the treatment planning sys-
(Figure 26.1, left, middle panel). For carbon ions, the tem, particularly the same CT-range calibration curve
correlation becomes more evident due to the peaked sig- used to transfer the well-characterized beam range in
nal in both distributions (Figure 26.1, right). However, water to the patient tissue. However, in comparison to
this improved correlation is somewhat outweighed by the dose calculation engine, more stringent require-
the lower amount of activity yield at the same range ments apply to the relevant nuclear models and/or used
and delivered dose for carbon ions against protons, as cross sections for positron emitter production, which
experimentally shown by [26], where a factor of about play a crucial role in the PET formation process, while
1/3 was observed for irradiation in a plastic (polymethyl having only a negligible contribution on the dose cal-
methacrylate, PMMA) target. The reason for this lies culation accuracy. Moreover, such a prediction of the
in the fact that a much lower fluence of carbon ions is irradiation-induced activity must also correctly model
required to compensate for their higher energy loss rate the patient-specific treatment plan, the fraction-specific
(and enhanced biological effectiveness, when making time course of irradiation and PET data acquisition, as
the comparison at the same biological dose) with respect well as the detector-dependent image formation pro-
to protons. The correlation is also sensitive to the tim- cess [13, 31]. Here, a further challenge specific to the
ing of the irradiation and imaging, which influences the PET calculation engine is the correct representation
amount of detected decays from the different produced of the patient anatomy also in terms of tissue compo-
isotopes, depending on their exact time of creation sition, particularly with respect to the main elements
(based on the irradiation sequence), their half-life and (e.g., carbon, oxygen) contributing to the target acti-
the acquisition time window. In particular, in case of the vation, especially for the light ions subject only to this
peaked activity signal for heavy ions, the peak location activation mechanism [33, 34]. Moreover, especially for
can be changing over time depending on the contribu- in-room and offline approaches, the prediction model
tion of different β+-emitting projectile fragments, as should also describe the above-mentioned washout pro-
experimentally observed for 16O ion beams [27], while cesses, which are typically accounted for with different
less relevant for primary 12C ion beams where the peak components (slow, medium, fast) of biological decay,
is dominated by 11C. For the lighter ions subject to β+- with tissue- (and ideally also ion-) specific parameters
activation through target fragmentation only, the fall- [12–14, 23, 34]. Finally, both reproducibility and accu-
off position strongly depends on the tissue composition, racy checks demand reliable methods to compare the
determining the type of β+-active target fragment, and two PET distributions in terms of their fall-off positions.
To this end, different methods have been proposed over performed [17]. Although it might be argued that such
the last decades, starting from simple point-like analy- anatomical and positional changes should be nowadays
sis of certain percentage levels of the distributions [13, detected more easily with in-room x-ray based volu-
35], up to more sophisticated and automated approaches metric image-guidance prior to treatment, the men-
making use of the entire distal fall-off, e.g., with analysis tioned on-site PET-based approaches did not deliver
of relative shifts in beam-eye-view or correlation coef- any additional radiation exposure, as using the patient
ficients in selected distal 3D regions of interest [36–39]. model based on the planning CT. Moreover, in-beam
For the extensive pioneering clinical experience of PET imaging could also detect transient modifications
actively scanned carbon ion and passively scattered which might only occur during treatment, and there-
proton irradiation monitored by the dedicated in- fore be missed by pre- or post-treatment imaging.
beam and on-board instrumentation at GSI and NCC, With the more recent in-room and offline PET(/CT)
respectively, no detailed quantitative range analysis has monitoring experiences for passively scattered as well as
been reported. However, several published results from actively scanned proton and carbon ion irradiation at
these centers indicated the ability of on-site PET moni- different facilities, quantitative range analysis showed
toring to detect deviations from the planned treatment that while reproducibility better than 1 mm could be
(via accuracy checks at GSI, [8]) and inter-fractional observed (e.g., for glioma patients treated with both ion
changes (via both accuracy checks at GSI [3, 40] and species [42]), accuracy was typically limited to within
reproducibility checks at NCC [17]). In particular, the 1–5 mm for most of the examined cases [13, 39, 42–45].
availability of the in-beam PET monitoring at GSI trig- However, this is mostly due to the remaining challenges
gered improvements of the CT-range calibration curve of having an accurate prediction model, given the con-
in the early phase of the pilot clinical project [41], due siderable uncertainties in the nuclear models/cross sec-
to systematic deviations observed between PET mea- tion data and patient composition for estimation of the
surements and predictions in the first treated fractions physical activity production as well as limitations of the
[8]. PET-based detection of inter-fractional changes was available washout models and related parameters, in
mostly ascribed to varying patient anatomy or position, addition to co-registration issues and statistical noise
which introduced density modifications in the beam of the imaged weak and broad activity concentrations.
path resulting in the detected under- or over-range Hence, improved results can be expected from the lat-
(Figure 26.3) [3, 17, 40]. In case of substantial anatomi- est-generation dedicated in-beam PET scanners, for
cal changes, as later confirmed by an additional CT at which initial promising phantom experiments at clini-
NCC triggered by their regular PET-based reproduc- cal doses suggested range retrieval capabilities within
ibility check, an adaptation of the treatment plan was 1–2 mm [46].
FIGURE 26.3 Example of discrepancy between measured (left) and predicted (right) activity (shown as colored isolines super-
imposed onto the grayscale planning CT), detected with the dedicated in-beam PET camera at GSI for a patient treated with
a scanned carbon ion beam entering the patient from the right (left in the picture). Arrows indicate mismatches both in the
distal maximum penetration depth of activity (right arrow), as well as within the nasal cavity, where a reduction of tissue has
been suspected (left arrow). (Taken from [3].)
26.5 PET-BASED DOSE RECONSTRUCTION most plausible modification served as the basis for per-
The ultimate clinical exploitation of PET monitoring forming a new dose calculation using the GSI treatment
should ideally provide a solution to the challenging ill- planning engine, to quantify the dosimetric deviation
posed inverse problem of retrieving the actual dose deliv- with respect to the original treatment plan and thus sup-
ery from the measured activity, as already envisioned in port physicians in making decisions whether to under-
one of the first pioneering studies on in vivo visualiza- take additional investigations (e.g., repeated CT in case
tion of ion irradiation-induced activation [11]. A practi- of a suspected anatomical change) prior to the delivery of
cal approximate solution to this very complex problem the next treatment fraction (Figure 26.4, [3]). Although
was first implemented in the pilot project at GSI, in order this method proved useful in some of the encountered
to provide the physicians with a rough estimation of the cases at GSI, as reported in [3, 8, 40], it cannot provide
dosimetric consequences associated to the observed dis- an accurate dose quantification nor can be performed
crepancies between measured and expected in-beam for each individual patient in the clinical routine.
PET signal [3, 40]. To this end, the proposed PET-guided More rigorous mathematical approaches were more
dose quantification relied on a trial-and-error approach, recently proposed, which rely on analytical models of
which iteratively manipulated the planning CT (e.g., to the correlation between the dose and the PET signal,
mimic plausible anatomical changes, such as nasal cav- and its inversion. More extensive investigations on
ity filling) or the treatment plan (e.g., change of isocen- this topic have been pursued for proton therapy, where
ter or beam angle to imitate translational or rotational the limitation to target fragments as well as the domi-
patient positional changes), performed a fast (~10–30 nant activation mechanism through primary and sec-
seconds) recalculation of the positron emitter distribu- ondary protons simplifies the relationship between
tions and, on demand, an analytical fast image recon- PET and dose. In fact, both quantities directly depend
struction, until the trend observed in the experimental on the proton energy distribution, weighted with the
data could be reproduced [3]. In this case, the identified energy dependent mass stopping power in the case of
FIGURE 26.4 Planned dose distribution (colored isolines) calculated on the initial CT shown in grayscale (a), and its recalcu-
lation on the shown artificially manipulated CT based on the PET observations of figure [3] (b), both delivering a maximum
physical dose of about 650 mGy. The difference of the two distributions results in minimum and maximum dose changes of
−200 mGy and 495 mGy, respectively, fortunately located in uncritical anatomical regions (c). A new CT taken shortly after-
wards (d) confirmed a very similar anatomical modification as conjectured (cf. CT shown in (b)). (Adapted from [3].)
the dose, or energy dependent cross section for the posi- steps [53]. Hence, the method does not provide a direct
tron emitter production contributing to the resulting absolute dose quantification, and heavily relies on the
PET signal. Different groups proposed analytical formu- accuracy of several physical quantities (e.g., reaction
lations of this relationship either with so called positron cross sections) used for the analytical calculation in a
emitters’ species matrices [47] or filter functions [48, 49], given target. Moreover, despite the promising results, it
specific for each given β+-active isotope and calculated has so far only been demonstrated with measured 1D
on the basis of analytical expressions or fits to Monte data for monoenergetic (12C, 14N, 16O, and 20Ne) and pas-
Carlo simulated data, respectively. Inversion of the rela- sively modulated (12C and 16O) ion delivery to homoge-
tionship for dose reconstruction was based on decon- neous targets, imaged for 500 s starting immediately
volution methods of the 1D–3D activity distribution as after end of irradiation with an experimental dual-head
imaged with a PET scanner, applied to both in silico data in-beam positron camera [52]. Interestingly, the method
for homogenous/heterogeneous phantoms and a clinical has been also applied to monoenergetic and spread-out
dataset, as well as offline PET measurements of passively Bragg peak proton irradiation in [53, 54]. Very recently,
scattered proton irradiation of an heterogeneous phan- another approach was proposed to predict dose distri-
tom and a clinical cranial indication [47, 49]. In the case butions from PET images in carbon ion therapy [55]. It
of heavier ion beams like 12C, the relationship becomes relies on a similar assumption as in [48], that the one-
more complicated due to the presence of both projectile dimensional PET distribution can be described as a
and target β+-active fragments, including activity pro- convolution of the depth dose distribution with a filter
duction through long ranging, non-positron-emitting kernel. An evolutionary algorithm was then introduced
fragmentation products, which especially contribute a to invert the relationship and predict the depth dose dis-
distal tail beyond the activity and dose maximum. Here, tribution from a measured PET distribution. Although
the first approach proposed for dose quantification of the filter kernels could be obtained from a previously
12C and 16O ions relied on the so called maximum likeli- calculated library, for this work they were created on-
hood estimation method, which was introduced in [50, the-fly, using Monte Carlo simulated predictions of the
51] to estimate the position of the range (for monoener- β+-decay and depth dose distributions, and the same
getic beams) as well as of the distal and proximal edges evolutionary algorithm. The approach was tested for
of a spread-out Bragg peak (for more realistic clinical monoenergetic and polyenergetic carbon ion irradiation
beams) given a certain distribution of measured posi- of homogenous and heterogenous targets, simulated as
tron annihilations. The most probable positions are well as imaged during (only in the pauses of the pulsed
determined by maximizing the likelihood between the delivery) and up to at least 30 min after irradiation with
measured data set and an expectation [50–53]. The latter the openPET scanner at NIRS. Moreover, a first attempt
is calculated starting from an analytical model of the for in silico clinical data simulating a broad monoener-
spatial distribution of positron emitters produced in a getic irradiation of a patient head was also reported. The
target, using parametrized reaction cross sections. Their method proved particularly very reliable for confirma-
measurable annihilation events are then calculated tion of the beam range, with an accuracy of 0.5 (simula-
analytically taking into account the individual isotope tion) and 1.0 mm (measurement) [55]. Moreover, very
half-lives as well as the timing of irradiation and imag- good agreement was also obtained for the dose recon-
ing. Finally, the detection process is modeled as the construction when using consistent simulated data in rather
volution with a position-dependent detector response, homogenous media, including the considered clinical
as obtained from Monte Carlo simulation. Since the case, while larger uncertainties and oscillations were
positron emitter calculation entails all quantities also observed for application to heterogeneous phantoms
needed for determination of the shapes of the physical with sharp interfaces as well as all experimental data.
and biological dose distributions, once the most proba- The latter finding was especially ascribed to limitations
ble positions are determined, also the shapes of the most of the physics models of the used Monte Carlo simula-
probable underlying dose distributions can be estimated tions, which failed to perfectly reproduce the measured
[52, 53]. A final calibration step is then undertaken activity distributions, thus resulting in inaccurate filter
using measured data from the beam monitor system functions, which generated unavoidable inconsistencies
during the irradiation, as well as additional calculation in the deconvolution process. Moreover, similar to the
FIGURE 26.5 PET-based dose reconstruction for a head-and-neck patient treated with passively scattered proton delivery. The left
panel shows the treatment planning dose in colorwash, superimposed onto the grayscale planning CT image. The top row shows
the predicted activity distribution, calculated according to the filter formalism of [48] (left) and the reconstructed dose (right). The
bottom row shows the PET signal measured offline (left) and the resulting dose reconstruction (right). (Reprinted with permission
from [49]. ©Institute of Physics and Engineering in Medicine. Reproduced by permission of IOP Publishing. All rights reserved.)
work of [50–54], the method was so far only used in 1D, more reliable direct PET-based dose reconstruction in
either applied to laterally integrated distributions, or by the future.
assembling data from 1D deconvolved depth dose dis-
tributions along each voxel line in beam direction for 26.6 CONCLUSION AND OUTLOOK
the 3D in silico clinical dataset (as also done in [49]). Despite the generally discussed shortcomings of PET
Hence, full extension of the method to 3D still needs to monitoring as relying on an intrinsically “delayed”
be accomplished. (depending on the isotope half-life) emission and being
In summary, although the different proposed meth- sensitive to biological washout, it still represents the
ods generally show very encouraging results with dose most mature technique for 3D (and even 4D [56]) imag-
agreement in some cases reaching accuracy within 2% ing with well-proven technologies for clinical imple-
[47] and mean relative errors (normalized to the maxi- mentation at relatively moderate efforts. As shown by
mum) below 1% [55], especially in the in silico studies, the initial experience at different institutions with differ-
the so far only reported attempt to directly solve the ent ion species and beam delivery techniques, promising
dose quantification problem for offline PET data of clinical results could be obtained even with suboptimal
a real clinical case showed considerable issues due to implementations, at least for favorable anatomical loca-
the very low counting statistics and biological washout tions. Following these promising results, latest-gener-
effects (Figure 26.5, [49]). Hence, improvements of the ation in-beam PET scanners were recently developed,
PET image quality with online imaging workflows at optimized for this specific application. Together with
next-generation in-beam detectors, along with a better the current clinical trends of hypo-fractionation regi-
understanding of remaining uncertainties in the knowl- mens, which call for increased fractional dose deliv-
edge of the reaction cross-sections, tissue stoichiometry ery, all these advances should help mitigate the major
and washout parameters will be crucial for enabling a encountered drawbacks of low counting statistics and
biological washout. Moreover, several activities are 2. K. Parodi, “In vivo dose verification,” in The Physics of
ongoing worldwide to improve the knowledge of cross Proton Therapy, edited by H. Paganetti (CRC Press), 1st
edition.
section data (e.g., [57, 58]) and reliability of the models
3. K. Parodi, On the feasibility of dose quantification with
of the used computational algorithms for positron emit- in-beam PET data in radiotherapy with 12C and proton
ter production, typically relying (directly or indirectly) beams, Ph.D. Thesis, Dresden University of Technology;
on Monte Carlo methods (e.g., [59, 60]). Also, different 2004, in Forschungszentrum Rossendorf Wiss-Techn-
imaging techniques such as dual-energy CT and mag- Ber FZR-415 (2004).
netic resonance imaging, which are receiving increas- 4. F. Fiedler, P. Crespo, K. Parodi, M. Sellesk, W. Enghardt,
“The feasibility of in-beam PET for therapeutic beams
ing attention in the context of ion beam therapy, might of 3He,” IEEE Transactions on Nuclear Science 53, 2252–
provide improved patient models in comparison to con- 2259 (2006).
ventional CT for better tissue classification and more 5. J. Pawelke, T. Bortfeld, F. Fiedler, T. Kluge, D. Möckel, K.
correct assignment of elemental composition, as sug- Parodi, F. Pönisch, G. Shakirin, W. Enghardt, “Therapy
gested by the first promising studies of [33, 34] for the monitoring with PET techniques,” in Proceedings of the
Ion Beams in Biology and Medicine (IBIBAM) Meeting,
brain region. And while the results of the ongoing clini-
Heidelberg, Germany, September 2007 (TÜV Media
cal trials, e.g., at CNAO, will allow better assessing the GmbH, Köln, 2007); ISBN 978-3-8249-1071-7, pp. 97–105.
full potential of PET when using appropriate in-beam 6. C. A. Tobias, E. V. Benton, M. P. Capp, A. Chatterjee, M.
instrumentation specifically tailored to this application, R. Cruty, R. P. Henke, “Particle radiography and autoac-
there are also interesting developments in the nuclear tivation,” International Journal of Radiation Oncology,
Biology, Physics 3, 35–44 (1997).
medicine PET community which might considerably
7. D. W. Townsend, “Physical principles and technology
impact the capability of PET-based ion therapy moni- of clinical PET imaging,” Annals of the Academy of
toring in the future. In particular, ongoing developments Medicine of Singapore 33, 133–145 (2004).
toward next-generation detectors of ultrafast time-of- 8. W. Enghardt, P. Crespo, F. Fiedler, et al., “Charged
flight capabilities with coincidence time resolution below hadron tumour therapy monitoring by means of PET,”
10 ps [61] might open the perspectives of direct, event- Nuclear Instruments and Methods in Physics Research
Section A 525, 284–288 (2004).
by-event reconstruction of the activity produced dur- 9. K. Parodi, H. Paganetti, E. Cascio, J. Flanz, A. Bonab,
ing the irradiation, with minimal degradation of image N. Alpert, K. Lohmann, T. Bortfeld, “PET/CT imaging
quality in the case of limited angle geometries [62] and, for treatment verification after proton therapy – A study
with proper data acquisition, real-time treatment and with plastic phantoms and metallic implants,” Medical
range monitoring during actual “beam-on.” Additional Physics 34, 419–435 (2007).
10. S. Vynckier, S. Derreumaux, F. Richard, et al., “Is it pos-
intriguing proposed approaches either rely on the possi- sible to verify directly a proton-treatment plan using
bility of exploiting irradiation-induced millisecond short- positron emission tomography?,” Radiotherapy &
lived positron emitters like 12N (T1/2 = 11 ms) for so called Oncology 26, 275–277 (1993).
“beam-on PET” aiming to provide real-time feedback 11. G. W. Bennett, J. O. Archambeau, B. E. Archambeau, J.
[63], or to resort to the initial idea of using β+-radioactive I. Meltzer, C. L. Wingate, “Visualization and transport
of positron emission from proton activation in vivo,”
beams for irradiation [64], now that facilities able to pro-
Science 200, 1151–1153 (1978).
duce sufficiently high intensities for clinical irradiation 12. H. Mizuno, T. Tomitami, M. Kanazawa, et al., “Washout
are on the horizon [65]. Although limited to heavy ions measurements of radioisotopes implanted by radio-
(e.g., 11C and 15O), this method would offer a much more active beams in the rabbit,” Physics in Medicine and
straightforward correlation between the PET activity Biology 48, 2269–2281 (2003).
13. K. Parodi, H. Paganetti, H. A. Shih, et al., “Patient study
maximum and the beam range [66], along with a con-
on in-vivo verification of beam delivery and range using
siderable increase of counting statistics [67], thereby also PET/CT imaging after proton therapy,” International
simplifying the challenging task of dose reconstruction. Journal of Radiation Oncology, Biology, Physics 68, 920–
934 (2007).
REFERENCES 14. C. Ammar, K. Frey, J. Bauer, C. Melzig, S. Chiblak, M.
Hildebrandt, D. Unholtz, C. Kurz, S. Brons, J. Debus, A.
1. M. Durante, R. Orecchia, J. S. Loeffler, “Charged- Abdollahi, K. Parodi, “Comparing the biological wash-
particle therapy in cancer: Clinical uses and future out of β+-activity induced in mice brain after 12C-ion
perspectives,” Nature Reviews Clinical Oncology 14, and proton irradiation,” Physics in Medicine and Biology
483–495 (2017). 59, 7229–7244 (2014).
15. K. Parodi, P. Crespo, H. Eickhoff, T. Haberer, J. Pawelke, 25. K. Parodi, F. Pönisch, W. Enghardt, “Experimental
D. Schardt, W. Enghardt, “Random coincidences during study on the feasibility of in-beam PET for accurate
in-beam PET measurements at microbunched thera- monitoring of proton therapy,” IEEE Transactions on
peutic ion beams,” Nuclear Instruments and Methods in Nuclear Science 52, 778–786 (2005).
Physics Research Section A 545, 446–458 (2005). 26. K. Parodi, W. Enghardt, T. Haberer, “In-beam PET
16. G. Sportelli, N. Belcari, N. Camarlinghi, G. A. Cirrone, G. measurements of β+-radioactivity induced by pro-
Cuttone, S. Ferretti, A. Kraan, J. E. Ortuño, F. Romano, ton beams,” Physics in Medicine and Biology 47, 21–36
A. Santos, K. Straub, A. Tramontana, A. Del Guerra, V. (2002).
Rosso, “First full-beam PET acquisitions in proton ther- 27. F. Sommerer, F. Cerutti, K. Parodi, A. Ferrari, W.
apy with a modular dual-head dedicated system,” Physics Enghardt, H. Aiginger, “In-beam PET monitoring of
in Medicine and Biology 59, 43–60 (2014). mono-energetic (16)O and (12)C beams: Experiments
17. T. Nishio, A. Miyatake, T. Ogino, et al., “The develop- and FLUKA simulations for homogeneous targets,”
ment and clinical use of a beam ON-LINE PET system Physics in Medicine and Biology 54, 3979–3996 (2009).
mounted on a rotating gantry port in proton therapy,” 28. A. Miyatake, T. Nishio, T. Ogino, “Development of
International Journal of Radiation Oncology, Biology, activity pencil beam algorithm using measured distri-
Physics 76, 277–286 (2010). bution data of positron emitter nuclei generated by pro-
18. M. G. Bisogni, A. Attili, G. Battistoni, N. Belcari, N. ton irradiation of targets containing (12)C, (16)O, and
Camarlinghi, P. Cerello, S. Coli, A. Del Guerra, A. (40)Ca nuclei in preparation of clinical application,”
Ferrari, V. Ferrero, E. Fiorina, G. Giraudo, E. Kostara, Medical Physics 38, 5818–5829 (2011).
M. Morrocchi, F. Pennazio, C. Peroni, M. A. Piliero, 29. K. Frey, J. Bauer, D. Unholtz, et al., “TPSPET-A TPS-
G. Pirrone, A. Rivetti, M. D. Rolo, V. Rosso, P. Sala, based approach for in vivo dose verification with PET
G. Sportelli, R. Wheadon, “INSIDE in-beam positron in proton therapy,” Physics in Medicine and Biology 59,
emission tomography system for particle range moni- 1–21 (2014).
toring in hadrontherapy,” Journal of Medical Imaging 4, 30. M. Priegnitz, F. Fiedler, D. Kunath, et al., “An exper-
011005 (2017). iment-based approach for predicting positron emitter
19. H. Tashima, E. Yoshida, N. Inadama, F. Nishikido, distributions produced during therapeutic ion irradia-
Y. Nakajima, H. Wakizaka, T. Shinaji, M. Nitta, S. tion,” IEEE Transactions on Nuclear Science 59, 77–87
Kinouchi, M. Suga, H. Haneishi, T. Inaniwa, T. Yamaya, (2012).
“Development of a small single-ring OpenPET proto- 31. F. Ponisch, K. Parodi, B. G. Hasch, W. Enghardt, “The
type with a novel transformable architecture,” Physics modelling of positron emitter production and PET
in Medicine and Biology 61, 1795–809 (2016). imaging during carbon ion therapy,” Physics in Medicine
20. V. Ferrero, E. Fiorina, M. Morrocchi, F. Pennazio, G. and Biology 49, 5217–5232 (2004).
Baroni, G. Battistoni, N. Belcari, N. Camarlinghi, M. 32. K. Parodi, A. Ferrari, F. Sommerer, H. Paganetti,
Ciocca, A. Del Guerra, M. Donetti, S. Giordanengo, G. “Clinical CT-based calculations of dose and posi-
Giraudo, V. Patera, C. Peroni, A. Rivetti, M. D. D. R. tron emitter distributions in proton therapy using the
Rolo, S. Rossi, V. Rosso, G. Sportelli, S. Tampellini, F. FLUKA Monte Carlo code,” Physics in Medicine and
Valvo, R. Wheadon, P. Cerello, M. G. Bisogni, “Online Biology 52, 3369–3387 (2007).
proton therapy monitoring: Clinical test of a Silicon- 33. B. Berndt, G. Landry, F. Schwarz, T. Tessonnier, F. Kamp,
photodetector-based in-beam PET,” Scientific Reports 8, G. Dedes, C. Thieke, M. Würl, C. Kurz, U. Ganswindt,
4100 (2018). F. Verhaegen, J. Debus, C. Belka, W. Sommer, M. Reiser,
21. X. Zhu, S. España, J. Daartz, N. Liebsch, J. Ouyang, H. J. Bauer, K. Parodi, “Application of single- and dual-
Paganetti, T. R. Bortfeld, G. El Fakhri, “Monitoring energy CT brain tissue segmentation to PET monitor-
proton radiation therapy with in room PET imaging,” ing of proton therapy,” Physics in Medicine and Biology
Physics in Medicine and Biology 56, 4041–4057 (2011). 62, 2427–2448 (2017).
22. Y. Hishikawa, K. Kagawa, M. Murakami, et al., 34. J. Bauer, W. Chen, S. Nischwitz, J. Liebl, S. Rieken, T.
“Usefulness of positron-emission tomographic images Welzel, J. Debus, K. Parodi, “Improving the model-
after proton therapy,” International Journal of Radiation ling of irradiation-induced brain activation for in vivo
Oncology, Biology, Physics 53, 1388–1391 (2002). PET verification of proton therapy,” Radiotherapy &
23. J. Bauer, D. Unholtz, F. Sommerer, et al., “Implementation Oncology 128, 101–108 (2018).
and initial clinical experience of offline PET/CT-based 35. C. H. Min, X. Zhu, K. Grogg, G. El Fakhri, B. Winey,
verification of scanned carbon ion treatment,” H. Paganetti, “A recommendation on how to analyze in-
Radiotherapy & Oncology 107, 218–226 (2013). room PET for in vivo proton range verification using
24. C. Kurz, J. Bauer, M. Conti, L. Guérin, L. Eriksson, K. a distal PET surface method,” Technology in Cancer
Parodi, “Investigating the limits of PET/CT imaging at Research & Treatment 14, 320–325 (2015).
very low true count rates and high random fractions 36. A. Knopf, K. Parodi. T. Bortfeld, H. A. Shih, H.
in ion-beam therapy monitoring,” Medical Physics 42, Paganetti, “Systematic analysis of biological and physi-
3979–3991 (2015). cal limitations of proton beam range verification with
offline PET/CT scans,” Physics in Medicine and Biology verification of proton radiotherapy,” Physics in Medicine
54, 4477–4495 (2009). and Biology 51, 1991–2009 (2006).
37. S. Helmbrecht, A. Santiago, W. Enghardt, P. Kuess, F. 49. S. Remmele, J. Hesser, H. Paganetti, T. Bortfeld, “A
Fiedler, “On the feasibility of automatic detection of deconvolution approach for PET-based dose recon-
range deviations from in-beam PET data,” Physics in struction in proton radiotherapy,” Physics in Medicine
Medicine and Biology 57, 1387–1397 (2012). and Biology 56, 7601–7619 (2011).
38. P. Kuess, S. Helmbrecht, F. Fiedler, W. Birkfellner, 50. T. Inaniwa, T. Tomitani, T. Kohno, T. Kanai, “Quantitative
W. Enghardt, J. Hopfgartner, D. Georg, “Automated comparison of suitability of various beams for range
evaluation of setup errors in carbon ion therapy using monitoring with induced beta+ activity in hadron
PET: Feasibility study,” Medical Physics 40, 121718 therapy,” Physics in Medicine and Biology 50, 1131–1145
(2013). (2005).
39. K. Frey, D. Unholtz, J. Bauer, J. Debus, C. H. Min, T. 51. T. Inaniwa, T. Kohno, T. Tomitani, “Simulation for
Bortfeld, H. Paganetti, K. Parodi, “Automation and position determination of distal and proximal edges
uncertainty analysis of a method for in vivo range veri- for SOBP irradiation in hadron therapy by using the
fication in particle therapy,” Physics in Medicine and maximum likelihood estimation method,” Physics in
Biology 59, 5903–5919 (2014). Medicine and Biology 50, 5829–5845 (2005).
40. W. Enghardt, K. Parodi, P. Crespo, F. Fiedler, J. Pawelke, 52. T. Inaniwa, T. Kohno, T. Tomitani, E. Urakabe, S. Sato,
F. Pönisch, “Dose quantification from in-beam positron M. Kanazawa, T. Kanai, “Experimental determination
emission tomography,” Radiotherapy & Oncology 73, of particle range and dose distribution in thick targets
S96–S98 (2004). through fragmentation reactions of stable heavy ions,”
41. E. Rietzel, D. Schardt, T. Haberer, “Range accuracy in car- Physics in Medicine and Biology 51, 4129–4146 (2006).
bon ion treatment planning based on CT-calibration with 53. T. Inaniwa, T. Kohno, T. Tomitani, S. Sato, “Monitoring
real tissue samples,” Radiation Oncology 23, 2–14 (2007). the irradiation field of 12C and 16O SOBP beams using
42. S. P. Nischwitz, J. Bauer, T. Welzel, H. Rief, O. Jäkel, T. positron emitters produced through projectile fragmen-
Haberer, K. Frey, J. Debus, K. Parodi, S. E. Combs, S. Rieken, tation reactions,” Physics in Medicine and Biology 53,
“Clinical implementation and range evaluation of in vivo 529–542 (2008).
PET dosimetry for particle irradiation in patients with 54. T. Inaniwa, T. Kohno, F. Yamagata, T. Tomitani, S. Sato, M.
primary glioma,” Radiotherapy & Oncology 115, 179–185 Kanazawa, T. Kanai, E. Urakabe, “Maximum likelihood
(2015). estimation of proton irradiated field and deposited dose
43. A. C. Knopf, K. Parodi, H. Paganetti, T. Bortfeld, J. distribution,” Medical Physics 34, 1684–1692 (2007).
Daartz, M. Engelsman, N. Liebsch, H. Shih, “Accuracy 55. T. Hofmann, M. Pinto, A. Mohammadi, M. Nitta, F.
of proton beam range verification using post-treatment Nishikido, Y. Iwao, H. Tashima, E. Yoshida, A. Chacon,
positron emission tomography/computed tomography M. Safavi-Naeini, A. Rosenfeld, T. Yamaya, K. Parodi,
as function of treatment site,” International Journal of “Dose reconstruction from PET images in carbon
Radiation Oncology, Biology, Physics 79, 297–304 (2011). ion therapy: A deconvolution approach,” Physics in
44. C. H. Min, X. Zhu, BA. Winey, K. Grogg, M. Testa, Medicine and Biology 64, 025011 (2019).
G. El Fakhri, T. R. Bortfeld, H. Paganetti, H. A. Shih, 56. C. Kurz, J. Bauer, D. Unholtz, D. Richter, K. Herfarth, J.
“Clinical application of in-room positron emission Debus, K. Parodi, “Initial clinical evaluation of PET-
tomography for in vivo treatment monitoring in proton based ion beam therapy monitoring under consideration
radiation therapy,” International Journal of Radiation of organ motion,” Medical Physics 43, 975–978 (2016).
Oncology, Biology, Physics 86, 183–189 (2013). 57. J. Bauer, D. Unholtz, C. Kurz, K. Parodi, “An experi-
45. J. Handrack, T. Tessonnier, W. Chen, J. Liebl, J. Debus, mental approach to improve the Monte Carlo modelling
J. Bauer, K. Parodi, “Sensitivity of post treatment pos- of offline PET/CT-imaging of positron emitters induced
itron-emission-tomography/computed-tomography to by scanned proton beams,” Physics in Medicine and
detect inter-fractional range variations in scanned ion Biology 58, 5193–213 (2013).
beam therapy,” Acta Oncologica 56, 1451–1458 (2017). 58. S. Salvador, J. Colin, D. Cussol, C. Divay, J.-M.
46. F. Pennazio, G. Battistoni, M. G. Bisogni, N. Camarlinghi, Fontbonne, M. Labalme, “Cross section measurements
A. Ferrari, V. Ferrero, E. Fiorina, M. Morrocchi, P. Sala, for production of positron emitters for PET imaging in
G. Sportelli, R. Wheadon, P. Cerello, “Carbon ions carbon therapy,” Physical Review C 95, 044607 (2017).
beam therapy monitoring with the INSIDE in-beam 59. G. Battistoni, J. Bauer, TT. Boehlen, F. Cerutti, M. P.
PET,” Physics in Medicine and Biology 63, 145018 (2018). Chin, R. Dos Santos Augusto, A. Ferrari, P. G. Ortega,
47. E. Fourkal, J. Fan, I. Veltchev, “Absolute dose recon- W. Kozłowska, G. Magro, A. Mairani, K. Parodi, P. R.
struction in proton therapy using PET imaging modal- Sala, P. Schoofs, T. Tessonnier, V. Vlachoudis, “The
ity: Feasibility study,” Physics in Medicine and Biology FLUKA code: An accurate simulation tool for particle
54(11), N217–N228 (2009). therapy,” Frontiers in Oncology 6, 116 (2016).
48. K. Parodi and T. Bortfeld, “A filtering approach based 60. A. Chacon, S. Guatelli, H. Rutherford, D. Bolst, A.
on Gaussian-powerlaw convolutions for local PET Mohammadi, A. Ahmed, M. Nitta, F. Nishikido, Y.
Iwao, H. Tashima, E. Yoshida, G. Akamatsu, S. Takyu, A. short-lived positron emitters during proton therapy,”
Kitagawa, T. Hofmann, M. Pinto, D. R. Franklin, K. Parodi, Physics in Medicine and Biology 62, 4654–4672 (2017).
T. Yamaya, A. Rosenfeld, M. Safavi-Naeini, “Comparative 64. J. Llacer, A. Chatterjee, E. L. Alpen, et al., “Imaging by
study of alternative Geant4 hadronic ion inelastic physics injection of accelerated radioactive particle beams,”
models for prediction of positron-emitting radionuclide IEEE Transactions on Medical Imaging 3, 80–90 (1984).
production in carbon and oxygen ion therapy,” Physics in 65. R. Augusto, T. Mendonca, F. Wenander, et al., “New devel-
Medicine and Biology 64, 155014 (2019). opments of 11C post-accelerated beams for hadron ther-
61. P. Lecoq, C. Morel, J. O. Prior, D. Visvikis, S. Gundacker, apy and imaging,” Nuclear Instruments and Methods in
E. Auffray, P. Križan, R. Martinez Turtos, D. Thers, Physics Research Section B 376, 374–337 (2016).
E. Charbon, J. Varela, C. de La Taille, A. Rivetti, D. 66. A. Mohammadi, H. Tashima, Y. Iwao, S. Takyu, G.
Breton, J.-F. Pratte, J. Nuyts, S. Surti, S. Vandenberghe, P. Akamatsu, F. Nishikido, E. Yoshida, A. Kitagawa, K.
Marsden, K. Parodi, J. M. Benlloch, M. Benoit, “Roadmap Parodi, T. Yamaya, “Range verification of radioactive
toward the 10 ps time-of-flight PET challenge,” Physics in ion beams of 11C and 15O using in-beam PET imaging,”
Medicine and Biology 65, 21RM01 (2020). Physics in Medicine and Biology 64(14), 145014 (2019).
62. P. Crespo, G. Shakirin, F. Fiedler, W. Enghardt, A. 67. R. S. Augusto, A. Mohammadi, H. Tashima, E. Yoshida,
Wagner, “Direct time-of-flight for quantitative, real- T. Yamaya, A. Ferrari, K. Parodi, “Experimental valida-
time in-beam PET: A concept and feasibility study,” tion of the FLUKA Monte Carlo code for dose and β+-
Physics in Medicine and Biology 52, 6795–6811 (2007). emitter predictions of radioactive ion beams,” Physics in
63. H. J. T. Buitenhuis, F. Diblen, K. W. Brzezinski, S. Medicine and Biology 63, 215014 (2018).
Brandenburg, P. Dendooven, “Beam-on imaging of
Chapter 27
Prompt Gamma Detection for

Proton Range Verification
Paulo Magalhaes Martins
German Cancer Research Center – DKFZ
Heidelberg, Germany
Instituto de Biofísica e Engenharia Biomédica – IBEB
Lisbon, Portugal
Riccardo Dal Bello and Joao Seco

German Cancer Research Center – DKFZ
University of Heidelberg
Heidelberg, Germany
CONTENTS
27.1 Introduction to Prompt Gamma Physics 428
27.1.1 Nuclear Reactions Leading to Prompt Gamma Generation 428
27.1.2 Detecting Prompt Gamma and Sources of Background 430
27.1.3 Improving the Signal-to-Noise Ratio 431
27.2 Ions and Their Nuclear Reactions 432
27.2.1 Monte Carlo Implementation 432
27.2.2 Cross Sections 433
27.3 PGI for Proton Therapy 433
27.3.1 Historical Development 433
27.3.2 Cesium Iodide – CsI(Tl) 434
27.3.3 Knife-Edge Slit Camera 434
27.3.4 Prompt Gamma Spectroscopy 434
27.3.5 Prompt Gamma Timing 435
27.3.6 Compton Camera 436
27.3.7 Oxygen Concentration 436
27.4 PGI for Ion Therapy 437
27.4.1 Time-of-Flight 437
27.4.2 Linac-Based Experiments 437
27.4.3 Synchrotron-Based Experiments 437
27.5 Perspectives and Future Challenges 438
References 439
427
27.1 INTRODUCTION TO PROMPT dose with the by-products of nuclear interactions exists

GAMMA PHYSICS but it is not direct and trivial. Second, the energies of the
27.1.1 Nuclear Reactions Leading to particles outgoing nuclear interactions can reach several
Prompt Gamma Generation MeV. This implies that the secondary radiation (e.g.,
In charged particle radiotherapy, moderately relativistic neutrons, protons or gamma-rays) has a higher poten-
proton and ion beams are delivered. The primary mech- tial to leave the patient and be measured. Third, the time
anism for energy loss is the collision with the atomic scale of nuclear interactions is the one of the strong force.
electrons. The direct interaction of the beam particles The de-excitation processes that follow have characteris-
with the target nuclei has only a minor contribution in tic times connected to each specific channel. These can
the stopping power. Still, the nuclear interactions are range from several minutes (e.g., T1/2 = 20.4 min for the
highly relevant for range verification methods. β+ decay of 11Cg.s.) down to sub-nanoseconds (e.g., T1/2 =
Range verification is based on nuclear inelastic colli- 18.4 ps for the γ decay of 16O*6.13 MeV). In this chapter, we
sions. These are violent events where the projectile may will focus on the fast processes that produce high ener-
change the target nucleus to a different isotope by knock- getic gamma radiation, i.e., prompt gamma.
ing out light fragments or modifying its energy state. Prompt gamma are one of the possible by-products
Because the produced nuclei are not at the ground level, of the inelastic nuclear reactions of the beam particles
they undergo a chain of de-excitation reactions. Secondary with the target nuclei. Figure 27.1 represents in a sche-
radiation is produced during these processes that can leave matic way two possible inelastic collisions producing
the patient and be measured with an external detector to final states differing from the initial ones. Generally,
retrieve the Bragg peak position. after such a collision the beam particle is lost and does
Few general aspects of range verification based on not contribute to the primary dose deposition in the
inelastic nuclear interactions have to be highlighted. Bragg peak. Typically, for a 16 cm deep Bragg peak,
First, the secondary radiation is produced by interaction about 20% of the primary protons and about 50% of
mechanisms that differ radically from the ones leading the primary 12C ions undergo inelastic nuclear inter-
to the primary energy loss and the production of the actions before the end of the range [1]. The cross sec-
Bragg peak. Therefore, the correlation of the deposited tions of such processes are energy dependent and
FIGURE 27.1 Schematic representation of nuclear interactions of the beam particles with the target nuclei. Top: proton–nucleus
inelastic collision, specifically 16 O( p, p ′α )12 C * . Bottom: nucleus–nucleus inelastic collision, specifically 16 O(12 C ,10B* d n)15 O * .
Prompt Gamma Detection for Proton Range Verification ◾ 429
FIGURE 27.2 Nuclear energy levels for 4He (left) and 16O (right). The ground state and the first five states above it are shown.
The principal de-excitation channels and their characteristic decay times are also reported. (Data from: National Nuclear Data
Center, Brookhaven National Laboratory.)
they generally increase at lower collision energies. de-excitation channel. The ground state is reached either
Therefore, the prompt gamma yield can be correlated with a single transition or with a photon emission chain.
with the residual energy of the projectile in the target, The net result is that a series of gamma quanta are emit-
i.e., with the residual range. ted, having well defined discrete energies. It is impor-
Inelastic nuclear interactions produce excited nuclear tant to notice that every single nucleus has its unique
states. Figure 27.2 shows the first five energy levels above energy levels. Therefore, the discrete energy spectrum
the ground state for two light nuclei relevant for charged of the de-excitation photons is a characteristic feature
particle therapy. Similar properties are observed unique for every given isotope.
also for other nuclei such as p or 12C. The principal It is now possible to interpret the prompt gamma
de-excitation mechanisms are depicted as well. The first emission in the case of heavier projectiles, such as 12C
property to be noticed is that the energy separation for or 16O. Here, both the target nucleus and the projectile
the nuclear levels is in the MeV range, which correlates can be excited and produce gamma radiation. As pre-
directly with the energy of the secondary radiation pro- sented schematically in Figure 27.1, the gamma emission
duced in the de-excitation processes. A major difference for the excited projectile happens while this nucleus is
is observed between the lightest nuclei and the heavier travelling away from the interaction zone in a relativis-
ones. Namely, the excited levels of p or 4He nuclei do not tic moving frame. Therefore, the energy of the gamma
relax by the emission of gamma radiation. The energy quanta emitted by the projectile has to be corrected the
separation between the ground state and the first excited relativistic Doppler factor. This does not apply for the
level is sufficient to open the channels for the emission of prompt gamma emission from the target nucleus. Figure
nucleons. A different situation is observed for 12C or 16O 27.2 also presents the characteristic times (T1/2) that
nuclei. Here, the energy separation between the ground describe the time intervals in which the excited nuclear
state and the excited levels is smaller and the preferred states exist after their production and before undergo-
de-excitation channel is the gamma isomeric transi- ing relaxation. These values are specific for every energy
tion. We observe that every nuclear level has a preferred level and we can generally observe that T1/2 = 1ns.
Finally, we should also point out that there is a second the detector. Such scattered events are recorded but are
mechanism for nuclear de-excitation replacing the single not correlated with the Bragg peak position.
discrete emission with multiple photon emission. This On the other hand, a photon that reaches the detec-
presents a continuum spectrum. However, for the ener- tor without previous interactions has still a significant
gies and the nuclei found in proton and ion therapy, the probability of delivering just partial information. After a
probability for the discrete emission is approximately an first interaction inside the detector, the scattered photon
order of magnitude higher than the continuous one [2]. can leave it and deposit its remaining energy elsewhere.
To summarize, the nuclear interactions of the par- In such a case, only a part of its energy is recorded
ticle beam with the target nuclei play a minor role in and therefore the original energy information is lost.
dose deposition but are important for range verifica- Moreover, the fraction of energy deposited in the detec-
tion, in particular, the inelastic nuclear collisions that tor may be smaller than the one minimally measurable,
produce excited nuclear states. These can de-excite with which leads to a complete loss of the event.
the emission of gamma radiation that leaves the patient Conversely, if there is a measurable energy in the
and can be measured with an external detector. The detector, the correct time and direction information of
gamma radiation is emitted by light nuclei, such as 12C the original prompt gamma are preserved. Therefore,
or 16O, but not by the lightest ones, such as p or 4He. The these events can be treated either as background or as
prompt gamma energy spectrum presents discrete lines signal depending on whether the energy information is
that are characteristic of the nuclei involved. These lines relevant for the specific PGI technique or not. We define
span up to approximately Eγ = 10 MeV with production such events as partial energy deposition events. Finally,
times T  1 ns. if after the first interaction, also all the following ones
happen inside the detector we observe a full energy
27.1.2 Detecting Prompt Gamma and deposition event. In this most desirable case, the sys-
Sources of Background tem retrieves all the original energy, time, and direction
The critical aspect for a successful application of Prompt information of the prompt gamma.
Gamma Imaging (PGI) is an efficient detection of the As mentioned before, prompt gammas are produced
secondary radiation that leaves the patient. The details within nanoseconds following the interaction of the par-
depend on the specific PGI technique used and will be ticle beam with the patient. We have seen how these par-
discussed in a following section. We will now focus on ticles are only one type among the multiple by-products of
some general physics aspects that apply to all the systems. nuclear interactions. The prompt gamma emission channel
Prompt gamma-rays have an energy spectrum that competes with the production of other secondary particles
goes up to 10 MeV. Gamma quanta with energies greater on a similar time scale. Among the secondary particles,
than 1 MeV interact with materials preferentially there are neutrons which are likely to reach the detectors.
through Compton scattering or pair production. The Such events generate a continuum energy spectrum which
photoelectric effect, which is usually exploited for imag- is hardly distinguishable from the one created by prompt
ing at lower photon energies, plays here only a minor gamma events. This spectrum contributes to the back-
role. It is therefore likely for a single prompt gamma to ground that we observe in the measurements with the sev-
undergo multiple interactions with partial energy depo- eral PGI systems.
sition before being fully absorbed. This has an impact The neutrons can also interact with nuclei in the
both on the transport of the gamma radiation from the patient and produce further gamma radiation. One of the
nucleus to the detector as well as on the interactions most important reaction is the hydrogen neutron cap-
within the detector. ture (10 n +11 p → 12d + γ ( 2.22MeV)). The line associated to
Those interactions may happen within the patient, this reaction can be clearly identified in Figure 27.3 (left),
with the patient table and other equipment, or with the since it is not correlated in time with the radio-frequency
walls of the treatment room [3]. In such interactions, (RF) of the cyclotron. The mono energetic photon pro-
the gamma radiation is not absorbed but scattered with duced by this reaction is not directly correlated to the
a different angle and its energy is reduced. This means Bragg peak position and contributes to the background
that a significant number of events lose their original as well. We define all these processes as neutron induced
energy, time, and direction information before reaching background.
FIGURE 27.3 Left: prompt gamma lines and corresponding single- and double-escape peaks visible as time correlated hori-
zontal stripes. The PMMA target was irradiated with a mono-energetic proton beam of 130 MeV in a setup without collima-
tion. Right: schematic geometries of an anti-coincidence detector system: primary CeBr3 with a BGO AC shielding (left);
primary CeBr3 with a CeBr3 AC shielding (right). (Adapted from [30] and [24].)
For what concerns the charged nuclear fragments, PGI system identifies two events in coincidence. On the
only the most energetic ones will leave the patient and other hand, an energy deposition in the primary detec-
reach the detector system. However, before leaving the tor without any signal in the secondary detector can be
patient they may also undergo nucleus–nucleus inter- interpreted as an event with full energy deposition and
actions. Such events produce a further generation of it should be recorded, i.e., in anti-coincidence.
gamma radiation, which is not directly correlated with The use of time-of-flight (TOF) is helpful to differ-
the Bragg peak position. This phenomenon is more likely entiate the prompt gamma from the scattered events,
for 12C ion beams, but not very relevant to proton treat- the neutrons and the fragment induced background [4].
ments. We define this as fragment induced background. Figure 27.4 (left) shows a typical TOF spectrum for a
carbon beam. The spectrum is obtained by measuring
27.1.3 Improving the Signal-to-Noise Ratio the difference between the time given by the accelerator
The detailed knowledge of the physics processes regu- RF and the time signal given by the PGI system. One
lating the production, transport and detection of the can identify a prominent peak, which is given by the
prompt gamma radiation can be exploited to differenti- prompt gamma radiation that travels without previous
ate the signal from the background. The implementation interactions towards the detector. Few nanoseconds
of the following techniques is specific to every PGI sys- after it, there is the delayed component caused by the
tem. We will focus on the physics properties of the back- scattered events. These photons originate together with
ground. The identification of the differences between the primary prompt gamma but undergo scattering
the signal and the background properties can highly before reaching the detector. Therefore, the total dis-
improve the quality of the recorded signal. tance travelled is greater compared to the straight line
An anti-coincidence detection configuration is from the point of origin in the patient to the PGI sys-
depicted in Figure 27.3 (right). Such a system aims to tem. This increases the time interval needed to reach the
suppress the events with partial energy deposition. The detector and the scattered photons appear in the delayed
principle relies on the fact that the photon, after a partial component of the TOF spectrum. The constant and time
energy deposition, leaves the primary detector carrying independent background is given by the target activa-
with it the residual energy. A secondary detector can tion and the neutron induced component. One further
be placed in the immediate proximity of the primary delayed component is observable in the case of the car-
detector with the purpose of recording the escaping bon ion irradiation, namely the fragment induced back-
photon. This second energy deposition happens imme- ground. In this case, the TOF difference with respect to
diately after the first interaction in the primary detec- the prompt gamma peak is greater if compared to the
tor, i.e., in coincidence. The event is discarded when the scattered events. We can then select only the events in
FIGURE 27.4 Left: TOF spectra measured for two target locations Z and for two selections on energy E deposited in a scintil-
lator: Z = 8 mm (line), Z = 26 mm (dots), E < 1 MeV, and E > 1 MeV. Origin of time scale: ion impact on target. Right: detection
rates (E > 1 MeV) as a function of longitudinal position of target, obtained for two different TOF ranges: 2 < T < 10 ns (prompt
γ-rays, squares) and T > 10 ns (neutrons, circles). (Adapted from [4].)
the TOF window in the prompt gamma peak region and interactions, but several solutions have been proposed.
suppress the background which is highly uncorrelated A distinction should be made between the simulation of
with the Bragg peak position. the prompt gamma production induced by proton irra-
The energy selection aims to simultaneously sup- diation or by heavier ions. In the former, we consider a
press a fraction of the scattered events and of the neu- nucleon–nucleus interaction that is generally treated by
tron induced component [5]. As discussed before, the an intra nuclear cascade (INC). In the latter, a more com-
prompt gamma-rays are emitted with a complex spec- plex nucleus–nucleus interaction takes place and can be
trum up to several MeV. If we take as an example the described by the quantum molecular dynamics (QMD)
16 O transitions presented in Figure 27.2, we observe model. In both cases, after the interaction and the energy
that the direct emission from this nucleus happens exchange, a pre-equilibrium and a de-excitation model
mainly above Eγ > 2.5 MeV, while the neutron induced describe the steps until reaching a stable state. The Monte
background is comprised mainly at lower energies. Carlo codes most commonly used for the simulation of
Moreover, we know that the scattered radiation usu- prompt gamma production are FLUKA [7], Geant4 [8],
ally reaches the detector after losing part of its energy. and MCNPX [9]. These three codes share the use of intra
Taking into account all these effects, it is convenient to nuclear cascade models to simulate nuclear interactions.
accept only the events above a certain energy thresh- The specific implementation differs for each software and
old. The higher the energy threshold the better the also within a given software by the exact choice of the
background rejection. This however decreases the parametrization. The following aspects are general and
available statistics. Therefore, a compromise has to be apply to all Monte Carlo software.
reached and an optimization is needed for every spe- The INC is used to describe the interaction of the pro-
cific PGI system. A further and more elegant approach ton with a nucleus. It is the basis for the implementa-
for the energy selection consists in adopting an energy tion of nuclear interactions in the modern Monte Carlo
window only for certain specific transitions, e.g., the codes. The target nucleus is treated as an ensemble of
Eγ = 6.13MeV emitted by the 16O*6.13 MeV nucleus [6]. quasi-free nucleons with an energy distribution approx-
The applicability of each technique depends on the imated by a Fermi gas. The projectile is transported
specific PGI system. Furthermore, multiple noise sup- inside an effective time independent nuclear potential
pression techniques may also be adopted simultaneously. and undergoes multiple two-body interactions. The final
state returns the single particles that escaped the nuclear
27.2 IONS AND THEIR NUCLEAR REACTIONS potential and the modified target nucleus. The excited
27.2.1 Monte Carlo Implementation nucleus is then given as an input to the pre-equilibrium
Conversely to the electromagnetic interactions, no rig- and de-excitation models to simulate the emission of
orous model exists to describe the nuclear inelastic prompt gamma radiation.
FIGURE 27.5 Differential cross sections of prompt gamma production induced by proton interactions with 16O nuclei. The
titles of the plots describe the nuclear channel and the corresponding energy of the gamma ray emitted in MeV. (Adapted
from [15].)
The QMD is used to describe the nucleus–nucleus partial and total energy deposition events. The most
interaction. The collision cannot be modeled with a modern Monte Carlo codes generally reproduce in a
series of two-body interactions but rather consider- correct way the integrated experimental data and can be
ing all the target and the projectile nucleons as simul- considered reliable for such applications.
taneous participants of the interaction. Each nucleon The direct comparison of the cross sections for single
is described by a Gaussian wave package and the total de-excitation lines is a more fundamental approach.
wave function is given by the direct product of all the However, great discrepancies have been observed
nucleon wave functions. The equation of motion of the between different Monte Carlo codes [11]. Therefore, an
single nucleons are derived from the Hamiltonian of experiment-based approach is preferred in such case.
the system and the nuclear potential is updated at every Figure 27.5 shows the cross sections for several reaction
step. This model has been proposed for the collision of lines obtained during proton irradiation. The experi-
heavy nuclei. However, several studies have optimized mental data is used as input parameter for the model-
the parameters in the Geant4 implementation for the ing of the cross sections. A common feature can be
application with 12C beams [2] and even protons [10]. observed for all the reaction lines. Namely, the maxima
of the cross sections are located at low proton energies
27.2.2 Cross Sections just before the drop to zero-values. The lower the pro-
The implementation of the nuclear models in the Monte ton energy, the shorter the residual range of the beam
Carlo codes has to be benchmarked against experimen- particle. Therefore, the maximum value of the prompt
tal data. This is necessary to assess the reliability of the gamma yield in Figure 27.4 (right) has to be interpreted
simulations, which are a fundamental component of the as a consequence of the location of the maxima at low
development of PGI systems. Multiple approaches can energies in Figure 27.5.
be used. Here we discuss the comparison between the
integrated profiles and between the cross sections. 27.3 PGI FOR PROTON THERAPY
The integrated profiles include the data over the full 27.3.1 Historical Development
prompt gamma energy spectrum above the threshold. The PGI as a technique for online verification of the
Multiple transition lines are detected simultaneously ion range in particle therapy was first proposed in 2003
at a given depth of the target. An example is shown in by Stichelbaut and Jongen at the PTCOG meeting [12].
Figure 27.4 (right). The total yield is the result of both We had to wait however until 2007 to have the first
experimental results from a 38 MeV proton beam [13]. collimator and a CsI(Tl) scintillator detector showed for
Until then, prompt gamma radiation emitted during the first time a clear correlation between the distribu-
particle therapy was always regarded as a noise source, tion of the prompt gammas and the location of the distal
mostly in the context of range verification through dose edge for a 38 MeV proton beam, thus confirming
positron emission tomography [14]. The PET signal dur- the previous simulations with MCNPX [5, 13]. A spa-
ing the spill time was highly corrupted by the prompt- tial resolution of 1–2 mm was already envisioned at that
gamma background and measurements of PET data time. In 2012, a full-scale multislit collimation system
were only possible during the inter-spill time and after based on CsI(Tl) was evaluated by Monte Carlo simu-
the treatment. lations for all therapeutic proton energies. This system
Since 2007, many groups emerged around the world requires however a two-dimensional distribution of the
trying to verify in vivo and online the range of the par- proton dose with 2D position sensitive gamma detectors
ticle beams inside the patient. The leading groups were [27] and lacks so far enough statistics when compared to
the ones in Boston [15], Belgium [16], Delft [17], Dresden the other available systems.
[18], Korea [13], Lyon [4], Maryland [19], Milan [20],
and Munich [21]. Each of these groups specialized in 27.3.3 Knife-Edge Slit Camera
different approaches to solve the same problem: how to In 2012, Smeets et al. presented a pioneering proto-
achieve by means of PGI a range uncertainty of 1 mm? type and methods for verification of the proton range
In recent years, the investment in this technology had through a slit camera [16]. The experimental work fol-
a leap mostly due to the successful efforts of the Belgian lowed thorough simulations with MCNPX that defined
and the Boston groups that rose the interest of major the main parameters for a proton beam at 160 MeV, such
companies in developing their own prototypes. The as: collimator material and thickness; collimator slit
knife-edge slit camera was first applied in the clinical width and angle; and scintillation material, thickness
setting in 2016 in Dresden [22]; while the prompt gamma and segment width. A range estimation of 1–2 mm stan-
spectroscopy (PGS) [15] was first tested for clinical use dard deviation has been achieved for a 160 MeV proton
in patients in 2019 in Boston [23]. beam impinging on a homogeneous PMMA phantom,
Meanwhile, the research has evolved to different corresponding to doses in water at the Bragg peak of just
clinical applications. While the former prototypes were 25 cGy.
mainly conceived for cyclotron-based facilities, a new This prototype has been acquired by IBA and evolved
device is being planned for synchrotron-based facilities to a clinical prototype that was tested in humans for the
[24]. Other competing devices are also heading towards first time at the OncoRay – Dresden in 2016 [22]. These
clinical application by taking advantage of different fea- two papers demonstrated great accuracy in the mea-
tures of the detection method, such as prompt-gamma surement of relative shifts between fractions. However,
timing [18] and the Compton camera [25]. many improvements are being carried out to increase
Another possibility explored by prompt gamma spec- the absolute range verification capability of the camera
troscopy is the determination of the elemental composi- [28]. Concurrently, further tests in humans took place
tion of irradiated tissues through characteristic prompt at the University of Pennsylvania that showed a median
gamma rays emitted from 16O and 12C [19]. This tech- precision in shift estimation after spot aggregation of
nique could open new doors to the direct evaluation of 2.1 mm and a median inter-fraction variation of 1.8 mm
the response of the tissues to particle therapy [26]. [29]. Figure 27.6 shows the principle behind this system
and the prototype itself mounted on a trolley, which is
27.3.2 Cesium Iodide – CsI(Tl) subject to the limitations of the camera positioning.
In 2006, C.-H. Min et al. proposed a multilayered col-
limator system to locate the distal dose falloff in pro- 27.3.4 Prompt Gamma Spectroscopy
ton therapy [5]. Such a system would capture neutrons Also in 2012, Verburg and Seco proposed for the first
and prevent unwanted gammas while measuring at time to use the energy spectrum of the prompt-gamma
right angles the prompt-gamma component result- emission to retrieve an absolute range verification of
ing from the interaction of the proton beam with the the proton beam. As previously mentioned and shown
patient tissue. One year later, a prototype composed by a in Figure 27.5, systematic simulation studies compared
FIGURE 27.6 Schematic setup of the prompt gamma knife-edge slit camera system for imaging of the vertex field. Dark lines
represent proton tracks, and light grey lines represent gamma rays selected by the collimator aperture. Upper right inset, from
left to right: (a) the table-mounted U-shaped range shifter, (b) the camera trolley positioning system, and (c) the camera knife-
edge slit collimator. (Adapted from [29].)
several Monte Carlo nuclear reaction models and avail- use of CeBr3 detectors for PGS at the proton facility in
able experimental data of discrete gamma line cross- Dresden [30]. A semi-collimated setup was proposed
sections [11]. The dedicated nuclear reaction codes, such for measuring the falloff region of the Bragg Peak. This
as TALYS and EMPIRE, reproduced the experimental configuration would be later adopted also by Hueso-
data more consistently than Geant4. González et al. [23].
Further experimental work demonstrated the fea-
sibility of measuring the discrete prompt-gamma rays 27.3.5 Prompt Gamma Timing
for in vivo range verification of clinical proton beams A novel concept of prompt gamma-ray timing (PGT)
[6]. The differential cross-sections were measured for 15 has been proposed by Golnik et al. [18]. This is an
prompt gamma-ray lines from proton-nuclear interac- uncollimated method that uses a gamma-ray detector
tions with 12C and 16O up to 150 MeV and an absolute of reasonable time resolution for measuring the finite
range was determined with a standard deviation of 1.0– transit time from entering the patient’s body until stop-
1.4 mm [15]. ping in the target volume. The underlying principle is
In 2018, a full-scale clinical prototype was presented the increasing transit time with the particle range that
[23]. Figure 27.7 shows the detection system, which causes measurable effects in the PGT spectra that can be
consists of eight LaBr3 scintillators and a collimator, used for range verification.
mounted on a rotating frame. By making use of the pre- This method has been demonstrated at a clinical pro-
viously experimentally determined cross-sections and a ton facility and range differences of 5 mm in defined het-
GPU-accelerated Monte Carlo simulation, an absolute erogeneous targets were identified with a single detector
proton range was determined for each spot of the distal and 108 incident protons [31]. However, PGT spectra were
energy layer with a mean statistical precision of 1.1 mm. observed to be smeared out by the bunch time spread
Upcoming clinical studies are envisioned to assess the and accelerator related proton bunch drifts against the
suitability of such system for range verification during RF of the accelerator have been detected [32]. A proton
patient treatments. bunch monitor has been therefore developed to correct
Concurrently, a recently formed group at the German for potential bunch drifts and increase the robustness of
Cancer Research Center – DKFZ demonstrated the the PGT method. Nevertheless, this technique is highly
FIGURE 27.7 Left: 3D model of the clinical PGS prototype, which can rotate around its axis according to the beam incidence
angle. The tungsten collimator is visible on the front plane, the eight scintillation detectors in the middle and the readout
electronics on the back plate. Right: illustration of the gantry, proton treatment head, patient, top x-ray flat panel and the PGS
system. The detector frame can rotate according to the gantry angle, and is mounted on a positioning robot, consisting of six
actuators. The robot stands on a platform on wheels that is moved into the treatment room. (Adapted from [23].)
dependent on the number of gamma rays detected per Draeger et al. are close to reach a clinical prototype
incident proton. Large detector loads and a high acquisi- [25]. Recent results reported prompt-gamma measure-
tion throughput are therefore mandatory to draw statisti- ments with a small CC prototype placed at three dif-
cally significant conclusions on range errors [33]. ferent locations along the proton beam path. The data
were combined to simulate measurements with a larger
27.3.6 Compton Camera scale, clinical CC capable of imaging Bragg peak shifts
The Compton camera (CC) belongs to the actively colli- for both hypo-fractionated and standard treatments. 3D
mated systems and it requires position-sensitive gamma images of the prompt gamma emission were produced
ray detectors with high resolution and efficiency, which and range shifts of 2 mm were detected for the delivery
are arranged in one scatterer, one absorber, or in several of a 2 Gy spot.
scatter planes. In contrast to the other PGI systems, it
allows for two-dimensional or even three-dimensional 27.3.7 Oxygen Concentration
images to be reconstructed as more gamma-rays and The spectroscopic capability of PGI is being further
directions can be detected. However, the event coinci- explored for measuring the elemental compositions of
dence needed between the different stages limits the over- irradiated tissues. The nuclei emit a unique energy spec-
all efficiency. The major pitfalls are its highly demanding trum during proton irradiation [26]. It has been shown
instrumentation in terms of spatial, time and energy reso- that the total emission from 16O per unit dose is pro-
lution and the computationally intensive reconstruction portional to the concentration of oxygen within an irra-
algorithms [33]. Several groups around the world have diated volume of tissue [19]. PG emissions could then be
been devoting significant efforts towards a clinical pro- used to accurately determine the concentration of key
totype for range verification [34–38] and some of them elements in tissue (O, C, N, Ca, K, etc.). Since the pro-
have abandoned it due to the technical complexity, elec- ton range highly depends on the density of the tissues,
tronics expense, low coincident efficiency, high detector variations in density and concentration of the several ele-
load, radiation background, and the elevated percentage ments could be tracked on a daily basis over the course
of random coincidences [33]. of the treatment, thus providing information both on the
tumor response and the Bragg peak position. Dedicated at synchrotrons. This has however been overcome with
measurements have been made with HPGe detectors [19, such novel single-particle trigger.
39, 40]. Verburg and Seco also reported the use of the
PGS system to determine the relative content of 12C and 27.4.2 Linac-Based Experiments
16O however with a large statistical uncertainty. In such Testa et al. were the first to perform prompt gamma
a system, these concentrations are left as free parameters experiments for heavier particles. By means of a TOF
for the optimization process [15]. technique, Testa et al. measured the longitudinal pro-
file of prompt gamma-rays emitted by 73 MeV/u 13C
27.4 PGI FOR ION THERAPY irradiating a PMMA target [4, 47]. The experiment
27.4.1 Time-of-Flight was performed at the GANIL facility (Caen, France)
Several groups reported the importance of TOF mea- with a pulsed beam (beam pulses of ≈1 ns every
surements for background reduction as the prompt 80 ns) with an intensity of 1 nA. The main detector, a
component can be clearly differentiated from the cylindrical NaI(Tl), was placed at right angles to the
delayed neutron-induced gamma component [6, 17, beam direction behind a 20 cm thick lead collimator.
18, 41]. This issue is even more relevant at synchro- The method used to reject the neutron component
tron facilities where the microstructure of the beam was similar to the one used in proton facilities. The
is destroyed during the extraction of the beam. In gamma rays and the neutrons were discriminated by
cyclotron-based facilities, the TOF information is eas- subtracting the given phase of the RF-signal from
ily retrieved from the RF of the accelerator despite the accelerator to the detection time provided by the
some challenges already mentioned concerning proton gamma ray detector. In Figure 27.4 (left), we showed
bunch drifts. a TOF spectra for two target locations (in the middle
Other groups doing experiments on synchrotron- of the Bragg peak and 12 mm beyond the Bragg peak)
based facilities have used external detectors as primary and two energy regions. This allowed for the dis-
trigger for TOF measurements [3, 42, 43]. The most crimination of the prompt gamma component, whose
commonly used is the plastic scintillator, e.g., EJ-200. detection rate as a function of the longitudinal posi-
These scintillators cope well with moderate intensities, tion Z was shown on the right.
provided they remain below 106 s −1. This allows single Further experiments were performed at GANIL
particles to be identified and their arrival time to be with a 95 MeV/u 12C ion beam and compared with the
correlated with the arrival time of the prompt gamma- results obtained at GSI with a 305 MeV/u 12C ion beam
rays on the secondary detector. In 2018, Martins et al. [3]. An external plastic detector has been used at GSI
have shown at PTCOG a first small-scale prototype of to provide TOF information. A BaF2 detector was cho-
a trigger system capable of tracking single particles at sen for gamma detection and a BC501 scintillator for
clinical intensities for all beam species accelerated at neutron detection. The results show similar results at
the Heidelberg Ion-Beam Therapy Center – HIT [44]. the two facilities both for the prompt gamma and the
This single-particle trigger was able to detect single, neutron components. The presented technique is appli-
double and triple hits within ion bunches. It also pro- cable also in case of a continuous spill structure, pro-
vided TOF information for prompt-gamma measure- vided the identification of the primary ions one by one
ments with sub-ns resolution. Such prototype should is achieved.
not be compared to other detectors used for position
tracking of the single particles, such as hodoscopes 27.4.3 Synchrotron-Based Experiments
[45]. The major demands for a time-tracker in hadron Testa et al. also presented a systematic work with results
therapy is its radiation hardness, single particle iden- based on the same TOF spectrum analysis, but consid-
tification at clinical intensities and a very small thick- ering the absolute prompt gamma yields and includ-
ness, since the minimum amount of material between ing the statistical and systematic uncertainties [48]. The
the nozzle and the patient remains a primary concern. experiments were performed at GANIL, GSI (Darmstadt,
Krimmer et al. [46] and Biegun et al. [17] consid- Germany), HIT (Heidelberg, Germany) and the WPE
ered the TOF neutron rejection not to be effective for Proton Center (Essen, Germany). Carbon ions showed
bunch widths larger than 5 ns (typically 80 ns) found a prompt gamma production a factor five higher than
protons. The values for the absolute prompt gamma yields verification. The slightly lower energy resolution of these
for carbon ions accelerated at GANIL confirm the pre- new CeBr3 detectors, if compared to the crystals used
vious results from Agodi et al. [49]. The value obtained by Verburg and Seco (LaBr3) [15], is compensated by the
for 310 MeV/u carbon ions was (79 ± 2stat ± 23sys) × 10−6. inexistent intrinsic activity and lower cost. The major
Pinto et al. have also compared the experimental data limitation of the Boston system is the cost of the LaBr3
with Geant4 using all applicable proton inelastic models. scintillating crystals [50].
They observed that Geant4 overestimates the prompt-
gamma emission yields by 40%. In order to improve the 27.5 PERSPECTIVES AND FUTURE
discrepancy observed, they used the QMD model with a CHALLENGES
wave packet width, L = 1.3 fm2 [10]. Around 15 years have passed since PGI was first time pro-
Vanstalle et al. also proposed to benchmark Geant4 posed until the first application in men. The use of PET
models for prompt gamma monitoring with carbon ions for monitoring charged hadron tumor therapy [51] has
at GSI [42]. They also used external plastic scintillators been shown to suffer from several effects, such as biologi-
for the trigger and BaF2 scintillator for the detection of cal washout, lower falloff correlation, low motion and tis-
the prompt gamma rays. They measured the total yields sue reproducibility, thus yielding a minimum uncertainty
and energy spectra of prompt gamma produced by the of 6 mm in identifying the proton range [50, 52, 53]. PGI
interaction of 220 MeV/u carbon ions with PMMA at 60° is today the most promising technique for real-time in
and 90° with respect to incoming beam. The reported vivo range verification. Either isolated or in combination
values for both angles are similar and a prompt gamma with PET [54], PGI is expected to be installed at most of
yield of 1.06 × 10 −2 was obtained at 90°. They observed the proton therapy centers around the world. Only a cen-
a good match between the shapes of both simulated and ter that can provide an imaging system for the position of
experimental spectra, especially with INCL++ physics the Bragg peak will be able to reduce the safety margins
list. Mattei et al. reported on the prompt gamma yields and combat the largest criticism to particle therapy: the
measured during the irradiation of a PMMA target by range uncertainties [26, 50, 53].
4He, 12C, and 16O ion beams [43]. The experiments were The phase III clinical trials are essential to clarify the
performed with LYSO scintillators at 60° and 90° and a advantage of particle therapy over conventional photon
plastic start counter was used as well. The trigger rate therapy [55–57]. The PGI systems may play an impor-
was very low (below 6 kHz) being mostly dictated by tant role in future clinical trials. In a first stage, such
the DAQ dead time. The results with carbon ions were systems may gather as much data from daily treatments
compared with the four previous experiments [20, 42, before drawing future necessary conclusions regarding
48, 49] and considerations were taken concerning the its application in the clinical setting. The data acquisi-
expected resolution of a slit camera for prompt gamma tion should not conflict with the daily clinical workflow.
monitoring of particle therapy with heavier particles. Very early, one of the proposed uses was the delivery of
A novel approach inherited from the PGS system small bunch of particles to the most distal spots or lay-
developed by Verburg and Seco is currently being devel- ers. This would demand a very short acquisition time
oped at the DKFZ. First experiments concerning the with enough statistics and would serve as a guide for the
performance of CeBr3 scintillators to be used at HIT have full treatment.
been previously undertaken at OncoRay – Dresden [30]. The ultimate goal is to achieve an automatic feedback
A Monte Carlo optimization of the detector was carried on the position of the Bragg peak and adjust the treat-
out in order to maximize the signal-to-noise ratio after ment accordingly. This demands very fast treatment
considering different detector geometries [24]. The pres- planning routines, some of these already available in the
ence of an active coincidence shield has been shown to research community [58–60]. However, the first studies
increase the signal-to-noise ratio up to a factor of 3.5. are expected to track only the variations due to inter-
Both papers investigated the capability of such detec- fractional changes [61].
tors to detect low-energy lines down to 511 keV typically We presented the basic physical principles of PGI and
emitted by irradiated metals. This opens new perspec- the historical development of this technique. There are
tives for the detection of metal inserts in the beam path currently two prototypes [22, 23] ready for daily clini-
and unveils additional energy lines to be used for range cal use and that tackle the range uncertainty problem in
different ways respectively: (1) fluence approach based 4. E. Testa, M. Bajard, M. Chevallier, et al., “Monitoring
on the knife-edge slit camera design by Smeets et al.; the Bragg peak location of 73 MeV/u carbon ions by
means of prompt γ-ray measurements,” Applied Physics
and (2) prompt-gamma spectroscopy approach based
Letters 93(9), 093506, 2008.
on the design by Verburg and Seco [15]. Both achieved 5. C. H. Min, C. H. Kim, M. Y. Youn, J. W. Kim, “Prompt
the desired 1–2 mm accuracy and are able to perform gamma measurements for locating the dose falloff
both absolute and relative range verification. The chal- region in the proton therapy,” Applied Physics Letters
lenge is now heading synchrotron facilities, where the 89(18), 183517 (2006).
beam structure is different and every ion beam species 6. J. M. Verburg, K. Riley, T. Bortfeld, J. Seco, “Energy- and
time-resolved detection of prompt gamma-rays for pro-
has its own features. Also here there are two different ton range verification,” Physics in Medicine and Biology
approaches to the same problem, either considering the 58(20), L37–L49 (2013).
absolute prompt gamma yields or the reaction channels 7. T. T. Böhlen, F. Cerutti, M. P. W. Chin, et al., “The
associated to specific energy lines. This latter approach FLUKA Code: Developments and challenges for high
demands very precise experimental data from cross- energy and medical applications,” Nuclear Data Sheets
120, 211–214 (2014).
sections, as a function of materials, ions species and
8. J. Allison, K. Amako, J. Apostolakis, et al., “Recent devel-
energy. Moreover, in order to use the TOF information, opments in Geant4,” Nuclear Instruments and Methods
an external trigger detector has to be placed between the in Physics Research Section A 835, 186–225 (2016).
nozzle and the patient. This is actually a problem shared, 9. C. J. Werner, J. S. Bull, C. J. Solomon, et al., “MCNP ver-
to a less extent, with the cyclotron-based facilities as sion 6.2 release notes,” Technical report, Los Alamos
National Lab. (LANL), Los Alamos, NM (United States)
proton bunch drifts along time have been reported [32].
(2018).
Two other prototypes – Compton camera [25] and 10. M. Pinto, D. Dauvergne, N. Freud, J. Krimmer, J. M.
PGT [62] – are reaching the clinical stage, each of which Létang, E. Testa, “Assessment of Geant4 prompt-gamma
standing on distinct proof-of-principles. The CC will emission yields in the context of proton therapy moni-
enable 2D and 3D images of the Bragg peak and the PGT toring,” Frontiers in Oncology 6, 10 (2016).
will be easier to translate to the clinics due to its uncolli- 11. J. M. Verburg, H. A. Shih, J. Seco, “Simulation of prompt
gamma-ray emission during proton radiotherapy,”
mated setup. Both have drawbacks, either the computa- Physics in Medicine and Biology 57(17), 5459–5472 (2012).
tion time and complicated instrumentation of the CC or 12. F. Stichelbaut and Y. Jongen, “Verification of the proton
the load and throughput demands of the PGT. Both the beams position in the patient by the detection of prompt
CC and the PGS system, due to its spectroscopic nature gamma-rays emission,” in 39th Meeting of the Particle
may be able to quantify the elemental compositions of Therapy Co-Operative Group (2003), p. 16.
13. C. H. Min, J. W. Kim, M. Y. Youn, C. H. Kim,
irradiated tissues. This could be used to measure in vivo “Determination of distal dose edge location by measur-
the tissue density and the 16O and 12C concentration of ing right-angled prompt-gamma rays from a 38 MeV pro-
irradiated tumors. ton beam,” Nuclear Instruments and Methods in Physics
The PGI is now a consolidated technique for range Research Section A 580(1), 562–565 (2007).
verification and we hope this will bring increased accep- 14. P. Crespo, T. Barthel, H. Frais-Kolbl, et al., “Suppression
of random coincidences during in-beam PET mea-
tance and applications to radiotherapy with protons,
4He, 12C, and 16O.
surements at ion beam radiotherapy facilities,” IEEE
Transactions on Nuclear Science 52(4), 980–987 (2005).
15. J. M. Verburg and J. Seco, “Proton range verification
REFERENCES through prompt gamma-ray spectroscopy,” Physics in
Medicine and Biology 59(23), 7089–7106 (2014).
1. A. C. Kraan, “Range verification methods in particle 16. J. Smeets, F. Roellinghoff, D. Prieels, et al., “Prompt
therapy: Underlying physics and Monte Carlo model- gamma imaging with a slit camera for real-time range
ing,” Frontiers in Oncology 5 (2015). control in proton therapy,” Physics in Medicine and
2. G. Dedes, M. Pinto, D. Dauvergne, et al., “Assessment Biology 57(11), 3371–3405 (2012).
and improvements of Geant4 hadronic models in the 17. A. K. Biegun, E. Seravalli, P. Cambraia Lopes, et al.,
context of prompt-gamma hadrontherapy monitoring,” “Time-of-flight neutron rejection to improve prompt
Physics in Medicine and Biology 59(7), 1747–1772 (2014). gamma imaging for proton range verification: A simu-
3. M. Testa, M. Bajard, M. Chevallier, et al., “Real-time lation study,” Physics in Medicine and Biology 57(20),
monitoring of the Bragg-peak position in ion therapy 6429–6444 (2012).
by means of single photon detection,” Radiation and 18. C. Golnik, F. Hueso-González, A. Müller, et al., “Range
Environmental Biophysics 49(3), 337–343 (2010). assessment in particle therapy based on prompt γ-ray
timing measurements,” Physics in Medicine and Biology 33. F. Hueso-González, F. Fiedler, C. Golnik, et al.,
59(18), 5399–5422 (2014). “Compton camera and prompt gamma ray timing: Two
19. J. C. Polf, R. Panthi, D. S. Mackin, et al., “Measurement methods for in vivo range assessment in proton ther-
of characteristic prompt gamma rays emitted from apy,” Frontiers in Oncology 6, 80 (2016).
oxygen and carbon in tissue-equivalent samples dur- 34. S. W. Peterson, D. Robertson, J. Polf, “Optimizing a three-
ing proton beam irradiation,” Physics in Medicine and stage Compton camera for measuring prompt gamma
Biology 58(17), 5821–5831 (2013). rays emitted during proton radiotherapy,” Physics in
20. I. Mattei, G. Battistoni, F. Bini, et al., “Prompt-γ production Medicine and Biology 55(22), 6841–6856 (2010).
of 220 MeV/u 12c ions interacting with a PMMA target,” 35. T. Kormoll, F. Fiedler, S. Schöne, J. Wüstemann, K.
Journal of Instrumentation 10(10), P10034–P10034 (2015). Zuber, W. Enghardt, “A Compton imager for in-vivo
21. P. G. Thirolf, C. Lang, S. Aldawood, et al., “Development dosimetry of proton beams—a design study,” Nuclear
of a Compton camera for online range monitoring of Instruments and Methods in Physics Research Section A
Laser-accelerated proton beams via prompt-gamma 626–627, 114–119 (2011).
detection,” EPJ Web of Conferences 66, 11036 (2014). 36. S. Kurosawa, H. Kubo, K. Ueno, et al., “Prompt gamma
22. C. Richter, G. Pausch, S. Barczyk, et al., “First clinical detection for range verification in proton therapy,”
application of a prompt gamma based in vivo proton Current Applied Physics 12(2), 364–368 (2012).
range verification system,” Radiotherapy & Oncology 37. J. Krimmer, J. L. Ley, C. Abellan, et al., “Development
118(2), 232–237 (2016). of a Compton camera for medical applications based
23. F. Hueso-González, M. Rabe, T. A Ruggieri, T. Bortfeld, on silicon strip and scintillation detectors,” Nuclear
J. M. Verburg, “A full-scale clinical prototype for proton Instruments and Methods in Physics Research Section A,
range verification using prompt gamma-ray spectros- 787, 98–101 (2015).
copy,” Physics in Medicine and Biology 63(18), 185019 38. P. G. Thirolf, S. Aldawood, M. Böhmer, et al., “A
(2018). Compton camera prototype for prompt gamma medi-
24. R. Dal Bello, P. Magalhaes Martins, J. Seco, “CeBr3 scin- cal imaging,” EPJ Web of Conferences 117, 05005 (2016).
tillators for 4He prompt gamma spectroscopy: Results 39. A. Schumann, J. Petzoldt, P. Dendooven, et al.,
from a Monte Carlo optimization study,” Medical “Simulation and experimental verification of prompt
Physics 45(4), 1622–1630 (2018). gamma-ray emissions during proton irradiation,”
25. E. Draeger, D. Mackin, S. Peterson, et al., “3D prompt Physics in Medicine and Biology 60(10), 4197–4207
gamma imaging for proton beam range verification,” (2015).
Physics in Medicine and Biology 63(3), 035019 (2018). 40. L. Kelleter, A. Wrońska, J. Besuglow, et al., “Spectroscopic
26. K. Parodi and J. C. Polf, “In vivo range verification in study of prompt-gamma emission for range verification
particle therapy,” Medical Physics 45(11), e1036–e1050 in proton therapy,” Physica Medica 34, 7–17 (2017).
(2018). 41. P. Cambraia Lopes, E. Clementel, P. Crespo, et al.,
27. C. H. Min, H. R. Lee, C. H. Kim, S. B. Lee, “Development “Time-resolved imaging of prompt-gamma rays for
of array-type prompt gamma measurement system for proton range verification using a knife-edge slit camera
in vivo range verification in proton therapy,” Medical based on digital photon counters,” Physics in Medicine
Physics 39(4), 2100–2107 (2012). and Biology 60(15), 6063–6085 (2015).
28. J. Petzoldt, G. Janssens, L. Nenoff, C. Richter, J. Smeets, 42. M. Vanstalle, I. Mattei, A. Sarti, et al., “Benchmarking
“Correction of geometrical effects of a knife-edge slit Geant4 hadronic models for prompt-γ monitoring in car-
camera for prompt gamma-based range verification in bon ion therapy,” Medical Physics 44(8), 4276–4286 (2017).
proton therapy,” Instruments 2(4), 25 (2018). 43. I. Mattei, F. Bini, F. Collamati, et al., “Secondary radia-
29. Y. Xie, E. H. Bentefour, G. Janssens, et al., “Prompt gamma tion measurements for particle therapy applications:
imaging for in vivo range verification of pencil beam scan- Prompt photons produced by 4He, 12C and 16O ion
ning proton therapy,” International Journal of Radiation beams in a PMMA target,” Physics in Medicine and
Oncology, Biology, Physics 99(1), 210–218 (2017). Biology 62(4), 1438–1455 (2017).
30. P. Magalhaes Martins, R. Dal Bello, A. Rinscheid, et al., 44. P. Magalhaes Martins, R. Dal Bello, G. Hermann, et al.,
“Prompt gamma spectroscopy for range control with “Proceedings of the 57th annual meeting of the particle
CeBr3,” Current Directions in Biomedical Engineering therapy cooperative group (PTCOG),” International
3(2), 113–117 (2017). Journal of Particle Therapy 5(2), 58–229 (2018).
31. F. Hueso-González, W. Enghardt, F. Fiedler, et al., “First 45. B. D. Leverington, M. Dziewiecki, L. Renner, R. Runze,
test of the prompt gamma ray timing method with het- “A prototype scintillating fibre beam profile monitor for
erogeneous targets at a clinical proton therapy facility,” ion therapy beams,” Journal of Instrumentation 13(05),
Physics in Medicine and Biology 60(16), 6247–6272 (2015). P05030–P05030 (2018).
32. J. Petzoldt, K. E. Roemer, W. Enghardt, et al., 46. J. Krimmer, D. Dauvergne, J. M. Létang, É. Testa, “Prompt-
“Characterization of the microbunch time structure gamma monitoring in hadrontherapy: A review,” Nuclear
of proton pencil beams at a clinical treatment facility,” Instruments and Methods in Physics Research Section A
Physics in Medicine and Biology 61(6), 2432–2456 (2016). 878, 58–73 (2018).
47. E. Testa, M. Bajard, M. Chevallier, et al., “Dose profile mon- 56. M. Durante, R. Orecchia, J. S. Loeffler, “Charged-particle
itoring with carbon ions by means of prompt-gamma therapy in cancer: Clinical uses and future perspec-
measurements,” Nuclear Instruments and Methods in tives,” Nature Reviews Clinical Oncology 14(8), 483–495
Physics Research Section B 267(6), 993–996 (2009). (2017).
48. M. Pinto, M. Bajard, S. Brons, et al., “Absolute 57. M. Durante, “Proton beam therapy in Europe: More
prompt-gamma yield measurements for ion beam centres need more research,” British Journal of Cancer
therapy monitoring,” Physics in Medicine and Biology 120(21), 777–778 (2019).
60(2), 565–594 (2014). 58. J. M. Verburg, C. Grassberger, S. Dowdell, J. Schuemann,
49. C. Agodi, F. Bellini, G. A. P. Cirrone, et al., “Precise mea- J. Seco, H. Paganetti, “Automated Monte Carlo simula-
surement of prompt photon emission from 80 MeV/u tion of proton therapy treatment plans,” Technology in
carbon ion beam irradiation,” Journal of Instrumentation Cancer Research & Treatment 15(6), NP35–NP46 (2016).
7(03), P03001–P03001 (2012). 59. S. Mein, K. Choi, B. Kopp, et al., “Fast robust dose cal-
50. J. Seco and M. F. Spadea, “Imaging in particle therapy: culation on GPU for high-precision 1H, 4He, 12C and 16O
State of the art and future perspective,” Acta Oncologica ion therapy: The FRoG platform,” Scientific Reports 8(1)
54(9), 1254–1258 (2015). (2018).
51. W. Enghardt, P. Crespo, F. Fiedler, et al., “Charged 60. L. Tian, G. Landry, G. Dedes, et al., “Toward a new treat-
hadron tumour therapy monitoring by means of PET,” ment planning approach accounting for in vivo proton
Nuclear Instruments and Methods in Physics Research range verification,” Physics in Medicine and Biology
Section A 525(1), 284–288 (2004). 63(21), 215025 (2018).
52. M. Moteabbed, S. España, H. Paganetti, “Monte Carlo 61. S. Schmid, G. Landry, C. Thieke, et al., “Monte Carlo
patient study on the comparison of prompt gamma and study on the sensitivity of prompt gamma imaging to
PET imaging for range verification in proton therapy,” proton range variations due to interfractional changes
Physics in Medicine and Biology 56(4), 1063–1082 (2011). in prostate cancer patients,” Physics in Medicine and
53. A. C. Knopf and A. Lomax, “In vivo proton range veri- Biology 60(24), 9329–9347 (2015).
fication: A review,” Physics in Medicine and Biology 62. T. Werner, J. Berthold, W. Enghardt, et al., “Range veri-
58(15), R131–R160 (2013). fication in proton therapy by prompt gamma-ray timing
54. K. Parodi, “On- and off-line monitoring of ion beam (PGT), steps towards clinical implementation,” in IEEE
treatment,” Nuclear Instruments and Methods in Physics Nuclear Science Symposium Conference Record 2017
Research Section A 809, 113–119 (2016). (2017), pp. 1–5.
55. D. K. Ebner and T. Kamada, “The emerging role of carbon-
ion radiotherapy,” Frontiers in Oncology 6, 140 (2016).
Chapter 28
Acoustic-Based Proton
Range Verification
Kevin C. Jones
Rush University Medical Center
Chicago, Illinois
CONTENTS
28.2 Underlying Phenomenon 444
28.3 Protoacoustic Emissions 445
28.3.1 Pulse Width and Rise Time 447
28.3.2 Arrival Time 449
28.3.3 Frequency 449
28.4 Current State of Protoacoustics 450
28.4.1 Protoacoustic Image Formation 450
28.4.2 Time-of-Flight Protoacoustic Range Verification 450
28.4.3 Heterogeneity 452
28.4.4 Challenges 453
References 453
28.1 INTRODUCTION actual depth, the dose deposited by a photon beam

The spatial dose deposition provided by particle therapy will change by < 1% per millimeter due to the low dose
is fundamentally different than that from photon ther- gradient beyond d max. For particle beams, if the radio-
apy. Megavoltage photons exhibit low skin dose, a peak at logical depth is different than planned, the dose to the
1–4 cm depth, and a monotonic decrease (inverse square distal edge of the tumor may change from 100% to 0%
and exponential) in dose thereafter. In contrast, heavy (undershoot) or 0% to 100% for normal tissue down-
charged particles deposit a large fraction of their total stream of the tumor edge (overshoot). Due to the steep
dose in the last few centimeters of travel in the Bragg peak. gradient downstream of the Bragg peak, even a couple
Due to the absence of exit dose beyond the Bragg peak, of millimeters of depth error may cause these drastic
the dose deposition provided by particle therapy is often changes in delivered dose, and the biological reper-
considered superior to that of photons because structures cussions are further exacerbated by the high relative
immediately downstream from the target may be spared. biological effectiveness for low energy protons at the
Although the lack of exit dose is advantageous, end of their range [3]. Although the energy of the beam
particle therapy irradiation is also associated with exiting the treatment head and the range in water are
increased risks due to its strong dependence on depth known to high accuracy (≤600 μm) [4], expected and
[1, 2]. If the expected tumor depth does not match the actual depths in a patient may differ for many reasons,
443
including depth calculation uncertainties, setup errors, most clinical accelerators compared to the ideal pulses
or patient anatomy changes [2]. One careful consider- for protoacoustic signal generation.
ation predicts an uncertainty in proton range of up to The goal of this chapter is to provide the reader with
4.6% of the range +1.2 mm [1]. an understanding of the underlying physics of proto-
The uncertainty in delivered particle range and the acoustics, the research that has been done in the field,
associated risks have prevented using particle therapy to and the challenges facing the technique.
its full potential. Clinically, an uncertainty of up to 3.5%
of the range +1–3 mm is assumed in planning for pro- 28.2 UNDERLYING PHENOMENON
ton beams. Due to the possibility of overshooting, beam As the name suggests, thermoacoustic phenomena con-
arrangements are chosen to avoid placing organs at risk vert thermal energy into acoustic pressure waves. Fields
directly downstream from the tumor [2]. In addition to of thermoacoustics differ based on the initial source of
considering the ideal, planned dose deposition, particle energy. For example visible light [27] (optoacoustics or
therapy plans are assessed for their robustness to possi- photoacoustics), radio wave radiation [28], x-ray radia-
ble range errors [5]. Robustness consideration increases tion [29], and particle beams [30] have all been used as
calculation time and adds complexity to physician deci- excitation sources [31]. The underlying analytical equa-
sion making. tions that govern the emission of acoustic waves in
Many researchers have focused on developing in vivo response to these various forms of heat deposition are
techniques for determining the particle dose deposition general to all of the aforementioned techniques.
delivered to the patient. These considered techniques In a lossless, homogeneous medium, the pressure p

have largely been focused on the most popular form of reaching a detector positioned at r at time t is given by
particle therapy – proton therapy – and the field of study two equations (Equations 28.1 and 28.2). For a delta-
has been termed proton range verification. Different function, instantaneous heating pulse, the pressure is
explored methods include positron emission tomogra- given by [18]:
phy (PET) using proton-activated nuclei that decay with
positron emission [6], prompt gamma detection using  
cβ d  E (t =| r − r ′ | /c ) 
2π π

proton-activated nuclei that decay with gamma emis-
sion (imaging [7, 8], spectroscopy [9], and time-of-flight
pδ (r , t ) = 
4π C p dt 
0 0
∫∫(tc )2 sinθ dθ dφ 3 D
tc



[10] techniques), and thermoacoustic-based techniques (28.1)
[11–26] which are the focus of this chapter.
In general, thermoacoustic techniques gather infor- where c (m s−1) is the sound speed, β is the thermal expan-
mation about deposited energy by detecting the pression coefficient (K−1), and Cp is the specific heat capacity

sure waves emitted as the energy is converted to heat. (J kg−1 K−1). E3D(r ′) is the three dimensional spatial energy
When applied to particle range verification, the general deposition (J m−3). The double integral in Equation (28.1)
thermoacoustic technique has been called ionoacous- describes a spherical surface integral centered at the

tics [15], RACT [16], and protoacoustics [13]. Because detector at r , where the radial distance variable has been
the majority of the literature is focused on determin- converted to the product of time and sound speed. The
ing the range of proton beams, the term protoacoustics pressure wave reaching the detector originates on the
will be used here. Protoacoustics is an attractive proton spherical surface, and its arrival time is determined by
range verification technique because it is non-invasive, its distance from the detector divided by the travel speed,
 
the signal may be collected in vivo in real-time, and | r − r ′ | /c. The magnitude of the pressure depends on how
the signal is linearly proportional to deposited dose efficiently the deposited energy is converted into a pres-

(although a heat defect must be considered). In addi- sure change at r′. This conversion efficiency is described
tion, the system is conceptually simple and is projected by the Grüneisen parameter, Γ, a dimensionless (Pa/(J/m3)
to be of low cost. The challenges facing clinical trans- cancels out) material-specific parameter equivalent to the
lation include low signal levels, potential errors due to prefactor in Equation (28.1) (once c is brought out of the
heterogeneous sound speeds in tissue, limited acoustic integral): Γ = c 2 β /C p . The pressure contributed by each
transmission through air pockets and bone, and the dif- point or voxel is further scaled by the inverse of the dis-
ference in time-structure of proton pulses produced by tance (tc in the denominator).
Acoustic-Based Proton Range Verification ◾ 445
The integral in Equation (28.1) is always positive, and Lp/c (the stress confinement threshold), thermal confine-
its maximum occurs at the time (or distance divided ment is generally not a concern for protoacoustics.
by c) that corresponds to the spherical shell that overlaps If the medium is not homogeneous, c, β, and Cp will

the most deposited energy E3D. Each point is expanded all vary in space, r ′. One can still conceivably apply
in response to the heating (unless β is negative). Due to Equation (28.1) with these varying material-specific
the time derivative, however, the detected pressure has properties brought into the integral, but the delay time
 
positive and negative peaks. Conceptually, these posi- between pressure emitted at r ′ and detected at r would
tive and negative peaks are caused by gradients in the depend on the speed of sound integrated along the line
 
heated volume. The largest magnitude pressure peaks between these two points t =| r − r ′ | /cavg . Equation (28.1)
will be observed when there are large spatial gradients does not consider attenuation, scattering, or diffraction
in deposited heat, E3D. of acoustic waves. Therefore, propagation of thermoa-
If the heating excitation is deposited in a finite time coustic waves in heterogeneous media is typically simu-
(the heating pulse is not a delta-function), then the pres- lated using more sophisticated techniques that solve the

sure detected is given by the convolution of pδ (r , t ) with underlying wave equation.
the temporal shape of the proton pulse Et(t) (s−1):
28.3 PROTOACOUSTIC EMISSIONS
 
p(r , t ) = pδ (r , t ) ⊗ Et (t ). (28.2) Proton dose depositions generate acoustic emissions [30].
The thermoacoustic origin of these emissions was
One can think of the excitation pulse Et as the linear proven in the seminal 1979 Sulak et al. paper [30], in
combination of many delta-function pulses, each of which the authors demonstrate a linear relationship
which initiates the emission of its own set of pressure between acoustic amplitude and β/Cp. This relation-
waves. The detected pressure wave is the sum of all of ship is shown by measuring proton-induced acoustic
these contributions. An interesting result of Equation signals in different materials and in water at different
(28.2) is that the magnitude of the detected pressure also temperatures. The most striking of these experimental
depends on the gradients in the excitation pulse Et along results is the disappearance and inversion of acous-
with the spatial gradients discussed above. Because pδ tic amplitude as the water temperature is decreased
has negative and positive peaks, convolution with a below 4°C, the temperature at which water reaches its
slowly varying Et gives low intensity, broadened pres- densest state and the expansion coefficient becomes
sure waves. negative. These results are shown in Figure 28.1. The
These relationships between spatial and temporal discrepancy between the observed 6°C and expected
gradients and the emitted pressure waves are elegantly 4°C inversion point is not resolved (although subse-
encapsulated in the idea of stress confinement [27]. If quent measurements [32] report 4°C). The authors also
the heat is deposited at a rate slower than the transit show that the acoustic emissions are not the result of
time across the heated volume, then spatial confinement microbubble formation – another potential underly-
is not achieved, pressure leaks out of the volume faster ing phenomenon – by demonstrating that the acoustic
than heat is deposited, and the temporal characteris- amplitude is independent of ambient pressure, nucleat-
tics of the heating pulse affect (generally broaden) the ing ion content, and N2 content.
detected pressure waves. A simple rule-of-thumb equa- In a homogeneous medium, protons deposit their
tion states that stress confinement is achieved when the dose in pencil beam kernels. A narrow pencil beam
excitation pulse is shorter in time than Lp/c, where Lp is will generate a dose deposition with an approximately
the characteristic linear dimension of the heated volume Gaussian radial profile – the width of which expands
E3D. Loss of stress confinement is usually detrimental, as with depth – and an integrated depth profile that is fairly
the associated broadening results in loss of the desired constant until the last couple centimeters of its range,
dosimetric spatial information that would otherwise when a large dose is deposited at the Bragg peak before
define the shape of the protoacoustic waves. Another abruptly stopping. Even if the protons are delivered in
consideration is that of thermal confinement, the idea a broad field, one may think of this broad field as being
that heat may diffuse out of the volume faster than Et. composed of a linear combination of pencil beams,
Due to the slow rate of thermal diffusion compared to each of which is depositing its own pencil beam kernel.
capped by a high dose disc (the Bragg peak). Each of

these shapes contributes to a macroscopic protoacous-
tic pressure wave. By placing a set of detectors in a line
parallel to the proton dose deposition, the origin of the
different pressure waves may be determined. Figure 28.2
shows the experimentally measured protoacoustic waves
arriving at such a row of detectors [34]. In this figure,
there are three sets of protoacoustic waves. The waves
along A-E are due to the entrance window, and are not
of particular interest to proton range verification. The
waves along C-B-D are termed the γ waves, and those
along A-B are the α waves [34]. Each wave is a bipolar
pressure wave in which the compression (positive) peak
precedes the rarefaction (negative) peak.
The γ wave is generated by the Bragg peak disc, and
its arrival time depends on the distance between the
Bragg peak and the detector, labeled as l. As a pseudo
point source, the pressure falls as 1/R with increasing
l [18, 30].
The α wave is generated by the cylindrical, pre-
Bragg peak plateau, and its arrival time is related to
FIGURE 28.1 The protoacoustic amplitude measured in the radial distance, labeled as s, between the proton
water as a function of initial temperature. An inversion point beam axis and the detector. As a pseudo line source,
is observed at 6°. (Reprinted with permission from Elsevier:
the pressure falls as 1/ R when the detector is close
Elsevier, Nuclear Instruments and Methods, Experimental
to the beam axis (s is on the order of the range of the
studies of the acoustic signature of proton beams travers-
ing fluid media, L. Sulak, T. Armstrong, H. Baranger, M. proton) [18]. When the detector is at the depth of the
Bregman, M. Levi, D. Mael, J. Strait. T. Bowen, A.E. Pifer, A. Bragg peak (for example, marked at B in Figure 28.2),
Polakos, H. Bradner, A. Parvulescu, W.V. Jones, J. Learned, the α wave overlaps with the γ wave (s = l). At detector
161, 203–217, 1979.) depths deeper than the Bragg peak, all protoacoustic
waves generated by the pre-Bragg peak cylinder con-
Clinical proton beams have a depth penetration of up tribute to the γ wave. Therefore, there is no distinct α
to 30 cm and radial full-width-half-max (FWHM) pro- wave at depths beyond the Bragg peak.
files of around 0.5–1.5 cm [33]. Clinical proton beams The hypothesis of protoacoustic proton range veri-
are commonly delivered as many pencil beams (pen- fication techniques is that the pressure waves carry
cil beam scanning – PBS) or as a broad beam (double information about the proton dose deposition. How
scattering). For PBS delivery, the depth and horizontal/ this information is used distinguishes particular proto-
vertical positioning is controlled by changing the energy acoustic approaches. At one extreme, Radiation-Induced
and rastering with magnetic fields. For double scatter- Acoustic Computed Tomography (RACT) collects the
ing delivery, the number of protons (current) is modu- protoacoustic pressure traces from many (ideally all)
lated as a function of time as a spinning wheel of varying detector positions surrounding the dose deposition, and
thickness shifts the energy/range of the delivered pro- reconstructs the 3D deposition though tomographic
tons to conform in depth to the far borders of the tar- back-projection approaches [16]. On the other extreme,
get. The double scattered beam aperture is controlled by single or a sparse set of detectors are used to measure the
collimating jaws and a field-specific compensator, which Bragg peak and beam axis positions through time-of-
further conform the depth dose to the far edge of the flight calculations based on the arrival time of the differ-
tumor. ent peaks [13, 15]. Specifically, l ≈ cτ γ and s ≈ cτ α , where
Conceptually, the proton dose deposition can be τγ and τα are the arrival times of the protoacoustic γ and
approximated by a cylinder (pre-Bragg peak plateau) α waves, respectively.
FIGURE 28.2 Protoacoustic emissions collected along a line parallel to the proton beam axis at offsets of 2.2 (I), 6.2 (II), and
13.6 (III) cm. (Reprinted with permission from Springer Nature: Springer Nature, Instruments and Experimental Techniques,
Measuring the Ultrasonic Field Generated in Water upon the Deceleration of a Proton Beam, V. I. Albul, V. B. Bychkov, S.
S. Vasil’ev, K.E. Gusev, V.S. Demidov, E.V. Demidova, N.K. Krasnov, A.F. Kurchanov, V.E. Luk’yashin, A. Yu. Sokolov, 47,
502–506, 2004.)
28.3.1 Pulse Width and Rise Time protoacoustic signal. Two important characteristics of
As indicated in Equation (28.2) and described above the proton pulse’s temporal shape are its pulse width,
with stress confinement, the temporal characteristics usually defined as the FWHM, and the rise time, usually
of the proton pulse affect the measured protoacous- defined by the time to rise from 10% to 90% or 20% to
tic signal. If stress confinement is not achieved, the 80% of the maximum. Two of the commonly considered
time derivative of the proton pulse is imprinted on the pulse shapes are rectangular and Gaussian.
For rectangular proton pulses, the rise and fall of middle portion of a rectangular proton pulse make it
the proton pulse generate a positive and negative pro- non-ideal.
toacoustic peak, respectively, whose separation corre- Gaussian proton pulses are easily modeled, and have
sponds to the pulse width of the rectangular pulse [13, been the focus of many simulations. Gaussian shapes
15, 30]. Interestingly, if the rise and/or fall times are have also been observed experimentally [17, 22, 25].
fast enough, the beginning and/or end of the proton Unlike rectangular pulses, in which the central, flat
pulse will satisfy stress confinement even if the pulse portion does not contribute to the protoacoustic sig-
width is too long. For example, when the rectangular nal, Gaussian shapes are advantageous because the
pulse width is increased the delay between positive fall immediately follows the rise in proton amplitude.
and negative acoustic peaks also increases. Even as the Studies investigating the ideal proton pulse have indi-
pulse width is increased, the positive and negative fea- cated that proton pulses with a shape that matches the
tures are unaffected, and retain their sharp features. spatially defined protoacoustic wave generated by a
In this case, the individual positive- and negative- delta-function proton pulse, pδ, are best suited to maxi-
associated protoacoustic peaks will themselves report mizing the protoacoustic signal-to-noise ratio (SNR) [18].
directly on the spatial dose deposition. Although this The reason for this is due to two effects. First, the effi-
might appear attractive – that quasi-stress confinement ciency – amplitude per deposited dose – of generating
may be achieved by abruptly turning on and off a rect- protoacoustic amplitude increases for shorter pulses
angular pulse even for a proton pulse with arbitrarily (Figure 28.3a). Second, the SNR increases linearly with
long pulse width – the dose deposited in the middle signal amplitude and with the square root of the num-
of the rectangular pulse generates no protoacous- ber of averages. Although the efficiency effect suggests
tic signal due to the equilibrium, constant heating. shorter pulses are better, optimal SNR is achieved by
Given that protoacoustic measurements are primarily generating the highest possible amplitude per pulse
limited by deposited dose (described later), the dose- when the cumulative dose limits the number of repeated
depositing – but protoacoustically non-contributing – irradiations (averages) that can be delivered (Figure 28.3b).
FIGURE 28.3 (a) The maximum simulated protoacoustic pressure (left axis) and protons per Gaussian pulse (right axis) is
plotted as a function of 10–90% rise time (300 nA peak proton current, detector 5 cm distal to Bragg peak). (b) The maxi-
mum simulated protoacoustic signal per deposited dose (single pulse) is plotted on the left axis versus Gaussian proton pulse
FWHM. The simulated SNR per total deposited dose is also plotted versus proton pulse length (b, right axis). Shorter pulses
deposit less dose per pulse and, therefore, allow more averaging, but the absolute protoacoustic amplitude generated per pulse
is lower than generated by longer pulses. As a result, for an equivalent total dose, the SNR is expected to peak at an intermedi-
ate pulse length. (c) As the proton pulse peak current is increased, the SNR per total deposited dose also increases. (Reprinted
from K.C. Jones, C.M. Sehgal S. Avery, How proton pulse characteristics influence protoacoustic determination of proton-
beam range: Simulation studies, Physics in Medicine and Biology 61, 2213–2242, 2016 © Institute of Physics and Engineering
in Medicine. Reproduced by permission of IOP Publishing. All rights reserved.)
The highest amplitude is generated when the proton pulse peak maximum and the initial protoacoustic posi-
pulse matches the pδ peaks because the convolution tive peak maximum, then an additional correction, Sinher,
overlap is greatest. It follows that because pδ depends on is required to accurately convert the arrival time into dis-
the spatial dimensions of the dose deposition, the ideal tance to the Bragg peak: l = τγc + Sinher [19]. This correc-
proton pulse will change depending on proton beam tion will depend on arrival time definition feature, proton
spot size and energy. For clinical beams, which have a pulse width and characteristics, and angular position of
spot size FWHM ~1 cm and penetrate ~1–30 cm, the the detector relative to the Bragg peak.
ideal proton pulse for maximizing SNR is a Gaussian-
like pulse with a FWHM ~ 5 μs. 28.3.3 Frequency
Both the spatial characteristics of the dose deposition
28.3.2 Arrival Time and the temporal characteristics of the excitation pro-
The protoacoustic pressure wave is bipolar, with positive ton pulse may shape the emitted protoacoustic waves.
and negative peaks. This bipolar structure complicates Therefore, the frequency of the protoacoustic spectrum –
analysis. For time-of-flight based proton range verifica- and the appropriate detectors – are affected by both the
tion, the arrival time of the protoacoustic wave must be proton pulse and the proton spot size and energy [16,
identified [18, 24]. Intuitively, it is not clear which feature 30]. For clinically relevant proton beams, the pδ has a
to identify as the arrival time (for example, positive and frequency that peaks at ~50 kHz (Figure 28.4) [18].
negative peaks are two possibilities). Simulations have Interestingly, the protoacoustic peaks emitted along
indicated that the true arrival time for pδ that gives l = the beam direction have higher frequency spectra than
τγc is located somewhere between the positive and nega- those emitted laterally because the Bragg peak is nar-
tive peak, but this “somewhere” depends on the measure- rower in the depth direction than it is radially. Due to
ment position relative to the Bragg peak [18]. Jones et al. the expected <100 kHz frequency spectrum, the typical
proposed calibrating the system for a particular arrival selected detector is the hydrophone, a water-proof trans-
time definition [18, 19]. For example, if one chooses to ducer that is sensitive to lower frequency pressure waves
define the arrival time as the delay between the proton than standard ultrasound imaging transducers.
FIGURE 28.4 (a) Protoacoustic frequency spectra generated by Gaussian proton pulses simulated for detectors placed sur-
rounding a 150 MeV dose deposition. (b) The protoacoustic frequency spectra simulated at a detector placed on the proton
beam axis but deeper than the Bragg peak. (c) As the Gaussian proton pulse FWHM is increased, the frequency spectrum
peaks at lower frequencies. (Reprinted from K.C. Jones, C.M. Sehgal S. Avery, How proton pulse characteristics influence pro-
toacoustic determination of proton-beam range: Simulation studies, Physics in Medicine and Biology 61, 2213–2242, 2016 ©
Institute of Physics and Engineering in Medicine. Reproduced by permission of IOP Publishing. All rights reserved.)
28.4 CURRENT STATE OF PROTOACOUSTICS 28.4.1 Protoacoustic Image Formation

Recent work in the field of protoacoustics has included There are two general approaches to protoacoustic range
both simulation and experiment. Simulations have been verification: (1) using many detectors to generate an
performed by numerically assessing the analytical equa- image and (2) using a few detectors to determine dis-
tion (Equations 28.1 and 28.2) [16, 18, 19, 30, 35] and tances between Bragg peak and detector. Work toward
with the k-Wave MATLAB toolbox [13, 15, 23], a ver- the first (imaging) approach is described in this section,
satile program that allows inclusion of attenuation and while the second (triangulation) is described in the next
heterogeneity by solving the wave equation with a psue- section.
dospectral model [36]. Although there have been no experimental reports,
Experimental measurements require a detector, a simulations suggest that a cup-like array of 71 × 179 =
triggered data collection instrument (oscilloscope), and 12709 detectors will provide sufficient protoacoustics
a pulsed proton source. The detectors that have been signal-to-noise ratio to reconstruct a 3D image of a 1.6
used include hydrophones [11, 17, 22, 30, 37], ultrasound cGy (at Bragg peak) clinical proton dose deposition with
transducers for >1 MHz frequencies (low energy and range verification accuracy of 1 mm [16] (homogeneous
short pulse proton beams) [15, 21, 26], strain gauges water medium, 1 μs rectangular pulse width, using typi-
[38], and laser vibrometers [38]. For triggering data col- cal noise for a hydrophone). Because the reconstruction
lection, direct electronic signals from the proton accel- relies on the number of projection viewpoints, many
erator are often preferred [17], but scintillator-based detectors are needed, and they must be positioned all
prompt gamma detectors are versatile and do not rely around the dose distribution. If a higher proton dose
on such access [22]. is acceptable, the measurement can be collected with a
Although the above experimental components smaller array of detectors that are rotated around the
(detector and oscilloscope) are relatively simple, com- beam. For example, a single L-shaped line of 71 detec-
mon, and/or inexpensive, the largest limitation on tors may be rotated around the beam at 2° intervals dur-
experimental studies has been (and will continue to be) ing irradiation with 179 pulses in order to build up an
access to appropriate proton beams. Limited access is equivalent dataset.
partially due to the sparsity of proton centers, but it is As shown in Figure 28.5, 2D protoacoustic dose
further magnified by the disconnect between typical deposition image reconstruction has been demonstrated
clinical proton beams and those required for efficient experimentally using ultrasound transducer arrays to
protoacoustic signal generation. For protoacoustic sig- collect the high frequency protoacoustic signal gener-
nal generation, the proton source must be pulsed and ated by low energy (50 MeV), short pulse (1.5 μs) pro-
the pulses should have rise times of <20 μs. Ideally, the ton beams (>2000 Gy, 1.5 mm accuracy) [21]. This work
beams should be of clinical energies (up to 230 MeV), is exciting because it produces 2D images of the Bragg
have instantaneous peak currents of 100s of nA, and peak using common ultrasound technology, but the dif-
have pulse widths ~5 μs. Generating proton pulses with ferences in protoacoustic frequency generated by these
<20 μs rise times has been challenging because the most low energy proton beams compared to clinical beams
common clinical proton accelerators – synchrotrons and prevent use of these ultrasound transducers in proton
isochronous cyclotrons – generate pulses with rise time therapy clinics.
of ~200 and 50 μs, respectively [39]. Therefore, research-
ers have utilized research accelerators with <1 μs short 28.4.2 Time-of-Flight Protoacoustic
pulses [11, 15, 37], chopped the beam output [21], or they Range Verification
have modified clinical accelerators, such as modulat- Instead of generating images based on signal collected
ing the output of an isochronous source [17]. Although from many detectors, the protoacoustic wave arrival
these modifications are not (currently) appropriate for time measured at one or a few detectors may be used to
generating proton beams for treating patients, the newer determine the Bragg peak position. This was first dem-
clinical proton synchrocyclotron accelerators (e.g., IBA onstrated experimentally by detecting the protoacoustic
S2C2 [22] and Mevion S250 [25]) generate ideal proton signal generated by low energy (20 MeV), short pulse
pulses (4–5 pC in 10 μs for the IBA S2C2 [40]) for proto- (≤ 4.3 μs rectangular pulse with 3 ns rise time) proton
acoustic signal generation without modification. beams (projected 1.6 Gy at the Bragg peak) with a single
FIGURE 28.5 (a–c) The protoacoustic data generated by a 50 MeV proton beam was collected with an ultrasound transducer.
The reconstructed Bragg peak (d–e) corresponds to the Monte Carlo calculated Bragg peak position (g–i). LDPE refers to
the tank wall. The measurements were repeated in three configurations, in which the transducer is moved 1 cm left (a, d, g),
centered (b, e, h), and 1 cm right (c, f, i) of the proton beam. (Reprinted from S.K. Patch, M.K. Covo, A. Jackson, Y.M.
Qadadha, K.S. Campbell, R.A. Albright, P. Bloemhard, A.P. Donoghue, C.R. Siero, T.L. Gimpel, S.M. Small, B.F. Ninemire,
M.B. Johnson, L. Phair, Thermoacoustic range verification using a clinical ultrasound array provides perfectly co-registered
overlay of the Bragg peak onto an ultrasound image, Physics in Medicine and Biology 61, 5621–5638, 2016 © Institute of
Physics and Engineering in Medicine. Reproduced by permission of IOP Publishing. All rights reserved.)
element 3.5 MHz ultrasound transducer [15]. With the proton current of 860 nA (for the 17 μs proton pulse), as
detector placed downstream of the Bragg peak, a range is shown in Figure 28.6. Although these measurements
verification accuracy of ≤ 100 μm was demonstrated. were performed at a clinical center with a clinical cyclo-
For proton beams of clinical energy and spot size tron, the proton pulsing was generated in an “off-label”
measured in a water tank, Jones et al. reported a pro- procedure – currently unsuitable for treating patients –
toacoustic range verification standard deviation of 2.0 mm to achieve short pulses and high currents.
based on data collected at many detector positions sur- Characterization of range verification in water
rounding the dose deposition (17 μs Gaussian-like pro- using protoacoustic signals generated by a clinical
ton pulse, up to 65 Gy) [19]. Extrapolation based on the synchrocyclotron was reported by Lehrack et al. [22].
detector noise in the treatment room led to a predic- Synchrocyclotrons generate clinical proton pulses with
tion of 2.2 mm range verification precision (standard FWHM of 1–10 μs, which is ideal for generating proto-
deviation) with 2 Gy of deposited dose assuming a high acoustic signals. For a detector placed at depth beyond
FIGURE 28.7 The proton beam range measured from

the protoacoustic arrival time is plotted versus the num-
FIGURE 28.6 For the experimentally generated 17 μs exci- ber of averages and dose. The proton beam is generated by
tation proton pulse, the projected precision (standard devia- a synchrocyclotron source. (Reprinted from S. Lehrack, W.
tion) of measuring the protoacoustic time-of-flight arrival Assmann, D. Bertrand, S. Henrotin, J. Herault, V. Heymans,
time is plotted versus total dose. If the peak proton current F. Vander Stappen, P.G. Thirolf, M. Vidal, J. Van de Walle,
can be increased to ~900 nA (10 × 107 protons/pulse), then a K. Parodi, Submillimeter ionoacoustic range determination
projected precision of 2.2 mm can be achieved with a dose of for protons in water at a clinical synchrocyclotron, Physics in
2 Gy. (Reprinted with permission from John Wiley and Sons: Medicine & Biology 62, L20–L30, 2017 © Institute of Physics
John Wiley and Sons, Medical Physics, Acoustic time-of- and Engineering in Medicine. Reproduced by permission of
flight for proton range verification in water, Kevin C. Jones, IOP Publishing. All rights reserved.)
François Vander Stappen, Chandra M. Sehgal, Stephen Avery
43, 5213–5224, 2016.) 28.4.3 Heterogeneity
the Bragg peak, submillimeter accuracy (root mean Tissue heterogeneity presents challenges for both proto-
square error of 0.8 mm) was achievable with <10 Gy of acoustic image reconstruction and time-of-flight range
maximum deposited dose (2.5–3.7 μs FWHM proton verification methods. Tissue heterogeneity affects both
pulses). Multiple beam energies and ranges were exam- protoacoustic generation – through material-dependent
ined, and a 1 mm accuracy is achievable at <3 Gy of Grüneisen parameters – and protoacoustic wave propa-
deposited dose, as is shown in Figure 28.7. gation – through material-dependent speed of sound
These initial water tank results investigating the rela- and attenuation. As a result, the initial protoacoustic
tionship between deposited dose and accuracy are prom- pressure distribution will be more complicated than
ising. The goal is to achieve 1 mm accuracy with doses the smooth pencil-beam dose distributions observed in
< 2 Gy. Looking forward, moving to heterogeneous tis- homogeneous material. Furthermore, the macroscopic
sue is expected to decrease measured protoacoustic sig- α and γ waves, which are the result of constructive inter-
nal amplitude due to acoustic attenuation, but there is ference from emissions all along the dose deposition,
also higher expected initial pressure due to the higher may be distorted due to the initial pressure distribu-
Grüneisen parameter (Figure 28.1). Also, low-frequency tion and/or by propagation through non-homogeneous
acoustic noise from the patient (e.g., breathing, blood material.
flow, digestion), which is not present for these studies, To characterize these effects, Jones et al. performed
is expected to decrease the protoacoustic SNR. One CT-based protoacoustic simulations of heteroge-
method for increasing the SNR without additional dose, neous material [23]. Material-dependent properties
is to use multiple detectors. were mapped based on Hounsfield unit values using
parameters drawn from literature. Proton beams irra- skull – brain. On the technical side, standard synchro-
diated the liver and prostate, and the protoacoustic tron and isochronous cyclotron beam accelerators do
pressure waves reaching detectors on the patient’s skin not generate ideal proton pulses for protoacoustic sig-
or in a transrectal probe were simulated. Under noise- nal generation. Newer synchrocyclotron proton sources,
less conditions, accuracies of ≤1.6 mm were reported however, do produce and treat patients with proton
(δ -function proton pulses) using multiple detectors pulses that are ideal for protoacoustic signal generation.
to triangulate the Bragg peak position. These calcula- If the protoacoustic range verification methods prove
tions were performed assuming the central axis of the successful, protoacoustic detectors must be integrated
proton beam was known, however. For a longer, 14 μs into the clinical workflow (perhaps with additional
FWHM proton pulse, the accuracy suffered (≤5.8 mm). ultrasound imaging), and the resulting range informa-
Attenuation in the liver case resulted in a loss in ampli- tion must be acted on to either adjust the treatment in
tude of a factor of two. Interestingly, time-of-flight dis- real-time or adapt for future fractions.
tance calculations varied minimally (≤0.7 mm) when a Current researchers in the field have not yet advanced
generic water sound speed was used and compared to to the stage of measuring protoacoustic signals in patients,
using a heterogeneity-corrected sound speed. Overall, an experiment performed in 1995 [12]. Unfortunately,
the simulations showed that amplitudes and range veri- the publication that resulted from this study is brief and
fication accuracy are detrimentally affected by hetero- the results are difficult to interpret. Looking forward, in
geneity, but there are situations in which accurate range vivo studies are necessary to characterize the protoacous-
verification may be possible. For the considered cases, tic detectability in the presence of biological acoustic
accuracy depended on detector placement. noise. Verifying the accuracy of proton range verifica-
Also through simulations, Patch et al. have demon- tion methods in vivo, however, is complicated by the lack
strated a method for accurately (≤ 2 mm) determining of a gold-standard technique against which they may be
the Bragg peak position in heterogeneous material based compared. Therefore, there are worthwhile, unanswered
on a two stage technique that combines beamforming questions – particularly validation accuracy – that may
and comparison to a database of dose distributions and be investigated through heterogeneous phantom or ex
their associated simulated protoacoustic emissions [25]. vivo studies.
One proposed [15] and demonstrated [21, 26] method
for neutralizing the effects of heterogeneous sound 28.5 CONCLUSIONS
speeds is to overlay the protoacoustically identified Bragg Proton therapy has not realized its full potential due to
peak on an ultrasound image. Because the heterogene- range uncertainty. When dose is deposited by pulsed
ity will similarly affect the propagation of protoacoustic proton beams, thermoacoustic pressure emissions (pro-
(lower frequency, one-way acoustic propagation) and toacoustics) are generated. By collecting these emissions,
ultrasound (higher frequency, two-way acoustic propa- the position of their origin – the Bragg peak – may be
gation) waves, any accumulated error will be shared by determined. Therefore, protoacoustics is a potential and
both modalities. Therefore, the Bragg peak position will promising proton range verification method. Challenges
be known relative to anatomy identified on the ultra- facing the technique include low acoustic signal and tis-
sound image. This method has been demonstrated using sue heterogeneity.
low energy proton beams (50 and 15 MeV), with excel-
lent results [21, 26]. REFERENCES
1. H. Paganetti, “Range uncertainties in proton therapy
28.4.4 Challenges
and the role of Monte Carlo simulations,” Physics in
The largest challenges facing protoacoustic range veri- Medicine and Biology 57, R99–R117 (2012).
fication methods are low signal-to-noise levels, errors 2. A.-C. Knopf and A Lomax, “In vivo proton range veri-
induced by heterogeneity, and limited anatomical sites fication: A review,” Physics in Medicine and Biology 58,
R131–R60 (2013).
of application. Because acoustic waves are not transmit-
3. H. Paganetti, “Relative biological effectiveness (RBE)
ted through air pockets, possible sites for development values for proton beam therapy. Variations as a function
include liver, pancreas, prostate, breast, and – if the low of biological endpoint, dose, and linear energy transfer,”
frequency protoacoustic waves can penetrate through Physics in Medicine and Biology 59, R419–R472 (2014).
4. B. Clasie, N. Depauw, M. Fransen, C. Gomà, H. R. in a water phantom,” Medical Physics 42, 937–946
Panahandeh, J. Seco, J. B. Flanz, H. M. Kooy, “Golden (2015).
beam data for proton pencil-beam scanning,” Physics in 17. K. C. Jones, F. Vander Stappen, C. R. Bawiec, G.
Medicine and Biology 57, 1147–1158 (2012). Janssens, P. A. Lewin, D. Prieels, T. D. Solberg, C. M.
5. J. Unkelbach and H. Paganetti, “Robust proton treat- Sehgal, S. Avery, “Experimental observation of acous-
ment planning: Physical and biological optimization,” tic emissions generated by a pulsed proton beam from
Seminars in Radiation Oncology 28, 88–96 (2018). a hospital-based clinical cyclotron,” Medical Physics 42,
6. A. Knopf, K. Parodi, T. Bortfeld, H. A. Shih, H. 7090–7 (2015).
Paganetti, “Systematic analysis of biological and physi- 18. K. C. Jones, C. M. Sehgal, S. Avery, “How proton pulse
cal limitations of proton beam range verification with characteristics influence protoacoustic determination
offline PET/CT scans,” Physics in Medicine and Biology of proton-beam range: Simulation studies,” Physics in
54, 4477–4495 (2009). Medicine and Biology 61, 2213–2242 (2016).
7. J. C. Polf, S. Avery, D. S. Mackin, S. Beddar, “Imaging 19. K. C. Jones, F. Vander Stappen, C. M. Sehgal, S. Avery,
of prompt gamma rays emitted during delivery of clini- “Acoustic time-of-flight for proton range verification in
cal proton beams with a Compton camera: Feasibility water,” Medical Physics 43, 5213–5224 (2016).
studies for range verification,” Physics in Medicine and 20. S. Kellnberger, W. Assmann, S. Lehrack, S. Reinhardt,
Biology 60, 7085 (2015). P. Thirolf, D. Queirós, G. Sergiadis, G. Dollinger, K.
8. M. Priegnitz, S. Barczyk, L. Nenoff, C. Golnik, I. Keitz, Parodi, V. Ntziachristos, “Ionoacoustic tomography of
T. Werner, S. Mein, J. Smeets, F. V. Stappen, G. Janssens, the proton Bragg peak in combination with ultrasound
L. Hotoiu, F. Fiedler, D. Prieels, W. Enghardt, G. Pausch, and optoacoustic imaging,” Scientific Reports 6, 29305
C. Richter, “Towards clinical application: Prompt (2016).
gamma imaging of passively scattered proton fields 21. S. K. Patch, M. K. Covo, A. Jackson, Y. M. Qadadha,
with a knife-edge slit camera,” Physics in Medicine and K. S. Campbell, R. A. Albright, P. Bloemhard, A. P.
Biology 61, 7881 (2016). Donoghue, C. R. Siero, T. L. Gimpel, S. M. Small, B. F.
9. J. M. Verburg and J. Seco, “Proton range verification Ninemire, M. B. Johnson, L. Phair, “Thermoacoustic
through prompt gamma-ray spectroscopy,” Physics in range verification using a clinical ultrasound array pro-
Medicine and Biology 59, 7089 (2014). vides perfectly co-registered overlay of the Bragg peak
10. C. Golnik, F. Hueso-González, A. Müller, P. Dendooven, onto an ultrasound image,” Physics in Medicine and
W. Enghardt, F. Fiedler, T. Kormoll, K. Roemer, J. Biology 61, 5621 (2016).
Petzoldt, A. Wagner, G. Pausch, “Range assessment in 22. S. Lehrack, W. Assmann, D. Bertrand, S. Henrotin, J.
particle therapy based on prompt γ-ray timing measure- Herault, V. Heymans, F. Vander Stappen, P. G. Thirolf,
ments,” Physics in Medicine and Biology 59, 5399 (2014). M. Vidal, J. Van de Walle, K. Parodi, “Submillimeter
11. J. Tada, Y. Hayakawa, K. Hosono, T. Inada, “Time ionoacoustic range determination for protons in water
resolved properties of acoustic pulses generated in water at a clinical synchrocyclotron,” Physics in Medicine &
and in soft tissue by pulsed proton beam irradiation – a Biology 62, L20 (2017).
possibility of doses distribution monitoring in proton 23. K. C. Jones, W. Nie, J. C. H. Chu, J. V. Turian, A. Kassaee,
radiation therapy” Medical Physics 18, 1100–1104 (1991). C. M. Sehgal, S. Avery, “Acoustic-based proton range
12. Y. Hayakawa, J. Tada, N. Arai, K. Hosono, M. Sato, verification in heterogeneous tissue: Simulation stud-
T. Wagai, H. Tsuji, H. Tsuji, Acoustic pulse generated ies” Physics in Medicine & Biology 63, 025018 (2018).
in a patient during treatment by pulsed proton radia- 24. W. Nie, K. C. Jones, S. Petro, A. Kassaee, C. M. Sehgal,
tion beam,” Radiation Oncology Investigations 3, 42–55 S. Avery, “Proton range verification in homogeneous
(1995). materials through acoustic measurements,” Physics in
13. K. C. Jones, A. Witztum, C. M. Sehgal, S. Avery, “Proton Medicine & Biology 63, 025036 (2018).
beam characterization by proton-induced acoustic 25. S. K. Patch, D. E. M. Hoff, T. B. Webb, L. G. Sobotka,
emission: Simulation studies,” Physics in Medicine and T. Zhao, “Two-stage ionoacoustic range verification
Biology 59, 6549 (2014). leveraging Monte Carlo and acoustic simulations to sta-
14. M. Ahmad, L. Xiang, S. Yousefi, L. Xing, “Theoretical bly account for tissue inhomogeneity and accelerator-
detection threshold of the proton-acoustic range veri- specific time structure – a simulation study,” Medical
fication technique,” Medical Physics 42, 5735–5744 Physics 45, 783–793 (2018).
(2015). 26. S. K. Patch, D. Santiago-Gonzalez, B. Mustapha,
15. W. Assmann, S. Kellnberger, S. Reinhardt, S. Lehrack, “Thermoacoustic range verification in the presence of
A. Edlich, P. G. Thirolf, M. Moser, G. Dollinger, M. acoustic heterogeneity and soundspeed errors – robust-
Omar, V. Ntziachristos, K. Parodi, “Ionoacoustic char- ness relative to ultrasound image of underlying anatomy,”
acterization of the proton Bragg peak with submillime- Medical Physics 46, 318–27 (2019).
ter accuracy,” Medical Physics 42, 567–574 (2015). 27. M. Xu and L. V. Wang, “Photoacoustic imaging in bio-
16. F. Alsanea, V. Moskvin, K. M. Stantz, “Feasibility of medicine,” Review of Scientific Instruments 77, 041101
RACT for 3D dose measurement and range verification (2006).
28. S. K. Patch, D. Hull, W. A. See, G. W. Hanson, “Toward the Ultrasonic Field Generated in Water upon the
quantitative whole organ thermoacoustics with a clini- Deceleration of a Proton Beam,” Instruments and
cal array plus one very low-frequency channel applied Experimental Techniques 47, 502–506 (2004).
to prostate cancer imaging, IEEE Transactions on 35. T. Terunuma, T. Sakae, Y. Hayakawa, A. Nohtomi, Y.
Ultrasonics, Ferroelectrics, And Frequency Control 63, Takada, K. Yasuoka, A. Maruhashi, “Waveform simu-
245–255 (2016). lation based on 3D dose distribution for acoustic wave
29. L. Xiang, B. Han, C. Carpenter, G. Pratx, Y. Kuang, generated by proton beam irradiation,” Medical Physics
L. Xing, “X-ray acoustic computed tomography with 34, 3642–3648 (2007).
pulsed x-ray beam from a medical linear accelerator” 36. B. E. Treeby and B. T. Cox, “k-Wave: MATLAB toolbox
Medical Physics 40, 010701 (2013). for the simulation and reconstruction of photoacoustic
30. L. Sulak, T. Armstrong, H. Baranger, M. Bregman, wave fields,” Journal of Biomedical Optics 15, 021314
M. Levi, D. Mael, J. Strait, T. Bowen, A. E. Pifer, P. A. (2010).
Polakos, H. Bradner, A. Parvulescu, W. V. Jones, J. 37. V. I. Albul, V. B. Bychkov, K. E. Gusev, V. S. Demidov,
Learned, “Experimental studies of the acoustic signa- E. V. Demidova, S. L. Konovalov, A. F. Kurchanov,
ture of proton beams traversing fluid media,” Nuclear V. E. Luk’yashin, V. I. Lyashuk, E. G. Novikov, A. A.
Instruments and Methods 161, 203–217 (1979). Rostovtsev, A. Y. Sokolov, U. F. Feizkhanov, N. A.
31. S. Hickling, L. Xiang, K. C. Jones, K. Parodi, W. Khaldeeva, “Measurements of the parameters of the
Assmann, S. Avery, M. Hobson, I. El Naqa, “Ionizing acoustic radiation accompanying the moderation of
radiation-induced acoustics for radiotherapy and diag- an intense proton beam in water,” Instruments and
nostic radiology applications,” Medical Physics 45, Experimental Techniques 44, 327–334 (2001).
e707–e721 (2018). 38. W. Nie, K. C. Jones, A. Lieth, T. TeBeest, J. Foley, A.
32. K. Graf, G. Anton, J. HÖSsl, A. Kappes, T. Karg, U. Katz, Kassaee, C. M. Sehgal, S. Avery, “TU-D-KDBRB1-3
R. Lahmann, C. Naumann, K. Salomon, C. Stegmann, comparison of acoustic detection methods for proton
“Testing thermo-acoustic sound generation in water range verification,” in American Association of Physicists
with proton and laser beams,” International Journal of in Medicine 2018 Annual Meeting (Nashville, TN, 2018).
Modern Physics A 21, 127–31 (2006). 39. “Prescribing, recording, and reporting proton-beam
33. L. Lin, C. G. Ainsley, T. D. Solberg, J. E. McDonough, therapy (ICRU Report 78),” Journal of the International
“Experimental characterization of two-dimensional Commission on Radiation Units and Measurements 7(2)
spot profiles for two proton pencil beam scanning (2007).
nozzles” Physics in Medicine and Biology 59, 493–504 40. S. Henrotin, M. Abs, E. Forton, Y. Jongen, W. Kleeven, J.
(2013). Van de Walle, P. Verbruggen, “Commissioning and test-
34. V. I. Albul, V. B. Bychkov, S. S. Vasil’ev, K. E. Gusev,V. ing of the first IBA S2C2,” in Proceedings of Cyclotrons
S. Demidov, E. V. Demidova, N. K. Krasnov, A. F. 2016 (Zurich, Switzerland, 2016).
Kurchanov, V. E. Luk’yashin, A. Y. Sokolov, “Measuring
Chapter 29
Proton Radiography and Proton

Computed Tomography
Xinyuan Chen and Tianyu Zhao
St. Louis, Missouri
CONTENTS
29.2 Overview of Physics of Proton Imaging 458
29.2.1 Proton Interactions with Matter 458
29.2.2 pRG/pCT Image Reconstruction 458
29.3 History of pRG/pCT Development 459
29.4 Current Technologies and Remaining Challenges 460
29.4.1 Proton Integrating System 460
29.4.2 Proton Tracking System 461
29.5 Conclusion 462
References 463
29.1 INTRODUCTION proton SPR from the CT image is degraded by a variety

Proton therapy has been widely accepted and applied in of factors, including acquisition parameters [2], imag-
cancer treatment because of its excellent 3D dose confor- ing noise [3], and imaging artifacts [4], in addition to
mity. Compared with traditional photon radiation ther- uncertainties associated with the stoichiometric cali-
apy, proton therapy is able to deposit most of its energy bration implicitly and explicitly, such as stoichiomet-
to a small targeted region without exiting dose due to ric parameterization [2, 5], variation in human tissue
the Bragg peak property of the protons. Therefore, radi- composition and mean excitation energy [6, 7] in the
ation is delivered to tumor conformally while normal calculation of theoretic proton SPR of human tissues.
tissues beyond the target are spared by precise control The overall uncertainty (1σ ) in proton SPR for differ-
of proton energy. ent tissue types has been estimated as 1.6% (soft tissue),
However, the superior accuracy in proton range can 2.4% (bone), and 5.0% (lung) [8].
only be achieved with precise knowledge of the proton To overcome the intrinsic uncertainty in converting
stopping power ratio (SPR) of different tissues inside the CT Hounsfield Unit (HU) to proton SPR, it has been
the patient. Currently, the most reliable and common suggested that proton can be used directly as the imag-
way of estimating proton SPR of a patient is stoichio- ing source to obtain in vivo integrated and pixelwise
metric calibration [1], which maps the electron density proton SPR, respectively, in proton radiography (pRG)
from a polyenergetic x-ray computed tomography (CT) and proton computed tomography (pCT). In prin-
image to the proton SPR. The accuracy of the estimated ciple, these imaging modalities can provide additional
457
imaging for patient setup and reduce the uncertainty in Increasing initial kinetic energy of the protons may
proton range to less than 1%. improve spatial resolution of pRG/pCT via reduced
In this chapter, the basic physics of proton imaging, MCS, and thus reduced angular straggling. However
including proton interactions with matter and pRG/ high initial kinetic energy may also lead to higher rate
pCT reconstruction physics is introduced. Brief history of inelastic nuclear interactions and more importantly,
of pRG/pCT development is reviewed. And finally, the reduce energy contrast through the patient [13].
most advanced technologies of pRG/pCT are introduced
and discussed. 29.2.2 pRG/pCT Image Reconstruction
The primary mechanism of pRG/pCT contrast is the
29.2 OVERVIEW OF PHYSICS OF energy loss of each proton, unlike the fluence loss of pho-
PROTON IMAGING tons in the x-ray radiography and x-ray CT. Therefore,
29.2.1 Proton Interactions with Matter the protons need to penetrate through the patient instead
Protons interact with matter mainly through of being stopped inside patient within the target area as
Coulomb force between the electric field of the pass- used in proton therapy. Accordingly, the initial kinetic
ing proton and the electric field of the orbital electrons energy of the protons in pRG/pCT is usually higher than
of the material. This interaction can be described by that used in proton therapy. While typical initial kinetic
the Bethe–Bloch equation [9, 10], where the energy energy for therapeutic applications ranges from 60 MeV
loss rate of a proton is approximately proportional to (~3 cm range in water) to 230 MeV (~33 cm range in
inverse square of its velocity. Thus when a proton slows water), the typical initial kinetic energy for pRG/pCT is
down in the material, its energy loss rate increases at or near the maximum energy of a medical accelerator,
drastically, inducing a pronounced peak, named Bragg i.e., 230–250 MeV (~38 cm range in water) [14]. Although
peak, toward the end of its range in the medium. The the 250 MeV proton beam doesn’t have enough energy
existence of Bragg peak restricts proton dose within a to completely pass through the hip region of a typical
small region, assuring the dose conformity of proton adult person, or the shoulder-to-shoulder distance in
therapy. Note that due to the statistical nature of pro- most male patients, it has sufficient range for scanning
ton interactions along the beam path, the Bethe–Bloch a human head and lung region of most people as long
equation only characterizes the mean energy loss rate as the person’s arms are raised out of the beam path.
of a proton. The energy spectrum of protons widens Therefore, many research efforts have been focused on
as protons go deeper into the medium, thus range of developing proton imaging modalities that scan mainly
individual proton varies. This phenomenon is called the human head. Since proton SPR has little dependency
energy straggling [11]. on the initial kinetic energy of the protons used in pRG/
Besides the Coulomb interaction with orbital elec- pCT, the SPR acquired by proton transmission imaging
trons, protons may also interact with nuclei of the is useful in the treatment planning for proton therapy.
atoms through elastic Coulomb scattering, called Recall that the stopping power (S) is defined as the
“Multiple Coulomb Scattering” (MCS). Although this rate of energy loss per unit path length,
interaction has neglectable effect on the energy change
of proton, it may change the direction of proton con- dE
S= − (29.1)
tinuously, causing the so-called angular straggling. dx
This is the major factor that limits the spatial resolu- and the relative SPR is the ratio of stopping power (S) at
tion of pRG/pCT since protons don’t travel in a straight a point of the material relative to that for water.
path as primary photons do in the x-ray radiography
and CT imaging. Smaterial
In addition to these processes, a proton may also SPR = (29.2)
Swater
go through inelastic nuclear interactions at a rate of
approximately 1% per cm in water at therapeutic ener- To estimate relative proton SPR, a set of values of water
gies. Such nuclear interactions remove the proton from equivalent path length (WEPL) is acquired by pRG/pCT.
the beam and reduce the primary fluence [12]. WEPL is defined as proton’s path length in water that
Proton Radiography and Proton Computed Tomography ◾ 459
can produce the same amount of energy loss as that pro-

duced by proton passing through the imaging object.
WEPL, written in Equation 29.3, is the line integral of
SPR along a specific path length, which is analogous to
ray projection in radiography.
 
WEPL (t ,θ ) =
∫ SPR(r )dr
Lt ,θ
(29.3)
where t and θ define the specific path that the proton


goes through, r is the unit length that has the same
direction as the passing proton.
In pRG, 2D images of mean WEPL can be used
directly for the verification or correction of x-ray plan-
ning CT scans. In pCT, an additional step of reconstruc-
tion is needed to obtain SPR in a 3D volume from the
WEPL measurements.
Based on different approaches to measure WEPL,
pRG/pCT can be classified into two categories. One cat- FIGURE 29.1 The first published proton radiograph from
egory is called proton integrating system, where WEPL 1968 [15]. The contract was generated by a 100 µm thick,
is calibrated from the signal in a detector averaged over pennant-shape aluminum foil. (Reprinted from Koehler [15]
a large number of protons. The other category is called with permission from The American Association for the
proton tracking system, where WEPL is calibrated from Advancement of Science.)
the residual energy or range of each proton. We will dis-
cuss them in the following sections.
difference (0.5%) [18]. The 158 MeV external beam of the
29.3 HISTORY OF pRG/pCT DEVELOPMENT Harvard cyclotron was used for that purpose. NaI scintil-
First examples of pRG were demonstrated in the 1960s. lators coupled to photomultiplier tubes were used as detec-
In 1968, Koehler published the first image generated by tors. The reconstruction was performed based on Abel’s
pRG (Figure 29.1) using the proton beam at the Harvard equation and the fact that the phantom was circularly sym-
Cyclotron (Cambridge, MA) [15]. The proton beam was metric. This system was considered as a proton integrat-
scattered and broadened to approximately 2 cm in diam- ing system, because all the protons within the proton beam
eter by a lead scatterer, which reduced the beam energy were tracked as a whole. Although the reconstructed radial
from 160 MeV to 137 MeV. The imaging contrast was distribution of density difference was presented, no recon-
induced by a 100 µ m thick, pennant-shape aluminum structed image was included in the publication.
foil. A photographic film (Polaroid type-52 film) was In 1978, a reconstructed image (Figure 29.2) using
placed close to the proton range to detect the sharp drop proton CT was first published by Hanson et al. at Los
in proton fluence at this location. In spite of relative poor Alamos Laboratory (Los Alamos, NM) [19]. A 240 MeV
spatial resolution, the ability of energetic protons from proton beam was used as the imaging source. The advan-
an accelerator to produce high contrast image has been tage of low dose compared to x-ray CT was pointed out.
brought to people’s attention. Many similar works on The detector was composed of two parts: a hyperpure
pRG were carried out afterwards. However they all used germanium (HPGe) detector and two layers of thick
the sharp fall off of proton fluence to create contrast, scintillation counters. The former was used to measure
thus the imaging objects were limited to have a small residual energy of each proton, the latter was used as a
thickness [16, 17]. trigger of the HPGe detector so that the HPGe detec-
Later on in 1976, together with Cormack, Koehler tor only recorded the events that happened within the
published the first pCT design to detect small density central region, and thus increased the spatial resolution.
average proton WEPL of a number of protons through

the imaging object. The structure of it is relatively simple
compared to that of proton tracking systems. Usually,
well confined proton pencil beams from a medical accel-
erator are used as the imaging source. A residual energy
detector, such as a gadolinium oxysulfide scintillator
coupled to an amorphous silicon matrix array is used
to measure the averaged residual energy of the protons
transmitted through the patient along a specific path.
The reading of the detector will then be calibrated to the
WEPL along the corresponding path. Finally, the proton
SPR distribution within the patient can be calculated
based on a set of values of WEPL.
Because of its compact structure, and its use of same
proton beams with similar energies as the therapeutic
proton beams, a proton integrating system has greater
potential to be incorporated in proton therapy for treat-
ment planning. However due to MCS, the spatial resolu-
FIGURE 29.2 First image reconstructed from the Los
Alamos proton CT scanner in 1978. ©1978 IEEE. (Reprinted tion of the proton integrating system is limited to a few
from Hanson et al. [19] with permission from IEEE.) millimeters, depending on the spot size and the thick-
ness of the imaging object, and inferior to the proton
The design concept of the detector has profound impact tracking system, especially for the materials with high
on the later on proton tracking systems. inhomogeneity, and at the edge region between different
In the following decade, Hanson’s group took over tissue types.
the development of pCT by publishing a series of arti- A recent work by Zhang et al. [25] showed pRG
cles [20–22]. The HPGe detector was replaced by a stack images (Figure 29.3) of an anthropomorphic head phan-
of plastic scintillators in the later experiments. Human tom, a range compensator and a frozen lamb’s head.
specimens (brain and heart) were scanned. A proton Time-resolved dose rate functions (DRFs) were mea-
rate in excess of 10 kHz was obtained. The possibility sured by an x-ray amorphous silicon flat panel. Three
of using curved projection path was also discussed. The methods of deriving SPR including root mean square
Los Alamos work was a significant step forward, both (RMS) of DRFs only, intensity of DRFs only, and inten-
conceptually and experimentally. sity-weighted RMS were implemented and compared.
In 1987, Takada et al. [23] at Tsukuba were able to com- Interfaces between different materials were proved to be
plete pCT scans in eight minutes using a multiple-pencil- enhanced by incorporating the intensity information of
beam scanning method in combination with moving slits, DRFs. SPRs derived from both RMS only and intensity-
and a large magnetic spectrometer to measure the resid- weighted RMS were within ±1 for most of the Gammex
ual energy. In 1999, Zygmanski et al. at the Massachusetts phantom inserts, and with a mean absolute percentage
General Hospital (MGH) [24] implemented a cone-beam error of 0.66% for all inserts.
CT system using scattered proton beam, and pointed out Moreover, a Monte Carlo study has verified the fea-
the advantages of obtaining stopping powers directly sibility of proton CT using multiple-layer ionization
with pCT. The two systems are proton integrating sys- chamber (MLIC) as the detector [26]. The MLIC detec-
tems, and assumed straight paths through the phantom. tor was able to record the residual energy of the proton
beam at different depths. Thus the depth dose curve
29.4 CURRENT TECHNOLOGIES AND with a complete Bragg peak was plotted from the MLIC.
REMAINING CHALLENGES The depth dose curve of the transmitted proton beam
29.4.1 Proton Integrating System was then calibrated to WEPL. SPR distribution of the
Proton integrating system is based on the assumption phantom was reconstructed with less than 1% deviation,
that signal measured in detector can be calibrated to and lower dose compared to x-ray CT.
FIGURE 29.3 Proton radiography images derived by RMS of DRFs only (a), intensity of DRFs only (b), and intensity-weighted
RMS (c) for an anthropomorphic head phantom, a range compensator and a frozen lamb’s head by Zhang et al. [25]. (Reprinted
from Zhang et al. [25] with permission from IOP.)
As demonstrated by the above-mentioned works, object. A schematic of an ideal proton-tracking pRG/

image quality of proton integrating systems has the pCT is shown in Figure 29.4. Based on the entrance and
potential to be improved in the future. The MCS effect exiting positions, trajectory of individual proton inside
may be corrected by the dose function recorded by the the patient can be a straight line, or a most-likely path
detector and more sophisticated statistical model of the modeled statistically from estimated distribution of dis-
proton beam. placement and scattering angle [27], or a polynomial
fitting path [28]. Studies [29] in homogeneous phantom
29.4.2 Proton Tracking System showed that most-likely path is superior in imaging
Proton tracking systems usually consist of two parts of quality with fastest convergence and least residual error.
detector: position sensitive detector (PSD) that tracks In 1999, an improved proton tracking system was
the direction and/or position of each proton, and developed by Pemler et al, where two scintillating fiber
residual energy range detector (RERD) to measure hodoscopes, one before and one after the patient, and
the residual energy of the protons exiting the imaging a range telescope made of plastic scintillator tiles were
FIGURE 29.4 Schematic of the ideal proton-tracking pRG/pCT.
used to indicate both position and range of 106 protons/s imaging system (Figure 29.6). Accuracy in SPR was
[30]. The image was scanned in an area of 22.0 × 3.0 cm2 measured to be ≤1.6% for all the inserts.
with 1 MHz proton rates [30]. Between 2003 and 2013, Proton tracking systems require high-speed tracking
a collaboration between Loma Linda University (LLU) of protons at the entrance and exit sides of the patient,
and University of California Santa Cruz (UCSC), devel- with continuous recording of the residual energy.
oped the first and the second generation of proton-track- Although this approach records adequate information
ing system using four silicon tracker planes (two before and is promising in theory for accurate pRG and pCT,
and two after the patient) to determine both direction the equipment currently employed must to be reduced
and position of the incoming and outgoing proton. The in complexity and size if it is to be suitable in a clinical
proton rate was improved from 10 to 20 kHz to over setting [36].
1 MHz, significantly reducing the acquisition time [31,
32]. By incorporating proton “most likely path” in an 29.5 CONCLUSION
advanced iterative reconstruction method, the error of Range uncertainty, originating from the uncertainty
SPR was brought down to less than 1% [33]. in the mapping of electron density to proton stopping
A newly emerged group working on an Italian project power, has been a unique challenge for proton therapy.
for a PRonton IMAging (PRIMA) device proposed two The lack of capability to pin down proton SPR in the
generations of pCT system [34]. The systems were made primary dataset for treatment planning, offsets the most
of a silicon microstrip tracker and a YAG:Ce crystal calo-
rimeter to capture single protons trajectory and residual
energy, respectively. The first generation pCT system was
constructed to have an active area of about 5 × 5 cm2 and
a data rate capability of 10 kHz. Two slices of tomographic
reconstructions of the test phantom were displayed in
Figure 29.5, with a sketch of the analyzed phantom on the
left. An extended field of view (up to ~ 5 × 20 cm2) and an
increased event rate capability up to 1 MHz was described
for the second generation system.
A novel proton tracking system was proposed by
the Proton Radiotherapy Verification and Dosimetry
Applications (PRaVDA) consortium [35]. The pCT
imaging system was based entirely on solid-state detec-
tor, making it possible to track multiple proton per read- FIGURE 29.5 A sketch of the imaging phantom (left). Two
out cycle, which leads to a potential reduction in proton slices of the tomographic reconstructions: one in the area with
CT scan time. A 75-mm diameter PMMA sphere with holes and one in the uniform region (right) [34]. (Reprinted
substitute inserts was imaged by the fully solid-state from Scaringella et al. [34] with permission from IOP.)
proton beam ranges in patients from dose and PET-

activity distributions,” Physics in Medicine and Biology
55, 7557–7571 (2010).
3. A. V. Chvetsov and S. L. Paige, “The influence of CT
image noise on proton range calculation in radiotherapy
planning,” Physics in Medicine and Biology 55 (2010).
4. K. M. Andersson, A. Ahnesjö, C. V. Dahlgren, “Evaluation
of a metal artifact reduction algorithm in CT stud-
ies used for proton radiotherapy treatment planning,”
Journal of Applied Clinical Medical Physics 15, 112–119
(2014).
5. B. Schaffner and E. Pedroni, “The precision of proton
range calculations in proton radiotherapy treatment
planning: Experimental verification of the relation
between CT-HU and proton stopping power,” Physics in
Medicine and Biology 43, 1579–1592 (1998).
6. A. Besemer, H. Paganetti, B. Bednarz, “WE-C-BRB-06:
Clinical impact of uncertainties in the mean excita-
tion energy of human tissues during proton therapy,”
Medical Physics 39, 3944 (2012).
FIGURE 29.6 A slice of reconstructed spherical phantom 7. H. Paganetti, “Range uncertainties in proton therapy
containing three substitute inserts (top left: water equivalent, and the role of Monte Carlo simulations,” Physics in
top right: adipose equivalent, bottom: average bone equiva- Medicine and Biology 57, R99 (2012).
lent) [35]. 8. M. Yang, X. Zhu, P. Parker, et al., “Comprehensive
analysis of proton range uncertainties related to patient
stopping-power-ratio estimation using the stoichio-
anticipated benefit of precise delivery and sparing dis-
metric calibration,” Physics in Medicine and Biology 57,
tal organs to a certain extent. Although the intrinsic 4095 (2012).
uncertainty can be reduced with more sophisticated CT 9. H. Bethe, “Zur Theorie des Durchgangs schneller
calibration or dual-energy CT scan, and its dosimetric Korpuskularstrahlen durch Materie,” Annals of Physics
impact could be mitigated with robust optimization, 397, 325–400 (1930).
it never goes away as long as the primary dataset used 10. F. Bloch, “Zur Bremsung rasch bewegter Teilchen beim
Durchgang durch Materie,” Annals of Physics 408, 285–
for dose calculation is acquired with x-ray tomography. 320 (1933).
Alternatively, proton provides a promising way of imag- 11. A. J. Lomax, “Charged particle therapy: The physics of
ing patients, providing either volumetric images with interaction,” Cancer Journal 15(4), 285–291 (2009).
accurate relative proton SPRs, or projectional radio- 12. J. F. Janni, “Energy loss, range, path length, time-of-
graphic images that could be used for patient setup and flight, straggling, multiple scattering, and nuclear
interaction probability: in two parts. Part 1. For 63 com-
range verification. The implementation of volumetric pounds Part 2. For elements 1 ≤ Z ≤ 92,” Atomic Data
proton imaging becomes increasingly attractive and and Nuclear Data Tables 27, 147–339 (1982).
practical with mounting interests and maturing tech- 13. J. Beringer, J. F. Arguin, R. M. Barnett, et al., “Review of
nologies. While multiple proton tracking systems show particle physics,” Physical Review D 86, 010001 (2012).
superior imaging quality and are closer to pre-clinical 14. M. Berger, J. Coursey, M. Zucker, Stopping-Power and
Range Tables for Electrons, Protons, and Helium Ions
test, proton integrating systems provide integrated solu-
(National Institute of Standards and Technology, 2000).
tions for both volumetric imaging and projectional 15. A. M. Koehler, “Proton radiography,” Science 160, 303–
radiography that are compatible with most commercial 304 (1968).
proton therapy cyclotrons. 16. D. R. Moffett, E. P. Colton, G. A. Concaildi, et al., “Initial
test of a proton radiographic system,” IEEE Transactions
REFERENCES on Nuclear Science 22, 1749–1751 (1975).
17. V. W. Steward and A. M. Koehler, “Proton radiography
1. U. Schneider, E. Pedroni, A. Lomax, “The calibration of as a diagnostic tool,” Physics in Medicine and Biology 18,
CT Hounsfield units for radiotherapy treatment plan- 591 (1973).
ning,” Physics in Medicine and Biology 41, 111 (1996). 18. A. M. Cormack and A. M. Koehler, “Quantitative pro-
2. S. España and H. Paganetti, “The impact of uncertain- ton tomography: Preliminary experiments,” Physics in
ties in the CT conversion algorithm when predicting Medicine and Biology 21, 560–569 (1976).
19. K. M. Hanson, J. N. Bradbury, T. M. Cannon, et al., “The 28. N. Krah, J. M. Létang, S. Rit, “Polynomial modelling
application of protons to computed tomography,” IEEE of proton trajectories in homogeneous media for fast
Transactions on Nuclear Science 25, 657–660 (1978). most likely path estimation and trajectory simulation,”
20. K. M. Hanson, “Proton computed tomography,” IEEE Physics in Medicine and Biology 64 (2019).
Transactions on Nuclear Science 26, 1635–1640 (1979). 29. T. Li, Z. Liang, J. V. Singanallur, T. J. Satogata, D. C.
21. K. M. Hanson, J. N. Bradbury, T. M. Cannon, et al., Williams, R. W. Schulte, “Reconstruction for proton
“Computed tomography using proton energy loss,” computed tomography by tracing proton trajectories: A
Physics in Medicine and Biology 26, 965–983 (1981). Monte Carlo study,” Medical Physics 33, 699–706 (2006).
22. K. M. Hanson, J. N. Bradbury, R. A. Koeppe, et al., 30. P. Pemler, J. Besserer, J. De Boer, “A detector system for
“Proton computed tomography of human specimens,” proton radiography on the gantry of the Paul-Scherrer-
Physics in Medicine and Biology 27, 25–36 (1982). Institute,” Nuclear Instruments and Methods in Physics
23. Y. Takada, K. Kondo, T. Marume, K. Nagayoshi, I. Okada, Research 432, 483 (1999).
K. Takikawa, “Proton computed tomography with a 250 31. R. F. Hurley, R. W. Schulte, V. A. Bashkirov, “Water-
MeV pulsed beam,” Nuclear Instruments and Methods in equivalent path length calibration of a prototype proton
Physics Research Section A 273, 410–422 (1988). CT scanner,” Medical Physics 39, 2438 (2012).
24. P. Zygmanski, K. P. Gall, M. S. Z. Rabin, S. J. Rosenthal, 32. R. P. Johnson, V. Bashkirov, V. Giacometti, et al., “Results
“The measurement of proton stopping power using from a pre-clinical head scanner for proton CT,” in 2014
proton-cone-beam computed tomography,” Physics in IEEE Nuclear Science Symposium and Medical Imaging
Medicine and Biology 45, 511–528 (2000). Conference (NSS/MIC) (2014), pp. 1–5.
25. R. Zhang, K.-W. Jee, E. Cascio, G. C. Sharp, J. B. Flanz, 33. S. N. Penfold, R. W. Schulte, Y. Censor, A. B. Rosenfeld,
H.-M. Lu, “Improvement of single detector proton radi- “Total variation superiorization schemes in proton
ography by incorporating intensity of time-resolved computed tomography image reconstruction,” Medical
dose rate functions,” Physics in Medicine and Biology Physics 37, 5887–5895 (2010).
63(1), 015030 (2017). 34. M. Scaringella, M. Bruzzi, M. Bucciolini, et al., “A pro-
26. X. Chen, R. Liu, S. Zhou, et al., “A novel design of proton Computed Tomography based medical imaging sys-
ton computed tomography detected by multiple-layer tem,” Journal of Instrumentation 9(12), 12009 (2014).
ionization chamber with strip chambers: A feasibility 35. M. Esposito, C. Waltham, J. T. Taylor, et al., “PRaVDA:
study with Monte Carlo simulation,” Medical Physics The first solid-state system for proton computed tomog-
47(2), 614–625 (2020). raphy,” Physica Medica 55, 149–154 (2018).
27. D. C. Williams, “The most likely path of an energetic 36. G. Poludniowski, N. M. Allinson, P. M. Evans, “Proton
charged particle through a uniform medium,” Physics radiography and tomography with application to proton
in Medicine and Biology 49, 2899–2911 (2004). therapy,” British Institute of Radiology 88, 20150134 (2015).
IV
Imaging Modalities
465
Chapter 30
Dosimetry of Imaging
Modalities in Radiotherapy
George X. Ding
Vanderbilt University School of Medicine
Nashville, Tennessee
CONTENTS
30.1.1 Planar Imaging Modalities 467
30.1.2 Volumetric Systems 468
30.1.3 Need of Dosimetry for Imaging Dose 469
30.2 Available Methods for Imaging Dosimetry 470
30.2.1 Measurements of Output of Beams Used for Imaging 470
30.2.2 Patient Imaging Dose Calculations 471
30.2.3 Patient Imaging Dose Estimations 471
30.3 Imaging Dose to Patients 473
30.4 Summary 474
References 474
30.1 INTRODUCTION be found in AAPM Medical Physics Monograph No 39

In modern radiation therapy, image-guided radiation [9]. The following is a brief introduction of the imaging
therapy (IGRT) has been adopted as the standard of care modalities and the dosimetry need for imaging devices.
to improve the geometric accuracy of patient position-
ing during radiotherapy [1–7] because it is capable of 30.1.1 Planar Imaging Modalities
significantly reducing target positioning errors, hence Planar imaging modalities include electronic portal
enabling highly conformal treatments. There are many imaging devices (EPIDs) and stereoscopic x-ray imaging
different imaging modalities used for image guidance systems. These systems allow matching of planar, kilo-
in radiotherapy. The currently available technologies voltage (kV) radiographs, fluoroscopy, or megavoltage
to perform IGRT can be divided, according to their (MV) images with digitally reconstructed radiographs
operation, into planar systems, volumetric systems, and (DRRs) from the planning CT [8].
non-radiographic systems [8]. The dosimetry of imag- Although the film was earlier commonly used as
ing modalities used in radiotherapy involves ionizing a verification method for field placement of treatment
radiation beams (such as x-rays) while non-radiographic portal in radiation therapy [10], it has been gradually
systems use non-ionizing radiation (such as magnetic replaced by the electronic systems when EPID systems
resonance imaging (MRI) systems, optical systems, ultra- became available for IGRT both online and offline
sound, etc). An overview of the state-of-the-art image- [11]. The developments driven by the need to image
guidance techniques employed in radiation therapy can patient position before treatment delivery without using
467
found in several review articles [21–23]. Currently medi-

cal accelerators made by Elekta and Varian for radio-
therapy are equipped with these devices.
Room-mounted stereoscopic devices utilize angled
x-rays tubes [24]. This type of device normally consists
of a pair of x-ray tubes and image detectors mounted in
the floor and on the ceiling. Examples of these devices are
used by the CyberKnife (Accuray) system (Figure 30.2)
and by the BrainLab ExacTrac X-ray (Brainlab) system
(Figure 30.3).
30.1.2 Volumetric Systems

The need to visualize patient 3D volumetric images for
accurate radiation delivery led to the clinical availabil-
ity of MV-CBCT [25], MVCT [26–28], and kV-CBCT
FIGURE 30.1 Conventional electronic portal imaging device
[29–32] in radiotherapy. Volumetric MV-CBCT images
(EPID). (Reproduced from AAPM TG-180 [1].)
are reconstructed using projections acquired with the
EPID and result in a greater dose than a pair of orthogo-
radiographic film [12–20] led to the EPIDs. The EPID nal MV portal images even with a low dose-rate therapy
device is attached to the medical accelerator used for beam [33].
radiotherapy and is capable of obtaining a pair of orthogo- Current kV imaging devices are generally integrated
nal projected digital images instantly using radiotherapy into linear accelerators and are capable of acquiring both
megavoltage beams, as shown in Figure 30.1. The details 2D radiographs and 3D volumetric kV-CBCT images
on development and the technology of the EPID can be [29–32]. Examples of kV-CBCT scanners integrated into
FIGURE 30.2 The CyberKnife™ system with its main components including IGRT systems. The stereotactic x-ray systems
have ceiling mounted x-ray tubes and floor-mounted detectors. The x-ray systems are supplemented with a ceiling-mounted
camera to determine the patient surface contour. (Reproduced from ICRU-91.)
Dosimetry of Imaging Modalities in Radiotherapy ◾ 469
FIGURE 30.3 The ExacTrac™ system with floor-mounted x-ray tubes and ceiling-mounted flat panel detectors. (Reproduced
from ICRU-91.)
a linac are Varian Medical Systems, Inc. (Palo Alto, CA) While the commonly adopted radiation protection
On Board Imaging (OBI) system (Figure 30.4) and an safety philosophy of As Low As Reasonably Achievable
Elekta (Stockholm, Sweden) X-Ray Volume Imaging (ALARA) is still applicable to imaging dose, minimiz-
(XVI) system (Figure 30.4). ing imaging doses, however, should not compromise
image quality needed for target localization [1].
30.1.3 Need of Dosimetry for Imaging Dose When there is reasonable expectation that the imag-
IGRT has now become the standard of care to improve ing dose will exceed 5% of the total prescribed dose, the
the geometric accuracy of patient positioning during imaging dose should be accounted for as part of the total
radiotherapy [1–7]. Unlike diagnostic imaging examina- dose to the patient [1]. The magnitude of imaging dose is
tions, the patient may be imaged multiple times during dependent on many factors, including the frequency of
any fraction in order to position the patient accurately imaging and the technique used. While the dose from
during radiotherapy delivery. The additional radiation a single imaging procedure is much lower than a frac-
exposure to patients resulting from these x-ray imaging tion of therapeutic dose, the integral dose of repeated
procedures may results in excessive dose to sensitive imaging procedures may not be negligible and should
organs and potentially increases the chance of second- be accounted for [2].
ary cancers and, therefore, needs to be managed in order In order to determine if the additional imaging doses
to minimize its risk [1, 2]. Based on considerations of exceed the recommended threshold, it is important to
clinical relevance, dose tolerances for critical organs and know the magnitude of additional imaging dose rela-
data available in the literature, AAPM TG-180 recom- tive to the therapeutic dose. It is known that the imag-
mended 5% of the therapeutic target dose as the thresh- ing dose to the patient differs significantly depending
old beyond which the imaging dose should be accounted on modalities and procedures used ranging from 0.1
for as part of the total dose to radiotherapy patients [1]. cGy to 5.0 cGy for a single image acquisition, with the
using Monte Carlo methods [48–58]. The imaging doses

resulting from MV beams can be calculated by using
commercial treatment planning systems except for the
beam that is only used for imaging, such as a the 2.5
MV beam in a Varian TrueBeam machine. However,
for such a beam once the beam output is determined
the imaging dose can be estimated with acceptable
accuracy [59].
Although kV imaging is commonly used in the
IGRT, it is relatively challenging to measure or to
calculate kV imaging doses to the patient in a radio-
therapy clinic. This is because measuring dose from
kV beams requires specific detectors and expertise.
AAPM TG-180 recommended two methods that can
be used to estimate dose resulting from imaging pro-
cedures. One requires patient-specific dose calcula-
tions while the other makes use of tabulated values to
estimate organ doses from a specific imaging proce-
dure referred to as non-patient-specific imaging dose
estimation [1].
30.2.1 Measurements of Output of
Beams Used for Imaging
It needs to be emphasized that it is essential to determine
the output of any beam that is used for imaging in order
to facilitate AAPM TG-180 recommended methods to
FIGURE 30.4 kV image devices integrated into linear accel- estimate the imaging dose to a patient. If the output
erators: (a) Varian OBI system on a Varian Trilogy treatment of an imaging beam does not meet the manufacturer’s
unit ; (b) Elekta XVI on an Elekta Synergy treatment unit. specifications, the accuracy of estimated patient doses
(Reproduced from AAPM TG-180.) cannot be guaranteed by using AAPM TG-180 recom-
mended methods.
When a therapy MV beam is used for imaging, its
exception of MV volumetric imaging [1]. Therefore, the beam output is calibrated according to the dosimetry
knowledge of dosimetry for imaging dose is necessary. protocols [60, 61] and monitored on a daily basis. For an
While the methods of determination of imaging dose MV beam that is used for imaging only, its output can
resulting from the therapeutic megavoltage beams are be measured according to the same dosimetry protocols
generally available in radiotherapy clinics, the dosim- [60, 61] based on measured values of percentage depth-
eters for determining imaging dose resulting from kilo- dose (PDD) curve at depth of 10 cm [62].
voltage beams are not generally available. The output for each image acquisition procedure, with
specified protocol parameters, should be measured in air or
30.2 AVAILABLE METHODS FOR in-phantom, according to the AAPM dosimetry protocols
IMAGING DOSIMETRY for kV [63] and MV [60] beams, to confirm that the mea-
The commonly used imaging procedures in IGRT are sured dose is within the manufacturer-stated specifications
kilovoltage radiographs and cone-beam computed at the time of acceptance of the imaging device. The expo-
tomography (kV-CBCT). kV imaging provides supe- sure or absorbed dose from a kV beam can be measured
rior image contrast compared to MV imaging. The by using a commercially available x-ray test devices, such
approaches used to determine kV imaging doses include as RaySafe X2, RaySafe Xi, which can measure kVp, half-
experimental measurements [34–47] and calculations value layer HVL and dose, etc. with a simple measurement
setup, or by using a calibrated ionization chamber accord- acquisition parameters and therefore it is able to pro-
ing to the dosimetry protocol [61, 63]. Although the meth- vide accurate personalized organ doses resulting from
odology for measuring the HVL and dose with a calibrated image guidance procedures from different patient size
ionization chamber is available from dosimetry protocols and image location [50, 72, 73].
[61, 63], the calibration conditions recommended in these Although Monte Carlo techniques are capable of pro-
dosimetry protocols are often not applicable to imaging viding accurate patient-specific organ dose calculations
acquisition procedures, especially when the x-ray source from different x-ray imaging procedures, the long com-
is moving during the scan [1]. Although water is the most putational times and expertise required for Monte Carlo
suitable medium for kV x-ray beam measurements, plas- simulations make it impractical in a clinical setting to
tic phantom materials are more convenient in practice. calculate imaging doses in daily clinical routine prac-
By using an ionization chamber, a method [64] to deter- tice. The success of Monte Carlo studies has provided
mine the absorbed dose output resulting from a specific detailed information about patient organ dose resulting
image acquisition procedure has been described based on a from different image guidance procedures and led to a
dosimetry protocol [61, 63]. practical method of accounting for the additional dose
to patients resulting from these imaging procedures,
30.2.2 Patient Imaging Dose Calculations referred to as non-patient-specific imaging dose estima-
When the imaging beam is a therapeutic MV beam, the tions [1].
patient-specific imaging dose can be calculated based Based on the studies reported in the literature [57,
on the beam monitor units and field sizes used during 58, 72, 73], the magnitude of patient organ doses from
image acquisition, unless the MV beam is an imaging- kV imaging is small relative to therapeutic dose. It has
only one [62]. These MV dose calculations are patient CT been shown that it is adequate to use simpler approaches
based and can be included to therapeutic dose directly that provide reasonable estimates of the imaging dose
during treatment planning when it is necessary [65]. If by using simple look-up tables to estimate organ doses
the total number of imaging procedures is known, the from repeated imaging procedures. The approach is
repeated imaging dose can be summed and accounted referred to as non-patient-specific imaging dose estima-
for the total dose to the target and OARs. Miften et tion [1] in which inter-patient variation and geometry
al. demonstrated this approach by showing optimized dependence are small in most cases and the dose esti-
IMRT plans with and without MV-CBCT included in mates could be provided in the form of simple look-up
the process [65]. tables, which may be accurate enough to estimate the
The dose calculation for kV beams is currently not dose from repeated imaging procedures.
available in a commercial treatment planning system,
although a model-based algorithm has been shown to 30.2.3 Patient Imaging Dose Estimations
achieve an acceptable accuracy for kV energy beams The method of estimating organ doses from a specific
[66]. Alaei et al. has shown that a commercial treat- imaging procedure, referred to as non-patient-specific
ment planning system can be modified by character- imaging dose estimation, is recommended by AAPM
izing kV beams to perform kV image dose calculations TG-180 [1] in the absence of patient-specific dose calcu-
[67]. However, the modification of adding a kV beam in lations. Tables 30.1–30.3 are the examples of tabulated
a commercial treatment planning system requires sig- organ doses for specified image acquisition procedures
nificant expertise and the dose calculation accuracy is associated with known mAs shown in Tables 30.4 and
limited with larger uncertainties to bones. 30.5. For planar kV beams (Table 30.5), only beam entry
To meet the dose calculation accuracy demand, doses are listed as the organ doses from planar beams are
Monte Carlo techniques have been extensively used in strongly dependent on the beam entry direction. Similar
radiotherapy in recent decades [68–70] and applied for tabulated organ dose data for the Elekta kV imaging sys-
kV imaging dose calculations [71]. Currently, patient tem are shown in Tables 30.6–30.8. These tabulated val-
organ doses resulting from kV imaging procedures ues are for specified mAs used for image acquisition are
reported in the literature are generally obtained from sufficient to estimate imaging dose to within 20% [72].
the Monte Carlo calculations [1, 57, 72]. This type of When users use a different mAs from listed in the image
dose calculation requires the details of the imaged beam acquisition procedure, the organ dose can be scaled by
TABLE 30.1 Organ Doses for the Head & Neck and Brain Treatment Sites from Varian OBI v1.4 Using Standard Head kV-CBCT Scan
Standard Head, Brain Standard Head, Head & Neck
Organ D50 Range (cGy) D10 Range (cGy) Organ D50 Range (cGy) D10 Range (cGy)
Brain 0.21–0.33 0.27–0.40 Brain 0.15–0.22 0.16–0.23
Brainstem 0.19–0.30 0.22–0.32 Larynx 0.21–0.29 0.25–0.33
Chiasm 0.08–0.26 0.09–0.26 Oral Cavity 0.13–0.26 0.20–0.31
Eyes 0.03–0.31 0.04–0.35 Parotids 0.26–0.42 0.31–0.48
Optic Nerves 0.05–0.27 0.05–0.27 Spinal Cord 0.16–0.25 0.19–0.32
Pituitary 0.07–0.24 0.08–0.25 Thyroid 0.07–0.23 0.11–0.32
Spinal Cord 0.26–0.33 0.29–0.34 Esophagus 0.07–0.16 0.14–0.26
Skin 0.19–0.41 0.39–0.63 Skin 0.18–0.27 0.34–0.44
Bones 0.45–1.11 1.13–1.67 Bones 0.25–0.65 0.64–1.07
Note: D50 and D10 are minimum dose delivered to 50% and 10% of the organ volume, respectively.
Source: Reference [72] and kV-CBCT scan parameters for Varian OBI 1.4 shown in Table 30.4.
TABLE 30.2 Organ Doses for the Chest Treatment Site from TABLE 30.3 Organ Doses for the Pelvis Treatment Site from
Varian OBI v1.4 Using Low-dose Thorax kV-CBCT Scan Varian OBI v1.4 Using Pelvis kV-CBCT Scan
Low-Dose Thorax Pelvis Scan, Prostate Isocenter
Organ D50 Range (cGy) D10 Range (cGy) Organ D50 Range (cGy) D10 Range (cGy)
Aorta 0.42–0.58 0.44–0.63 Bladder 1.36–2.20 1.72–2.69
Lungs 0.30–0.61 0.43–0.72 Bowel 1.54–1.91 2.04–2.65
Small Bowel 0.33–0.54 0.39–0.61 Femoral Heads 2.40–3.60 3.22–4.88
Esophagus 0.29–0.60 0.35–0.74 Prostate 1.19–1.79 1.33–1.89
Kidney 0.43–0.54 0.49–0.59 Rectum 1.51–1.99 1.70–2.22
Heart 0.31–0.55 0.41–0.63 Skin 1.80–1.96 2.26–2.92
Liver 0.31–0.51 0.38–0.61 Bone 2.93–3.96 4.61–5.72
Spinal Cord 0.32–0.57 0.35–0.78 Source: Reference [72] and kV-CBCT scan parameters for Varian
Spleen 0.32–0.52 0.36–0.60 OBI 1.4 shown in Table 30.4.
Stomach 0.28–0.57 0.31–0.62
Trachea 0.36–0.71 0.47–1.04
Skin 0.46–0.57 0.64–0.89
Bones 1.06–1.74 1.47–2.25
Source: Reference [72] and kV-CBCT scan parameters for Varian
OBI 1.4 shown in Table 30.4.
TABLE 30.4 Parameters for kV-CBCT Specified Acquisition Techniques in Varian OBI 1.4
kV-CBCT Name Bow-Tie Filter (kV) (mAs) Gantry Rotation (degrees)
OBI Standard-dose Head Full fan 100 145 200
OBI Low-dose Head Full fan 100 72 200
OBI High-Quality Head Full fan 100 720 200
OBI Pelvis Half fan 125 700 360
OBI Pelvis Spot Light Full fan 125 720 200
OBI Low-Dose Thorax Half fan 110 262 360
TrueBeam Head Full fan 100 147 200
TrueBeam Pelvis Half fan 125 1056 360
TrueBeam Spotlight Full fan 125 733 200
TrueBeam Thorax Half fan 125 264 360
Source: From reference [58].
TABLE 30.5 Parameters for kV Radiographs for Specified the ratio of mAs used and that list for the image acquisi-
Acquisition Techniques in Varian OBI 1.4 tion. The tabulated organ doses can assist the clinician
Name (kV) (mAs) Entry Dose in: (1) determining if the imaging doses are expected
Head-AP 100 8 0.1 cGy to be close to the 5% threshold, (2) choosing a suitable
Head-Lat 70 5 0.05 cGy
IGRT protocol, and (3) accounting for the organ dose
Thorax-AP 75 5 0.1 cGy
resulting from a specific image acquisition procedure
Thorax-Lat 95 40 0.5 cGy
Pelvis-AP-Med 75 10 0.1 cGy over the course of treatment [1].
Pelvis-Lat-Med 105 80 1.6 cGy It is worth repeating here that it is essential to mea-
Note: The clinical default OBI blades are set to X1 = X2 = 13.3 cm and
sure the output of any beam that is used for imaging in
Y1 = Y2 = 10.3 cm in all acquisition techniques. All six tech- order to use this method to estimate the imaging dose to
niques were modeled with and without full-fan bow-tie filter. a patient. When the beam output of an imaging device
Source: From Supplementary data in reference [58]. exceeds the manufacturer’s specifications, the imaging
dose to the patient can be underestimated because the
TABLE 30.6 Organ Doses for the Head & Neck Treatment Site tabulated patient organ doses are based on the specific
from Elekta XVI kV-CBCT Scan Using S Cassettes, 100 kVp, 0.1 clinical default acquisition protocols.
mAs/Acquisition, 360 Acquisitions, 345–190 Degree (IEC) Rotation
Head and Neck 30.3 IMAGING DOSE TO PATIENTS
Organ D50 Range (cGy) Depending on the imaging modality and acquisition
Brainstem 0.06–0.08 procedure, the dose resulting from single image guid-
Rt Eye 0.08–0.09
ance procedure varies significantly ranging from 0.1
Lt Eye 0.13–0.13
cGy to 15 cGy [1]. In general organ doses from imag-
Rt Parotid 0.05–0.06
Lt Parotid 0.16–0.17 ing decrease from MV-CBCT, to MVCT, to MV portal
Rt Cochlea 0.04–0.05 images, to kV-CBCT, to kV radiographs [1, 58].
Lt Cochlea 0.09–0.12 Table 30.9 summarizes the range of imaging dose to
Oral Cavity 0.09–0.11 soft tissues from different acquisition procedures. The
Source: From reference [1]. listed range of imaging doses are based on reported data
in the literature and are applicable to the default clini-
cal image acquisition protocols set by manufacturers. If
TABLE 30.7 Organ Doses for the Pelvis Treatment Site from user’s image acquisition parameters differ from the clin-
Elekta XVI kV-CBCT Scan Using M Cassette with Bowtie Filter, 120
kVp, 1.6 mAs/Acquisition, 650 Acquisitions, Full 360 Degree ical default settings, the image doses can be obtained by
Rotation scaling the listed dose with MV monitor unit for MV
Pelvis beams or mAs for kV beams [1]. For room-mounted
Organ D50 Range (cGy)
Bladder 0.9–2.0 TABLE 30.9 Summary of the Range of Imaging Dose to Soft
Rectum 1.1–1.9 Tissues Resulting from a Single Imaging Procedure by Acquiring 3D
Small Bowel 1.0–1.8 or 2D Images for Patient Treatment Positioning [1]
Source: Reference [1]. Imaging Dose to Soft Tissues
Acquisition Procedure Low (cGy) High (cGy)
MV-CBCT (Linac-based) 1.0 16.0
TABLE 30.8 Organ Doses for the Pelvis Treatment Site from MVCT Lungs (TOMO unit) 0.8 2.5
Elekta XVI kV-CBCT Scan Using M Cassette Without Bowtie Filter, A pair of portal images (6 MV) 2.0 5.0
120 kVp, 1.0 mAs/Acquisition, 650 Acquisitions, Full 360 Degree A pair of portal images (2.5 MV) 0.5 2.0
Rotation kV-CBCT (Linac-based) head 0.1 0.5
Pelvis kV-CBCT (Linac-based) chest 0.3 1.0
Organ D50 Range (cGy) kV-CBCT (Linac-based) pelvis 1.0 3.0
Bladder 1.1–2.5 A pair of radiographs (kV) head 0.01 0.2
Rectum 1.3–2.4 A pair of radiographs (kV) chest 0.01 0.5
Small Bowel 1.1–2.3 A pair of radiographs (kV) pelvis 0.01 2.0
Source: Reference [1]. Note: Organs are within the imaged region of interest.
stereoscopic devices equipped with angled x-rays tubes stereotactic body radiation therapy for lung tumors:
used by the CyberKnife (Accuray) or BrainLab ExacTrac Initial experience,” Lung Cancer 56(1), 77–88 (2007).
4. D. A. Jaffray and J. H. Siewerdsen, “Cone-beam com-
X-ray (Brainlab) system, the dose from a pair of kV radio-
puted tomography with a flat-panel imager: Initial
graphs are generally less than that from linac-based kV performance characterization,” Medical Physics 27(6),
systems due to extended distances between x-ray source 1311–1323 (2000).
and the patient. 5. D. Letourneau, J. W. Wong, M. Oldham, M. Gulam, L.
Watt, D. A. Jaffray, J. H. Siewerdsen, A. A. Martinez,
30.4 SUMMARY “Cone-beam-CT guided radiation therapy: Technical
implementation,” Radiotherapy & Oncology 75(3), 279–
In this chapter, the imaging guidance doses to radio- 286 (2005).
therapy patients from commonly used imaging 6. M. Oldham, D. Letourneau, L. Watt, G. Hugo, D. Yan,
modalities and their dosimetry were briefly reviewed. D. Lockman, L. H. Kim, P. Y. Chen, A. Martinez, J.
Techniques and methods for imaging dose determi- W. Wong, “Cone-beam-CT guided radiation therapy:
nations were briefly described. Tabulated organ doses A model for on-line application,” Radiotherapy &
Oncology 75(3), 271–278 (2005).
from commonly used image acquisition procedures
7. C. Thilmann, S. Nill, T. Tucking, A. Hoss, B. Hesse, L.
were provided for a quick reference. The tabulated data Dietrich, R. Bendl, B. Rhein, P. Haring, C. Thieke, U.
can be used to estimate if the accumulated imaging Oelfke, J. Debus, P. Huber, “Correction of patient posi-
dose from multiple imaging procedures exceeds the tioning errors based on in-line cone beam CTs: Clinical
threshold of 5% of the therapeutic target dose beyond implementation and first experiences,” Radiation
Oncology 1, 16 (2006).
which imaging dose should be considered in the treat-
8. ICRU-91, “ICRU REPORT 91: Prescribing, recording,
ment planning process as recommended in AAPM and reporting of stereotactic treatments with small pho-
TG-180 [1]. The provided data can be used by clinicians ton beams,” Journal of the ICRU 14(2) (2014).
to make informed decisions in selecting an appropriate 9. P. Alaei and G. X. Ding (eds.), “Image guidance in radi-
imaging procedure regarding the risk and benefits of ation therapy: Techniques, accuracy, and limitations,”
the x-ray image guidance. in AAPM Medical Physics Monograph No 39 (Medical
Physics Publishing, Madison, WI, 2018), Vol. 39, ISBN
The goal of imaging guidance is to position the 978-1936366-62-0 (Hardcopy), ISBN 978-1936366-63-7
patient accurately so that the radiotherapy beams can be (eBook).
delivered to the target precisely. A variety of techniques 10. V. Dyk (ed.), The modern technology of radiation oncol-
to reduce the imaging dose are available. Although ogy: A compendium for medical physicists and radiation
ALARA should always be the guiding principle in prac- oncologists (Medical Physics Publishing Corporation,
Madison, WI, 2005), Vol. 2, ISBN-13 978-1930524255,
tice, balancing ALARA principle with the requirement ISBN-10 1930524250.
for effective target localization requires that imaging 11. A. L. Boyer, L. Antonuk, A. Fenster, M. Van Herk,
dose be managed on the consideration of weighing risks H. Meertens, P. Munro, L. E. Reinstein, J. Wong, “A
and benefits to the patient [1]. review of electronic portal imaging devices (EPIDs),”
Medical Physics 19(1), 1–16 (1992). https://doi.
REFERENCES org/10.1118/1.596878
12. M. Strandqvist and B. Rosengren, “Television-controlled
1. G. X. Ding, P. Alaei, B. Curran, R. Flynn, M. Gossman, pendulum therapy,” The British Journal of Radiology
T. R. Mackie, M. Miften, R. Morin, X. G. Xu, T. C. Zhu, 31(369), 513–514 (1958). https://doi.org/10.1259/0007-
“Image guidance doses delivered during radiotherapy: 1285-31-369-513
quantification, management, and reduction: Report of 13. H. Wallman and N. Stålberg, “A television-Röntgen
the AAPM Therapy Physics Committee Task Group system for pendulum therapy,” The British Journal
180,” Medical Physics 45(5), e84–e99 (2018). of Radiology 31(370), 576–577 (1958). https://doi.
2. M. J. Murphy, J. M. Balter, J. A. BenComo, I. Das, org/10.1259/0007-1285-31-370-576
S. Jiang, C. M. Ma, G. Olivera, R. F. Rodebaugh, K. 14. J. R. Andrews, R. W. Swain, P. Rubin, “Continuous
Ruchala, H. Shirato, F. F. Yin, “The management of visual monitoring of 2 mev. roentgen therapy,” The
imaging dose during image-guided radiotherapy: American Journal of Roentgenology, Radium Therapy,
Report of the AAPM Task Group 75,” Medical Physics and Nuclear Medicine 79(1), 74–78 (1958).
34(10), 4041–4063 (2007). 15. S. Benner, B. Rosengren, H. Wallman, O. Netteland,
3. D. M. Duggan, G. X. Ding, C. W. Coffey, 2nd, W. Kirby, “Television monitoring of a 30 MV x-ray beam,” Journal
D. E. Hallahan, A. Malcolm, B. Lu, “Deep-inspiration of Scientific Instruments 7(1), 29–34 (1962). https://doi.
breath-hold kilovoltage cone-beam CT for setup of org/10.1088/0031-9155/7/1/302
16. N. A. Baily, R. A. Horn, T. D. Kampp, “Fluoroscopic Harari, P. Reckwerdt, B. Paliwal, M. Ritter, H. Keller, J.
visualization of megavoltage therapeutic x ray beams,” Fowler, M. Mehta, “Image guidance for precise confor-
International Journal of Radiation Oncology, Biology, mal radiotherapy,” International Journal of Radiation
Physics 6(7), 935–939 (1980). https://doi.org/10.1016/0360- Oncology, Biology, Physics 56(1), 89–105 (2003).
3016(80)90341-7 29. D. A. Jaffray, D. G. Drake, M. Moreau, A. A. Martinez,
17. H. Meertens, M. Van Herk, J. Weeda, “A liquid ioni- J. W. Wong, “A radiographic and tomographic imag-
sation detector for digital radiography of therapeutic ing system integrated into a medical linear accelera-
megavoltage photon beams,” Plasma Sources Science tor for localization of bone and soft-tissue targets,”
and Technology 30(4), 313–321 (1985). https://doi. International Journal of Radiation Oncology, Biology,
org/10.1088/0031-9155/30/4/004 Physics 45(3), 773–789 (1999).
18. J. Leong, “Use of digital fluoroscopy as an online verifica- 30. A. S. Shiu, E. L. Chang, J. S. Ye, M. Lii, L. D. Rhines, E.
tion device in radiation therapy,” Physics in Medicine and Mendel, J. Weinberg, S. Singh, M. H. Maor, R. Mohan,
Biology 31(9), 985–992 (1986). https://doi.org/10.1088/ J. D. Cox, “Near simultaneous computed tomogra-
0031-9155/31/9/004 phy image-guided stereotactic spinal radiotherapy:
19. M. van Herk and H. Meertens, “A matrix ionisation An emerging paradigm for achieving true stereotaxy,”
chamber imaging device for on-line patient setup verifi- International Journal of Radiation Oncology, Biology,
cation during radiotherapy,” Radiotherapy and Oncology Physics 57(3), 605–613 (2003).
11(4), 369–378 (1988). https://doi.org/10.1016/0167- 31. M. Uematsu, T. Fukui, A. Shioda, H. Tokumitsu, K.
8140(88)90208-3 Takai, T. Kojima, Y. Asai, S. Kusano, “A dual computed
20. A. Ezz, P. Munro, A. T. Porter, J. Battista, D. A. Jaffray, A. tomography linear accelerator unit for stereotactic radi-
Fenster, S. Osborne, “Daily monitoring and correction ation therapy: A new approach without cranially fixated
of radiation field placement using a video-based portal stereotactic frames,” International Journal of Radiation
imaging system: A pilot study,” International Journal Oncology, Biology, Physics 35(3), 587–592 (1996).
of Radiation Oncology, Biology, Physics 22(1), 159–165 32. K. M. Yenice, D. M. Lovelock, M. A. Hunt, W. R.
(1992). https://doi.org/10.1016/0360-3016(92)90995-T Lutz, N. Fournier-Bidoz, C. H. Hua, J. Yamada, M.
21. A. L. Boyer, and H. Meertens, “A review of electronic Bilsky, H. Lee, K. Pfaff, S. V. Spirou, H. I. Amols, “CT
portal imaging devices (EPIDs),” Medical Physics 19(1), image-guided intensity-modulated therapy for para-
1–16 (1992). https://doi.org/10.1118/1.596878 spinal tumors using stereotactic immobilization,”
22. P. Munro, “Portal imaging technology: past, present, and International Journal of Radiation Oncology, Biology,
future,” Seminars in Radiation Oncology 5(2), 115–133 Physics 55(3), 583–593 (2003).
(1995). https://doi.org/10.1016/S1053-4296(95)80005-0 33. J. Pouliot, A. Bani-Hashemi, J. Chen, M. Svatos, F.
23. L. E. Antonuk, “Electronic portal imaging devices: Ghelmansarai, M. Mitschke, M. Aubin, P. Xia, O.
A review and historical perspective of contempo- Morin, K. Bucci, M. Roach, 3rd, P. Hernandez, Z.
rary technologies and research,” Physics in Medicine Zheng, D. Hristov, L. Verhey, “Low-dose megavoltage
and Biology 47(6), R31–R65 (2002). https://doi. cone-beam CT for radiation therapy,” International
org/10.1088/0031-9155/47/6/201 Journal of Radiation Oncology, Biology, Physics 61(2),
24. M. Murphy, M. Balter, J. BenComo, I. Das, S. Jiang, C. 552–560 (2005).
M. Ma, G. Olivera, R. Rodebaugh, K. Ruchala, H. F. 34. J. R. Perks, J. Lehmann, A. M. Chen, C. C. Yang, R. L. Stern,
Yin, “The management of imaging dose during image- J. A. Purdy, “Comparison of peripheral dose from image-
guided radiotherapy: Report of the AAPM Task Group guided radiation therapy (IGRT) using kV cone beam
75,” Medical Physics 34, 4041–4063 (2007). CT to intensity-modulated radiation therapy (IMRT),”
25. E. C. Ford, J. Chang, K. Mueller, K. Sidhu, D. Todor, G. Radiotherapy & Oncology 89(3), 304–310 (2008).
Mageras, E. Yorke, C. C. Ling, H. Amols, “Cone-beam CT 35. M. Stock, A. Palm, A. Altendorfer, E. Steiner, D. Georg,
with megavoltage beams and an amorphous silicon elec- “IGRT induced dose burden for a variety of imaging pro-
tronic portal imaging device: Potential for verification tocols at two different anatomical sites,” Radiotherapy &
of radiotherapy of lung cancer,” Medical Physics 29(12), Oncology 102(3), 355–363 (2011). https://doi.org/S0167-
2913–2924 (2002). https://doi.org/10.1118/1.1517614 8140(11)00607-4 [pii] 10.1016/j.radonc.2011.10.005
26. T. R. Mackie, J. Balog, K. Ruchala, D. Shepard, S. 36. D. E. Hyer and D. E. Hintenlang, “Estimation of organ
Aldridge, E. Fitchard, P. Reckwerdt, G. Olivera, T. doses from kilovoltage cone-beam CT imaging used
McNutt, M. Mehta, “Tomotherapy,” Seminars in during radiotherapy patient position verification,”
Radiation Oncology 9(1), 108–117 (1999). Medical Physics 37(9), 4620–4626 (2010).
27. T. R. Mackie, T. Holmes, S. Swerdloff, P. Reckwerdt, J. O. 37. A. Palm, E. Nilsson, L. Herrnsdorf, “Absorbed dose and
Deasy, J. Yang, B. Paliwal, T. Kinsella, “Tomotherapy: dose rate using the Varian OBI 1.3 and 1.4 CBCT sys-
A new concept for the delivery of dynamic conformal tem,” Journal of Applied Clinical Medical Physics 11(1),
radiotherapy,” Medical Physics 20(6), 1709–1719 (1993). 3085 (2010).
28. T. R. Mackie, J. Kapatoes, K. Ruchala, W. Lu, C. Wu, 38. S. Kim, S. Yoo, F. F. Yin, E. Samei, T. Yoshizumi,
G. Olivera, L. Forrest, W. Tome, J. Welsh, R. Jeraj, P. “Kilovoltage cone-beam CT: Comparative dose and
image quality evaluations in partial and full-angle scan image-guided radiotherapy procedure,” International
protocols,” Medical Physics 37(7), 3648–3659 (2010). Journal of Radiation Oncology, Biology, Physics 73(2),
39. M. W. Kan, L. H. Leung, W. Wong, N. Lam, “Radiation 610–617 (2009).
dose from cone beam computed tomography for image- 51. G. X. Ding, P. Munro, J. Pawlowski, A. Malcolm, C. W.
guided radiation therapy,” International Journal of Coffey, “Reducing radiation exposure to patients from
Radiation Oncology, Biology, Physics 70(1), 272–279 kV-CBCT imaging,” Radiotherapy & Oncology 97(3),
(2008). https://doi.org/S0360-3016(07)04081-3 [pii] 585–592 (2010). https://doi.org/S0167-8140(10)00468-8
10.1016/j.ijrobp.2007.08.062 [pii] 10.1016/j.radonc.2010.08.005
40. A. Shah, E. Aird, J. Shekhdar, “Contribution to normal 52. Y. Zhang, Y. Yan, R. Nath, S. Bao, J. Deng, “Personalized
tissue dose from concomitant radiation for two com- assessment of kV cone beam computed tomography
mon kV-CBCT systems and one MVCT system used in doses in image-guided radiotherapy of pediatric cancer
radiotherapy,” Radiotherapy & Oncology 105(1), 139– patients,” International Journal of Radiation Oncology,
144 (2012). https://doi.org/S0167-8140(12)00221-6 [pii] Biology, Physics 83(5), 1649–1654 (2012). https://doi.org/
10.1016/j.radonc.2012.04.017 S0360-3016(11)03480-8 [pii] 10.1016/j.ijrobp.2011.10.072
41. N. Wen, H. Guan, R. Hammoud, D. Pradhan, T. Nurushev, 53. Y. Zhang, Y. Yan, R. Nath, S. Bao, J. Deng, “Personalized esti-
S. Li, B. Movsas, “Dose delivered from Varian’s CBCT to mation of dose to red bone marrow and the associated leu-
patients receiving IMRT for prostate cancer,” Physics in kaemia risk attributable to pelvic kilo-voltage cone beam
Medicine and Biology 52(8), 2267–2276 (2007). computed tomography scans in image-guided radiother-
42. C. Walter, J. Boda-Heggemann, H. Wertz, I. Loeb, A. apy,” Physics in Medicine and Biology 57(14), 4599–4612
Rahn, F. Lohr, F. Wenz, “Phantom and in-vivo measure- (2012). https://doi.org/10.1088/0031-9155/57/14/4599
ments of dose exposure by image-guided radiotherapy 54. B. R. Walters, G. X. Ding, R. Kramer, I. Kawrakow,
(IGRT): MV portal images vs. kV portal images vs. “Skeletal dosimetry in cone beam computed tomogra-
cone-beam CT,” Radiotherapy & Oncology 85(3), 418– phy,” Medical Physics 36(7), 2915–2922 (2009).
423 (2007). https://doi.org/S0167-8140(07)00527-0 [pii] 55. E. Spezi, P. Downes, E. Radu, R. Jarvis, “Monte Carlo
10.1016/j.radonc.2007.10.014 simulation of an x-ray volume imaging cone beam CT
43. S. C. Jeng, C. L. Tsai, W. T. Chan, C. J. Tung, J. K. Wu, unit,” Medical Physics 36(1), 127–136 (2009).
J. C. Cheng, “Mathematical estimation and in vivo dose 56. P. Downes, R. Jarvis, E. Radu, I. Kawrakow, E. Spezi,
measurement for cone-beam computed tomography “Monte Carlo simulation and patient dosimetry for a
on prostate cancer patients,” Radiotherapy & Oncology kilovoltage cone-beam CT unit,” Medical Physics 36(9),
92(1), 57–61 (2009). 4156–4167 (2009).
44. G. X. Ding and A. W. Malcolm, “An optically stimulated 57. E. Spezi, P. Downes, R. Jarvis, E. Radu, J. Staffurth, “Patient-
luminescence dosimeter for measuring patient expo- specific three-dimensional concomitant dose from cone
sure from imaging guidance procedures,” Physics in beam computed tomography exposure in image-guided
Medicine and Biology 58(17), 5885–5897 (2013). https:// radiotherapy,” International Journal of Radiation Oncology,
doi.org/10.1088/0031-9155/58/17/5885 Biology, Physics 83(1), 419–426 (2011). https://doi.org/
45. R. M. Al-Senan and M. R. Hatab, “Characteristics of an S0360-3016(11)02860-4 [pii] 10.1016/j.ijrobp.2011.06.1972
OSLD in the diagnostic energy range,” Medical Physics 58. G. X. Ding, and P. Munro, “Radiation exposure to
38(7), 4396–4405 (2011). patients from image guidance procedures and tech-
46. V. Schembri and B. J. Heijmen, “Optically stimulated niques to reduce the imaging dose,” Radiotherapy &
luminescence (OSL) of carbon-doped aluminum oxide Oncology 108(1), 91–98 (2013). https://doi.org/10.1016/j.
(Al2O3:C) for film dosimetry in radiotherapy,” Medical radonc.2013.05.034
Physics 34(6), 2113–2118 (2007). 59. G. Ding and P. Munro, “The imaging dose to patients from a
47. N. Tomic, S. Devic, F. DeBlois, J. Seuntjens, “Comment 2.5 MV imaging beam.” International Journal of Radiation
on “Reference radiochromic film dosimetry in kilovolt- Oncology Biology Physics 99(2), E654 (2017). https://www.
age photon beams during CBCT image acquisition” redjournal.org/article/S0360-3016(17)33233-9/pdf.
[Medical Physics 37, 1083–1092 (2010)],” Medical Physics 60. P. R. Almond, P. J. Biggs, B. M. Coursey, W. F. Hanson,
37(6), 3008 (2010). M. S. Huq, R. Nath, D. W.O. Rogers, “AAPM’s TG-51
48. J. C. Chow, M. K. Leung, M. K. Islam, B. D. Norrlinger, protocol for clinical reference dosimetry of high-energy
D. A. Jaffray, “Evaluation of the effect of patient dose photon and electron beams,” Medical Physics 26(9),
from cone beam computed tomography on prostate 1847–1870 (1999).
IMRT using Monte Carlo simulation,” Medical Physics 61. P. Andreo, D. T. Burns, K. Hohlfeld, M. S. Huq, T.
35(1), 52–60 (2008). Kanai, F. Laitano, V.G. Smyth, S. Vynckier, “Absorbed
49. G. X. Ding, D. M. Duggan, C. W. Coffey, “Accurate dose determination in external beam radiotherapy:
patient dosimetry of kilovoltage cone-beam CT in radi- An international code of practice for dosimetry based
ation therapy,” Medical Physics 35(3), 1135–1144 (2008). on standards for absorbed dose to water,” in Technical
50. G. X. Ding and C. W. Coffey, “Radiation dose from Report Series, IAEA (International Atomic Energy
kilovoltage cone beam computed tomography in an Agency, Vienna, 2000), Vol. 398.
62. G. X. Ding and P. Munro, “Characteristics of 2.5MV 69. D. W. Rogers, “Fifty years of Monte Carlo simulations
beam and imaging dose to patients,” Radiotherapy for medical physics,” Physics in Medicine and Biology
& Oncology 125(3), 541–547 (2017). https://doi. 51(13), R287–R301 (2006).
org/10.1016/j.radonc.2017.09.023 70. J. Seco and F. Verhaegen (eds.), Monte Carlo Techniques
63. C. M. Ma, C. W. Coffey, L. A. DeWerd, C. Liu, R. Nath, in Radiation Therapy (CRC Press, 2016), ISBN
S. M. Seltzer, J. P. Seuntjens, “AAPM protocol for 40-300 9781138199903.
kV x-ray beam dosimetry in radiotherapy and radiobi- 71. E. Spezi, P. Alaei, G. X. Ding, “Calculations of the
ology,” Medical Physics 28(6), 868–893 (2001). imaging dose with emphasis on kilovoltage imaging,”
64. G. X. Ding and C. W. Coffey, “Beam characteristics and in Image Guidance in Radiation Therapy: Techniques,
radiation output of a kilovoltage cone-beam CT,” Physics in Accuracy, and Limitations, AAPM Monogram No.39,
Medicine and Biology 55(17), 5231–5248 (2010). https://doi. edited by P. Alaei and G. X. Ding (Medical Physics
org/S0031-9155(10)51181-3 [pii] 10.1088/0031-9155/55/17/022 Publishing, Madison, WI, 2018), ISBN 978-1936366-
65. M. Miften, O. Gayou, B. Reitz, R Fuhrer, “IMRT plan- 62-0 (Hardcopy) ISBN 978-1936366-63-7 (eBook).
ning and delivery incorporating daily dose from mega- 72. A. P. Nelson and G. X. Ding, “An alternative approach
voltage cone-beam computed tomography imaging,” to account for patient organ doses from imaging guid-
Medical Physics 34(10), 3760–3767 (2007). ance procedures,” Radiotherapy & Oncology 112(1),
66. J. M. Pawlowski and G. X. Ding, “An algorithm for 112–118 (2014). https://doi.org/S0167-8140(14)00265-5
kilovoltage x-ray dose calculations with applications in [pii] 10.1016/j.radonc.2014.05.019
kV-CBCT scans and 2D planar projected radiographs,” 73. P. Alaei, E. Spezi, M. Reynolds, “Dose calculation and
Physics in Medicine and Biology 59(8), 2041–2058 (2014). treatment plan optimization including imaging dose
https://doi.org/10.1088/0031-9155/59/8/2041 from kilovoltage cone beam computed tomography,”
67. P. Alaei, G. X. Ding, H. Guan, “Inclusion of the dose Acta Oncologica 53(6), 839–844 (2014). https://doi.org/
from kilovoltage cone beam CT in the radiation therapy 10.3109/0284186X.2013.875626
treatment plans,” Medical Physics 37(1), 244–248 (2010).
68. F. Verhaegen and J. Seuntjens, “Monte Carlo model-
ling of external radiotherapy photon beams,” Physics in
Medicine and Biology 48(21), R107–R164 (2003).
Index
Note: Locators in italics represent figures and bold indicate ArcCHECK, 44, 45, 178, 297, 298, 339
tables in the text. Area monitoring, 188–191
calibration of, 188–189
neutron calibration, 189–190
A
photon calibration, 189
Absolute detectors, in proton therapy dosimetry, 396–398 properties of, 190–191
Absolute dose calibration, 285, 285–287 calibration and long-term stability, 190
Alfonso formalism, 286–287 directional dependence, 190
beam quality, 285 discrimination in mixed fields, 190–191
perturbation and volume averaging effects, 285–286, 286 energy dependence, 190
Absorbed dose, 3–5 linearity of response, 190
Absorbed-dose calibration factor, 264 measurement range, 190
Accelerator-based neutron sources (ABNS), 186 measurement uncertainties, 190
Acceptance and commissioning; see also Quality assurance response time, 190
clinical commissioning, 370 sensitivity, 190
equipment for, 181 Australian Clinical Dosimetry Service (ACDS), 103, 272
and QA measurement, 380
test for, 180–181
B
Acceptance testing for kilovoltage x-ray beam dosimetry,
367–369 Beam(s)
Accredited Dosimetry Calibration Laboratories (ADCLs), 21, collimation, 329–332, 332
233, 264 data acquisition for ZAP-X, 310–311, 311
Acoustic-based proton range verification, 443–453 dose profiles, 262, 262
current state of protoacoustics, 450–453 dosimetric considerations with FFF, 267–273
introduction to, 443–444 energy spectrum and corrections with FFF, 271
protoacoustic emissions, 445–449, 446–449 non-uniformity, 270
protoacoustic image formation, 450, 451 orientation and cavity size, 321–322, 322
thermoacoustic phenomena, 444–445 production and monitoring in helical tomotherapy,
TOF, 450–452, 452 327–329, 331
Active monitoring, 188, 191–194; see also Passive monitoring quality, 269, 269–270, 270, 285
gas-filled detectors, 191 scanned, 402–403, 403
GM counters, 192–193 scattered, 401–402, 402
ionization chambers, 191–192, 192, 193, 194 shape, 268, 268
in photon monitoring techniques, 191–195 spectrum, 281
proportional counters, 193–194 Beamlet, 336, 337
Active neutron monitoring, 198–202; see also Passive neutron BEAMnrc, 215, 220
monitoring Beam-quality dependence, 26
dose equivalent meters, 199–202, 200, 201 Beta emitting sources, 246–247
fluence meters, 198–199 clinical user calibrations, 246
neutron spectrometers, 202 clinical user measurements, 246
ADCL brachytherapy calibrations, 236–237, 237 dose calculations, 246–247
Air density effects, 26–27 Beta particle brachytherapy standards, 235
Ambient dose equivalent, 187 Boron neutron capture therapy (BNCT), 186
American Association of Physicists in Medicine (AAPM), 11, 21, Brachytherapy dosimetry, 23, 231-252
63, 77, 233, 343, 346, 376, 405 calibrations and measurements
Angular straggling, 458 ADCL and SSDL, 235–237, 237
Annual QA, 182, 349; see also Quality assurance NIST/PSDL, 232–235
Anti-coincidence detection configuration, 431, 431 source vendor, 238
479
480 ◾ Index
clinical user calibrations real-time filed verification, 145

for beta emitting sources, 246–247 real-time in vivo dose verification, 146
for photon emitting sources, 240–242, 242 weakness
clinical user measurements, 242–244 limited to MV photon and electron therapy
dose calculations, 244–246 beams, 146
algorithms, 245–246 surface-limited dosimetry, 146
for permanent and temporary implants, 246 Cherenkov radiation in optical fibers, 128–131, 129, 130
TG-43 reports, 244, 244–245 Clinical user calibrations
introduction to, 232 for beta emitting sources, 246–247
investigators, 238–239 for photon emitting sources, 240–242, 242
model-based vs. TG-43, 247 Clinical user measurements, 242–244
new radiation dosimeters, 247–248 for beta emitting sources, 246
safety and earning, 248 for photon emitting sources, 242–244
societal recommendations, 239–240 Cobalt-60, 321, 344
consensus data, 240 Collecting electrode potential, 25
data dissemination, 240 Collimator scatter factor (Sc), 256–258, 257, 258, 260, 263, 296
potential future directions, 240 definition of, 256–257
rationale, 239–240 COMPASS (software tool), 180
standards, see Brachytherapy standards Compton camera (CC), 434, 436
in vivo quality assurance, 247 Compton scattering effect, 8
Brachytherapy standards Cone-based output factors, 290, 290
beta particle, 235 Consensus data, 240
electronic, 235, 235 Continuous slowing down approximation (CSDA), 6
high-energy HDR, 234, 234–235 Correction factors, 265, 270
high-energy LDR, 234 application of, in OSLD, 101–103
low-energy LDR, 233, 233–234 beam quality, 398
retired, 235 charge, 16
BrainLab ExacTrac X-ray (Brainlab) system, 469 detector specific, 287, 290–292, 291, 291, 292
Bubble detectors, 204, 204 field output, 291, 292
IAEA TRS 483, 307
overall chamber, 361
C
perturbation, 285
Calibrated units (CU), 110 for PMMA Sheaths, 361
Calibration coefficient, 21, 101 Creep-up effect, 46
Calibration factor, 110 CyberKnife®, 155, 163, 305-310
Calorimeter/calorimetry, 397–398 introduction to, 306
applications of, 36–37 OARs, 308–310, 309
basis of, 31 outputs factors, 307, 307–308
components of, 32 TG-51 formalism for, 306–307
graphite, 33–34 TPRs, 308, 309
interferometer, 36 CyberKnife (Accuray) system, 468
primary standards, 31
ultrasonic, 35
D
water, 34–36, 34–36
Cavity theory, 11–13 Daily QA, 181; see also Quality assurance
Cesium iodide – CsI(Tl), 434 Data, 333
Characteristics curve, 64 consensus, 240
Charged particle equilibrium (CPE), 256, 257 dissemination, 240
Charged particles therapy dosimetry, 62 PDD, 311
Charged particle transport, 5–7 DAVID (PTW, Freiburg, Germany), 179
Charge measurements corrections, 264–265 DBSCAN, 217
Cherenkov imaging dosimetry, 139–147 Delta4 DISCOVER (ScandiDos, Uppsala, Sweden), 179
Cherenkov light, 140–141 Delta4 PHANTOM, 178
introduction to, 139–140, 140 Detector, see also specific types
linac pulsing and time gated real-time imaging, 143–145, arrays for patient-specific QA, 177–179, 178
144, 145 MOSFETs, 47
strength and reader theory, 97–99, 98, 99
no extraneous dose, 145 for relative dosimetry, 287–289, 288
nuclear medicine applications, 145 scintillation fiber optic dosimetry, 127–128
Index ◾ 481
specific correction factors, 290–292, 291, 291, 292 Electrometers, 24–25, 289
for stereotactic radiosurgery, 285 for kilovoltage x-ray beam dosimetry, 361
Diamonds, 48–51 Electron contamination, 273
construction and functioning, 48, 48–50 Electron dosimetry, 375–388
dosimetry with, 50, 50–51 properties of electron beams, 375–376
Digital electrometer, 25 reference dosimetry, 376–380
Diode arrays, 298, 299 for non-standard electron modalities, 378–380
Diodes, 40–44, 152–154 verification, 378
construction and functioning, 40, 40–42 review of, 375
dosimetry with, 42–44, 44 solid phantom measurements, 387–388
Directional dose equivalent, 187 film measurements, 387–388
Dissemination, data, 240 other measurements, 388
Dolphin (IBA, Schwarzenbruck, Germany), 179–180 QA measurements, 387
Dose in vivo dosimetry, 388
absorbed, 3–5 water-phantom dosimetry, 376–386
calculations, see Dose calculations commissioning and QA measurements, 380
defined, 3, 4 detector choice, 380–381
at depth, 364–365, 366, 367 lateral profiles, 384–385, 385
medium, 4 output factors, 385–386
on surface, 363–364, 364, 365 PDD, 381–384, 383, 384
Dose calculations, 4, 99, 99–103, 101 Electronic brachytherapy (eBT) standards, 235, 235
for beta emitting sources, 246–247 Electronic personal dosimeters, 197–198
for photon emitting sources, 244–246 Electronic portal imaging devices (EPIDs), 109–119, 156–158, 299;
Dose enhancement effect, 47 See also EPID-based dosimetry
Dose equivalent, 187 Electron return effect, quantifying, 321, 321
Dose-equivalent meters, 199–202, 200, 201 Elekta Versa HD, 279, 280
Dose horns, 262 EMPIRE, 435
Dose rate, 269 Energy dependence, 26, 190
Dosimeters, see also specific types FFF beams, 272–273
electronic personal, 197–198 Energy selection, 432
for kilovoltage x-ray beam dosimetry, 360–361, 362 Energy spectrum and corrections with FFF beams, 271
optically stimulated luminescence, see Optically stimulated EPDL97, 220
luminescence dosimeters EPID-based dosimetry, 109–119
Dosimetry, see also specific types applications, 110
detectors used for relative, 287–289, 288 introduction to, 109
electron, see Electron dosimetry linearity of, 111, 111
EPID-based, see EPID-based dosimetry for patient-specific QA, 179
with FFF beams, 267–273 technology and characteristics of, 109–111, 110, 111
film, see Film dosimtery use of, for linac QA, 111–117
gamma knife, see Gamma knife dosimetry flatness and symmetry, 116
of imaging modalities in radiotherapy, 467–474 imager QA, 112
in vivo, see In vivo dosimetry isocentricity, 114–115
kilovoltage x-ray beam, see Kilovoltage x-ray beam dosimetry jaw positioning, 113–114
linac-based SRS/SBRT, see Linac-based SRS/SBRT dosimetry MLC positioning, 112–113
Magnetic field, see Magnetic field dosimetry output, 115–116
Photon beam, see Photon beam dosimetry variable axis angles, 115
Proton therapy, see Proton therapy dosimetry wedge, 116–117
scintillation fiber optic, see Scintillation fiber optic dosimetry use of, for patient QA, 117–119
scintillation imaging, see Scintillation imaging dosimetry EPIgray (DOSIsoft), 157
small field, see Small field dosimetry Equivalent dose, 187
thermoluminescence, see Thermoluminescence dosimetry External beam radiation therapy (EBRT), 155
water-phantom dosimetry, see Water-phantom dosimetry
DOSXYZnrc, 218, 220
F
Failure modes effect analyses (FMEAs), 248
E
Faraday Cup, 398
EBRT, See External beam radiation therapy FFF beams, see Flattening filter-free (FFF) beams
EGS4/EGSnrc, 220 Film dosimeters, 61-71, 197
Elastic collisions, 5, 394 use in in vivo study, 160–162
482 ◾ Index
Film dosimetry, 61–71 beamlets, 336, 337

electron dosimetry, 387–388 dose calculation by convolution/superposition, 337–338
introduction to, 62 optimization process, 336–337
radiochromic, 64–67, 65, 66, 399 user-defined parameters, 335–336
radiographic, 62–64, 63, 64 introduction to, 326–327, 326–331
Flatness (F), defined, 116, 262 major components of, 326
Flattening filter-free (FFF) beams, 267–273 motion synchronization, 334–335
characteristics of, 268–270 MVCT imaging, 332–333
beam quality, 269, 269–270, 270 patient positioning, 334
beam shape, 268, 268 quality assurance
dose rate, 269 absolute dose measurement and calibration, 338–339
characteristics of commercially available, 280 Hollow waveguides, 125, 125
introduction to, 268
other clinical considerations
I
electron and neutron contamination, 273
motion and interplay effects, 273 IBA StealthCHAMBERTM, 293
small fields, 273 Image-guided radiation therapy (IGRT)
reference dosimetry in, 270–271 imaging dose
beam non-uniformity, 270 need of dosimetry for, 469–470
energy spectrum and corrections with, 271 to patients, 473–474
recombination, 270–271 imaging in, 470–473
routine clinical measurements, 271–273 introduction to, 467–470
volume averaging effects, 285–286, 286 measurements of output of beams used for, 470
FLUKA, 201, 217, 218, 221, 432 patient imaging dose calculations, 470
Fluke Biomedical7 Neutron Survey Meter Model 190N, 200 patient imaging dose estimations, 471–473, 472–473
Fragment induced background, 431 planar imaging modalities, 467–468, 468, 469
Frank, Ilya, 128 volumetric systems, 468–469
Full energy deposition event, 430 In-air output ratio, 256
Incident learning systems (ILS), 248
INCISE MLC, 306
G
Inelastic collisions, 5
G4RunManager, 219 Intensity-modulated radiation therapy (IMRT), 62, 284
G4VUserActionInitialization, 219 introduction to, 174
Gafchromic film, 287, 298, 399 optimization of, 322
Gamma Knife dosimetry, 343–352 plans
absolute dosimetry, 345–347, 346 beamlets, 336, 337
end-to-end testing, 350–352, 351 dose calculation by convolution/superposition, 337–338
introduction to, 343–344 optimization process, 336–337
measurement, 345 user-defined parameters, 335–336
relative dosimetry quality assurance
off-axis distributions, 349–350, 351 acceptance and commissioning, 180–181
output factors, 347–349, 348, 350 annual QA, 182
safety, 344–345 daily QA, 181
Gas-filled detectors, 188, 191 detector arrays, 177–179, 178
GATE, 220 devices and tools used for, 178
Geant4, 218–220, 219, 432, 435 dosimeters and devices, 175–180
Geant4-DNA, 216, 217 EPID-based analysis, 179
GM counters, 192–193 log file analysis, 179
Graphite calorimeter, 33–34, 33 measurement using film, 176–177
monthly QA, 181–182
patient-specific, 174–178
H
point measurements, 175–176
H&D curve, 64, 64 quantitative evaluation tools, 175
Helical tomotherapy, 325–339, 326 rationale, 174–175
beam collimation, 329–332, 332 rationale, 180
beam production and monitoring, 327–329, 331 3D dosimeters, 180
head and neck tomotherapy plan, 327 Interferometer calorimeter, 36
IMRT breast plan, 328 International Atomic Energy Association (IAEA), 233
IMRT plans International Bureau of Weights and Measures (BIPM), 233
Index ◾ 483
International Commission on Radiation Units and Measurements routine quality assurance, 370–371
(ICRU), 365, 394 safety tests, 368
International Commission on Radiological Protection (ICRP), x-ray beam performance checks, 369
153, 187 dose at depth, 364–365, 366, 367
International Electrotechnical Commission (IEC), 282 dose on surface, 363–364, 364, 365
Interplay effects of FFF beams, 273 dosimeters for, 360–361, 362
Intra nuclear cascade (INC), 432 electrometers for, 361
Intraoperative radiation therapy (IORT), 155, 355 formalisms for, 358, 358–360, 359–361
In vivo dosimetry, 151–166 output factors for, 362–363
advantages and disadvantages, 165 PDD and lateral dose profiles, 363
dose verification in particle therapy, 413–414 for phantoms, 361–362
electron dosimetry, 388 protocol, 357–360, 358
offline review dosimeters relative dosimetry measurement, 362–363
film dosimeters, 160–162 review of, 355–357
OSLDs and TLDs, 163–164 specifications, 362, 362
RPLDs, 162–163 uncertainty, 365–366, 368
online review dosimeters, 152–160 Knife-edge slit camera, 434, 435
diodes, 152–154 KURBUC, 216
EPIDs, 156–158
ion chambers, 159–160
L
MOSFET, 154–156
scintillation fiber optic dosimetry, 158–159 Lateral dose profiles, see Percentage depth dose (PDD) and profiles
Ionization chambers, 19–21, 191–192, 192, 193, 194, 288 Leakage currents, 21, 28, 49, 351
arrays, 298–299 Light field versus x-ray field, 369
basic components of, 21–23 Linac-based SRS/SBRT dosimetry, 277–301
brachytherapy, 23 detectors for stereotactic radiosurgery, 285–297
characteristics of, 25–28 absolute dose calibration, 285–287
electrometers, 24–25 MLC characterization, 296–297
introduction to, 19–20 PDD and profiles, 292–295
other components of, 23–25 relative output factors, 287
parallel-plate chambers, 22, 22–23 tissue phantom ratio (TPR), 295–296
proton therapy dosimetry, 396–397 field size definition for, 282–284
thimble chambers, 22–23 jaws, 282–283, 283, 284
triaxial cable, 23, 23–24 MLCs, 283–284
types of, 20–21 stereotactic cones, 284
use in in vivo study, 159–160 introduction to, 278–281
Ionization quenching, 124, 131–132, 132 common treatment platforms, 278
Ionizing radiation, quantities used to describe, 8–10 Elekta Versa HD, 279, 280
Ion recombination factor, 280–281 flattening filter-free, high dose rate modes, 280–281
Isocentricity, 114–115 Novalis Tx, 278–279
iViewDose (Elekta), 157 Varian EDGE and Truebeam, 279–280
patient specific QA measurement, 298–301
Linearity of response, 190, 200
J
Log file analysis for patient-specific QA, 179
Jaw and MLC output factors, 289–290, 290 Log file verification, 300–301
Jaw-based PDD and profiles, 294, 296 Luxel+ dosimeter badge, 196
Jaw positioning, 113–114
Jaws, 282, 282–283, 283, 284
M
Magnetic field dosimetry, 315–323
K
considerations for photons, 316
KERMA (Kinetic Energy Released in Matter), 256, 260 considerations for protons, 316, 317
Kilovoltage x-ray beam dosimetry, 355–371 electron return effect dependencies, 316–318, 317
clinical implementations of, 366–371 equipment and performance, 318, 318–319, 319, 320
acceptance testing, 367–369 introduction to, 315
clinical commissioning, 370 physics of interactions, 316
mechanical tests, 368–369 potential clinical impact, 321–322
radiation dosimetry tests, 369 beam orientation and cavity size, 321–322, 322
radiation tests, 369 optimization of IMRT, 322
484 ◾ Index
treatment planning, 319–321 Neutron contamination, 273

dose calculation/optimization, 319–320 Neutron induced background, 430
quantifying electron return effect, 321, 321 Neutron monitoring techniques, 198–204
unmet needs and new directions, 322–323 active, 198–202
MatriXX Evolution, 298 passive, 202–204
MCNPX, 432, 434 Neutron spectrometers, 195, 202
Measurement range, 190 NIST/PSDL calibration, 232–235
Measurement uncertainties, 191 NOREC, 216
Megavoltage computed tomography (MVCT) imaging, 332–333 Numerical current integrator, 25
Metal Oxide Semiconductor Field Effect Transistors (MOSFETs),
44–48, 45, 154–156, 242, 247, 248, 273, 399
O
construction and functioning, 44–47, 45
dosimetry with, 47–48, 48 OCTAVIUS 1000 SRS, 299
limitations, 47 Off-axis ratios (OARs), 262, 262, 308–310, 312, 309
Multi-leaf collimators characterization, 296–297 for cones, 308
Monaco TPS, 297 for IRIS, 308–309
Pinnacle TPS, 297 for MLC, 309
positioning, 112–113 Off-center ratios (OARs) for ZAP-X, 312, 312
Varian Eclipse TPS, 297 1D–point measurement, 142, 143
Modulation factor, 335–337 1/f noise and fading, 46–47
Monaco TPS, 297 Optical fibers, 124–125
Monitor Unit calculations, 262–263 Cherenkov radiation in, 128–131, 129, 130
Monte Carlo, 211-227 commercially available, 132–133
Monte Carlo codes, 432–433; see also specific codes hollow waveguides, 125, 125
Monte Carlo N-Particle (MCNP) transport code, 220 solid-core fibers, 124, 124–125
Monte Carlo simulation, 131–132, 285–286 Optically stimulated luminescence detectors (OSLD),
advantages of using, 287 97–106
applications, 213–218 proton therapy dosimetry, 399
in dosimetry instruments, 214–215 use in in vivo study, 163–164
in microdosimetry, 215–217, 216 Optically stimulated luminescence dosimeters, 97–106, 196,
in nuclear medicine, 217 196–197
in patient dose calculation, 213–214 detector and reader theory, 97–99, 98, 99
for x-ray imaging, 217–218 dose calculation, 99, 99–103, 101
codes and systems practical applications
EGS4/EGSnrc, 220 brachytherapy, 105
FLUKA, 221 electrons, 105
Geant4, 218–220, 219 external photon beams, 104
MCNP, 220 imaging, 106
PENELOPE, 220–221 out of field, 104–105
introduction, 211–213, 212 protons, 105
limitations, 221 quality assurance, 106
PET-based dose reconstruction, 419, 419–421, 421 small fields, 104
Monthly QA, 181–182; see also Quality assurance surface dosimetry, 104
Motion and interplay effects of FFF beams, 273 precision, 103
Multi-detector arrays, 298–299 reuse of, 103–104
diode arrays, 298 within and without plastic casing, 98
EPID, 299 Organic scintillators, 126
film, 298 Output factors
ion chamber arrays, 298–299 for cones, 290, 290, 294–295, 307–308, 307
Multielement detectors, proton therapy dosimetry, 400 CyberKnife M6 FIM, 307–308, 307
Multi-Leaf Collimators, 160, 263, 283–284 electron dosimetry, 385–386
Multiple Coulomb Scattering (MCS), 376, 385, 394, 395, 458 for gamma knife dosimetry, 347–349
for IRIS, 308
Jaw and MLC, 289–290, 290
N
for kilovoltage x-ray beam dosimetry, 362–363
National Institute of Standards and Technology (NIST), 34, 233, for linac-based SRS/SBRT dosimetry,
264, 346 287–292, 288
Neutron calibration, 189–190 for MLC, 308
Index ◾ 485
for ZAP-X, 311, 312 Photomultiplier tubes (PMTs), 127–128

water-phantom dosimetry, 385–386 Photon beam dosimetry, 255–265
Oxygen concentration, 436–437 absolute dosimetry protocol, AAPM TG-51, 263–265
beam dose profiles and off-axis ratios, 262, 262
collimator scatter factor, 256–258, 257, 258
P
introduction to, 255–256
Parallel-plate chambers, 22, 22–23 monitoring unit calculations, 262–263
Partial energy deposition events, 430 SAD calculations, 263
Particle dosimetry, applications of calorimeters in, 36–37 SSD calculations, 262–263
PARTRAC, 216 percent depth dose curve, 259, 259–260, 260
Passive monitoring, 188, 195–198 phantom scatter factor (Sp), 258, 258–259
calibration, 195–196 TAR, 260–261
electronic personal dosimeters, 197–198 TPR and TMR, 261, 261
film dosimeters, 197 Photon beam loses lateral CPE (LCPE) condition, loss of,
individual monitoring, 195 281–282, 282
characteristics of, 196 Photon calibration, 189
optically stimulated luminescence dosimeters, 196, 196–197 Photon–electron interactions, 7
thermoluminescent dosimeters, 196 Photon emitting sources, 240–242, 242
Passive neutron monitoring, 202–204 Photon interactions, 7–8
bubble detectors, 204, 204 Photon monitoring techniques, 191–198
film, 202–203 active monitoring, 191–195
SSNTDs, 203 passive monitoring, 195–198
TLDs, 203 Picket Fence test, 113
Patient specific QA measurement, 174–178, 298–301 Pinnacle TPS, 297
3D calculation on patient geometry, 299–301 Pitch, defined, 335
log file verification, 300–301 Plastic scintillator detectors (PSD), 126, 289
multi-detector arrays, 298–299 Polarity effects, 28
recommendations for, 300, 301 Polarization in dosimetry, 49, 50, 128, 140
transmission measurement and back projection, 301 Polyethylene (PE), 21, 124, 199, 200, 202, 203,
PENELOPE, 213, 217, 220–222, 339 Polymethylmethacrylate (PMMA), 51, 124, 128–132, 361, 387,
Percentage depth dose (PDD) and profiles, 292–295, 293, 294 417, 434
cone-based profiles, 294–295 Positron emission tomography (PET), 413–421
jaw-based, 294, 296 -based dose reconstruction, 419, 419–421, 421
for kilovoltage x-ray beam dosimetry, 363 -based range monitoring, 417–418, 418
reference detectors, 293–294, 295 signal formation, 414, 414
water-phantom dosimetry, 381–385, 383, 384, 385 signal imaging, 415–417, 416
ZAP-X, 311–312, 312 in vivo dose verification in particle therapy, 413–414
Percent depth dose (PDD) curve, 259, 259–260, 260 Pressure/temperature effects, 26–27
defined, 259 Primary standards dosimetry laboratory (PSDL), 233
Personal dose equivalent, 187 Projections, defined, 336
Perturbation factors, 285–286 Prompt gamma detection
PGI for ion therapy, 437–438 cross sections, 432, 433, 433
linac-based experiments, 437 improving signal-to-noise ratio, 431, 431–432, 432
synchrotron-based experiments, 437–438 introduction to, 428–432
time-of-flight, 437 ion and their nuclear reactions, 432–433
PGI for proton therapy, 433–437 Monte Carlo implementation, 432–433
cesium iodide – CsI(Tl), 434 nuclear reactions to, 428, 428–432, 429
Compton camera (CC), 436 perspectives and future challenges, 438–439
historical development, 433–434 PGI
knife-edge slit camera, 434, 435 for ion therapy, 437–438
oxygen concentration, 436–437 for proton therapy, 433–437
prompt gamma-ray timing (PGT), 435–436 for proton range verification, 427–439
prompt gamma spectroscopy, 434–435, 436 and sources of background, 430–431
Phantoms for kilovoltage x-ray beam dosimetry, 361–362 Prompt Gamma Imaging (PGI), 430
Phantom scatter factor (Sp), 258, 258–259 for ion therapy, 437–438
defined, 258 for proton therapy, 433–437
Photodiodes, 128 Prompt gamma-ray timing (PGT), 435–436
Photoelectric effect absorption coefficient, 7–8 Prompt gamma spectroscopy, 434–435, 436
486 ◾ Index
PRonton IMAging (PRIMA) device, 462 imager, 112

Proportional counters, 193–194 IMRT, dosimeters and devices for, 173–182
Protoacoustic emissions, 445–449 monthly, 181–182
arrival time, 449 OSLD, 106
current state of, 450–453 patient-specific, 174–178
frequency, 449 periodic, 180–182
pulse width and rise time, 447–449 for proton therapy, 404–405, 406, 407
Protoacoustics, 444 TLD, 85
heterogeneity, 452–453 Quantum molecular dynamics (QMD), 432
image formation, 450, 451
range verification, time-of-flight of, 450–452, 452
R
Proton beam delivery techniques
scanned beams, 402–403, 403 RACT, See Radiation–Induced Acoustic Computed Tomography
scattered beams, 401–402, 402 Radiation detectors for ZAP-X surgical unit, 310
Proton integrating system, 460–461, 461 Radiation dosimetry, 10–16
Proton radiography and proton computed tomography cavity theory, 11–13
(pRG/pCT), 457–463 of TG-21, 13–15
current technologies and challenges, 460–462 of TG-51, 15–16
history of, 459, 459–460, 460 Radiation-Induced Acoustic Computed Tomography (RACT), 446
image reconstruction, 458–459 Radiation monitoring
introduction to, 457–458 active, 188
overview of, 458–459 area monitoring, 188–191
proton integrating system, 460–461, 461 continuous, pulsed and mixed radiation fields, 204–207, 205
protons interactions with matter, 458 neutron monitoring techniques, 198–204
proton tracking systems, 461–462, 462, 463 passive, 188
Proton Radiotherapy Verification and Dosimetry Applications photon monitoring techniques, 191–198
(PRaVDA) consortium, 462 protection and operational dose quantities, 186–188
Proton range verification, prompt gamma detection for, 427–439 radiation weighting factors, 187
Proton therapy dosimetry, 393–408 Radiochromic film, 64–67, 65, 66, 298, 399
detectors in, 396–401 calibration procedures, 69, 69–70
absolute, 396–398 factors affecting, 67, 67–69, 69
calorimetry, 397–398 batch, 68
ionization chambers, 396–397 beam quality and energy, 67
multidimensional, 400–401 dose and dose rate, 67
multielement, 400 external magnetic field, 68
OSLDs, 399 linear energy transfer, 67–68
radiochromic film, 399 orientation, 68
relative, 398–400 post irradiation time and temporal response, 68
scintillation, 400–401 storage, 68
solid state diode detectors, 399–400 temperature and humidity, 67
TLDs, 399 film handling, 69
fundamentals of, 394–396 response to radiation, 65, 65–67, 66
introduction to, 393–394 scanners, 69
measurement techniques, 403–404 uncertainty budget of, 70–71, 71
PGI for, 433–437 Radiographic film, 62–64, 63, 64
protocol, 404 Radiometric calorimeter, 32, 32
proton beam delivery techniques, 401–403 Radiophotoluminescent dosimeters (RPLDs), use in in vivo study,
QA for, 404–405, 406, 407 162–163
uncertainties, 405–406, 408 RADPOS system, 155
Proton tracking systems, 461–462, 462, 463 Recombination effects, 27–28
Recombination correction factor, 27, 270–271
Reference detectors, 293–294, 295
Q
Reference dose calibration for ZAP-X, 311
Quality assurance (QA) Reference dosimetry in FFF beams, 270–271
annual, 182, 349 Refractive index, 36, 128
daily, 181 Relative dosimetry, detectors used for, 287–289, 288
defined, 151 Relative output factors, 287–292
electron dosimetry, 387 from common treatment platforms, 289–292
helical tomotherapy, 338–339 cone-based, 290, 290
Index ◾ 487
detector Solid-core fibers, 124, 124–125

specific correction factors, 290–292, 291, 291, 292 Solid-state detectors, 194–195, 288–289
used for, 287–289, 288 proton therapy dosimetry, 399–400
electrometers, 289 scintillators, 194–195
Jaw and MLC, 289–290, 290 semiconductors, 195
Retired brachytherapy standards, 235 Solid-state nuclear track detectors, 203
Routine clinical measurements with FFF beams, 271–273 Source size occlusion, 282
dose rate and recombination, 271–272 Source super-sampling, 336
energy dependence, 272–273 Source vendor calibrations and measurements, 238
profile shape, 271 Spherical chambers, 23, 23
SRS MapCHECK, 298, 299
Stereotactic body radiation therapy (SBRT), 62, 155
S
Stereotactic cones, 282, 284
Safety of gamma knife dosimetry, 344–345 Stereotactic radiosurgery (SRS), 155
Scintillation detectors, 194–195 detectors in, 285–297
proton therapy dosimetry, 400–401 Stopping power (S), defined, 458
Scintillation fiber optic dosimetry, 123–133 Symmetry, defined, 116
characteristics of commercially available, 132–133
Cherenkov radiation in, 128–131, 129, 130
T
detectors, 127–128
photodiode, 128 TALYS, 435
photomultiplier tubes (PMTs), 127–128 Technical Reports Series (TRS), 21
introduction to, 123–124 TEPC, 217
ionization quenching, 131–132, 132 TG-21, 13–15
LET dependencies, 131–132, 132 TG-51, 15–16
optical fibers, 124–125 Thermoacoustic phenomena, 444–445
scintillators, 125–127, 127 Thermo Eberline ASP/2e NRD Neutron Survey Meter, 199, 200
use in in vivo study, 158–159 Thermoluminescence dosimetry, 75–93
Scintillation imaging dosimetry, 139–147 annealing and reuse, 85, 86
introduction to, 139–140, 140 applications in radiotherapy, 85–91, 86, 87
linac pulsing and time gated real-time imaging, 143–145, brachytherapy, 89–90
144, 145 dosimetric audits, 90–91, 91
scintillation light, 141–143 phantom measurements, 87–88, 88
1D–point measurement, 142, 143 radiation protection, 86–87
2D–surface and planar measurements, 142, 143 in vivo dosimetry, 88–89, 89, 90
3D–volumetric measurement, 142–143, 143 chips, 83
strength definition of, 75–76, 76
insensitivity to optical properties, 146–147 equipment required for, 82–85
mature technology of scintillators, 146 evaluation, quality assurance, and analysis, 85
stability of signal and linearity with dose, 146 fundamental principles of LiF:Mg,Ti
weakness basic theory, 77
complications of mixing Cherenkov and scintillation, 147 dose response, 79, 79–80
direct view limitations, 147 glow curves, 77–79, 78
Secondary Standards Dosimetry Laboratory (SSDL), 21, 233 other influences on dosimetry, 80
brachytherapy calibrations, 235–237, 237 handling, 83, 84
Semiconductor dosimeters, 39–40, 40 history of, 76
diamonds, 48–51 limitations of, 91–92, 92
diodes, 40, 40–44, 44 literature of, 77
MOSFET, 44–48, 45, 48 materials, 80, 80–82, 82, 83
Semiconductors, 40, 194, 195 reader, 83–85, 84
Sensitometric curve, 64, 64, 176, 177 scope of, 76
Signal-to-noise ratio, improving, 431, 431–432, 432 Thermoluminescent detectors (TLDs), 75–108, 399
Single program multiple data (SPMD) model, 213 proton therapy dosimetry, 399
Sinogram, 336, 337 use in in vivo study, 163–164
Small field dosimetry, 281–284 Thermoluminescent dosimeters, 75–108, 196
field size definition for Linac SRS/SBRT, 282–284 Thimble chambers, 21, 22–23
loss of LCPE, 281–282, 282 3D–volumetric measurements, 142–143, 143
source size occlusion, 282 Time-of-flight (TOF) spectra, 431–432, 432
Small field measurements with FFF beams, 273 Tissue-air ratio (TAR), 260–261
488 ◾ Index
Tissue maximum ratio (TMR), 261, 261 V

Tissue phantom ratio (TPR), 261, 261, 295–296
Volume averaging effects, 285–286, 286
conversion, 295–296
Volumetric modulated arc therapy (VMAT), 62, 109, 154, 174,
CyberKnife, 308, 309
284, 322
measurement, 295
TOPAS, 220, 221
Total body irradiations (TBI), 88, 153 W
Total skin electron therapy (TSET), 105, 142, 153
Traceability, 233 Water calorimeter, 34–36, 34–36
Transient charged particle equilibrium (TCPE), 260 Wedge, 116–117
Treatment planning system (TPS), 104, 116–118, 154, 175, 255, 293, Well chamber, 237
295, 312, 386, 405, 406, 470–471 Wheatstone bridge circuitry, 32
convolution/superposition dose calculation, 337–338 Wide Energy Neutron Detection Instrument Wendi-2, 200, 200
CyberKnife, 308 Winston–Lutz test, 114–115
Monaco, 297
Pinnacle, 297 X
Varian Eclipse, 297
ViewRay, 321 X-ray, see Kilovoltage x-ray
Triaxial cable, 23, 23–24
2D–surface and planar measurements, 142, 143
Z
ZAP-X, 310, 310–312
U
beam data acquisition, 310–311, 311
Ultrasonic calorimeter, 35 measuring beam data before importing into TPS, 312
Uncertainties OARs, 312, 312
ADCL calibration, 237 output factors, 311, 312
dosimetry, 365–366, 368 PDD, 311–312, 312
measurement, see Measurement uncertainties radiation detectors for, 310
proton therapy dosimetry, 405–406, 408 reference dose calibration, 311

Radiation Therapy Dosimetry - A Practical Handbook

Uploaded by

Copyright:

Available Formats

Radiation Therapy Dosimetry - A Practical Handbook

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Radiation Therapy Dosimetry - A Practical Handbook

Uploaded by

Copyright:

Available Formats

Library of Congress Cataloging‑in‑Publication Data

Names: Darafsheh, Arash, editor. Title: Radiation therapy dosimetry : a

ISBN: 9781138543973 (hbk)

Part I Radiation Dosimeters and Dosimetry Techniques

Chapter 1 ◾ Fundamentals of Radiation Physics and Dosimetry 3

Chapter 2 ◾ Ionization Chamber Instrumentation 19

Chapter 4 ◾ Semiconductor Dosimeters 39

Chapter 5 ◾ Film Dosimetry 61

Chapter 6 ◾ Thermoluminescence Dosimetry 75

Chapter 7 ◾ Optically Stimulated Luminescence Dosimeters in Clinical Practice 97

Chapter 8 ◾ EPID-Based Dosimetry 109

Chapter 9 ◾ Scintillation Fiber Optic Dosimetry 123

Chapter 10 ◾ Cherenkov and Scintillation Imaging Dosimetry 139

Chapter 11 ◾ Clinical Considerations and Dosimeters for In Vivo Dosimetry 151

Chapter 12 ◾ Dosimeters and Devices for IMRT QA 173

Chapter 13 ◾ Area and Individual Radiation Monitoring 185

Chapter 14 ◾ Monte Carlo Techniques in Medical Physics 211

Chapter 15 ◾ Brachytherapy Dosimetry 231

Part III External Beam Radiation Therapy

Chapter 16 ◾ Photon Beam Dosimetry of Conventional Medical Linear Accelerators 255

Chapter 17 ◾ Dosimetric Considerations with Flattening Filter-Free Beams 267

Chapter 18 ◾ Linac-Based SRS/SBRT Dosimetry 277

Chapter 19 ◾ CyberKnife and ZAP-X Dosimetry 305

Chapter 20 ◾ Dosimetry in the Presence of Magnetic Fields 315

Chapter 21 ◾ Helical Tomotherapy Treatment and Dosimetry 325

Chapter 22 ◾ Gamma Knife Dosimetry 343

Chapter 23 ◾ Kilovoltage X-Ray Beam Dosimetry 355

Chapter 24 ◾ Electron Dosimetry 375

Chapter 25 ◾ Proton Therapy Dosimetry 393

Chapter 27 ◾ Prompt Gamma Detection for Proton Range Verification 427

Chapter 28 ◾ Acoustic-Based Proton Range Verification 443

Chapter 29 ◾ Proton Radiography and Proton Computed Tomography 457

Part IV Imaging Modalities

Chapter 30 ◾ Dosimetry of Imaging Modalities in Radiotherapy 467

John A. Antolak Xinyuan Chen

Giordano Biasi Christopher L. Deufel

George X. Ding Nicholas Hardcastle

Tomas Kron Jessica Lye

Stephen F. Kry Paulo Magalhaes Martins

Francisco J. Reynoso Matthew T. Studenski

Mark J. Rivard Baozhou Sun

Anatoly B. Rosenfeld Reza Taleei

Joao Seco Irwin I. Tendler

It is also important to specify the medium that the 1  ∆E   ∆E 

understanding of the types of interaction that can occur,

 dE  Z z 2   Emax  2  and 15, respectively. Note that the above expression is

c   Restructuring Equation 1.9 to accommodate a

kinetic energy of the light charged particle. For an ini-

1.3 PHOTON INTERACTIONS dψ σ a ψN A ρ

The scattered electron is then emitted with an energy of where P ≈ 28

attenuation cross sections for the three types of photon

The current challenge at hand is to relate the measured  dE 

well satisfied. The work of Spencer and Attix [3, 12]

= = = (1.45) where d is the portion of electrons generated from the

the chamber. Using Burlin cavity theory, we can work

Ionization Chamber Instrumentation

2.1 INTRODUCTION TO there is variation among individuals, this was a subjec-

with a series of correction factors allows the measured

Part I Radiation Dosimeters and Dosimetry Techniques

Chapter 1 ◾ Fundamentals of Radiation Physics and Dosimetry 3

Chapter 2 ◾ Ionization Chamber Instrumentation 19

Chapter 4 ◾ Semiconductor Dosimeters 39

Chapter 5 ◾ Film Dosimetry 61

Chapter 6 ◾ Thermoluminescence Dosimetry 75

Chapter 7 ◾ Optically Stimulated Luminescence Dosimeters in Clinical Practice 97

Chapter 8 ◾ EPID-Based Dosimetry 109

Chapter 9 ◾ Scintillation Fiber Optic Dosimetry 123

Chapter 10 ◾ Cherenkov and Scintillation Imaging Dosimetry 139

Chapter 11 ◾ Clinical Considerations and Dosimeters for In Vivo Dosimetry 151

Chapter 12 ◾ Dosimeters and Devices for IMRT QA 173

Chapter 13 ◾ Area and Individual Radiation Monitoring 185

Chapter 14 ◾ Monte Carlo Techniques in Medical Physics 211

Chapter 15 ◾ Brachytherapy Dosimetry 231

Part III External Beam Radiation Therapy

Chapter 16 ◾ Photon Beam Dosimetry of Conventional Medical Linear Accelerators 255

Chapter 17 ◾ Dosimetric Considerations with Flattening Filter-Free Beams 267

Chapter 18 ◾ Linac-Based SRS/SBRT Dosimetry 277

Chapter 19 ◾ CyberKnife and ZAP-X Dosimetry 305

Chapter 20 ◾ Dosimetry in the Presence of Magnetic Fields 315

Chapter 21 ◾ Helical Tomotherapy Treatment and Dosimetry 325

Chapter 22 ◾ Gamma Knife Dosimetry 343

Chapter 23 ◾ Kilovoltage X-Ray Beam Dosimetry 355

Chapter 24 ◾ Electron Dosimetry 375

Chapter 25 ◾ Proton Therapy Dosimetry 393

Chapter 27 ◾ Prompt Gamma Detection for Proton Range Verification 427

Chapter 28 ◾ Acoustic-Based Proton Range Verification 443

Chapter 29 ◾ Proton Radiography and Proton Computed Tomography 457

Part IV Imaging Modalities

Chapter 30 ◾ Dosimetry of Imaging Modalities in Radiotherapy 467

2.1 INTRODUCTION TO there is variation among individuals, this was a subjec-