Paper

PROJECTION X-RAY IMAGING: RADIOGRAPHY,

MAMMOGRAPHY, FLUOROSCOPY

James Anthony Seibert1

INTRODUCTION
Abstract—Recent advances in digital detector technology for med-
ical radiography and fluoroscopy have resulted in improved PROJECTION X-RAY imaging represents 60–70% of medical
workflow efficiency, operational flexibility, image quality, and di-
Downloaded from http://journals.lww.com/health-physics by BhDMf5ePHKbH4TTImqenVGrjdWQL4F/wrPiNC82OyOACdy4vzVCsB8au3iIX9FIo on 01/08/2019

imaging examinations performed at typical community

agnostic accuracy. This is attributed to the implementation of por-
table flat-panel x-ray detector devices able to provide real-time and academic-based hospitals, as shown in Fig. 1. Although
readout, processing, and display of medical images. As a result, radiation doses associated with these exams are on the low
digital radiography flat-panel detectors are rapidly replacing com- end when compared to computed tomography and radio-
puted radiography passive detectors for projection imaging exams pharmaceutical imaging on a per procedure basis, there is
and image intensifier detectors for dynamic fluoroscopy exams. Ad-
vanced exam capabilities now include digital tomosynthesis and a substantial impact on the collective population dose, esti-
dual-energy radiography; the former is based on rapid acquisition mated in 2006 to be 0.3 mSv for radiography and 0.4 mSv
of multiple angle-dependent image projections to synthesize tomo- for interventional procedures (NCRP 2009). The proper, ef-
graphic slabs at selectable depths within the patient, and the latter fective, and safe use of digital radiography and fluoroscopy
is based on rapid back-to-back acquisition of the same anatomy at
low and high x-ray energies to generate separate soft tissue and technology for patients is therefore an important responsi-
bone images. In both situations, superimposition of anatomy is re- bility of medical imaging users. Given the wide exposure
duced or eliminated, with the possibility of enhanced diagnostic latitude of digital detectors, undetected overexposures and
confidence. Increased x-ray absorption efficiency and lower elec- excessive radiation dose to the patient can occur, often with-
tronic noise of digital radiography detectors compared to computed
radiography detectors enable equal image quality at lower patient out knowledge of the technologist or the radiologist (Seibert
dose; however, because of a disconnect between image appearance 2008). Ensuring the appropriate image quality at the lowest
and radiation dose, lower patient dose is not always achieved. Ed- possible radiation dose to the patient requires understanding
ucation, training, and implementation of standards such as the detector technology improvements and trends, implementing
International Electrotechnical Commission 62494-1 Digital Radi-
ography Exposure Index are needed to ensure image quality at technical advances for image acquisition and processing, op-
the lowest appropriate radiation dose. The National Council on timizing protocols and technique factors, monitoring and
Radiation Protection and Measurements can contribute to radia- tracking patient radiation dose metrics, and developing
tion responsibility in medical imaging by providing guidance on examination-specific diagnostic reference level (DRL) indi-
use of digital radiography, including recommendations for acqui-
sition protocols and exposure index standards, for development of cations to identify outlier procedures. Some of these aspects
radiographic exam diagnostic reference levels, and for oversight are discussed in the following sections of this article.
of retake and reject analysis.
Health Phys. 116(2):148–156; 2019
Key words: National Council on Radiation Protection and Mea- X-RAY DETECTOR TECHNOLOGY TRENDS
surements; fluoroscopy; radiation dose; x rays
Computed radiography
The process of two-dimensional (2D) projection x-ray
imaging has been unchanged since 1895, with the first re-
ported radiograph of the hand of Professor Roentgen’s wife.
Over its first century of use, refinements of analog film,
1 screen-film detectors, and chemical processing technology
Department of Radiology, University of California Davis Health,
4860 Y Street, Suite 3100, Sacramento, CA 95817. resulted in a reproducible film image at an appropriate opti-
For correspondence contact the author at the above address, or email cal density to ensure proper diagnostic content to make a
at [email protected].
(Manuscript accepted 11 October 2018)
confident diagnosis. Of course, this only occurred when the
0017-9078/19/0 acquisition technique factors and film processing attributes
Copyright © 2019 Health Physics Society were correct to deliver the appropriate dose to the detector.
DOI: 10.1097/HP.0000000000001028 In the early 1990s, the introduction of photostimulable
148 www.health-physics.com

Copyright © 2018 Health Physics Society. Unauthorized reproduction of this article is prohibited.
 Projection x-ray imaging c J.A. SEIBERT 149

imaging plates are exposed to x rays, imprinting an electronic

latent image on the storage phosphor. Subsequently the plates
are processed using a CR reader in which a very fine,
small-spot laser beam scans the exposed plate and
generates a photostimulated luminescence signal with an
intensity proportional to the number of local incident x rays.
A light guide directs the stimulated light to a photomultiplier
for amplification and then to an analog-to-digital converter
and conversion to a digital number stored in the image
matrix (Bushberg et al. 2012). By scanning in a raster
fashion, an array of 2,000
2,000 measurements creates
the digital x-ray image. The downsides include labor-
intensive handling of the cassette and about a minute of
processing time before image viewing is possible.

Digital radiography
In the early 2000s, an x-ray detector technology based
on thin-film-transistor (TFT) amorphous silicon arrays, ini-
Fig. 1. Relative percentage of exam volumes for imaging modalities tially designed to display images on laptop computers and
at the University of California Davis Medical Center over several years. flat-panel monitors, was introduced for detection, digital ac-
MRI: magnetic resonance imaging; CT: computed tomography.
quisition, and storage of the radiographic image. With this
technology, the flat-panel detector is constructed with an
storage-phosphor imaging plates, cassettes, and the com- x-ray absorbing material that converts local incident x-ray
puted radiography (CR) reader (Fig. 2a) provided the first intensity into a proportional charge layered on a TFT array
digital image acquisition capability for radiography, leading of approximately 2,000
2,000 elements. Conversion of
to the widespread implementation of picture archiving the incident x-ray fluence to a digital signal is performed ei-
and communications systems for display, diagnosis, and ther by a direct method using a semiconductor material such
storage of digital images. With CR technology, a direct as amorphous selenium to directly convert the absorbed en-
replacement of screen film, individual cassettes containing ergy into a corresponding charge or by an indirect method

Fig. 2. (a) CR imaging plate and cassette holder, top left. CR reader and processing station with multiple input slots for queueing exposed CR cas-
settes. Image extraction and processing time is ~1 min. (b) Integrated flat-panel DR detector cassette and battery charging station, top right. Com-
puter processing station and wireless connectivity to DR cassette electronics, bottom right. Image readout and processing for display is available
within 10 s.
www.health-physics.com

Copyright © 2018 Health Physics Society. Unauthorized reproduction of this article is prohibited.
150 Health Physics February 2019, Volume 116, Number 2

using a scintillator material such as cesium iodide to pro- by flat-panel TFT arrays, and more recently by CMOS
duce light and subsequently generate a proportional charge arrays. Initial flat-panel fluoroscopy systems were installed
by a photoconductor material (Bushberg et al. 2012). The in interventional radiology and cardiology laboratories only,
digital image is produced with no need for user handling chiefly due to high cost and secondarily due to limited
or intervention. In some detectors, real-time rates can be signal gain provided by the early flat panels that suffered
achieved (30 images s−1) for fluoroscopy applications. Al- from high electronic noise and poor image quality at very
though there are differences between detector conversion, low dose rates. With low electronic noise properties and
these devices are collectively known as digital radiography recent manufacturing capability of larger area CMOS
or direct radiography (DR) detectors. Early DR technology detector arrays, many manufacturers are now building
was bulky, required direct connections to x-ray generators, portable C-arm fluoroscopy systems (Fig. 3b) and small
and was typically built into tables and wall-stand devices portable minifluoroscopy units with these detectors for
without any capability for flexible positioning. However, low dose-rate fluoroscopic imaging. An increase in fluoro
the capability of displaying a processed image in 10 s or less and radiography systems with CMOS detectors is a likely
generated a tremendous advantage in workflow and through- expanding trend.
put compared to CR detectors. What about the future? Photon-counting, energy-
In the 2005–2010 timeframe, the portable TFT flat- sensitive detectors have been introduced for mammography
panel cassette made its debut, with a tethered communication imaging using silicon-strip detectors in combination with
cable connected to the x-ray generator and image processing photon-counting electronics in a slot-scan geometry (Weigel
computer. This allowed for more flexible positioning of the et al. 2014). These detectors discretely count individual x-ray
free cassette but with cable limitations. Subsequent im- photon events and assess absorbed energy over two or more
provements in lithium battery technology coupled with dig- threshold levels over the x-ray energy range used in mam-
ital Wi-Fi communication links has created a TFT flat-panel mography (20–35 keV). Spectral differences of the transmitted
cassette that can provide positioning functionality of CR, x rays through the anatomy are preserved, and energy-
with the rapid readout and display of DR. This innovation weighted tissue decomposition processing can be achieved
is now the state-of-the-art DR flat-panel detector design (see next section). In addition, lower dose is also achievable
available today; an example is shown in Fig. 2b. The disad- through the energy thresholding process. Future energy-
vantages include high cost and less resilience to damage sensitive area detector modules made of cadmium zinc tel-
when mishandled, as compared to CR. luride (CdZnTe) or cadmium telluride (CdTe) materials
coupled to digital signal processing integrated circuits are
Emerging detector technologies under investigation for radiography and computed tomogra-
Emerging x-ray detector technologies for future radiog- phy (Taguchi and Iwanczyk 2013). These detectors promise
raphy detectors to be considered include complementary significant advances in photon counting and energy binning
metal-oxide semiconductor (CMOS) and energy-sensitive of incident x-ray photons, with the advantage of rendering
photon-counting detectors. CMOS detectors are made of
crystalline silicon, using a production process like that of
random-access memory chips for computers. CMOS de-
vices in very small sizes are now ubiquitous in photo cam-
eras, cell phones, and other light-sensitive sensors. Until
recently, the ability to scale the size to larger area was not
possible. Now available is very large area CMOS wafer pro-
duction for digital radiography (Farrier et al. 2009). X-ray
CMOS array detectors are manufactured with x-ray con-
verters (semiconductors and scintillators) like that described
for TFT flat panels. Advantages over amorphous silicon
TFT detectors include flexible addressing of the individual
CMOS detector elements (with nondestructive reading of
the detector charge) and lower electronic noise, potentially
leading to lower radiation dose to the patient for a given
exam and for use in fluoroscopic applications.
The image intensifier, in use for over 60 y in fluoros-
Fig. 3. (a) Image intensifier detector, top left. C-arm fluoroscopy sys-
copy systems (Fig. 3a), is a high-gain detector but is very tem with image intensifier detector, bottom left. (b) CMOS solid-state
bulky and prone to geometric distortion. This real-time detector, top right. C-arm fluoroscopy system with CMOS detector,
acquisition detector technology is rapidly being displaced bottom right.
www.health-physics.com

Copyright © 2018 Health Physics Society. Unauthorized reproduction of this article is prohibited.
 Projection x-ray imaging c J.A. SEIBERT 151

material-specific analyses of a patient’s anatomy, among such gridless radiography options on their radiography sys-
many other benefits. tems. Functionally, a scatter kernel database and convolu-
tion with the unprocessed image is used to estimate the
TECHNICAL ADVANCES IN RADIOGRAPHY scatter distribution image, which is subtracted from the ac-
AND MAMMOGRAPHY quired primary-plus-scatter image. The difference image,
now mainly comprised of primary radiation, is scaled and
Hand in hand with digital detector implementation are adjusted for optimal grayscale rendition. In practice, a sig-
technical advances made possible by the wide dynamic- nificant dose reduction is not achieved, but the technologist
range response of the digital detector and subsequent image does benefit from not having to position a heavy grid device
manipulation/processing capabilities afforded by access to combined with the detector. Also, a common reason for
the digital data. Examples described in this section include poor image quality due to grid cutoff nonuniformities is
x-ray scatter compensation, energy subtraction radiography, eliminated, and retakes are reduced. The images shown in
and digital tomosynthesis. These capabilities can result in Fig. 4 illustrate the similarity of bedside radiography chest
lower dose, similar dose, or higher radiation dose, with the images acquired with and without a grid, the latter with scat-
potential improvement in productivity, improved image ter processing implemented.
quality through reduced superimposition of the projection
radiograph, and/or improved detectability. Radiation re- Dual-energy radiography
sponsibility in medicine dictates that there must be adequate Dual-energy radiography is an imaging method that
justification and resultant exam-specific optimal image typically acquires two distinct measurements of the anat-
quality attained when these advances are implemented. omy at high and low effective x-ray energies (Bushberg et al.
Gridless radiography 2012). This can be accomplished with the use of rapidly
During the acquisition of a projection image, the fate of switched x-ray tube voltages and exposures synchronized
the incident x rays on the patient are no interaction (a pri- to the recording and readout of a real-time TFT flat-panel
mary x ray), total absorption (a photoelectric interaction), detector. Scaling and subtracting the images allows for em-
and scatter (most often a Compton interaction). Ideally, only phasis of one type of anatomy (e.g., soft tissue) or another
the first two events occur, and transmitted primary radiation (e.g., bone) based upon the energy dependence of the
is totally absorbed by the detector and generates signal in x-ray attenuation coefficients. The images shown in Fig. 5
the detector to produce the image. Unfortunately, x rays illustrate the power of eliminating bone in the soft tissue im-
are also scattered in isotropic directions through collisions age and likewise eliminating the soft tissue in the bone im-
with electrons and redirection of the scattered photons, age of a chest x ray. In addition, the ability to discern the
many of which interact with the detector and cause a loss composition of a pulmonary nodule as with or without cal-
of subject contrast created by the differential attenuation cification has implications on the differential diagnosis and
of the anatomy. Antiscatter grid devices made of thin lead ability to predict benignancy (with calcifications) or
strips separated by radiolucent interspaces are typically used
to preserve the subject contrast by preferentially allowing
primary photons to pass and by attenuating scattered pho-
tons (Bushberg et al. 2012). Unfortunately, there is a need
to increase patient dose by a factor that compensates for
the removal of primary radiation by lead strips to achieve
an adequate number of x rays to the detector. In addition, be-
cause the grid is focused and must be aligned to the focal
spot, misplacement can result in grid artifacts and nonuni-
form transmission of x rays, resulting in suboptimal or un- Fig. 4. Comparison of grid and no-grid with scatter processing im-
usable images, ultimately requiring a retake image. This is ages. (Left) Patient image with an antiscatter grid and grid technique
a major problem with bedside radiography and the difficul- (105 kV, 1.2 mA s) for one manufacturer’s portable x-ray system.
ties encountered by the technologist in aligning the grid de- (Right) Subsequent image later in day of the same patient acquired
on a different manufacturer’s portable x-ray system without a grid
vice with the focal spot and diverging x-ray beam. One (96 kV, 1.4 mA s) and with scatter compensation processing. For the
solution is to not use a grid allowing the scatter to be de- image on the right, low-frequency estimates of the scatter contribu-
tected and devising a postprocessing algorithm that takes tion, image manipulation using convolution processing, and subtrac-
advantage of the wide detector dynamic range of the DR de- tion selectively removes the scatter component. While the images
are shown to be similar in image quality, systems with scatter process-
tector (Mentrup et al. 2016). Many manufacturers, includ- ing benefit from a slightly lower dose to the patient, elimination of
ing Philips Healthcare (Best, Netherlands) and Fujifilm possible grid artifacts, and no added weight to the portable detector—
Medical Systems (Stamford, Connecticut, US) now have which is helpful in bedside radiography examinations.
www.health-physics.com

Copyright © 2018 Health Physics Society. Unauthorized reproduction of this article is prohibited.
152 Health Physics February 2019, Volume 116, Number 2

Fig. 5. (Left) Composite chest x ray acquired at 120 kV (not shown, image acquired at 80 kV). Note suspicious lesion indicated by white circle.
(Center) Soft tissue image, created by weighting the low- and high-energy images to eliminate bone. The suspicious lesion in composite image is
not present. (Right) Bone image with weighting of dual-energy image pair to eliminate soft tissue. The lesion is present in this image, indicating a
benign calcification.

malignancy (soft tissue only). The justification of added ra- and 50% adipose tissue, each manufacturer designs their
diation dose and risk is based on the premise of improved system to deliver a fraction of the accumulated dose for each
diagnostic information and potential benefits. acquired projection equal to approximately 1/X for X acqui-
sitions. One manufacturer uses 15 individual image acquisi-
Digital tomosynthesis tions over an angular range of 15° and within the same
A major obstacle to projection radiography occurs dur- acquisition, also acquires a standard, conventional 2D mam-
ing the acquisition, namely the integration of x-ray attenua- mogram in a combination mode under 3 mGy. While this
tion information of the patient volume into a 2D array of represents an increase in radiation dose to the patient by a
corresponding digital values. This results in the superimpo- factor of ~2, the increased enhancement of lesions or reduc-
sition of anatomy and the loss of depth cues in the image. tion in overlapping tissues masquerading as lesions assists
Overlying anatomy can hide pertinent information such as in the sensitivity and specificity of the examination. An
lesions or other pertinent anatomical findings, and serendip- ability to create a synthesized 2D projection image from
itous arrangement can similarly mimic findings that are the tomosynthesis data set has enticed many practices to
false. Geometric tomography in the predigital age was a eliminate the conventional 2D mammogram altogether,
method widely used to reduce superimposition in projection even though there are some differences in the presentation
radiography by moving the source and detector in opposite of the anatomy. Thus, the radiation dose of a screening exam
directions about a pivot during a continuous acquisition to using tomosynthesis and 2D composite image processing is
render an in-focus slab of the patient’s anatomy at a specific like a typical conventional screening exam with all the ad-
depth and to smear out the underlying and overlying anat- vantages of having improved depth information. Digital
omy on the screen-film detector. With fast digital x-ray tomosynthesis has also been used in other areas of the body
acquisition capabilities with DR detectors, a similar ac- such as the chest and abdomen but with less success, due to
quisition process over a range of projection angles pro- competition with computed tomography and its high acqui-
vides multiple discrete images. The image data set, once sition speed, elimination of anatomical superimposition,
acquired, is manipulated by computer reconstruction provision of image reformatting capabilities (axial, sagittal,
methods by a shift-and-add method to render in-focus coronal, oblique reconstructions), and rendition of various
slabs of the anatomy over the full range of thickness slice thicknesses (McCollough 2019).
(Bushberg et al. 2012). This capability, known as digital
tomosynthesis, has been successfully applied to breast
RADIATION DOSE AND
imaging as an adjunct to and possible replacement for PROTOCOL MANAGEMENT
digital mammography. A simple schematic of the acquisi-
tion, reconstruction, and image rendering process is illus- There are many current deficiencies with the way dig-
trated in Fig. 6. ital radiography systems are designed, implemented, clini-
While there are many technical issues in the implemen- cally used, and evaluated for proper and optimal use. The
tation, the aspects of radiation dose are particularly impor- chief issue is that of the appearance of the radiographic im-
tant in the context of this article. As the Mammography age and the disconnect with radiation dose, which is often
Quality Standards Act requires the estimated average glan- misunderstood, even by experienced technologists. This is
dular dose to not exceed 3 mGy for a compressed breast partly due to a lack of training on how a digital detector re-
thickness of 4.2 cm with a composition of 50% glandular ports appropriate feedback with respect to proper acquisition
www.health-physics.com

Copyright © 2018 Health Physics Society. Unauthorized reproduction of this article is prohibited.
 Projection x-ray imaging c J.A. SEIBERT 153

Fig. 6. (a) Acquisition sequence for digital breast tomosynthesis—projection images acquired over ~5 s at multiple angles are used for reconstruc-
tion and synthesis of tomograms throughout the breast. Shown is a simple shift-and-add reconstruction that demonstrates separate focal planes in
the breast near the detector (black sphere) and near the top surface (white square). (b) A sequence of images illustrates a conventional 2D breast
mammogram, four individual tomograms (out of about 50 images) at various in-focal planes in the breast, and a synthesized composite recon-
structed 2D mammogram from the acquired image data set. In tomosynthesis image 3 above, a distinct speculated pattern indicative of a malignant
tumor is detected, not readily observable in the conventional mammogram (Fig. 6a recreated from Bushberg et al. 2012 with permission.)

techniques and to the myriad number of proprietary expo- the use of digital radiography because of its wide detector
sure indices that have little similarity with each other. Ac- dynamic range and ability to rescale the digital information
quisition protocols are often not evaluated nor tested for received prior to creating the final output image. This man-
proper setup. Retake rates with digital radiography are ifests as a disconnect between image appearance and the ra-
underreported because of the ease on some systems of sim- diation dose to the detector (and thus to the patient); in other
ply deleting poor quality images, coupled with the lack of words, it is difficult to discern when too much radiation is
defined methods for reporting rejected images. There is used during the acquisition of an image, as shown in Fig. 7.
very little guidance in terms of DRLs for various radio- When too little radiation is used, the underexposed image
graphic exams. Initial education and on-going training is appears noisy and will likely be rejected by a radiologist
certainly a key aspect of overcoming these deficiencies, as who depends on a certain contrast-to-noise ratio to make a
explained in this section. differential diagnosis. Thus, to avoid repercussions, a sim-
ple fix is for the technologist to increase the technique fac-
Disconnect between digital image appearance and tors, which can lead to dose creep (Seibert et al. 1996) and
radiation dose result in unnecessary and potentially undetected overexpo-
Technical advances in digital projection imaging can sures to the patient. Most digital radiography vendors
improve patient safety, provide enhanced image quality, and provide proprietary exposure index values generated by
aid the medical decision-making process. However, there is analyzing the digital detector response for specific exams
also the potential for unnoticed radiation overexposure in to identify a proper exposure range and outliers. This is a
www.health-physics.com

Copyright © 2018 Health Physics Society. Unauthorized reproduction of this article is prohibited.
154 Health Physics February 2019, Volume 116, Number 2

Fig. 7. Characteristics of the response to various levels of radiation (0.5–5 times the proper exposure) for a fixed-speed (film) detector and a var-
iable-speed (digital) detector of an anthropomorphic chest phantom. While the images appear to be of similar quality for the digital detector, a zoom
of a contrast-detail phantom (lower row) illustrates the excessive noise at low dose that causes detection loss for small objects. At higher doses,
diminishing return of contrast sensitivity occurs when comparing an image captured at 2.5 times vs. 5 times the appropriate dose (adapted from
Seibert 2008 with permission).

good practice for users to implement, understand, and use. the deviation index DI provides feedback in terms of achiev-
However, with inadequate training, and particularly in a ing the desired exposure to the detector:
multivendor situation, the lack of universal reporting stan-  
dards makes it difficult for the technologist and radiologist EI
DI ¼ 10 log : ð1Þ
to understand what the feedback means. The numbers are EI T
widely different: some exposure indices increase with in-
creased dose to the detector for one manufacturer’s algo- The logarithm (base 10) of the ratio of EI to EIT provides a
rithm, while other manufacturer’s exposure indices increase mechanism to directly identify underexposures as negative
with decreased dose to the detector. values, overexposures as positive values, and DI = 0 as the
ideal dose to the detector. A DI value scale and relationship
to under- or overexposure is shown in Fig. 8 as implemented
International exposure index standard at University of California Davis Medical Center. Initial im-
In 2008, an effort to implement a universal exposure in- plementation of the standard requires substantial work to
dex standard was undertaken by the American Association of identify the EIT values, which are dependent on the x-ray
Physicists in Medicine and the International Electrotechnical detector type used, the body part examined, and the image
Commission (IEC) in conjunction with the digital radiogra- quality for the examination deemed necessary by radiologist
phy manufacturers, resulting in the publication of the Expo- consensus. Our initial experience with implementation of
sure Index of Digital X-Ray Imaging Systems (IEC 62494-1 the standard demonstrates a wide variance of the reported
[IEC 2008; Shepard et al. 2009]). This standard describes a DI values. With further training, experience, and oversight
logical, exam-based method for inputting a desired target ex- of radiologic technologists, the expectation is a reduction
posure index EIT into the acquisition protocol. The EIT may in the spread of DI values and more consistent image quality
depend on the type of detector, on the type of examination, at optimal detector dose.
on the diagnostic question, and on other parameters deter- Acquisition protocol and image retake evaluation
mined by an individual facility’s criterion for image quality. Most fixed and portable digital radiography units have
The exposure index EI in the acquired image is based on many built-in acquisition protocols, often exceeding 100 se-
exam-specific histogram evaluation of relevant regions, and lections. Within these protocols are details related to patient
www.health-physics.com

Copyright © 2018 Health Physics Society. Unauthorized reproduction of this article is prohibited.
 Projection x-ray imaging c J.A. SEIBERT 155

Communications in Medicine (DICOM) Radiation Dose

Structured Report (DICOM 2018). Unfortunately, some of
the add-on DR detectors used in existing rooms with a pas-
sive trigger to initiate image capture do not have access to
the x-ray generator and x-ray tube parameters or technique
settings like most CR detector systems. The lack of data in-
terfaces and information regarding the x-ray tube and geom-
etry settings can preclude accurate estimates of patient dose,
even if the detector dose metrics are available.
Image retake and reject analysis is crucial for identify-
ing opportunities to improve image quality and patient
safety by analyzing trends and outliers (ACR 2017). While
most radiography systems and associated image processing
computers/databases have software for implementing retake
and reject analysis, there are no well-defined methods for
reporting, nor are there requirements from the states or ac-
creditation standards bodies mandating such analyses. On
many digital imaging systems, image rejects can simply
be thrown away without a trace, and no record of radiation
exposure is recorded. Reported retake rates below 5%
should be highly suspicious and indicative of a need for ad-
ministrative oversight, particularly for bedside radiography
exams. In this important area of practice evaluation, there
are roles for the DICOM x-ray radiation dose structured
report information object definitions (IOD) templates
(DICOM 2018), for system interoperability through the In-
tegrating the Healthcare Enterprise radiation exposure mon-
itoring effort (IHE 2010), and for all radiography system
Fig. 8. (a) Guidelines for using the deviation index feedback values at manufacturers to provide tools and methods for accurate
UC Davis Medical Center for bedside radiography examinations. (b) DI and robust accounting/reporting of retake rates.
values acquired for a subsample of studies—a wide range of DI values DRLs defined as the 75th percentile of the distribution
is evident at an early adoption stage of IEC 62494-1 (IEC 2008). of radiation doses for a radiographic exam (ICRP 2017) can
be used to identify high outlier radiation exposures for spe-
size, technique factors (kilovolt, milliampere, exposure cific exams. Currently, however, there are DRLs only for
time), and automatic exposure control settings (where avail- three adult radiographic exams in the United States (NCRP
able), among others. In addition, with a combination of CR 2012; ACR 2013). This is a deficiency that should be ad-
and DR detectors that might be used interchangeably for a dressed through the US government, perhaps sponsored
radiography system, there is a high probability of inappro- by the US Food and Drug Administration, the states, and
priate settings that could result in underexposure, overexpo- large organizations (professional societies, hospitals), to set
sure, inappropriate image processing, or other suboptimal out a plan for acquisition of information and determination
outcomes. Thus, it is important to stress the need for initial of DRLs and achievable doses (at the 50th percentile value)
and periodic evaluation of the protocols of each system for the most widely used digital radiographic procedures.
throughout the clinical service of the unit and particularly
during annual medical physics inspections. With implemen- ROLE OF THE NCRP
tation of the IEC 62494-1 exposure index standard (IEC
2008), EIT values must be input for each exam and each de- The National Council on Radiation Protection and
tector type (CR or DR). Ideally, with each output image or Measurements (NCRP) was chartered by Congress in
image sequence, all acquisition parameters and dose metric 1964 to collect, analyze, develop, and disseminate in the
values including kV, mA s, half-value layer (HVL), source- public interest information and recommendations about radia-
to-image distance (SID), source-to-skin distance (SSD), tion protection, radiation measurements, quantities, and units,
air kerma at a reference point (Ka,r), air kerma-area prod- and to develop basic concepts about radiation protection. Re-
uct (PKA), collimation dimensions, EI, EIT, and DI can cent examples of excellent guidance and recommendations
be provided as specified in the Digital Imaging and in medical imaging applications are the publication of
www.health-physics.com

Copyright © 2018 Health Physics Society. Unauthorized reproduction of this article is prohibited.
156 Health Physics February 2019, Volume 116, Number 2

NCRP Report 168, Radiation Dose Management for Available at https://www.acr.org/-/media/ACR/Files/

Fluoroscopically-Guided Interventional Medical Proce- Practice-Parameters/Rad-Digital.pdf. Accessed 26 July 2018.
Bushberg JT, Seibert JA, Leidholdt EL, Boone JM. Essential
dures (NCRP 2011), and NCRP Statement 11, Outline of physics of medical imaging. Riverwoods, IL: Wolters
Administrative Policies for Quality Assurance and Peer Re- Kluwer-Lippincott Williams and Wilkins; 2012.
view of Tissue Reactions Associated with Fluoroscopically- Digital Imaging and Communications in Medicine. X-ray radia-
Guided Interventions (NCRP 2014). In addition, NCRP is tion dose SR IOD templates. Template identif ication
10001–10007 [online]. Rosslyn, VA: National Electrical Man-
preparing a report or commentary on dose optimization in ufacturers Association, DICOM; 2018. Available at http://
computed tomography, through Scientific Committee 4-8 dicom.nema.org/dicom/2013/output/chtml/part16/sect_
of Program Area Committee 4. NCRP, however, has little XRayRadiationDoseSRIODTemplates.html. Accessed 26
guidance on digital projection radiography. From this con- July 2018.
Farrier M, Achterkirchen TG, Weckler GP, Mrozack A. Very large
text, a proposal to NCRP, based upon the many issues de- area CMOS active-pixel sensor for digital radiography. IEEE
scribed, is to consider a statement or commentary titled Trans Electron Devices 56:2623–2631; 2009.
“Radiation Dose Management in Digital Projection Radiog- Integrating the Healthcare Enterprise. Radiation exposure moni-
raphy” to address the deficiencies and needs noted in toring (REM) profile. Oak Brook, IL: IHE; 2010.
International Commission on Radiological Protection. Diagnostic
the practice of projection imaging. Guidance from NCRP reference levels in medical imaging. Thousand Oaks, CA:
can undoubtedly assist in the appropriate oversight and im- Sage Publications; ICRP Publication 135; 2017.
plementation of mechanisms for dose management in ra- International Electrotechnical Commission. Medical electrical
diography and mammography in the spirit of radiation equipment—exposure index of digital x-ray imaging systems.
Geneva: IEC; 62494-1, Ed. 1; 2008.
responsibility for practical and safe imaging. McCollough CH. Computed tomography technology—and
dose—in the 21st century. Health Phys 116:157–162; 2019.
SUMMARY Mentrup D, Jockel S, Menser B, Neitzel U. Iterative scatter correc-
tion for grid-less bedside chest radiography: performance for a
Radiography, mammography, and fluoroscopy play a chest phantom. Radiat Protect Dosim 169:308–312; 2016.
large role in the safe and effective medical care of patients. National Council on Radiation Protection and Measurements. Ion-
izing radiation exposure of the population of the United States.
Advances in digital detector technology, when adopted, can Bethesda, MD: NCRP; Report 160; 2009.
deliver more efficient workflow, better image quality, im- National Council on Radiation Protection and Measurements. Ra-
proved diagnostic reliability, and lower patient radiation diation dose management for fluoroscopically-guided inter-
dose. While lower radiation dose for radiographic exams ventional medical procedures. Bethesda, MD: NCRP; Report
168; 2011.
is certainly achievable, this goal is not always achieved for National Council on Radiation Protection and Measurements.
many reasons. Examples include insufficient education Reference levels and achievable doses in medical and dental
and training on new technological capabilities and func- imaging: recommendations for the United States. Bethesda,
tions, ignoring or not adopting exposure index standards, MD: NCRP; Report 172; 2012.
National Council on Radiation Protection and Measurements.
inadequately providing radiation dose feedback informa- Outline of administrative policies for quality assurance and peer
tion, and poor or nonexistent protocol and retake analysis review of tissue reactions associated with fluoroscopically-
mechanisms, among others. Defining checklists, establish- guided interventions. Bethesda, MD: NCRP; Statement
ing rules, reviewing protocols, and monitoring radiation 11; 2014.
Seibert JA. Digital radiography: image quality and radiation dose.
dose metrics can improve delivery of safe and high-quality Health Phys 95:586–598; 2008.
medical images. NCRP can contribute to radiation responsi- Seibert JA, Shelton DK, Moore EH. Computed radiography x-ray
bility in medical imaging by recommending guidance that exposure trends. Acad Radiol 3:313–318; 1996.
addresses the needs of medical projection imaging as de- Shepard SJ, Wang J, Flynn M, Gingold E, Goldman E, Krugh K,
Leong DL, Mah E, Ogden K, Peck D, Samei E, Willis CE. An
scribed in this article. exposure indicator for digital radiography: AAPM task group
116 (executive summary). Med Phys 36:2898–2914; 2009.
REFERENCES Taguchi K, Iwanczyk JS. Vision 20/20: single photon counting
x-ray detectors in medical imaging. Med Phys 40:100901-
American College of Radiology. ACR-AAPM practice parameter 1–100901-19; 2013.
for diagnostic reference levels and achievable doses in medical Weigel S, Berkemeyer S, Girrus R, Sommer A, Lenzen H,
x-ray imaging. Reston, VA: ACR; 2013. Available at https://www. Heindel W. Digital mammography screening with photon-
acr.org/-/media/ACR/Files/Practice-Parameters/diag-ref-levels. counting technique. Radiol 271:345–355; 2014.
pdf. Accessed 26 July 2018.
American College of Radiology. ACR-AAPM-SIIM-SPR practice
parameter for digital radiography. Reston, VA: ACR; 2017. ■■

www.health-physics.com

Copyright © 2018 Health Physics Society. Unauthorized reproduction of this article is prohibited.