Journal of Advanced Biomedical & Pathobiology Research Vol.3 No.

1, May 2013, 8-18

Monte Carlo simulation of Electron Article Info

Beams; Influential Parameters and
Potential Sources of Simulation Faults Received: 12.01.2013
Accepted: 02.02.2013
Published online:01.03.2013
Hiba Baha Eldin Sayed Omer, Abdelmoneim Adam Sulieman
and Omaima Nasir, College of Medicine, University of
Dammam Kingdom of Saudi Arabia (KSA);
Ahfad University for Women, Omdurman Sudan, Omdurman
Sudan, College of Applied Medical Sciences, Salman Bin
AbdulAziz University, Al-kharj, KSA, Department of Medical
Laboratory, Faculty of Applied Medical Sciences Taif University
KSA

ISSN: 2231-9123

ABSTRACT
High-energy photon and electron beams were accurately simulated by Monte Carlo code. Electron
simulation is not easy due to their light weight and their scattering nature. Moreover, because of the
commercial value of the detailed specifications of the accelerator parts, manufacturers are usually
reluctant to provide the information with the necessary details for modelling.The aim of this report is
to study the effect of slight alterations in simulating the intrinsic beam parameters and Linac
configurations. The electron beams of the ELEKTA SL18 were simulated for different energies and
field sizes using BEAMnrc. The output phase space files were used as input sources on a water
phantom created using DOSXYZnrc. Different sources of systematic sources of errors in the
simulation were investigated by slight alterations in beam and phantom parameters including the
electron energy width of the beam, the source to surface distance and the voxel size, in addition to
alterations in some Linac components such as, the walls of the scattering foils, the configuration of
ionization chamber and the mirror, the jaws opening in addition to the materials of applicators. This
was performed for a 10x10 field for a low energy (4MeV), moderate electron energies (8 MeV and 15
MeV) and high electron energy (18 MeV) at different source to surface distances (SSD). This work
shows the necessity of the accuracy of simulating some of the beam parameters, the water phantom
and the different Linac configurations and explains systematic errors that arise from slight alterations
in these values. Percent depth dose curves are more influenced by accurate modelling of the different
parts of the linear accelerator, especially the scattering foils, the monitor ionization chambers and the
applicator materials. Beam parameters, source-to-surface distance, voxel sizes as well as some Linac
components such as scattering foil walls, jaws opening and material of the applicators affect the off-
axis dose distributions especially in the plateau and the shoulder regions.

Key words: Monte Carlo, electron beams, beam parameters, LINAC components.

1. Introduction

Electron beam therapy is an important radiation therapy modality for treatment of

superficial tumors. Treatment planning of electron beams is more complicated than photon
beams due to variations in beam productions, the scattering of low energy electrons and the
presence of contaminant photons [ICRU 2004, Gerbi 2006, Khan 1992 and Hogstrom 2006].
 Journal of Advanced Biomedical & Pathobiology Research Vol.3 No.1, March 2013, 8-18

Therefore accurate beam modeling is very crucial for accurate dose calculations within the
patients. Monte Carlo (MC) calculations are widely accepted as the most accurate means for
dose calculations and are expected to produce errors within 2%/2mm in dose as opposed to
analytical methods where errors of 5%/5mm or more were reported. [Rogers 2002, ICRU
1987, Anoltak 2002, Chetty 2007, Ding 2005, Jiang 2000, Ma 1999]
The electron beam energy spectrum emerging from the accelerator treatment head is
complex and may differ appreciably among accelerators from different manufacturers and
among models of accelerator from the same manufacturer depending on the design of the
treatment head. [Followill 2004] Because of the commercial value of the detailed
specifications of the accelerator parts, manufacturers are usually reluctant to provide the
information with the necessary details for modelling. Thus, approximation of the initial beam
is often used in MC simulations, although this can lead to discrepancies with measured data.
[Jiang 2000, Followill 2004]
The most basic information required for a MC simulation of a treatment head is the
specifications of the accelerator parts, such as their locations, dimensions and materials. The
major sources of error are statistical errors or uncertainties resulting from simulation of the
accelerator treatment head and those resulting from the fluctuations in phantom/patient dose
calculations. In this paper the effects of minor alterations in the simulation of beam and
phantom parameters as well as Linac configurations on both the percent depth dose and dose
profiles are investigated. This would help future researchers in the field to point out the
possible mistakes in their simulations).

2. Materials and Methods:

Linac Simulation: The electron beams from ELEKTA SL18 were simulated according to the
manufacturers’ data on the treatment head geometry using BEAMnrc/EGSnrc [Rogers 2004,
Kawrakow 2004, Kawrakow 2000]. The accelerator head was built from the following
component modules (CMs) CONESTACK, which consists of a stack of truncated cones,
surrounded by a cylindrical wall, was used to simulate the exit window and the primary
scattering foils. It was used for the secondary scattering foils for low and moderate-energy
beams. CONS3R was used for the primary collimator and the secondary scattering foils for
high-energy beams. The distance between the primary and secondary scattering foils varied
with the energy of the incident beam. The ionization chamber, jaws and applicators were
simulated using IONCHAMB, JAWS and APPLICAT component modules respectively

The output phase space file was used as an input source on a water phantom created
using DOSXYZnrc [Walters 2004] to produce a dose output file “.3ddose” which contains
the information about the simulation geometry and the calculation results in a format that can
be read by STATDOSE to generate xvgr/xmgr plots (percent depth dose (PDD), normalized
to Dmax, and the dose profiles at Dmax) for comparison purposes.

The choice of energy threshold below which the histories of electrons and photons (global
ECUT and PCUT) or their product knock-on electrons and Bremsstrahlung photons (AE and
AP) are terminated, was the same for both Linac and phantom simulations. ECUT and AE
were both set to 0.521 MeV. PCUT and AP were set to 0.01MeV. Boundary crossing and

9
 Journal of Advanced Biomedical & Pathobiology Research Vol.3 No.1, March 2013, 8-18

electron step-size algorithms were chosen to be PRESTA1 and PRESTA2 respectively. A

mono-energetic beam from (ISOURCE =19), which involves a circular beam with 2D
Gaussian distribution was chosen for our beam simulations. No photon or electron splitting
was performed. In order not to reject any low energy electrons the global ESAVE was set to
zero. A cluster of five computers with different processor speeds was used for parallel
processing. The number of particles used for simulation ranged from 200-500 million
particles depending on the energy and field size. This number decreased with increasing
beam energy and field size. In phantom simulations the histories were chosen so that the
particles from the phase space files were not recycled more than 30 times to avoid
unnecessary uncertainties. Fine-tuning of the beam was performed until a good match in
relative depth dose and dose profiles between measurements and calculation was reached.
This included alteration in the beam energy as well as energy distribution. It has been
reported that the full width at half maximum (FWHM) of the intrinsic energy spectrum is
about 5% and 10% of the most probable energy for accelerators of the travelling wave and
standing wave types respectively [Bjork 2002]. This value did not produce a satisfactory
match between the measured and the calculated percent depth dose. The value was increased
until a better match was produced. The result Linac and component modules are referred to as
verified Linac, or verified (named) component modules, respectively

Effects of Beam and Linac Configurations Alterations: The effect of small alterations
in initial beam parameters, and configuration of some accelerator components was
investigated for 10x10 fields using 4 MeV, 8 MeV, 15 MeV and 18 MeV electron beams as
follows:

2.1 Energy Distribution of the Initial Beam: The effect of decreasing and increasing the
FWHM beyond the verified values was studied on both PDD and dose profiles. This is done
because the value that produced a good match between measured and calculated central axis
PDD was more than the suggested 10% of the most probable energy [Bjork 2002].

2.2 Source-To-Surface Distance (SSD): from a practical point of view, a longer distance
between the collimator and the patient is often preferable; otherwise the collimator might
collide with the shoulder of the patient when tumours in the neck region are to be treated
[ICRU 2004]. The effect of SSD was investigated by locating the phantom immediately after
the last scraper of the applicator i.e. at a SSD of 95 cm and at 100 cm and 105 cm
respectively.

2.3 Voxel Size: Calculated dose is affected by the size of the scoring voxel. For MC
calculations typical values in the scoring dimensions are voxel sides of 2-5 mm for field sizes
greater than 3x3 cm2. Increasing the voxel size reduces the simulation time as well as the size
of the output 3ddose files but may affect the statistical uncertainty [Chetty 2007]. A water
phantom consisting of Cartesian voxels of water was simulated using DOSXYZnrc. Different
voxel side dimensions: 2.5 mm referred to as small voxels, 5 mm referred to medium voxel
size and 10 mm referred to as big voxels were used to study the effect of voxel size on both
PDDs and dose profiles.

10
 Journal of Advanced Biomedical & Pathobiology Research Vol.3 No.1, March 2013, 8-18

2.4 The Scattering Foils: SL18 has dual scattering foils that consist of a high-z (atomic
number) material to broaden the beam into a Gaussian profile and a lower-z Gaussian-shaped
foil that minimally scatters the electrons at the tail of the profile and maximally scatters
electrons near the centre to flatten the field in the area of interest with a suitably sharp fall-off
at the field edges. The effects of materials and dimensions of the scattering foils for mono-
energetic and poly-energetic beams have been extensively studied elsewhere. [Bieda 2001,
Schreiber 2005, Kainz 2005] No literature was found explaining the influence of walls of the
scattering foils. For low and moderate energies of SL18 (up-to 15 MeV), the shape of the
foils descends in an increasing diameter pattern, which allows simulation using the
CONESTAK component module. This allows simulation of the walls. The situation is
different for higher energies, 18 MeV and above, where it was necessary to use CONS3R that
does not allow simulation of the walls or FLATFILT option that allows simulation of
overlapping cones with varying diameters and simulation of the surrounding walls. In this
paper the height of the walls of the scattering foils was altered and the foils were simulated
without the walls to compare the PDDs and profiles generated with and without alterations.

2.5 The Ionization Chambers: Most Linacs have segmented ion chambers that allow monitor
both beam intensity and uniformity [Brown 1999]. The monitor ionization chamber of the
SL18 is made in form of thin layers of a light material separated by air gaps. This is
embedded in a metal disk. Accurate simulation of the ionization chamber is not necessary in a
simulation of photons were a good match between measured and calculated PDD and profiles
was possible with a 1 cm thick water slab simulated by SLAB CM, which is relatively easy to
simulate. This is not the case for electron beams, where all components on the beam direction
are expected to influence the dose. Simulation of the ionization chamber was performed using
IONCHAMB component module, SLAB of Mylar 0.072 cm thick, which represents the total
thicknesses of the Mylar films within the ionization chamber, and CONSTAK. SLAB option
was used to evaluate the possibility of simplifying the simulations and CONSTAK option
was used because IONCHAMB module does not allow accurately simulating the different
materials of the inner walls surrounding the different segments. Finally simulation was
performed without the ionization chamber and the results were compared.

2.6 The Mirror: The light localizing system, which consists of the light source and a thin
mirror, is used for field size definition within the patients. Despite the very small thickness of
the mirror, being on the path of the electron beams, it may have scattering effects. The
influence of the mirror was investigating by simulating the Linac with and without mirrors.

2.7 The Jaws: The jaws opening, which indicates the field size at a named SSD, usually 100
cm and the inclination of the jaws are automatically defined at the Linac monitor according to
the applicator used and the energy. Using these values for x and y openings from the monitor
of the Linac did not always produce a good match between measured and calculated profiles.
Neither did calculations of the openings according to the opening of the upper scraper of the
applicators or according to the required field size, 10x10 for example. The effect of altering
the jaws opening was studied in this work.

11
 Journal of Advanced Biomedical & Pathobiology Research Vol.3 No.1, March 2013, 8-18

2.8 The Applicators: In radiotherapy with high-energy electron beams, scattered radiation
from the applicator influences the dose distributions in patients. The amount of radiation is
dependent on the applicator design [van Battum 2003]. Electron applicators provided by
ELEKTA are in the form of sets of diaphragms or scrapers with a thick base that is attached
to the treatment head. This reduces the intensity outside the useful beam to less than 2%. In
this paper the effect of two different materials: lead and aluminum, which are the materials
for ELEKTA applicators, are investigated.

3. RESULTS

The effect of slight alterations on beam parameters and Linac configurations are summarized
below:

3.1 Energy Distribution of The Initial Beam: Increasing the beam width from 10% of the
most probable energy (referred to as small FWHM in the graphs) to the verified FWHM MeV
slightly decreased the dose gradient and slightly increased the Bremsstrahlung tail for the
different energies. More effect is noticed on the PDD of the high energy (18 MeV). A much
smaller alteration is noticed with further increase in the FWHM. The increase in FWHM
slightly affects the dose profiles where the flatness is reduced and the shoulder is depressed
with a further increase in beyond the verified value, referred to as big FWHM. This can be
seen more clearly for the small energy (4 MeV). The effect of energy width on PDD and dose
profiles is demonstrated in Figure 1 below for 4 MeV and 18 MeV electron beams.

Fig. 1: Effect of alterations of the energy width on pdd and profiles

3.2 Source-To-Surface Distance (SSD): From Figure 2 below it is evident that increasing
the SSD from 95 cm to 105 cm slightly reduced the surface dose with almost no effect in the
remaining parts of the PDD. This is basically because the difference in distance from the
source is not big enough to cause the effect of the inverse square law for distances. The big

12
 Journal of Advanced Biomedical & Pathobiology Research Vol.3 No.1, March 2013, 8-18

effect is witnessed on the dose profiles, where the penumbra increases drastically with
increase in SSD, as a result of the air gap.

Fig. 2: Effect of alterations of the source-to-surface distance on pdd and profiles

3.3 Voxel Size: Figure 3 below demonstrates the effect of voxel size on both PDD and dose
profiles. It was determined that using small voxel dimensions affect the surface dose. A
difference of more than 10% is realised between the small voxel size and the medium or big
voxel sizes. The effect on the profiles is more witnessed with big voxel sizes, where the
shoulders become narrower and the penumbra much wider. With small voxels the profile is
less flat and the percentage error of the highest doses is much larger than the other values.
The maximum error of the highest doses exceeds 1% only with small voxels.

Fig. 3: Effect of alterations of the size of voxels on pdd and profiles

3.4 The Scattering Foils: The presence of the walls of the secondary scattering foils affect
both PDDs and beam profiles. More effect is noticed on the PDD of the high energy (18
MeV). Dmax and the whole fall-off region are shifted towards the surface and the dose

13
 Journal of Advanced Biomedical & Pathobiology Research Vol.3 No.1, March 2013, 8-18

gradient is slightly increased when the walls were simulated using FLATFILT component
module. The walls improve the flatness of the profiles, especially at the shoulder region. No
effect is seen on the penumbra. This is illustrated in Figure 4 below

Fig. 4: Effect of the secondary scattering foil walls on pdd and profiles

3.5 The Ionization Chambers: Little effect is seen on the PDD when simulating the Linac
without the ionization chamber or when altering the thickness of its walls or when simulating
it using CONSTAK rather than IONCHAMB CM for all energies. A big different though is
noticed when simulating it using SLAB, especially for the low energy (4 MeV) beam. The
whole fall-off region is shifted towards the centre and the dose gradient is slightly decreased.
Regarding the dose profiles, a slightly higher dose is noticed around the central axis when
simulating the Linac without the ionization chamber. SLAB option also causes depression of
the profiles at the shoulder region. Less effect is noticed as the energy of the beam increases.
The effect of alteration of the ionization chamber is illustrated in Figure 5 below for 4 MeV
and 18 MeV.

Fig. 5: Effect of alterations of the ionization chamber design on pdd and profiles

3.6 The Mirror: The presence of the mirror does not affect high energy beams. It slightly
reduced the surface dose of the PDDs for low energy beams without affecting the dose
14
 Journal of Advanced Biomedical & Pathobiology Research Vol.3 No.1, March 2013, 8-18

gradient. A slight depression in the shoulder region of the profile of the low energy beams is
noticed when the Linac is simulated without the mirrors. No effect is realized on the
penumbra as the result of absence of the mirror. Figure 6 below illustrates the effects of the
mirror on the PDD and dose profiles for 4 MeV and 18 MeV.

Fig. 6: Effect of the mirror on pdd and profiles

3.7 The Jaws: The jaws opening obtained from the Linac monitor, referred to as monitor
jaws slightly underestimates the surface dose of the PDD for low energy beams. No effect is
noticed for higher energies, or in the fall-off region or Bremsstrahlung tail of any of the
energies. More effect is realised on the dose profiles where “horns” are produced with the
monitor jaws openings. The effects of the jaws opening is illustrated in Figure 7 below.

Fig. 7: Effect of alterations of the jaws opening on pdd and profiles

3.8 The Applicators: The effect of the applicator materials is very evident for high energy
electron beams as seen on Figure 8 below. Using Aluminum applicators completely deforms
the PDD by reducing the “plateau” of uniform dose and decreases the dose gradient. It also
affects the dose profiles by reducing the dose near the central axis and increases the dose at
the edges. Less effect is seen for the low (4 MeV) beam.

15
 Journal of Advanced Biomedical & Pathobiology Research Vol.3 No.1, March 2013, 8-18

Fig. 8: Effect of applicator materials on pdd and profiles

4. DISCUSSION:
Accurate simulation of the beam parameters and linear accelerator configurations is
very crucial for electron beams since slight alterations may affect the output dose distribution.
This paper studies some of the effects caused by slight alterations on both percent depth dose
and lateral dose profiles as summarized below.
4.1 Energy Distribution of The Initial Beam: The energy width of the electron beam that
produces a good match between measured and calculated PDDs and dose profiles ranges
from 20% to 30% of the most probable energy, depending on the energy. The percentage is
higher for lower electron beam energies. An increase in this value affects the flatness of the
beam profiles and a decrease may increase the dose gradient resulting in miss-match between
the measurements and calculations.
4.2 Source to Surface Distance (SSD): When electrons are simulated for clinical
applications the actual SSD should be accounted for because slight alterations were found to
influence the surface dose as well as the lateral dose distributions.
4.3 Voxel Size: Voxel sizes of 0.5x0.5x0.5 cm3 seem to give a better match between
measured and calculated data. Decreasing the voxel size increases the uncertainty because
fewer particles deposit dose in a small volume. Increasing the voxel size reduces the
uncertainty but may introduce errors due to reduced spatial resolution. This is seen clearly in
the results where the dose at the edges was not accurately measured.
4.4 The Secondary Scattering Foils: FLATFILT component module is recommended for
simulating the secondary scattering foils because it allows simulating the foils for different
energies and includes the walls. The walls of the secondary scattering foils were found to
influence the dose distributions especially for higher energies electron beams, were walls can
be simulated with FLATFILT.
4.5 The Ionization Chambers: Although IONCHAMB component module does not account
for the difference in the materials of the inner walls, it allows accurate simulation of the
ionization chambers as shown in the results. The segmentation of the ionization chamber
reduces scatter effects [21]. This is probably why simulating the ionisation chamber as a
SLAB of Mylar had more influence on the PDD than not simulating it at all. Absence of this
component module did not have a big influence on the PDD but reduced the flatness of the
beam profiles.

16
 Journal of Advanced Biomedical & Pathobiology Research Vol.3 No.1, March 2013, 8-18

4.6 The Mirrors: It is necessary to accurately simulate the mirrors for low energy electron
beams because despite its thin nature, it can affect the scattering of the low energy beams.
Mirrors do not affect the moderate or high energy beams.
4.7 The Jaws: The jaws opening values obtained from the Linac monitor, referred to as
monitor jaws are bigger than the calculated openings, especially for 4 MeV. This results in
more scattered radiation at the edges of the beam. It is thus necessary to calculate the jaws
openings according to the required field size.
4.8 The Applicators: The applicators play the most important role in beam shaping and have
a major role in dose distribution due to their scattering effects on the electron beams at the
water phantom. The amount of scattered radiation from applicators is dependent on the
applicator design and materials. Applicators made from high Z materials such as lead are the
ones of choice especially when dealing with high electron energies. Low Z materials such as
aluminum can be used only for low energy electron beams.

5. CONCLUSIONS:

Accurate simulation of the linear accelerators is crucial for proper Monte Carlo calculations
of beam distributions in patients. This work shows the necessity of the accuracy of simulating
some of the beam parameters, the water phantom and the different Linac configurations and
explains systematic errors that arise from slight alterations in these values. Percent depth dose
curves are more influenced by accurate modelling of the different parts of the linear
accelerator, especially the scattering foils, the monitor ionization chambers and the applicator
materials. Beam parameters, source-to-surface distance, voxel sizes as well as some Linac
components such as scattering foil walls, jaws opening and material of the applicators affect
the off-axis dose distributions especially in the plateau and the shoulder regions.

ACKNOWLEDGEMENT:

This work was sponsored by: the State Scholarships Foundation (IKY), in Greece.

References
Anoltak J. A.; Beida M.; and Hogstrom K. R. (2002) Using Monte Carlo Methods to
Commission Electron Beams: A Feasibility Study Med. Phys.: 29 (5) ; 771-789
Bieda M. R.; Anoltak J. and Hogstrom K. B (2001). Effect of Scattering Foil Parameters on
Electron-Beam Monte Carlo Calculations Med. Phys: 28(12); 2527-2534.
Bjork P.; Knoos T. and Nilsson P. Influence of Initial Electron Beam Characteristics on
Monte Carlo Calculated Absorbed Dose Distribution for Linear Accelerator Electron
Beams Phys. Med Biol. 2002: 47; 4019-404.
Brown B. H.; Smallwood R. H.; Barber D. C.; Lawford P. V. and Hose D. R. (1999) Medical
Physics and Biomedical Engineering, Bristol and Philadelphia, Institute of Physics
Publishing.
Chetty I. J.; Curran B.; Cygler J. E.; DeMarco J. J.; Ezzell G.; Faddegon B. A.; Kawrakow I.;
Keall P. J.; Liu H.; Ma C. M; Rogers D. W. O.; Seuntjens J.; Sheikh-Bagheri D.; Siebers
J. V. (2007) Report of the AAPM Task Group No. 105: Issues associated with clinical

17
 Journal of Advanced Biomedical & Pathobiology Research Vol.3 No.1, March 2013, 8-18

implementation of Monte Carlo-based photon and electron external beam treatment

planning Med. Phys.: 34 (12); 4818-4853.
Ding G. X.; Cylger J. E.; Yu C. W.; Kalach N. I.; and Daskalov G. A (2005) Comparison of
Electron Beam Dose Calculation Accuracy Between Treatment Planning Systems Using
Either A Pencil Beam Or Monte Carlo Algorithm Int. J. Radiation Oncology Biol. Phys.:
63 (2); 622-633.
Followill D. S.; Davis D. S.; and Ibbott G. S.( 2004) Comparison of Electron Beam
Characteristics from Multiple Accelerators Int. J. Radiation Oncology Biol. Phys.: 59 (3);
905-910.
Gerbi B. J. (2006) Clinical Applications of High-Energy Electrons: Technical Basis of
Radiation Therapy 4th Edition Springer, Berlin, Heidelberg.
Hogstrom K. K.; and Almond P. (2006) Review of Electron Beam Therapy Physics Phys
Med. Biol: 51; R455- R498.
International Commission on Radiation Units and Measurements (2004). Physical
Characteristics of Electron Beams Journal of the ICRU: 4 (1)
International Commission on Radiation Units and Measurements. (1987) ICRU Report 42:
Use of Computers in External Beam Radiotherapy Procedures with High Energy Photon
and Electrons.
Jiang S. B.; Kapur A. and Ma C-M (2000) Electron Beam Modelling and Commissioning for
Monte Carlo Treatment Planning Med. Phys: 27(1); 180-191.
Kainz K. K.; Antolak J. A.; Almond P. R.; Bloch C. D.; and Hogstrom K. R. (2005) Dual
Scattering Foil Design for Poly-energetic Electron Beams Phys. Med. Biol.: 50; 755-767
Kawrakow I. (2000) Accurate Condensed History Monte Carlo Simulation of Electron
Transport. I. EGSnrc, The New EGS4 Version Med. Phys.: 27(3); 485-498.
Kawrakow I. and Rogers D. W. O. (2004) The EGSnrc Code System: Monte Carlo
simulation of electron and photon transport, NRC Report PIRS–701
Khan F. M. (1992) The Physics of Radiation Therapy 2d ed. Baltimore Maryland Williams &
Wilkins.
Rogers D. W. O. (2002) Monte Carlo Techniques in Radiotherapy Physics in Canada: 58;2
63-70. 2002
Ma C-M. and Jiang S. B. (1999) Topical Review: Monte Carlo Modelling of Electron Beams
From Medical Accelerators Phys. Med. Bol. R157-R189.
Rogers D. W. O.; Walters B.; and Kawrakow I. (2004) BEAMnrc Users Manual NRC Report
PIRS 509(a) rev I,
Schreiber E. C. and Faddegon B. A.( 2005) Sensitivity of Large-Field Electron Beams to
Variations in a Monte Carlo Accelerator Model Phys. Med. Biol.: 50(5); 769-78
van Battum L. J.; van der Zee W. and Huizenga H. (2003) Scattered Radiation from
Applicators in Clinical Electron Beams” Phys. Med. Biol.: 48; 2493-2507
Walters B. R. B.; and Rogers. D. W. O (2004). DOSXYZnrc Users Manual. NRC Report
PIRS 794 (rev B)

18