Dose Optimization in Digital Radiography and Computed Tomography An Essential Guide 3031228707 9783031228704 - Compress

Euclid Seeram
Dose Optimization
in Digital
Radiography
and Computed
Tomography
An Essential Guide
Dose Optimization in Digital Radiography
and Computed Tomography
Euclid Seeram
Dose Optimization in Digital

Radiography and Computed
Tomography
An Essential Guide
Euclid Seeram
University of Canberra
Burnaby, BC, Canada
ISBN 978-3-031-22870-4 ISBN 978-3-031-22871-1 (eBook)

https://doi.org/10.1007/978-3-031-22871-1
© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature
Switzerland AG 2023
This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether
the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of
illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and
transmission or information storage and retrieval, electronic adaptation, computer software, or by similar
or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors, and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the
editors give a warranty, expressed or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.
This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
This book is dedicated to my two smart,
precious, and overall cute granddaughters,
Claire and Charlotte, with love and blessings
to you both forever.
Preface
Digital radiography (DR) and computed tomography (CT) are based on physics,
engineering, and computer science principles that have influenced their survival as
useful clinical tools for imaging the patient. These two modalities have experienced
several technical innovations in recent years. These innovations have played major
roles in not only reduction of the radiation dose to the patient but improving the
image quality needed for diagnostic interpretation. For example, the fundamental
concepts of the wide exposure latitude of DR systems and the standardized expo-
sure indicator established by the International Electrotechnical Commission (IEC)
have provided researchers with the motivation to operate DR systems with the goal
of optimization of the dose and image quality. Additionally, CT technical innova-
tions such as, for example, new detector technology, iterative reconstruction (IR)
algorithms, and artificial intelligence-based image reconstruction are now offered
by several CT vendors, and play a significant role in dose reduction and optimiza-
tion in CT. Furthermore, the introduction of photon counting detectors (PCDs) has
solved the major problem of image noise during low-dose CT imaging. Dose opti-
mization has now become an integral part of DR and CT practice, especially in
radiation protection of the patient.
This book, Dose Optimization in Digital Radiography and Computed
Tomography, provides yet another useful resource to meet the educational require-
ments of professional radiologic technology associations including the American
Society of Radiologic Technologists (ASRT), the American Registry of Radiologic
Technologists (ARRT), the Canadian Association of Medical Radiation
Technologists (CAMRT), the College of Radiographers in the United Kingdom, as
well as those professional medical imaging organizations in Africa, Asia, Australia,
and continental Europe. Additionally, this book may serve as a resource for bio-
medical engineering technology programs that include DR and CT imaging systems
in their curriculum; residents in radiology, and medical physics students studying
the use of DR and CT in medical imaging.
vii
viii Preface
There are seven chapters in this book as follows:

Chapter 1 provides the motivation for dose optimization in DR and CT, with current
and relevant references.
Chapter 2 deals with a technical review of DR technologies, including computed
radiography (CR) and flat-panel digital radiography (FPDR), with current and
relevant references.
Chapter 3 identifies and describes the major technical factors affecting the dose in
DR, and various optimization strategies in DR, with current and relevant
references.
Chapter 4 addresses the technical aspects of CT, with current and relevant references.
Chapter 5 takes a close examination of the technical factors affecting the dose in
CT, dose reduction, and dose optimization strategies, with current and relevant
references.
Chapter 6 explains the nature of dose optimization research in medical imaging and
presents an overview of observer performance methods in image quality assess-
ment, with current and relevant references.
Chapter 7 is the final chapter and provides a set of self-assessment multiple-choice
questions (with answers) to check your understanding of the materials studied.
Enjoy the content and questions that follow and best wishes for any examination
that you have to write and pass. Remember, your patients will benefit from
your wisdom.
Burnaby, BC, Canada Euclid Seeram

Acknowledgments
One of the most satisfying tasks in writing a book of this nature is to express grati-
tude to all those who have done the original work. These are the medical physicists,
biomedical engineers, computer scientists, radiologists, and technologists who are
working in both research and practice to conceptualize and invent better ways to
image the patient with reduced dose and improved image quality.
This book deals with two major imaging modalities, digital radiography (DR)
and computed tomography (CT), and in this respect, it is indeed a pleasure to express
sincere thanks to all scientific and clinical experts in the field of DR and CT.
The content of this book is centered around the published works and expertise of
several noted medical physicists, radiologists and imaging technologists, computer
scientists, and biomedical engineers, whose research on dose optimization have
been quoted here to support the use of the ALARA (as low as reasonably achiev-
able) principle in radiation protection of the patient, established by the International
Commission on Radiological Protection (ICRP).
I am most grateful to Dr Rob Davidson, PhD, MAppSc (MI), BBus, FASMIRT,
Professor of Medical Imaging, University of Canberra, Australia. Dr. Davidson has
taught DR and CT physics and instrumentation for decades. He was one of my PhD
supervisors who guided me through the experiments on dose optimization in DR
imaging. He has also been responsible for my adjunct professorship appointment at
the University of Canberra, and has supported all my other projects linked to DR
and CT physics and technology. Thanks mate.
Furthermore, two medical physicists to whom I am truly grateful for my DR
education are Dr. Anthony Siebert, PhD, of the University of California at Davis and
Dr. Charles Willis, PhD, formerly of the University of Texas, MD Anderson Cancer
Center, from whom I have learned the physics and technical aspects of digital radi-
ography through their seminars and workshops that I have attended.
My CT physics education stems from not only attending conferences and work-
shops but also personal communications with a number of notable medical physi-
cists. In particular, I owe a good deal of thanks to Dr. Godfrey Hounsfield and Dr.
Allan Cormack, both of whom shared the Nobel Prize for Medicine and Physiology
for their work in the invention and development of the CT scanner. I have been in
ix
x Acknowledgments
personal communication with Dr. Hounsfield, and he graciously provided me with

details of his biography and his original experiments, and signed his Nobel Lecture,
which he sent to me. Other CT experts from whose published works I have learned
a great deal and to whom I am indebted include Jiang Hsieh, PhD, former Chief
Scientist with General Electric Healthcare; world-expert CT physicist, Professor
Willi Kalender, PhD, Institute of Medical Physics in Germany; Mahadevappa
Mahesh, PhD, Chief Medical Physicist, Johns Hopkins Hospital in Baltimore;
Michael McNitt-Gray, PhD, University of California; Cynthia McCollough, PhD,
Mayo Clinic; and Thomas Flohr, PhD, Siemens Healtineers, Germany.
Another notable individual who has made an impact on my professional journey
is Valentina Al Hamouche, MRT, MSc, and CEO for Vision, Compassion, Awareness
(VCA) Education Solutions for Health Professionals Inc. Toronto, Ontario, Canada.
Valentina has given me the opportunity to be a regular guest lecturer in not only
face-to-face presentations but also live webinars on CT physics and instrumenta-
tion, and radiographic sciences as well as live webinars on dose optimization in CT
and DR to technologists all over the world. She has brought continuing education to
health professionals through her Vision, Compassion, and Awareness, the name of
her educational organization (www.VCAeducation.ca)
I am particularly grateful to Merry Stuber, Senior Editor, Cell Biology &
Biomedical Engineering at Springer, a part of Springer Nature, New York, NY,
USA, who did all the hard work in reviewing the proposal herself but also in getting
external reviews of the proposal that led her to accept it for publication. Merry has
provide the needed continuous support and encouragement to bring this work to
fruition. Additionally, I am grateful to members of the production team at Springer
Nature, who have worked exceptionally hard during the production of this book,
especially in the page-proof stages.
I humbly acknowledge the support and praise that I get from my beautiful family.
First my lovely wife, Trish, a warm, smart, caring, and a very special person in my
life, thanks babe. Secondly, my caring and very brilliant son David, the best dad on
the planet to his two most precious daughters, my granddaughters, to whom this
book is dedicated. Thanks for your enduring love, support, and encouragement.
Last but not least, I must thank my students, not only in Canada but also all over
the world, who have diligently completed my DR and CT physics and technology
courses, both at the diploma and degree levels. Thanks for all the challenging ques-
tions, which have always “kept me on my toes.”
Keep on learning and have fun answering the questions that follow, and remem-
ber, your patients will benefit from your wisdom.
Burnaby, BC, Canada Euclid Seeram

Contents
1
Dose Optimization: A Major Principle of Optimization�� 1
1.1 Biological Effects of Radiation Exposure�� 1
1.2 Fundamental Principles of Radiation Protection�� 2
1.2.1 The Principle of Justification�� 3
1.2.2 The Principle of Optimization�� 3
1.3 Motivation for Dose Optimization in Digital Radiography
and Computed Tomography�� 6
1.3.1 Radiation Risks �� 6
1.3.2 Dose Creep in Digital Radiography�� 6
1.3.3 Dose in Computed Tomography �� 8
1.4 Optimization Strategies: An Overview �� 9
References�� 9
2 Digital Radiography: A Technical Review �� 13
2.1 Digital Radiography: Two Notable Facts Related to Dose
Management�� 13
2.2 Digital Radiography Systems: A Technical Review �� 15
2.2.1 Computed Radiography: Physical Principles and
Technology�� 16
2.2.2 Flat-Panel Digital Radiography: Physical Principles
and Technology �� 17
2.3 Other Considerations Related to CR and FPDR Systems�� 19
2.3.1 Image Processing�� 19
2.3.2 The Standardized Exposure Indicator
and the Deviation Index�� 20
2.4 Factors Affecting the Dose in Digital Radiography:
An Essential Review �� 22
2.5 Diagnostic Reference Levels�� 22
2.5.1 Definitions and Major Guidelines�� 23
References�� 23
xi
xii Contents
3
Optimization Strategies in Digital Radiography�� 25
3.1 Introduction�� 25
3.2 Optimization of Exposure Technique Factors�� 26
3.2.1 Optimization of kV and mAs�� 26
3.3 Optimization of the Exposure Indicator�� 27
3.4 Optimization of the Deviation Index�� 30
3.5 Optimization Using Image Postprocessing Algorithms�� 31
3.5.1 Optimization Studies Using Multifrequency Processing
and Noise Reduction Algorithms�� 32
3.6 Diagnostic Reference Levels: A Useful Tool
in Dose Optimization�� 36
References�� 37
4 Computed Tomography: A Technical Review �� 41
4.2 The CT Scanner: Fundamental Physics and Major System
Components�� 42
4.2.1 Attenuation Physics: An Essential Overview�� 42
4.2.2 Attenuation and CT Numbers �� 43
4.2.3 Data Acquisition: System Components and Principles�� 44
4.2.4 Image Reconstruction �� 46
4.2.5 Image Display/Storage/Communication�� 49
4.3 Multi-slice CT Principles at a Glance�� 50
4.3.1 MSCT Detectors: An Overview�� 51
4.3.2 Selectable Scan Parameters�� 53
4.4 Radiation Protection�� 54
References�� 55
5
Dose Reduction and Optimization Strategies in Computed
Tomography�� 57
5.2 Motivation for Dose Optimization in CT�� 58
5.2.1 Literature on Cancer Risks from CT�� 59
5.2.2 Effects of Low-Dose Chest CT on Chromosomal DNA�� 60
5.2.3 In Summary�� 60
5.3 Radiation Protection Principles�� 61
5.4 Elements of CT Dosimetry at a Glance�� 61
5.5 Factors Affecting the Dose in CT: An Overview�� 63
5.5.1 Image Quality Considerations�� 63
5.5.2 Significant Dose Parameters �� 63
5.6 Dose Optimization Strategies in CT �� 65
5.7 Other Useful Tools for Dose Reduction
and Optimization in CT�� 66
5.7.1 The Use of DRLs�� 66
5.7.2 The Use of AI-Based Image Reconstruction�� 68
5.7.3 The Use of PCDs�� 70
Contents xiii
5.8 The Role of the Technologist in CT Dose Reduction

and Optimization�� 71
References�� 72
6
Optimization Research in Medical Imaging�� 77
6.1.1 Strategies Used in Dose/Image Quality Optimization
Research�� 78
6.1.2 The Nature of Image Quality�� 79
6.1.3 Image Quality Assessment Tools for Clinical CT Images:
An Overview �� 80
6.1.4 Receiver Operating Characteristics: A Brief Overview�� 81
6.1.5 Visual Grading Analysis: An Overview�� 82
6.1.6 Example of a Dose Optimization Study in Digital
Radiography�� 84
References�� 85
7 Review Questions �� 87
Index�� 101
Chapter 1
Dose Optimization: A Major Principle
of Optimization
Keywords Biological effects of radiation exposure: International Commission on

Radiological Protection (ICRP) · Principles of radiation protection · Justification ·
Optimization · Dose optimization · Dose creep in digital radiography · Doses in
computed tomography (CT) · Optimization strategies
1.1 Biological Effects of Radiation Exposure
Biological effects of radiation exposure are sometimes referred to as health effects,

harmful effects, or simply radiation risks. These effects have been studied exten-
sively and documented in what is popularly known as epidemiologic studies [1].
These studies have identified several sources which have been placed into the fol-
lowing classes: Early radiation workers such as radiologists and physicists who
were exposed to high doses; workers from radiation and nuclear industries; survi-
vors from Hiroshima and Nagasaki (H-N) atomic bomb explosions; and workers
and inhabitants exposed to nuclear reactor accidents such as those at Chernobyl and
Three Mile Island. Other significant resources on biological effects are available
from the Biological Effects of Ionizing Radiation (BEIR) Reports, which also
includes data on the exposure of patients to high doses in medical imaging.
Furthermore, data on the H-N survivors have been provided the Radiation Effects
Research Foundation (RERF).
There are at least two significant and important points on biological effects has
been identified and stated by Hendee and O’Connor [1]. The first point is as follows:
The RERF studies of the Japanese atomic bomb survivors are the major source of what is
known about the health consequences to individuals exposed to ionizing radiation. The
RERF data and the models of radiation injury developed by the RERF scientists form the
back bone of the BEIR VII report. It is from the summary tables of radiation risk in the
BEIR VII report that projections of cancer incidence and death are made for medical expo-
sures in the United States. Hence an analysis of the assumptions and limitations of risk
estimates derived from BEIR VII must include a review of the RERF studies from which
BEIR VII is derived. (p. 314)
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 1

E. Seeram, Dose Optimization in Digital Radiography and Computed Tomography,
https://doi.org/10.1007/978-3-031-22871-1_1
2 1 Dose Optimization: A Major Principle of Optimization
The biological effects identified in the above-mentioned reports have been placed
into two categories, namely; stochastic effects and deterministic effects. Stochastic
effects are those effects for which the probability of occurrence in the individual
exposed, increases with increasing dose and for which there is no threshold dose.
Examples of stochastic effects include cancer, leukemia, and hereditary effects.
Deterministic effects on the other hand, are those effects for which the severity of
the effect in the exposed individual increases with increasing dose, and for which
there is a threshold dose. Examples of deterministic effects include cataracts,
radiation-induced skin burns, tissue damage, and organ disfunction [2].
The second significant and important point is that:
RERF data provide statistically significant evidence of increased cancers in Japanese survi-
vors who received doses of 100 mSv and higher with cancer incidence appearing to increase
linearly with dose. At less than 100 mSv, an increase in radiation-induced cancers, if any is
too small to be distinguishable from cancer incidence due to all causes. Consequently, a
model must be deployed to extrapolate from radiation-induced cancers at doses greater than
100 mSv to a hypothetical and much smaller number of cancers induced by doses of a few
millisieverts delivered during medical imaging [1].
Subsequently, several dose-response models for extrapolating from a high dose situ-
ation (for example 100 mSv) to the low doses used in diagnostic x-ray imaging have
been proposed, however, only two will be briefly outlined in this chapter, that is; the
linear dose-response model without a threshold dose, known as the LNT model; and
the linear dose-response model with a threshold dose. While the former implies that
there is no safe amount of radiation dose, and that any dose no matter how small
carries a degree of risk; the latter implies the effect of radiation exposure will mani-
fest itself at a certain dose referred to as the threshold dose. Below this threshold no
effect is observed [2].
Currently, there is ongoing debate about these models and their use in medical
imaging, however, Hendee and O’Connor [1] state that “the model used most widely
is the LNT model. This model is not chosen because there is solid biologic or epi-
demiologic data supporting its use. Rather, it is used because of its simplicity and
because it is a conservative approach (i.e., if it is not correct, then it probably over-
estimates the risk of cancer induction at low dose). For the purpose of establishing
radiation protection standards for occupationally exposed individuals and members
of the public, a conservative model that overestimates the risk is preferred over a
model that underestimates risk.” (p. 316).
1.2 Fundamental Principles of Radiation Protection
In order to protect patients undergoing medical imaging procedures using ionizing

radiation, the ICRP has recommended a comprehensive framework of radiation
protection against ionizing radiation. This framework is intended to prevent
1.2 Fundamental Principles of Radiation Protection 3
deterministic effects from occurring by keeping the radiation doses below threshold
values; and to minimize the probability of stochastic effects by taking all reasonable
efforts.
The framework is a recommendation that addresses three fundamental principles
of radiation protection [3]; justification, optimization, and dose limitation. While
justification and optimization focus on exposed individuals, the third principle
examines both occupational and public exposures, and does not include medical
exposures. These principles are now accepted by various national radiation protec-
tion organizations such as for example; the National Council on Radiation Protection
and Measurements (NCRP) in the United States; Radiation Protection Bureau-
Health Canada; the National Radiological Protection Board (NRPB) in the United
Kingdom; and other radiation protection organizations around the world.
This textbook will only address the major elements of optimization.
1.2.1 The Principle of Justification
The principle of justification is described as follows by the ICRP:

Any decision that alters the radiation exposure situation should do more good than harm.
This means that, by introducing a new radiation source, by reducing existing exposure, or
by reducing the risk of potential exposure, one should achieve sufficient individual or soci-
etal benefit to offset the detriment it causes [3].
Justification therefore focusses on the concept of net benefit, that is, there must be a
benefit related to the introduction of any new imaging modality, procedure, or radia-
tion exposure. The responsibility for justification falls upon not only the physician
who requests x-ray examinations based on good clinical reasons, and the individual
who performs the examination [4].
1.2.2 The Principle of Optimization
Following justification, the principle of optimization is applied. The ICRP states

that the essential goal of optimization during the conduct of the examination is “to
adjust imaging parameters and institute protective measures in such a way that the
required image is obtained with the lowest possible radiation dose, and net benefit
is maximised” [3]. In this regard, the ICRP recommends that the ALARA (as low as
reasonably achievable) philosophy should be applied. Furthermore, the ICRP [3]
emphasizes that ALARA “is only part of the concept of optimisation. The entire
concept implies, more precisely, keeping patient exposure to the minimum neces-
sary to achieve the required medical objective (diagnostic or therapeutic). In diag-
nostic imaging and x-ray-guided interventions, it means that the number and quality
of images are sufficient to obtain the information needed for diagnosis or

intervention….”
Optimization involves at least four elements [5], namely; the imaging equip-
ment, the adequacy of the imaging equipment, technical parameters of the examina-
tion, and diagnostic reference levels, and requires that:
• The imaging equipment is in good working order to deliver correct exposures for
respective examinations; and complies with required performance standards
established by various regulatory and professional standards. Such equipment is
subject to acceptance testing when it is installed; and subsequently, routine qual-
ity control tests should be conducted during routine use.
• The adequacy of the equipment takes into consideration not only the design and
use of a wide range of settings, but also specific pre-installed protocols and vari-
ous approaches for dose reduction when imaging pediatric patients.
• Technical parameters for the examination refer to several dose reduction meth-
ods, such as the use of correct exposure technique factors (mAs and kV) for the
examination, collimation, filtration, removable grids, and pulsed fluoroscopy, to
mention but a few.
• Diagnostic reference levels (DRLs) be implemented in the assessment of the
medical exposure of patients. There are several definitions of the DRL, however,
only two will be provided in here. The DRLs as defined by the ICRP “are a form
of investigation level applied to an easily measured quantity, usually absorbed
dose in all, or tissue-equivalent material at the surface of a simple standard phan-
tom or a representative patient” [6].
The third principle of radiation protection is dose limitation. According to the ICRP
this principle deals with the application of dose limits, and requires that “the total
dose to any individual from regulated sources in planned exposure situations other
than medical exposure of patients should not exceed the appropriate limits recom-
mended by the Commission” [3]. Since this principle specifically addresses doses to
occupationally-exposed individuals (technologists for example), members of the
public, and doses to other body parts such as the skin, lens of the eyes, and extremi-
ties, it will not be discussed further in this book.
Another characteristic feature of optimization is that the ALARA philosophy
includes image quality, that is the required image quality should be obtained with
the lowest possible radiation dose. Therefore, dose optimization strategies and
research must consider image quality and dose.
igital Radiography and Computed Tomography: System Components

D
at a Glance
Dose optimization in digital radiography and computed tomography (CT) requires

a thorough understanding of not only the major technical components but also the
factors affecting the dose to the patient, in each of these two imaging systems. Such
1.2 Fundamental Principles of Radiation Protection 5
understanding provides the basis for how the dose from each of these imaging tech-
nologies can be optimized.
While digital radiography (DR) essentially includes two major and popular sys-
tems, commonly referred to as computed radiography (CR) and flat-panel digital
radiography (FPDR) [2, 6], the technical components of which are illustrated in
Fig. 1.1; CT systems are more complex, not only in terms of physical principles, but
also the technologies involved in creating the CT image [7]. Major CT system com-
ponents are illustrated in Fig. 1.2. The principles and technology of DR and CT
systems will be described in Chaps. 2 and 4 respectively.
Fig. 1.1 The major technical components of two commonplace digital radiography imaging sys-
tems; (a) computed radiography (CR) and (b) flat-panel digital radiography (FPDR). (See text for
further explanation)
Fig. 1.2 The primary technical components of a CT Imaging system, includes data acquisition,
image reconstruction, and image display, storage, and communications. (See text for further
explanation)
1.3 Motivation for Dose Optimization in Digital

Radiography and Computed Tomography
1.3.1 Radiation Risks
Radiation risks provide the first and most important motivation for dose optimiza-
tion in diagnostic imaging using x-rays. These risks are real and have been
described extensively in the literature [8, 9] and fall into two categories, namely;
stochastic effects and deterministic effects. These risks have been summarized in
1 above.
1.3.2 Dose Creep in Digital Radiography
The motivation for dose optimization in digital radiography has its roots in what has
been referred to as dose creep or exposure creep, which has been identified in the
literature as an increase in the exposure over time as technologists use these DR
imaging systems [10–14].
This practice has been shown to occur because of the wide exposure latitude and
the image processing features of DR imaging systems. Such image processing com-
pensates for underexposure and overexposure while producing images that have
acceptable gray scale image quality. This will be described in detail in Chap. 2.
While underexposed images appear noisy (quantum mottle), overexposed images
do not exhibit quantum noise [15]. Noise-free images are always preferred by radi-
ologists. The significant and important problem with the use of overexposure is an
increase in the dose to the patient. Therefore, it is mandatory that users operate with
the ALARA philosophy in mind. This provides the opportunity for dose optimiza-
tion in digital radiography.
An important and meaningful study demonstrating that dose creep is a real phe-
nomenon is one by Gibson and Davidson [12]. The researchers conducted a longi-
tudinal study which examined and categorized the Exposure Indicators (EIs) s from
more than 17,000 mobile chest radiographs from intensive and critical care units at
one hospital over a 26-month period. This study showed that dose creep occurred
when x-ray exposure parameters were set manually, and that overexposure index
values increased over time (Fig. 1.3a). The optimal exposures also decreased during
the same period. To demonstrate that the increase in overexposures was not caused
by external factors, an intervention was needed. The researchers introduced an addi-
tional requirement for mobile chest radiographs in the intensive care unit; technolo-
gists were to record exposure factors of kilovolt (kV), milliampere-seconds (mAs),
and source-to-image receptor distance, as well as the EI value for each examination.
After the intervention, the initially observed trend of overexposures was gradually
halted over a 14-month period. Figure 1.3b illustrates the halt of the overexposure
1.3 Motivation for Dose Optimization in Digital Radiography and Computed... 7
ICU CXR Exposures

Aug 07 - Dec 09
70
Examinations(%)
60
50
40
30
20
Optimal Exposure
Over Exposure
10
0
Aug-07 Nov-08 Feb-08 Jun-08 Sep-08 Dec-08 Mar-09 Jul-09 Oct-09 Jan-10
ICU CXR Exposures

Post Intervention: Feb 10 - Mar 11
70
Examinations(%)
60
50
40
Optimal Exposure
Over Exposure
30
20
10
0
jan-10 Mar-10 May-10 Jun-10 Aug-10 Sep-10 Nov-10 Jan-11 Feb-11
Fig. 1.3 Graphs comparing the exposure index values for mobile chest radiographs performed in
the intensive care unit from before and after intervention. Before intervention, overexposure index
values increased over time, indicating dose creep in digital radiography (a). Optimal exposures
decreased over time, stopping only after an intervention was applied (b). (Reproduced with per-
mission from Professor Rob Davidson, Medical Imaging; School of Health Sciences, Faculty of
Health, University of Canberra, Australia)
trend. This intervention demonstrated conclusively that dose creep is a challenge

that can be addressed at the systemic level.
In another study of dose creep, Dalah [16], quantified dose creep for skull and
chest examinations using dose area product and entrance surface dose. The results
showed that the chest examination had the highest estimated dose creep of 51%.
Dalah [16] concluded that “dose-creep can be considerably high raising the need for
thorough investigations and society awareness”. This is yet another dose study that
supports the need for dose optimization in digital radiography.
1.3.3 Dose in Computed Tomography
The first clinically useful CT scanner was invented by Godfrey Newbold Hounsfield,
who worked at Electric and Musical Industries (EMI) in the United Kingdom.
Another notable pioneer in the is the evolution of the CT scanner is Allan MacLeod
Cormack, in South Africa. Hounsfield and Cormack shared the 1979 Nobel Prize in
Medicine or Physiology for their significant contributions such as providing math-
ematical solutions to the problem in CT [7].
The use of CT in in clinical medicine has increased significantly ever since its
introduction in the 1970s, due to several technical advances in CT scanner design
and performance. Several studies have shown that patient doses from CT examina-
tions are high relative to other radiography examinations [17–19] and that CT doses
typically range from 5 to 50 mGy to each organ within the image field [20]. As of
2011, CT contributed the highest collective amount of medical radiation exposure
in the United States compared with any other medical imaging modality [21]. These
relatively high doses from CT examinations, especially in pediatric imaging, have
raised concerns related to the risks of cancer associated with CT imaging [1,
2, 17–22].
In addition, continued technological advances in CT, such as for example, more
sensitive detectors, iterative and deep learning reconstruction algorithms, have
played a key role in reducing the dose from CT scanners, for specific use in what is
popularly referred to as low-dose CT imaging.
The Food and Drug Administration (FDA) [23] in the United States, in an article
entitled “What are the Radiation Risks from CT” states that “the effective doses
from diagnostic CT procedures are typically estimated to be in the range of 1 to
10 mSv”. Additionally, the World Nuclear Association [24] in 2022, states that the
effective dose from an abdomen and pelvis CT examination is 10 mSv.
More recently, Smith-Bindman and collaborators [25] conducted an extensive pro-
spective cohort study, “to determine patient, institution, and machine characteristics
that contribute to variation in radiation doses used for computed tomography (CT)”.
The researchers found that “CT protocols and radiation doses vary greatly across
countries and are primarily attributable to local choices regarding technical parame-
ters, rather than patient, institution, or machine characteristics. These findings suggest
that the optimization of doses to a consistent standard should be possible”.
References 9
1.4 Optimization Strategies: An Overview
In addressing optimization strategies, and operating within the ALARA principle,

the literature most often uses two terms: reduction and optimization. It is important
to note that while reduction means to “reduce or diminish in size, amount, extent, or
number”, optimization means “an act, process, or methodology of marking some-
thing (as a design, system, or decision) as fully perfect, functional, or effective as
possible” [26].
The optimization strategies that will be presented in Chaps. 3 and 5 for DR and
CT respectively, will therefore focus on the specific research methodologies that
address the dictionary definition of optimization.
Optimization strategies for DR and CT examinations have been identified and
described in the literature and they are focused on how technical factors can be used
to optimize the dose and not compromise the image quality [27]. This is the basic
tenet of the ALARA philosophy.
Technical factors related to DR imaging are several and include for example,
exposure technique factors (mAs and kV); the exposure indicator; the deviation
index; image processing and noise reduction algorithms; and diagnostic reference
levels [28–30]. Optimization strategies are therefore focused on these factors. While
technical factors will be described further in Chap. 2, optimization strategies for DR
imaging will be described in Chap. 3.
The image acquisition and image processing technologies are totally different in
CT compared to DR imaging systems. As a result, there are significant technical
factors that must be considered when describing optimization strategies in
CT. Technical factors are numerous and include for example, kV; mAs; noise; pixel
size; and slice thickness. Furthermore, there are several other factors that affect the
dose in CT imaging. These include the pitch; the scan field-of-view; beam collima-
tion; noise-reducing image reconstructive algorithms, particularly iterative recon-
struction; for example [2, 7–9, 31–33].
While Chap. 4 will describe the essential physics and technology of CT imaging,
including factors affecting the dose in CT; Chap. 5 will elaborate on specific optimi-
zation strategies that are noteworthy, and have received attention in the literature.
References
1. Hendee WR, O’Connor MK (2012). Radiation risks in medical imaging. Separating fact from
fantasy. Radiology 204 (2): 312–320
2. Bushong, S. (2022). Radiologic science for technologists (12th ed.). St. Louis: Elsevier Mosby.
3. International Commission on Radiological Protection (2007). The 2007 Recommendations
of the International Commission on Radiological Protection. Orlando Florida, Elsevier. ICRP
Publication 103. Ann ICRP; 37: 1–332.
4. Matthews K, Brennan PC (2008). Justification of x-ray examinations: General principles and
an Irish perspective. Radiography 14, 249–35
5. ICRP (2013). Radiological protection in paediatric diagnostic and interventional radiology.

ICRP Publication 121. Ann. ICRP 42(2).
6. Seeram E (2019). Digital Radiography: Physical Principles and Quality Control. Second
Edition, Singapore, Springer.
7. Seeram E (2023). Computed Tomography: Physical Principles, Patient Care. Clinical
Applications, and Quality Control (5th Edition). St Louis, MO., Elsevier
8. Seeram E, Brennan P (2017). Radiation Protection in Diagnostic X-Ray Imaging, Burlington,
MA., Jones and Bartlett Learning.
9. Bushberg JT, Seibert JA, Leidholdt EM, Boone JM (2021). The Essential Physics of Medical
Imaging. 4th ed. Philadelphia, PA., Wolters Kluwer.
10. Willis CE, Slovis TL (2004). The ALARA concept in pediatric CR and DR: dose reduction in
pediatric radiographic exams—a white paper conference executive summary. Pediatr Radiol;
34 (suppl 3):S162–S164
11. Seibert JA (2008), Digital radiography: image quality and radiation dose. Health Phys. Nov;
95 (5): 586–98) PubMed https://doi.org/10.1097/01.HP.0000326338.14198.a2
12. Gibson DJ, Davidson RA (2012). Exposure creep in computed radiography: a longitudinal
study. Acad Radiol. Apr;19(4):458–62. doi: https://doi.org/10.1016/j.acra.12.003. Epub. Jan
5. PubMed
13. Davidson, R. (2020). Dose creep in digital radiography. In C. Hayre, & W. Cox (Eds.), General
radiography: Principles and practice (pp. 59–69). CRC Press. https://doi.org/https://doi.org/1
0.1201/9781003047278-4
14. Benfield S, J.D. Hewis, C.M. Hayre (2021) Investigating perceptions of ‘dose creep’
amongst student radiographers: A grounded theory study, Radiography, Volume 27, Issue 2,
Pages 605–610
15. Seibert, J. A., & Morin, R. L. (2011). The standardized exposure index for digital radiography:
An opportunity for optimization of radiation dose to the pediatric population. Pediatr Radiol
41, 573–581.
16. Entesar Zawam Dalah (2020). Quantifying dose-creep for Skull and chest radiography using
dose area product and entrance surface dose: Phantom study, Radiation Physics and Chemistry,
Volume 167, 108231
17. Brenner DJ, Hall EJ (2007). Computed tomography: an increasing source of radiation expo-
sure. N Engl J Med;357(22), 2277–2284.
18. Berrington de Gonzalez A, Mahesh M, Kim KP, et al (2009). Projected cancer risks
from computed tomographic scans performed in the United States in 2007. Arch Intern
Med;169(22):2071–2077. doi:https://doi.org/10.1001/archinternmed.440.
19. Van der Molen AJ, Stoop P, Prokop M, Geleijns J (2013). A national survey on radiation
dose in CT in The Netherlands. Insights Imaging;4(3):383–390. doi:https://doi.org/10.1007/
s13244-013-0253-9
20. Mathews JD, Forsythe AV, Brady Z, et al (2013). Cancer risk in 680,000 people exposed to
computed tomography scans in childhood or adolescence: data linkage study of 11 million
Australians. BMJ; 346:1–18. doi:https://doi.org/10.1136/bmj.f2360.
21. Hricak H, Brenner DJ, Adelstein SJ, Frush DP, et al (2011). Managing radiation use in medical
imaging: a multifaceted challenge. Radiology; 258(3):889–905. doi:https://doi.org/10.1148/
radiol.10101157
22. Amis ES Jr. (2011). CT radiation dose: trending in the right direction. Radiology;261(1):5–8.
doi:https://doi.org/10.1148/radiol.11111319
23. FDA (2022). What are the radiation risks from CT? (https://www.fda.gov/radiation-emitting-
products/medical-x-ray-imaging/what-are-radiation-risks-ct) Accessed June 2022.
24. World Nuclear Assoiation (2022). Radiation and Health Effects. https://www.world-nuclear.
org/information-library/safety-and-security/radiation-and-health/radiation-and-health-effects.
aspx Accessed June 2022
References 11
25. Smith-Bindman R, Wang Y, Chu P, Chung R, Einstein AJ, Balcombe J, et al (2019) International
variation in radiation dose for computed tomography examinations: prospective cohort study.
BMJ; 364: k4931
26. Reduction, Optimization. Merriam-Webster Dictionary. Online English Language Dictionary.
www.merriam-webster.com/dictionary. Accessed June 2021.
27. Samei E, Järvinen H, Kortesniemi M, Simantirakis G, Goh C, Wallace A, et al (2018). Medical
imaging dose optimisation from ground up: expert opinion of an international summit.
J. Radiol. Prot. 38 967
28. Mothiram, U., Brennan, P. C., Lewis, S. J., Moran, B., & Robinson, J. (2014). Digital radiogra-
phy exposure indices: A review. Journal of medical radiation sciences, 61(2), 112–118. https://
doi.org/https://doi.org/10.1002/jmrs.49
29. Marcelo B. Freitas, Ricardo B. Pimentel, Laura F. Braga, Francisco S.A. Salido, Rodrigo
F.C.A. Neves, Regina B. Medeiros (2020). Patient dose optimization for computed radiog-
raphy using physical and observer-based measurements as image quality metrics, Radiation
Physics and Chemistry, Volume 172.
30. Seeram E (2022). Dose Optimization in Digital Radiography. American Society of
Radiological Technologists (ASRT), Essential Education; Pages 1–14. https://apps.asrt.org/
DirectedReading/DirectedReading.aspx (accessed June 30, 2021)
31. Wolbarst AB, Capasso P, Wyant AR (2013). Medical Imaging: Essentials for Physicians.
Hoboken, NJ: Wiley-Blackwell.
32. Goo HW (2012). CT radiation dose optimization and estimation: an update for radiologists.
Korean J Radiol.;13(1):1–11. doi:https://doi.org/10.3348/kjr.2012.13.1.1.
33. Maldjian PD, Goldman AR (2013). Reducing radiation dose in body CT: primer on dose met-
rics and key CT technical parameters. AJR Am J Roentgenol.; 200(4):741–747. doi:https://doi.
org/10.2214/AJR.12.9768
Chapter 2
Digital Radiography: A Technical Review
Keywords Digital radiography · Detector response to radiation · Computed

radiography (CR) · Photostimulable luminescence · Flat-panel digital radiography
(FPDR) · Indirect conversion digital radiography · Direct conversion digital
radiography · Image processing · Standardized exposure indicator · Deviation
index · Dose in digital radiography · Diagnostic reference levels
2.1 Digital Radiography: Two Notable Facts Related

to Dose Management
Digital Radiography (DR) imaging systems fall into two popular imaging modali-
ties, namely; computed radiography (CR), and flat-panel digital radiography
(FPDR). These systems have been described in detail in the literature [1–3].
Furthermore, the major system components and processes common to of each these
two systems were illustrated in Fig. 1.1 (Chap. 1). These include image acquisition
by the detectors after exposure; image preprocessing; image postprocessing; and
image display for viewing and interpretation.
The radiation transmitted through the patient creates a latent image on the detec-
tors. The latent image is rendered visible through the use of sophisticated software.
When imaging patients, operators must be cognizant of the following two notable
facts which play a role in achieving the ALARA (as low as reasonably achievable)
principle of the International Commission on Radiological Protection (ICRP):
1. The response of digital detectors to radiation exposure, is linear (Fig. 2.1) com-
pared to a sigmoidal response of film-screen detectors (which are now obsolete).
The consequence of a linear response is that image processing ensures that the
density and contrast of the displayed image appear acceptable to the observer,
whether the exposure is low or high. This is illustrated in Fig. 2.2. The graphs are
histograms showing frequency distributions of digital code values and character-
istic curves that convert digital values into optimized images that all appear with
the same brightness and density. “The VOILUT is adjusted to the histogram to

14 2 Digital Radiography: A Technical Review
Very Good
Optimum
High
Exposure
Image Quality
High Noise
Optimum
Exposure
Low
Exposure
LOW Radiation Exposure to the Detector HIGH
Fig. 2.1 The response of digital detectors to radiation exposure, is linear, compared to a sigmoidal
response of film-screen detectors (which are now obsolete). (See text for further explanation)
Fig. 2.2 The image processing features of DR imaging systems adjust images that are overex-
posed, underexposed, and correctly exposed to all appear visually acceptable in terms of image
gray scale
2.2 Digital Radiography Systems: A Technical Review 15
Fig. 2.3 Low exposures in digital radiography can produce noisy images (a); while high expo-
sures produce excellent noise-free image (b); however, high exposures also increase the dose to
the patient
achieve optimal rendering of the image content” [4]. Low exposure can produce
noisy images (Fig. 2.3a) while high exposures produce excellent noise-free
image (Fig. 2.3b); however, high exposures also increase patient exposure to
radiation [1–3].
2. An exposure indicator (EI), a numerical value is displayed on the image to indi-
cate that the correct exposure technique factors were used for the examination.
The EI is not the dose to the patient. The more current numerical value known as
the deviation index (DI) is now used as a cue for the exposure to the detec-
tor [4–6]
2.2 Digital Radiography Systems: A Technical Review
The physical principles and technology of DR imaging systems have been described
in the literature [1–4, 7]. As noted Sect. 2.1, two popular and commonplace systems
include CR and FPDR. The major system components and processes involved in
creating images were illustrated in Fig. 1.1 (Chap. 1). This section will review the
overall structure and function of each of these two imaging modalities.
2.2.1 Computed Radiography: Physical Principles

and Technology
CR imaging is based on the physical principle of photostimulable luminescence, a

characteristic of photostimulable phosphors. The phosphors typically used in CR
technology are barium fluorohalides (BaFX) where X represents the halide (iodide,
bromide). Another phosphor used is cesium bromide (CsBr).
Figure 2.4 shows an overview of the CR imaging process, from data acquisition
(x-ray exposure of the image receptor); image processing; and image display. In the
first step, image acquisition CR uses an imaging plate (IP) to capture radiation
transmitted through the patient during the examination. The IP is coated with the
appropriate phosphor. When exposed to x-rays, electrons from the valence band
(ground state) of the phosphor, are raised to a higher energy level (the conduction
band) where they are trapped, hence creating a latent image on the IP. To render the
latent image visible, the IP is inserted into a CR image reader, where it is scanned
by a laser beam. This process forces the trapped electrons in the conduction band to
return to the valence band, thus creating a bluish-purple light. Such emission is
called photostimulable luminescence. The light emitted is proportional to energy of
the x-ray photons striking the IP. The light is captured by a photomultiplier and
converts it into electrical signals. These signals are digitized by an analog-to-digital
converter, and subsequently sent to the computer. In the final stage of this process,
the IP is exposed to a high intensity light to remove any residual latent image (plate
erasure). The IP can now be used again for another examination.
The second notable step in the CR imaging process is image processing which
consists of preprocessing and post processing as shown in Fig. 2.4. Processing will
be reviewed briefly in Sect. 2.3.1.
Fig. 2.4 A general overview of the CR imaging process, from data acquisition (x-ray exposure of
the image receptor); to image processing; and image display. (See text for further explanation)
2.2 Digital Radiography Systems: A Technical Review 17
2.2.2 Flat-Panel Digital Radiography: Physical Principles

and Technology
The shortcomings of CR, such as poor x-ray detection efficiency (affects image
quality and dose), poorer spatial resolution compared to film-screen detectors, and
the potential for damage to the IP as it is transported to the CR image reader for
processing; have resulted in the development of other systems to solve these prob-
lems. One such system is the FPDR imaging system. The major components of a
flat-panel detector and associated image processing is illustrated in Fig. 2.5. There
are two types of FPDR detectors, based on how they convert x-rays to electrical
signals; indirect and direct conversion detectors. For more details of the structure
and function of these detectors, the reader should refer to the literature [1–3, 7]. A
brief description of each of these two types of FPDR systems will be presented in
Sects. 2.2.2.1 and 2.2.2.2.
Indirect Conversion Digital Radiography
Figure 2.6 illustrates the major components of a flat-panel digital radiography

system based on indirect conversion detector. First, x-rays fall upon a scintillator
(cesium iodide) and are converted into light using. Secondly. the emitted light
from the phosphor falls on a matrix array of electronic elements consisting of an
amorphous silicon (a-Si) photodiode flat-panel layer, with a thin-film transistor
(TFT) to create and store electrical charges in direct proportion to the x-ray expo-
sure. The charges produce electrical signals which are them subject to image pro-
cessing to produce a FPDR image displayed on a monitor for viewing and
interpretation.
Fig. 2.5 The major components of a flat-panel detector and associated image processing. (See text
for further explanation)
Fig. 2.6 The major components of a flat-panel digital radiography system based on indirect con-
version detector
Fig. 2.7 A schematic of the components of a direct conversion digital radiography system. First,
x-rays fall a photoconductor (selenium) and are converted directly into electrical charges that fall
upon a matrix array of electronic elements and stored for subsequent readout. Secondly, the electri-
cal signals are digitized by the ADC and processed by a digital computer to produce an image
Direct Conversion Digital Radiography
A schematic of the components of a direct conversion digital radiography system is

shown in Fig. 2.7. First, x-rays fall a photoconductor (selenium) and are converted
directly into electrical charges that fall upon a matrix array of electronic elements
and stored for subsequent readout. Secondly, the electrical signals are digitized by
the ADC and processed by a digital computer to produce an image.
2.3 Other Considerations Related to CR and FPDR Systems 19
2.3 Other Considerations Related to CR and FPDR Systems
Dose optimization in DR requires other considerations such as image processing,

the standardized exposure indicator, and the factors affecting dose in DR. Each of
these will be reviewed briefly in the appropriate sections below. Optimization strate-
gies will be described separately in Chap. 3.
2.3.1 Image Processing
Digital image processing is an essential feature of digital radiography imaging sys-

tems, and consists of two processes, namely; image preprocessing and postprocess-
ing which are applied to the original images. The purpose of image processing is to
improve the visibility of different anatomical structures and image densities in an
effort to enhance the diagnostic interpretation skills of the observer (radiologist, for
example). Preprocessing and postprocessing operations have been described in pre-
viously published materials, and readers should consult the literature for more
details [1–3, 7].
In digital radiographic imaging, preprocessing is mainly used to not only to iden-
tify, correct, and scale the raw image data prior to image display for the purpose of
viewing and interpretation. Postprocessing on the other hand is intended to manipu-
late the image displayed on the viewing monitor. Such processing can be applied to
sharpen and reduce noise in displayed images, as well as to alter the brightness and
contrast of the image to suit the viewing needs of the observer.
One such common image processing operation is referred to as windowing, using
controls such as the window width (WW) and the window level (WL) to change the
image contrast and brightness respectively [8]. Another commonly used image post-
processing operation is spatial frequency processing, specifically to change the
sharpness (detail) of an image.
Spatial frequency processing operates on the frequency domain illustrated in
Fig. 2.8. The spatial domain image of the hand is converted into a frequency domain
image using the mathematical rigor of the Fourier Transform (FT). As illustrated the
frequency domain image consists of high frequencies (sharpness) and low frequen-
cies (contrast). This processing operation is intended to change the sharpness of an
image by adjusting the frequency components of the image. Using a high pass digi-
tal filter, low frequencies are suppressed to produce a sharper image, as shown in
Fig. 2.9. Note the sharpness of the wrist bones compared to the original image. The
use of image processing as a dose optimization strategy will be discussed in Chap. 3.
Fig. 2.8 Spatial frequency processing operates on the frequency domain as illustrated. The spatial
domain image of the hand is converted into a frequency domain image using the mathematical
rigor of the Fourier Transform (FT). (See text for further explanation)
Fig. 2.9 An illustration of the use of a high pass digital filter which supresses low frequencies to
produce a sharper image. (See text for further explanation)
2.3.2 The Standardized Exposure Indicator

and the Deviation Index
The Exposure Indicator (EI) and the Deviation Index (DI) are numerical values that
appears on the digital radiography image to indicate to the operator that the correct
exposure technique was used for the examination. In the past, various vendor-
specific EIs were used. Furthermore, two scales were prevalent, namely; an inverse
scale and a proportional scale, as well as different algorithms and detector calibra-
tion techniques. These two scales relate EI to dose to the detector. The inverse scale
states that the EI in inversely proportional to the dose (EI α 1/Dose) that is, if the
detector dose is 5 μGy and the EI value for this dose is 400, then for 10 μGy and
20 μGy, the EI decreases from 200 to 100 respectively. The proportional scale states
that the EI is directly proportional to the dose (EI α Dose), that is, if the EI value for
detector dose of 5 μGy is 500, then the EIs for doses 10 μGy and 20 μGy, are 1000
and 2000 respectively.
These differences created some degree of confusion for operators. Subsequently,
the International Electrotechnical Commission (IEC) [5] and American Association
2.3 Other Considerations Related to CR and FPDR Systems 21
of Physicists in Medicine (AAPM) [9] in collaboration with the Medical Imaging

and Technology Alliance, standardized the EI, to eliminate confusion in clinical
imaging. This standardization requires that operators understand four parameters
for effective use of the standardized EI in clinical practice well. These include the
EI, target EI (EIT), Deviation Index (DI), and Volume of Interest (VOI).
Furthermore, the IEC requires that the standardized EI uses a linear proportional
scale related to the detector exposure/signal, that is, doubling the detector dose,
doubles the standardized EI value. Additionally, there are other conditions that must
be considered [5].
The IEC [5] definitions are as follows:
1. EI is a “measure of the detector response to radiation in the relevant image region
of an image acquired with a digital x-ray imaging system”.
2. EIT is the “expected value of the exposure index when exposing the x-ray image
receptor properly”.
3. DI is a “number quantifying the deviation of the actual exposure index from a
target exposure index”.
4. VOI is the “central tendency of the original data in the relevant image region.
The central tendency is a statistical term depicting generally the center of a dis-
tribution. It may refer to a variety of measures such as the mean, median, or
the mode”.
Presently all digital radiography imaging equipment use the standardized EI. The
following demonstrates the linear proportional scale of the standardized EI:
• If a detector dose of 2.5 μGy generates an EI of 250, then detector doses of
5 μGy; 10 μGy and 20 μGy, will generate EIs of 500; 1000 and 2000 respectively.
Once the EI is generated, the DI which quantifies the deviation of the actual EI
from the expected EIT and is calculated using the following algebraic expression:
DI 10 log10 EI / EI T
The results of this calculation indicate the following [5, 9]:

• A DI number of +1 is equal to an overexposure of 26% more than the desired
exposure.
• A DI number of −1 is equal to an underexposure of 20% less than the desired
exposure.
• A DI number of +3 is equal to 100% more than the desired exposure.
• A DI number of −3 is equal to 50% less than the desired exposure.
• The acceptable range of DI numbers is approximately −1 to +1, and the DR
system is able to deliver the EIT established by the department.
• Furthermore, numbers greater than +1 and less than −1 would indicate gross
overexposure and underexposure, respectively,
2.4 Factors Affecting the Dose in Digital Radiography:

An Essential Review
There are a host of factors in digital radiography that affects the dose to the patient
[1–3, 6, 10]. These include exposure factors (mAs and kV); collimation; antiscatter
grids; automatic exposure control systems; source-to-skin distance; and so forth.
Since the research on dose optimization in digital radiography focuses primarily on
exposure factors, this section will only highlight the essential mathematical rela-
tionship between dose and mAs; and dose and kV. There are at least two important
relationships for the operator to understand between dose and exposure factors, and
these include:
1. The dose is directly proportional to the mAs; doubling the mAs doubles the dose
[1–3, 6, 10]
2. The dose is also directly proportional to the square of the ratio of the kV. This
relationship means that doubling the kV would result in an increase in x-ray
intensity (dose) by a factor of 4 [2]. Furthermore, at 100 kV, the dose is 1.5 times
greater than 80 kV, and 2.5 times higher at 120 kV [10]. This relationship
between dose and kV demonstrates that a relatively small reduction of kV can
result in a substantial reduction in total radiation dose.
The literature also shows that there are other considerations that have an impact on
dose optimization. These include the EI, DI, image processing, and diagnostic refer-
ence levels. As noted by Seeram [11], “each approach has the potential to improve
patient dose levels, and many of these strategies can be combined to further improve
the balance of radiation dose and image quality in digital imaging”. Diagnostic
reference levels will be described briefly in Sect. 2.5.
2.5 Diagnostic Reference Levels
The Diagnostic Reference Level (DRL) is a useful tool for optimization of radiation
protection used to evaluate patient doses in medical imaging [12–14]. In radiation
protection, dose limits have been established by various radiation protection author-
ities. The ICRP for example have recommended dose limits for occupationally
exposed individuals (technologists and radiologists for example), pregnant workers,
and members of the public [2, 3, 6]. No dose limits have been established for
patients. Therefore, the ICRP, as early as 1990, introduced the notion of DRLs to be
used for patients. A significant point to note is that DRLs and dose limits are not the
same thing.
References 23
2.5.1 Definitions and Major Guidelines
The ICRP and the American College of Radiology (ACR) have defined the DRL. The
ICRP’s definition is that DRLs “are a form of investigation level applied to an easily
measured quantity, usually the absorbed dose in air or tissue equivalent material at
the surface of a simple standard phantom or a representative patient” [12].
Additionally, the ICRP states that DRLs “are not for regulatory or commercial pur-
poses, not a dose restraint and not linked to limits or constraints” [12]. The ACR has
defined the DRL as “an investigation level to identify unusually high radiation dose
or exposure levels for common diagnostic medical x-ray procedures” [15].
DRLs will not be described further in this text and the reader should refer to the
literature [6, 12, 15] for more details. A few major guidelines from the ICRP how-
ever, are noteworthy for the purposes of this chapter. DRLs are an advisory, and not
a regulatory measure, and are not related to dose limit. Their purpose is to identify
high radiation doses delivered to patients using easy to measure dose quantities such
as for example, skin entrance exposures. A significant feature of the DRL is that its
selection is established by professional organizations using a percentile point. A
notable method is to use a diagnostic reference range, where the upper limit of the
range is established at the 75th percentile, and the lower limit is set at the 25th per-
centile of the calculated patient dose. Furthermore, while the levels below the 25th
percentile may compromise image quality, levels above the 75th percentile may
reflect excessive dose, and therefore the potential for dose optimization must be
considered in this respect.
The ACR-AAPM-SPR [15] for example, recommends a DRLs for an adult PA
chest (23 cm thickness) with grid; adult AP abdomen (22 cm thickness) and an adult
AP lumbosacral spine (22 cm thickness) of 0.15 mGy; 3.4 mGy and 4.2 mGy
respectively. Readers should consult the guidelines of their respective countries for
established DRL values for medical x-ray examinations.
References
1. Seeram E (2019), Digital Radiography: Physical Principles and Quality Control. Second
Edition, Singapore, Springer.
2. Bushong S (2021). Radiologic Science for Technologists: Physics, Biology and Protection.
12th ed. St Louis, MO: Mosby-Elsevier Inc.
Imaging. 4th ed. Philadelphia, PA: Lippincott Williams & Wilkins.
4. Seibert, J.A., Morin, R.L (2011). The standardized exposure index for digital radiography: an
opportunity for optimization of radiation dose to the pediatric population. Pediatr Radiol; 41,
573–581. https://doi.org/10.1007/s00247-010-1954-6 PubMed
5. International Electrotechnical Commission (2008). IEC 62494-1 ed.1 Medical electrical

equipment exposure index of digital x-ray imaging Systems part 1: definitions and require-
ments for general radiography. Geneva, Switzerland (2008)
7. Carter CE, Vealé BL (2019) Digital Radiography and PACS. 3rd Edition. St Louis,
MO. Elsevier, Inc
8. Seeram E (2023). Computed Tomography: Patient Care, Clinical Applications, and Quality
Control. St Louis, MO. Elsevier
9. Shepard, J. S., Wang, J., & Flynn, M., et al. (2009). An exposure indicator for digital radiogra-
phy: AAPM Task Group 116 (executive summary). Med Phys 36(7), 2898–2914.
10. Sharma R, Sharma SD, Pawar S, Chaubey A, Kantharia S, Babu DA (2015. Radiation dose to
patients from X-ray radiographic examinations using computed radiography imaging system.
J Med Phys. Jan-Mar; 40(1): 29–37. doi: https://doi.org/10.4103/0971-6203.152244
11. Seeram, E (2022). Dose Optimization in Digital Radiography. Radiologic Technology; Directed
Readings Website, American Society of Radiologic Technologists (ASRT), New Mexico
12. ICRP Publication 135 (2017). Diagnostic Reference Levels in Medical Imaging. Annals of the
ICRP Volume 46, Issue 1, October, Pages 1–144
13. Seeram, E. and Brennan, P (2006). Diagnostic Reference Levels in Radiology. Radiologic
Technology, May/June 2006, Vol. 77/No. 5; 373–384
14. Seeram, E. (2020). Rad Tech’s Guide to Radiation Protection, 2e. Oxford: Wiley, 2020
15. ACR-AAPM-SPR (2018) Practice Parameter for Diagnostic Reference Levels and Achievable
Doses in Medical X-Ray Imaging. https://www.acr.org/-/media/ACR/Files/Practice-
Parameters/Diag-Ref-Levels.pdf Accessed June 2022.
Chapter 3
Optimization Strategies in Digital
Radiography
Keywords Dose optimization in digital radiography · Exposure technique factors

· kV · mAs · Exposure indicator · Deviation index · Image post processing ·
Multifrequency processing · Noise reduction algorithms · Diagnostic
reference levels
3.1 Introduction
As reviewed in Chap. 1, optimization is a radiation protection principle, first pro-

vided by the International Commission on Radiological Protection (ICRP) [1] and
now is a mandatory requirement of various radiation protection organizations
around the world. In summary, during the conduct of the examination, optimization
entails not only adjusting the “imaging parameters” [1], but also to “institute protec-
tive measures” [1]. These two tasks are intended to ensure that the first part of opti-
mization, the ALARA (as low as reasonably achievable) principle is observed. The
second part of optimization, as noted by the ICRP [2], involves at least four ele-
ments, namely; the imaging equipment, the adequacy of the imaging equipment,
technical parameters of the examination, and diagnostic reference levels.
This Chapter will outline several strategies that have been used successfully in
digital radiography, to optimize the dose to the patient without compromising image
quality needed for diagnosis. These strategies include optimization of the exposure
technique factors (mAs and kV); the Exposure Indicator (EI); the Deviation Index
(DI); image post processing algorithms; and finally, the use of diagnostic reference
levels [3]. As noted by Seeram [3], “each approach has the potential to improve
patient dose levels, and many of these strategies can be combined to further improve
the balance of radiation dose and image quality in digital imaging”.

26 3 Optimization Strategies in Digital Radiography
3.2 Optimization of Exposure Technique Factors
The evolution and continued development of digital radiography imaging systems

can be traced back to the introduction of Computed Radiography (CR) imaging
systems to replace film screen radiography (FSR). The limitations imposed by CR
lead to the introduction of Flat-Panel Digital Radiography imaging systems which
includes indirect conversion imaging systems and direct conversion imaging sys-
tems [4]. FSR has now been replaced by digital radiography imaging systems.
The correct use of these digital imaging systems in clinical practice provided the
motivation conduct research on several areas of performance. A significant topic for
research in the early stages of use focussed on the correct exposure technique fac-
tors to use for various examinations, since users were well versed in exposure tech-
nique factors for FSR. Section 3.3 of this Chapter will explore the optimization of
exposure technique factors.
3.2.1 Optimization of kV and mAs
One of the earliest reviews entitled “Optimization research exposure technique

approaches in CR imaging” [5] assessed three major optimization approaches,
namely, optimization of kV; optimization of mAs; and optimization of the Exposure
Indicator (EI) in practice. This section will focus on the kV and mAs optimization
approaches. The EI optimization will be discussed in Sect. 3.2.
The basis for exploring the optimization of the kV in CR imaging relates to the
response of the CR detector to the beam energies used in diagnostic radiology com-
pared to that of FSR detector [6].
The k-edge for a typical CR phosphor such as BaFBr/I (barium fluorobromide/
iodide) is about 37 keV (compared to FSR detector such as Gd2O2S which has a
k-edge of about 50 keV). Moore et al. suggest therefore “that lower tube voltages
relative to those used for film-screen should be used for CR” [6]. This physics fact
provided the motivation for research on optimization of the kV in CR imaging. With
this in mind, the literature is replete with studies focusing on optimization of kV
using different CR imaging systems and different body parts to the extent that these
studies have produced “conflicting results” [6, 7]. This was followed by additional
research to include indirect and direct flat panel detectors.
A study by Sun et al. [8] on “Optimization of chest radiographic imaging param-
eters: a comparison of image quality and entrance skin dose for digital chest radi-
ography systems”, the researchers explored the use of kV values of 100, 110, and
120, and mA settings of 1, 2, 4, 8, and 10. The found that a reduction of kV from
120 to 100 kV resulted in a dose reduction of up to 44% and without compromising
image quality. A recent study by Moore et al. [9] investigated the use of an optimum
range of kV for imaging the abdomen and pelvis. Their results showed that while
for the abdomen a range of 70 to 120 kV was optimum, a range of 80 to 120 kV was
3.3 Optimization of the Exposure Indicator 27
the optimum for imaging the pelvis. Furthermore, these investigators also demon-
strated that an optimum range for imaging the lumbar spine was of 80 to 120 kV [9].
These studies suggest that the kV can be reduced without affecting the image qual-
ity necessary for diagnosis [8, 9].
The need to explore optimization of the mAs in digital radiography stems from
the fact that “manipulation of the operating kV cannot stand alone even with digital
systems, and concomitant compensation of the applied mAs, together with adequate
scatter control are necessary” [10]. This notion suggests that optimization must con-
sider evaluating image quality with respect to the dose per image. An increase in the
dose per image will reduce the noise, hence image quality improves [11].
Optimization should therefore consider the dose per image in assessing image
quality.
The literature suggests that this approach, that is using the lowest mAs settings
“without excessive noise degradation of image quality” in optimizing the dose-
image quality in CR Imaging [10]. In this regard, Fauber et al. [7] investigated the
“insufficient and excessive radiation exposure on CR image quality”, using five
images at exposure groups of 1 mAs; 2 mAs; 4 mAs; 8 mAs (baseline value);
16 mAs; 32 mAs; 64 mAs; and 125 mAs. The researchers explained that “the mAs
value to achieve an optimal quality QC phantom image (line pairs) was ultimately
decreased by 300% for the lowest mAs value and increased 400% for the highest
mAs value”. An important finding of this study is that at low exposures it was dif-
ficult to see the line pairs/mm compared to the baseline value of 8 mAs, and image
mottle (noise) was pronounced at low exposures below the baseline value
of 8 mAs.
The above studies are intended to set the stage for further research into the opti-
mization of exposure technique factors, as digital radiography detectors continue to
be developed with the goal of improving performance in clinical imaging of patients.
3.3 Optimization of the Exposure Indicator
As noted by Seeram et al. [5], “Optimization of the EI is closely linked with optimi-
zation of exposure technique factors (kVp and mAs)”. The EI was described in
detail in Chap. 2 [3, 4, 5, 12].
In summary, the EI is a numerical parameter devised by CR manufacturers to
provide the user with a visual cue as to the amount of exposure to the CR IP. It pro-
vides the technologist with an indication of whether the appropriate exposure tech-
nique factors are used for the particular x-ray examination. It is quite important to
understand that the EI is not the patient dose. Patient dose depends on several fac-
tors such as kV, mAs, beam filtration, Source-to-Image Receptor Distance (SID),
body part being imaged, and collimation, for example.
The literature on optimization of the EI is extremely sparse in this area of
research. Two notable early studies that set the pace for further research in this con-
text are one by Peters and Brennan [13], and the other by Warren-Forward [14].
While the former study addressed the notion of an optimized EI as a dose manage-
ment strategy via establishing an optimized mAs; the latter sought to explore the use
of the EI by radiographers in clinical practice, as well as exposure creep, and the
relationship between EIs and radiation dose. The fundamental conclusions of these
two studies are as follows:
1. Peters and Brennan [13], reported that “although dose levels to the patient have
not been directly measured, it is safe to assume that a strong correlation exists
between higher exposure indices and higher patient dose”.
2. Warren-Forward [14], (who used two Kodak CR systems) reported that “a small
increase in the EI produced a large increase in the entrance surface dose …..an
EI of 2000 produced at 125 kVp can deliver the same patient dose as an EI of
1700 produced at 70 kVp, where the EI difference of 300 represents a doubling
of the dose to the detector”
The results of these two studies prompted ideas for future investigations and in this
regard, Fauber et al. [7], to conclude in their study that “further research is neces-
sary to understand the significance of the exposure indicator and its relationship to
exposure techniques and patient exposure”.
To explore the optimization of the EI, Seeram et al. [15], designed a noninterven-
tional descriptive correlational research study to correlate the EI with the entrance
skin dose (ESD) using a range of mAs values, while holding the kilovoltage peak
(kV) constant. The study compared the optimized EI with the manufacturer’s rec-
ommended values for the antero-posterior (AP) pelvis and AP lumbar spine
projections.
The methodology used in this study, involved the following steps using the ESD
and EI as dependent variables, and mAs as the independent variable:
1. The ESD for the AP pelvis and the AP lumbar spine using the digital radiography
vendor’s (Fuji) suggested exposure technique factors including mAs and kV.
2. The ESD measurements were measured with a calibrated dosimeter (Unfors
Instruments) free-in-air (ie, an air environment free of any absorbing or scatter-
ing objects) for the AP pelvis and the AP lumbar spine.
3. The mean milligray (mGy) per mAs value was calculated for all mAs values for
the AP pelvis and AP lumbar spine.
4. Descriptive statistics (eg, such as sample size, mean, standard deviation, and
range) were used for the dosimetry and EI data.
5. The Pearson correlation was applied to examine the relationship between the
ESD and the EI as well as the ESD and mAs.
6. The results were graphed and plotted showing the mean dose for the AP pelvis
and the AP lumbar spine as a function of mAs (Fig. 3.1); the inverse EI; the mAs;
and the doses were compared. The inverse EI shows a strong positive linear rela-
tionship (r = 0.999) for both the AP pelvis and the AP lumbar spine.
7. At doses of 5 mGy, 10 mGy, and 20 mGy, the EIs were 400, 200, and 100,
respectively.
3.3 Optimization of the Exposure Indicator 29
Fig. 3.1 Graphs of the mean dose in mGy plotted as a function of mAs, for the AP pelvis and the
AP lumbar spine
Fig. 3.2 The Reference image of the AP pelvis obtained at 25 mAs with a dose of 3.4 mGy. The
optimized image recorded at a reduction of 16 mAs at 2.09 mGy, without loss of image quality.
The noise test tool on the right and left of the image set shows the visual appearance of the mottle
or graininess of the images
Furthermore, the dosimetry phase of this experiment showed a strong positive linear
relation-ship (r = 0.999) between mAs and dose, mAs and the inverse EI, and the
inverse EI and dose for the AP pelvis and AP lumbar spine. It is important to note
that under the controlled conditions used in this study, the EI values were stable.
The vendor’s reference values of 25 mAs (EI = 86) for the AP pelvis and 50 mAs
(EI = 88) for the AP lumbar spine {were optimized to 16 mAs for the AP pelvis
(EI = 136)}as shown in Fig. 3.2; and 32 mAs for the AP lumbar spine (EI = 139), as
shown in Fig. 3.3.
Expert observers’ assessment of image quality showed that the vendor’s recom-
mended dose can be reduced by 36% for the AP pelvis and AP lumbar spine without
compromising image quality.
This study clearly demonstrates that operators can optimize the dose and image
quality in computed radiography imaging system using the mAs and associated EIs.
Fig. 3.3 The Reference image of the AP lumbar spine obtained at 50 mAs with a dose of 6.36 mGy.
The optimized image recorded at a reduction of 32 mAs at 4.07 mGy, without loss of image qual-
ity. The noise test tool on the right and left of the image set shows the visual appearance of the
mottle or graininess of the images
3.4 Optimization of the Deviation Index
The deviation index (DI) is an important parameter in the standardized EI paradigm,

and it has been described in detail in Chap. 2. In summary, the DI is related to a
target EI (EIT) which must be established by the imaging department using the
ALARA (as low as reasonably achievable) principle. The DI indicates the amount
of deviation between the actual EI and the EIT and is shown on the final image as a
measure of dose optimization.
The literature is sparse on research studies on the use of the DI as an optimization
strategy, however, only three studies will focus attention on the notion of using the
DI as an optimization strategy in digital radiography with the ALARA principle in
mind. The first study is one entitled “Optimising default radiographic exposure fac-
tors using Deviation Index” by Creeden and Curtis [16], and sets the stage for fur-
ther research on using the DI. Having performed an analysis of the exposure logs for
six DR units on five different examinations, the results showed that by reducing the
mAs values for these examinations were reduced by between 22% and 50% using
“multiple optimisation cycles the number of examinations outside the vendor’s sug-
gested optimal range had decreased by 28.4%”. The researchers concluded that
there was “a marked reduction in patient doses can be achieved through a depart-
mental programme of DI value monitoring and targeted optimisation of default
exposure settings” [16].
The second study is one examining the efficient use of the DI in optimizing dose
using keywords such as “dose indicator; dose optimization; and radiation exposure”
[17]. The results showed inefficient use of both the EIT (target EI set by the imaging
department) and the DI. This study provides the motivation for yet another investi-
gation of the use of the DI in optimizing the dose in digital radiography, since the
researchers point out that the implications for practice is that “Current
3.5 Optimization Using Image Postprocessing Algorithms 31
recommendations on targeting the mean DI of 0 need to be reinforced. Theoretical

knowledge and training need to be improved” [17].
The third and final study in this chapter, is one is one by Park et al. [18], that
examined the use of the EI, EIT and DI as tools for patient dose optimization/Part of
the methodology involved analyzing how the DI changes with EIT established by the
department, based on the Korean national DRLs. for posterior-anterior chest, a lat-
eral and anterior-posterior of the abdomen in clinical practice. The conclusion of
this study suggests that “as the exposure conditions and DRLs varied, the clinical
EI, EIT and DI also varied. These results reveal that the clinical EI, EIT and DI can
be used as tools for optimizing the patient dose if EIT is periodically and properly
updated” [18].
In addition to the above strategies, the use of image processing techniques and
noise reduction algorithms (to improve visualization of not only various anatomical
structures but also pathology); have been explored as optimization strategies, add-
ing to the traditional optimization strategies for digital radiography.
3.5 Optimization Using Image Postprocessing Algorithms
In Chap. 2, image processing was described briefly to set the stage for using it as an
optimization tool in digital radiography. Image processing is a complex topic and
will not be described in any detail in this book, however, an understanding of the
following aspects of image processing are essential to aid its use as an optimization
strategy: Why image processing of digital radiography images? And two major pro-
cesses of image processing.
Image processing for Radiologic Technologists has been described in the litera-
ture [19, 20]. This topic is now a part of radiologic technology curriculum that ii has
been described by Seeram in a textbook on Digital Radiography [4]. Image process-
ing in Digital Radiography consists of two processes, namely preprocessing and
postprocessing, as illustrated in Fig. 3.4.
Preprocessing operations are used to identify, correct, and scale the raw image
data apply appropriate corrections, such as artifact corrections for example, to the
Fig. 3.4 Image processing in digital radiography consists of two processes, namely preprocessing
and postprocessing. (See text for further explanation)
Fig. 3.5 The overall goal of an algorithm referred to as Flat-Field Correction is to correct for
artifacts in the raw data before the image is displayed for viewing by an observer
raw data. Each DR vendor offers proprietary algorithms for their systems and there-
fore, it is not within the scope of this book to describe these algorithms. One such
algorithm however, is a standard calibration operation known as flat-field correction
as illustrated in Fig. 3.5.
Postprocessing, on the other hand consists of a wide range of image processing
operations intended to change the image contrast, reduce image noise and enhance
the sharpness of the image displayed in an effort to enhance diagnostic interpreta-
tion. These include point processing operations such as grayscale processing (win-
dowing, image subtraction, and temporal averaging), local processing operations
(such as, spatial filtering, edge enhancement, and smoothing), and global operations
such as the Fourier transform (FT) [4, 21, 22]. The use of the FT coupled with a high
pass digital filter to sharpen an image is shown in Fig. 2.9. Examples of noise con-
trol image postprocessing algorithms include Fujifilm Flexible Noise Control (FNC)
and Multi-objective Frequency Processing (MFP); Philips Healthcare-UNIQUE
(UNIfied Image QUality Enhancement); and the advanced noise reduction image
processing engine (S-Vue™; Samsung Healthcare).
3.5.1 Optimization Studies Using Multifrequency Processing

and Noise Reduction Algorithms
There are several examples of image postprocessing as an optimization strategy in

digital radiography [3] and the general findings show that image postprocessing
algorithms, namely multifrequency processing can be successfully used as an opti-
mization strategy by demonstrating that a low-dose image with poor image quality
(eg, high noise) can be processed using specialized software to improve visibility of
structures [23–27].
This sub-section of this chapter will review briefly the studies by Precht et al. [23];
Lee et al. [25]; Feghali et al. [26]; and by Oh et al. [27], as summarized in an article in
Radiologic Technology: Essential Education Directed Readings article by Seeram [3].
The American Association of Physicists in Medicine (AAPM) [24], explains that

multifrequency processing software is designed to divide an image into separate
frequency ranges for processing before they are recombined into an enhanced out-
put image. In this subsection, a number of direct quotes will be cited from the
research literature, so as not to detract from the investigators’ original meaning.
As early as 2012, Precht et al. [23], explored the use of multifrequency process-
ing as an optimization strategy in digital radiography in pediatric radiographic
imaging. Using an anthropomorphic phantom, a total of 110 digital radiography
images were obtained with a Canon DR imaging system with a CXDI-50 C detector
and MLT[S] software. Furthermore, 3500 images taken of a contrast-detail phantom
(CDRAD 2.0- Artinis Medical Systems [23]) were obtained. While the former set of
images were assessed by three pediatric radiologists using Visual Grading Analysis,
an image quality assessment tool, the latter image sets were evaluated to provide an
objective image-quality assessment. The results of this investigation showed that
“optimal image-quality was maintained at a dose reduction of 61% with a multi-
Laplacian transformation sub-band pyramidal enhancement-technology [MLT(S)]
optimized images. Even for images of diagnostic quality, MLT(S) provided a dose
reduction of 88% as compared to the reference image. Software impact on image
quality was found significant for dose (mAs), dynamic range dark region and fre-
quency band” [23]. The researchers concluded that by optimizing image processing
parameters, a significant dose reduction is possible without significant loss of image
quality [23].
In another study utilizing multifrequency processing as an optimization strategy
is one by Lee et al. [25], entitled “Radiation dose reduction and improvement of
image quality in digital chest radiography by new spatial noise reduction algo-
rithm”. The researchers used an advanced noise reduction image processing engine
(S-Vue™-Samsung Healthcare). Figure 3.6 illustrates the basic comparison between
a conventional algorithm (S-Vue™ 3.00) and new image processing algorithm
(S-Vue™ 3.02). The methodology involved 69 patients, each of whom was exposed
to a baseline dose of 4.2 μGy using conventional image processing engine. Secondly,
images were obtained using a dose reduction protocol of 2.5 μGy; 2.11 μGy; and
1.78 μGy using the new image processing engine. Images were assessed by two
thoracic radiologists on “radiolucency of unobscured lung, pulmonary vascularity,
trachea, edge of rib, heart border, intervertebral disc space, and pulmonary vessels
in the retrocardiac area” and scored on a 5-point scale of these anatomical land-
marks. The results showed that “the image quality of the low-dose image was not
inferior to that of the baseline dose image even if the maximum average dose reduc-
tion rate was reduced to 47.8% of the baseline dose” [25]. Example images with
optimized doses are shown in Figs. 3.7 and 3.8.
In 2021, Feghali et al. [26], published a dose optimization study in Diagnostic
Interventional Imaging entitled “New image quality and dose reduction technique
for pediatric digital radiography”. In this study, the S-Vue™ (Samsung
Healthcare) advanced noise reduction image processing engine was used. Having
obtained images from 174 pediatric patients; “artificial noise was added to the
images to simulate acquisitions at 50%, 32% and 12.5% of the routine dose levels.
Fig. 3.6 A schematic illustration comparing the conventional and new algorithms of a dose opti-
mization study. (Reproduced with permission under the Creative Commons Attribution 4.0
International license. Lee W, Lee S, Chong S, Lee K, Lee J, Choi JC, et al. Radiation dose reduc-
tion and improvement of image quality in digital chest radiography by new spatial noise reduction
algorithm. PLOS ONE. 2020; 15 (2): e0228609. creativecommons.org/licenses/by/4.0/)
Fig. 3.7 Chest radiographs demonstrating a well-defined, small nodule in right upper lung(arrows).
A. Base-line image with a standard dose of 61.9 μGy ESE. B. Optimized, low-dose image of
30.2 μGy ESE. (Reproduced with permission under the Creative Commons Attribution 4.0
International license. Lee W, Lee S, Chong S, Lee K, Lee J, Choi JC, et al. Radiation dose reduc-
tion and improvement of image quality in digital chest radiography by new spatial noise reduction
algorithm. PLOS ONE. 2020; 15 (2): e0228609. creativecommons.org/licenses/by/4.0/)
Fig. 3.8 The two chest radiographs show a subsegmental consolidation in a retrocardiac region of
a left lower lung zone (arrows). For image A, the radiation dose of ESE was 54.2 μGy. The opti-
mized image in B recorded with 25.0 μGy, which was reduced to 46.1% of the baseline dose.
(Reproduced with permission under the Creative Commons Attribution 4.0 International license.
Lee W, Lee S, Chong S, Lee K, Lee J, Choi JC, et al. Radiation dose reduction and improvement
of image quality in digital chest radiography by new spatial noise reduction algorithm. PLoS
ONE. 2020; 15 (2): e0228609. creativecommons.org/licenses/by/4.0/)
A total of 696 images corresponding to four dose levels were post-processed

using S-Vue™ and further blindly scored by three pediatric radiologists using a
scoring grid of 4-6 criteria specifically defined per anatomical area” using appro-
priate statistical analysis [26]. The results showed the use of the S-Vue™ image
post-processing engine (software) reduced the dose by a factor of two, without
compromising the diagnostic image quality in pediatric digital radiographic
imaging.
A fourth dose optimization study in digital radiography is one by Oh et al. [27],
using an advanced spatial noise reduction algorithm (ASNR) in pediatric digital
radiography. Having obtained 9 sets of 30 images at “different levels of low-dose
image sets of these 270 images were generated by a noise simulation tool after
validation testing using phantoms. Each image set was obtained with both the
ASNR and conventional algorithm, and grouped randomly and blinded”. Image
assessment required that each of three radiologists select the “image with opti-
mum dose” keeping the ALARA principle in mind. The results showed that the
“average of the calculated ESE (entrance skin exposure) was lower with the
ASNR algorithm than with the conventional algorithm group” demonstrating that
the ASNR algorithm can be a useful tool in dose optimization in digital radiogra-
phy of pediatric patients.
3.6 Diagnostic Reference Levels: A Useful Tool

in Dose Optimization
A recent article by Damilakis and Vassileva [28] entitled “The growing potential of
diagnostic reference levels as a dynamic tool for dose optimization” and published
in Physica Medica, pointed out that Diagnostic Reference Levels (DRLs) are useful
tools in optimization od radiation protection in medical imaging. DRLs have been
identified as useful tools in radiation protection of patients in medical imaging [29]
and as such, the literature is replete with articles on the DRLs, in terms of what they
are; why they are needed; how to establish DRLs; and DRL values for various digi-
tal radiography examinations.
The concept of the DRL was first introduced by the ICRP in 1990 [30], based on
the fact that several studies have revealed large fluctuations in patient doses for the
same types of examinations [31]. These fluctuations varied between high levels to
low levels of radiation doses. DRLs are intended to focus attention on how to
address these variations. It is not within the scope of this chapter to describe details
of the DRL, such as for example how DRLs are established, and readers should
refer to the ICRP publication of DRLs [30]. The following summary points are
essential for a reasonable understanding of the DRL:
1. The definition of the DRL has been provided by the ICRP and the American
College of Radiology (ACR). While the ICRP definition states that DRLs “are a
form of investigation level, applied to an easily measured quantity, usually the
absorbed dose in air, or tissue-equivalent material at the surface of a simple stan-
dard phantom or a representative patient” [30], the ACR defines a DRL as “an
investigation level to identify unusually high radiation dose or exposure levels
for common diagnostic medical x-ray procedures.” [32]. The DRL is seen as an
advisory measure and not a regulatory one, and it is not related to dose limits
established for occupationally-exposed individuals (radiation workers), and
members of the public.
2. When dealing with the optimization of radiation protection, the DRL addresses
the issue of patient dose, and it is intended to identify high patient doses for vari-
ous common examinations using specific imaging equipment. It is also recom-
mended that when conducting a DRL study, the dose quantities and procedure
used to measure the doses should be easy (simple). The entrance skin exposure
(ESE) is a suggested dose quantity for example.
3. Professional organizations, such as the AAPM, ACR, the National Council on
Radiation Protection and Measurements (NCRP), the Canadian Association of
Radiologists (CAR) in North America for example, should establish DRLs,
using a percentile point on the observed distribution for patients, and specific to
a country or region. One pragmatic method is to use what has been referred to as
a diagnostic reference range. While the upper level of the range is established at
75th percentile, the lower level is set at 25th percentile of the calculated patient
dose. Below a 25th percentile level, image quality may be compromised, above
the 75th percentile may indicate excessive dose.
References 37
4. For actual DRL values for various imaging examinations, the reader should con-
sult their respective national radiation protection organizations. The ACR for
example provides DRLs for the PA chest and AP abdomen to be 0.22 mGy and
5.3 mGy respectively [32].
As described in the above sections, optimization in digital radiography involves a
number of significant strategies such as optimization of the exposure technique fac-
tors (kV/mAs); exposure indicator; deviation index; use of multifrequency process-
ing and noise reduction algorithms; and finally, the use of DRLs as a tool in the
optimization of radiation protection.
Following a literature review not only to “identify the reasons why optimization
is important and the steps to be followed for a successful optimization process in
digital radiology, Computed Tomography, interventional radiology and mammogra-
phy”, Tasapaki [29] proposed that there are five steps to a dose optimization pro-
cess, including:
1. “establishment of a quality assurance programme; a mistake, misuse or malfunc-
tion of an X-ray machine can potentially affect the health or life of thousands
of people,
2. establishment of a dose optimization team consisting of a radiologist, medical
physicist and radiation technologist,
3. determination of baseline dose levels and image quality as well as comparisons
with benchmarks to decide which exam protocols should be optimized,
4. modification of protocols by the medical physicist and,
5. evaluation of the optimization process and its effect on patient dose and image
quality” [29].
It is important to stress that in the optimization of dose and image quality in medical
imaging, the technologist must be an active participant in the process.
References
1. International Commission on Radiological Protection (2007). The 2007 Recommendations

of the International Commission on Radiological Protection. Orlando Florida, Elsevier. ICRP
Publication 103. Ann ICRP; 37: 1–332.
2. ICRP (2013). Radiological protection in paediatric diagnostic and interventional radiology.
ICRP Publication 121. Ann. ICRP 42(2).
3. Seeram E (2022). Dose Optimization in Digital Radiography. American Society of
Radiological Technologists (ASRT), Essential Education; Pages 1–14. https://apps.asrt.org/
DirectedReading/DirectedReading.aspx (accessed June 30, 2022)
4. Seeram E (2019). Digital Radiography: Physical Principles and Quality Control. 2nd Edition.
Springer Nature Singapore Pte Ltd. https://doi.org/10.1007/978-981-13-3244-9
5. Seeram E, Davidson R, Bushong S, Swan H (2013). Radiation dose optimization research:
Exposure technique approaches in CR imaging: A literature review, Radiography; 19: 331e338
6. Moore CS, Saunderson JR, Beavis AW (2008). Investigating the exposure class of a com-
puted radiography system for optimisation of physical image quality for chest radiography.
The British Journal of Radiology; 82(981):771e7
7. Fauber TL, Cohen TF, Dempsey MC (2011). High kilovoltage digital exposure techniques and
patient dosimetry. Radiologic Technology; 82(6):501e9.
8. Sun Z, Lin C, Tyan Y, Ng KH (2012). Optimization of chest radiographic imaging parameters: a
comparison of image quality and entrance skin dose for digital chest radiography systems. Clin
Imaging. 2012 Jul–Aug;36(4):279–86. doi: https://doi.org/10.1016/j.clinimag.2011.09.006
Epub; Jun 8. PubMed
9. Moore CS, Wood T, Balcam S, Needler L, Guest T, Ngu WP, Chong LW, Saunderson J, Beavis
A (2020). Optimisation of tube voltage range (kVp) for AP abdomen, pelvis and spine imaging
of average patients with a digital radiography (DR) imaging system using a computer simula-
tor. Br J Radiol. Oct 1;93(1114):20200565. Doi: https://doi.org/10.1259/bjr.20200565 Epub
2020 Aug 12. PubMed; PMCID: PMC7548356
10. Matthews K, Brennan PC (2009). Optimisation of X-Ray examinations: General principles
and an Irish perspective. Radiography; 15:262e8.
11. Marshall NW (2001). Optimisation of dose per image in digital imaging. Radiation Protection
Dosimetry; 94 (1e2):83e7.
12. Seibert JA (2008). Digital radiography: image quality and radiation dose. Health Phys.
Nov;95(5):586–98) PubMed https://doi.org/10.1097/01.HP.0000326338.14198.a2
13. Peters SE, Brennan PC (2002). “Digital radiography: are the manufacturers” settings too high?
Optimisation of the Kodak digital radiography system with the aid of the computed radiogra-
phy dose index. European Radiology; 12 (9):2381e7.
14. Warren-Forward H, Arthur L, Hobson L, Skinner R, Watts A, Clapham K, et al (2007). An
assessment of exposure indices in computed radiography for the posterioranterior chest and
the lateral lumbar spine. The British Journal of Radiology; 80 (849):26e31.
15. Seeram E, Davidson R, Bushong S, Swan H (2016). Optimizing the Exposure Indicator as a
Dose Management Strategy in Computed Radiography. Radiologic Technology, March/April,
Volume 87, Number 4; 380–391 PubMed
16. Creeden A, Curtis M (2020). Optimizing default radiographic exposure factors using Deviation
Index. Radiography; 26: 308–313 PubMed https://doi.org/10.1016/j.radi.2020.02.009
17. Guðjónsdóttir J, Paalsson KE, Sveinsdóttir GP (2021). Are the target exposure index and
deviation index used efficiently? Radiography (Lond). Aug;27(3):903–907. doi: https://doi.
org/10.1016/j.radi.2021.02.012. Epub 2021 Mar 9. PMID: 33707050.
18. Park H, Yoon Y, Kim J, Kim J, Jeong H, Tanaka N, Morishita J (2020). Use of clinical exposure
index and deviation index based on national diagnostic reference level as dose optimization
tools for general radiography in Korea. Radiat Prot Dosimetry; Nov 17: ncaa185. doi: https://
doi.org/10.1093/rpd/ncaa185. Epub ahead of print. PMID: 33201240.
19. Seeram E (2004). Digital image processing. Radiol Technol. Jul–Aug;75 (6):435–52; 453–5.
PMID: 15352557.
20. Seeram E, Seeram D (2008). Image Postprocessing in Digital Radiology-A Primer for
Technologists. J Med Imaging Radiat Sci;39 (1):23–41. doi: https://doi.org/10.1016/j.
jmir.2008.01.004. Epub 2008 Mar 22. PMID: 31051771.
21. Gonzalez RC, Woods RE (2018). Digital image processing. 4th ed. Toronto, Ontario: Pearson.
22. Baxes GA (1994). Digital image processing: principles and applications. New York: John
Wiley & Sons, Inc.
23. Precht H, Gerke O, Rosendahl K, Tingberg A, Waaler D (2012). Digital radiography: optimiza-
tion of image quality and dose using multi-frequency software. Pediatr Radiol; 42(9): 1112–8.
doi: https://doi.org/10.1007/s00247-012-2385-3. Epub 2012 Apr 17. PubMed
24. American Association of Physicists in Medicine (2006). Acceptance testing and quality con-
trol of photostimulable storage phosphor imaging systems. Report No 93. College Park, MD
25. Lee W, Lee S, Chong S, Lee K, Lee J, Choi JC, et al (2020). Radiation dose reduction and
improvement of image quality in digital chest radiography by new spatial noise reduction algo-
rithm. PLoS ONE 15 (2): e0228609. https://doi.org/10.1371/journal.pone.0228609 PubMed
26. Feghali JA, Chambers G, Delépierre J, Chapeliere S, Mannes I, Adamsbaum C (2021) New
image quality and dose reduction technique for pediatric digital radiography. Diagn Interv
References 39
Imaging;102(7–8):463–470. doi: https://doi.org/10.1016/j.diii.2021.01.009. Epub 2021 Feb

3. PubMed
27. Oh S & Kim JH, Yoo SY, Jeon T, Kim Y. (2021). Evaluation of the image quality and dose
reduction in digital radiography with an advanced spatial noise reduction algorithm in pediat-
ric patients. European Radiology. https://doi.org/10.1007/s00330-021-07942-6. PubMed
28. Damilakis J, Vassileva J (2021). The growing potential of diagnostic reference levels as a
dynamic tool for dose optimization, Physica Medica, Volume 84, Pages 285–287, https://doi.
org/10.1016/j.ejmp.2021.03.018
29. Tsapaki V (2020). Radiation dose optimization in diagnostic and interventional radiology:
Current issues and future perspectives, Physica Medica, Volume 79, Pages 16–21, https://doi.
org/10.1016/j.ejmp.2020.09.015
ICRP Volume 46, Issue 1, Pages 1–144
32. American College of Radiology ACR-AAPM-SPR (2018). Practice parameter for diagnostic
reference levels and achievable doses in medical x-ray imaging. https://www.acr.org/−/media/
ACR/Files/Practice-Parameters/Diag-Ref-Levels.pdf (accessed 30 August 2020)
Chapter 4
Computed Tomography: A Technical
Review
Keywords Computed tomography · Physical principles · Data acquisition · Image

reconstruction algorithms · Iterative reconstruction algorithms · Deep learning
image reconstruction algorithms in CT · Image display · Image storage ·
Attenuation · CT numbers · Multislice CT principles · Multislice CT detectors ·
Photon counting detector · Selectable scan parameters · Pitch · Radiation
protection
4.1 Introduction
The first clinically useful Computed tomography (CT) scanner was invented by
Godfrey Newbold Hounsfield in England. Later Hounsfield and Alan Cormack
working in South Africa shared the 1979 Nobel Prize in Medicine or Physiology
[1]. CT has better contrast resolution compared to digital radiography since it uses
a computer to process attenuation data acquired from the patient (data acquisition).
These attenuation values (μs) are subsequently converted into integers (0, a positive
number, a negative number) referred to as CT numbers [1, 2–4] and all values are
normalized to the attenuation of water (μwater). Attenuation values are captured by
special electronic detectors which convert electrical signals to digital data for pro-
cessing. The computed uses a sophisticated mathematical process referred to as
image reconstruction [5], which uses specialized algorithms to build up and display
images of a patient’s internal anatomy for diagnostic interpretation. CT images are
essentially cross-sectional images of slices that are perpendicular to the long axis of
the patient [6, 7]. Today, current CT scanners are referred to as Multislice CT
(MSCT) scanners, since they are based on acquiring volume of tissue rather than
one slice at a time.
The increasing use of CT has led to widespread concern about high patient radia-
tion doses and current radiation protection in CT has focused on dose optimization,
that is how to reduce patient dose and operate within the as low as reasonably

42 4 Computed Tomography: A Technical Review
achievable (ALARA) principle. Dose optimization will be described in detail in

Chap. 5.
The purpose of this chapter is to describe the essential technical system compo-
nents of CT, including data acquisition and image reconstruction, principles, as well
as image post processing and image display considerations. Furthermore, the basics
of MSCT scanners will be reviewed briefly. For a more detailed description of CT
principles and technology, readers should refer to a comprehensive textbook on CT
[1], and Chapters dedicated to CT principles and equipment, in other textbooks [2–4].
4.2 The CT Scanner: Fundamental Physics and Major

System Components
The major system components of a CT imaging are illustrated in Fig. 4.1. These
consist of data acquisition; image reconstruction; and image display, storage, and
communication.
While data acquisition collects x-ray transmission readings by scanning the
patient; image reconstruction uses these readings to build up a CT image of the
patient’s scanned anatomy. The transmitted radiation consists of attenuation values
(μs) which are subsequently converted by the detector electronics into digital data
for processing by a computer. This section will first explain the fundamental physics
of attenuation, followed by a description of the major system components of data
acquisition.
4.2.1 Attenuation Physics: An Essential Overview
Radiation attenuation is defined as the reduction of the intensity of a beam of radia-

tion as it passes through matter or the patient’s body. Attenuation behaves differ-
ently for a homogeneous beam (all photons in the beam have the same energy) and
for a heterogeneous beam (all photons have different energies). The original experi-
ments in CT performed by Hounsfield used a homogeneous beam from Americium
(a gamma radiation source) because this beam holds true for a law of attenuation
referred to as Beer-Lambert Law [3, 4, 6], described by the equation:
Fig. 4.1 The major system components of a CT imaging consist of data acquisition; image recon-
struction; and image display, storage, and communication
4.2 The CT Scanner: Fundamental Physics and Major System Components 43
I I0 e x
where I is the transmitted intensity; I0 is the original intensity; e is Euler constant

(2.718); μ is the linear attenuation coefficient, or the fractional reduction in the
intensity of a beam of radiation per unit thickness of the medium traversed; and x is
the thickness of the object. The mathematical problem in CT is to calculate the lin-
ear attenuation coefficient, expressed as μ, which indicates the amount of attenua-
tion that has occurred. The procedure to solve for μ is as follows:
I I0 e x
ln I 0 / I x
I / x • ln I 0 / I
where ln is the natural logarithm. In CT, however, the beam is attenuated by a given
amount of tissue with a specific thickness Δx. Hence, the attenuation is expressed as:
I I 0 e x
4.2.2 Attenuation and CT Numbers
Attenuation values are converted into integers (0, a positive number, a negative
number) referred to as CT numbers [1–4]. The system normalizes all tissue attenu-
ation values to the attenuation of water (μwater). CT numbers are computed using the
following algebraic expression:
tissue water
CTNumber ·K
water
where K is the scaling factor (contrast factor) of the CT manufacturer. In general, K

is equal to 1000. CT numbers always are computed with reference to the attenuation
of water. The CT number for water is 0, while it is +1000 for bone and −1000 for
air on the Hounsfield scale. Furthermore, CT number ranges for bone, muscle, white
matter, gray matter, blood, tumors, water, fat, lungs, and air, are 800–3000; 35–50;
36–46; 20–40; 13–18; 5–35; 0; −100; −150–400; and − 1000 respectively.
The overall process of obtaining CT numbers and converting them into a gray
scale image is illustrated in Fig. 4.2, where higher CT numbers are assigned white,
lower numbers black, and gray shades between black and white. This assignment is
related to the attenuation characteristics of tissues. Bone attenuates more radiation
and therefore is assigned white (the bone’s appearance is the same on a film-screen
Fig. 4.2 The overall process of obtaining CT numbers and converting them into a gray scale
image. (See text for further explanation)
image as it is on a digital image). Air attenuates very little radiation and appears
black on film-screen and digital images. Digital image processing is applied to these
integers through a process referred to as windowing to change the brightness and
contrast of the displayed image. The range of CT numbers is defined as the window
width and the center of the range is defined as the window level. Finally, the radiolo-
gist can manipulate the window width to alter image contrast, and the window level
to alter image brightness.
4.2.3 Data Acquisition: System Components and Principles
The data acquisition components of the CT scanner are illustrated in Fig. 4.3. These
components are housed in what is referred to as the CT Gantry, and include: an
x-ray tube, a beam shaping filter called a bow-tie filter, pre-patient collimators. The
x-ray tube is coupled to an array of detectors which collect radiation transmitted
through the patient. The detectors send electrical signals to analog-to-digital con-
verters (ADCs) which convert analog signals into digital data for processing by the
CT host computer.
Data acquisition is a systematic collection of attenuation data from the patient
through various scanning methods, which are designed to scan one slice per one
rotation of the tube and detectors (single-slice acquisition) to multiple slices per
single rotation of the x-ray tube and detectors. The latter scheme is referred to multi-
slice CT scanning (MSCT) scanners which use a fan-beam geometry to scan a vol-
ume of tissue. The term geometry, or data acquisition geometry, refers to the size,
shape, motion, and path traced by the x-ray beam. In MSCT scanning, the path
Fig. 4.3 The data

acquisition components of
the CT scanner. (See text
for further explanation)
Spiral path traced

by the continuous
rotation of
Direction of
the x-ray tube and
continuous
detectors
movement of
the patient
through the
CT gantry
Fig. 4.4 In MSCT scanning, the path traced by the x-ray beam as the patient moves through the
gantry aperture during scanning and is called a spiral or helical path
traced by the x-ray beam as the patient moves through the gantry aperture during
scanning and is called a spiral or helical path (Fig. 4.4) MSCT scanners have evolved
scanning 4 to 640 slices per revolution of the x-ray tube and detectors.
The purpose of the beam-shaping filter (bow-tie filter) as shown in Fig. 4.3, is to
make the beam more uniform (appear to be homogeneous) at the detector and thus
satisfy the Beer-Lambert law for calculating the linear attenuation coefficients.
Additionally, collimation directs the beam through the slice of interest. The
detectors measure the transmitted photons, and convert them to electrical signals
which are subsequently converted into digital data by the analog-to-digital convert-
ers. The digital data are sent to the CT host computer for image reconstruction.
4.2.4 Image Reconstruction
Image reconstruction in CT is based on sophisticated mathematics and has evolved

from simple back projection (BP) to filtered back projection (FBP) to more complex
iterative reconstruction (IR) and currently artificial intelligence-based (AI-based)
algorithms. This evolution illustrates basically that the problems associated with
former algorithms are solved by the more current state-of-the-art algorithms. For
example, while BP produced images that were blurred and were not diagnostic, the
FBP image reconstruction removed this blurring using digital image processing fil-
tering such as the convolution filtering [8–10].
FBP image reconstruction algorithms became the workhorse for decades, how-
ever, with the need for low dose CT imaging, the FBP convolution filters increased
image noise and produced image artifacts. These problems as well as others (such
as failure of the FBP algorithm to model the CT imaging system accurately [8])
provided the motivation to develop new algorithms that can reduce image noise
while maintaining image sharpness and contrast, especially in low-dose CT imag-
ing. IR algorithms were subsequently introduced into CT.
Iterative Reconstruction Algorithms
IR algorithms involve complex mathematics that model the CT imaging system

accurately in order to operate with low dose techniques. Examples of these algo-
rithms include Adaptive Statistical Iterative Reconstruction (ASiR) and Model-
Based Iterative Reconstruction (MBIR) from GE Healthcare [10]; Iterative
Reconstruction in Image Space (IRIS) and Sinogram Affirmed Iterative
Reconstruction (SAFIRE) from Siemens Healthineers. Additionally, while Philips
offers iDose4, Canon Medical Systems offers AIDR Adaptive Iterative Dose
Reduction, along with AIDR 3-D Adaptive Iterative Dose Reduction [10].
The flowchart shown in Fig. 4.5, illustrates the fundamental steps of a typical IR
algorithm. First, measured projection data sets are reconstructed with the FBP algo-
rithm, to produce an initial CT image. This image is forward projected (mathemat-
ics) to produce simulated data. Next, the measure projection data is compared with
the simulated data to create an updated image, which must meet a preset image
quality criterion established by the user and is a part of the algorithm. If this crite-
rion is not met, the iteration process continues over and over, until the difference
between the measured and simulated data sets is very small. The iterative process
stops when the final CT image matches the pre-set image quality criterion.
INPUT
Measured projection data
FBP
initial CT Image
Forward Projection
Artificial Raw Data
Compare artificial and

measured projection data
Calculate difference Iterative

reconstruction
Pre- Loop
defined Updated
Quality image
criterion No
met?
Sharper and less noisy image
Yes compared to the FBP image
Current CT Image
Stop
OUTPUT Final CT Image
Fig. 4.5 Flow chart showing the primary steps of a typical iterative reconstruction algorithm,
without modelling
Artificial Intelligence in CT Image Reconstruction
The literature is replete with articles on Artificial Intelligence (AI) describing what
AI is; its roots; and what are its applications [11]. In this subsection, AI will be
introduced briefly through a definition, and its application in CT image reconstruc-
tion. This is necessary in order to relate the use of AI-based CT image reconstruc-
tion to dose optimization. It is important for the reader interested in a further
exploration of AI in general and AI in medical imaging, to refer to the relevant lit-
erature for more details [11–19].
There are several authoritative definitions of AI, however, it is beyond the scope
of this book to address all of these. Cholet [17], defines AI as the “effort to automate
intellectual tasks typically performed by humans”. These tasks are performed by

computers using programs (software) that can sense, reason, act, and adapt. Such as
definition fits within the scope of other definitions offered in the literature as well
[18–24]. While machine learning is a subset of artificial intelligence, deep learning
is a subset of machine learning [11]. In machine learning, algorithms are trained to
use patterns learned from data to perform tasks, rather than by explicit program-
ming [12]. Deep learning on the other hand, uses algorithms that are characterized
by the use of multilayered neural networks; this network is a unique type of artificial
neural network (ANN) that resembles the multilayered human cognition system
[13]. ANN is comprised of hundreds of basic computing units, and a learning algo-
rithm, such as backpropagation, presents pairs of input signals and desired output
decisions to train the weights (parameters) of the network. This mimics the process
where the brain uses external sensory stimuli to learn to accomplish particular
tasks [13].
The motivation for the development and use of AI techniques in CT image recon-
struction stems from the limitations of FBP and IR algorithms. These limitations are
as follows: the FBP algorithm, image quality is adequate when the dose is high;
however, image noise and artifacts result when dose is lowered; IR algorithms solve
this problem, producing adequate image quality when using low-dose CT (LD-CT)
however, the noise texture often has been reported to appear blotchy, plastic-looking,
or unnatural [25]. AI-based reconstruction algorithms attempt to solve these
limitations.
The development and use of deep learning algorithms in CT image reconstruc-
tion is gaining widespread attention, with the goal of providing improved image
quality, dose performance, and reconstruction speed compared with iterative recon-
struction methods. It is not within the scope of this book, and hence this chapter to
describe the basic principles of deep learning. For a basic understanding of deep
leaning, the reader should refer to the works of Do et al. [21] and Chartrand et al.
[22] as well as one by Seeram [23], which is specifically intended for radiologic
technologists.
Deep Learning CT Image Reconstruction Algorithms
There are several examples of the use of deep learning image reconstruction (DLIR)
in CT. For example, two such DLIR algorithms were approved by the U.S. Food and
Drug Administration in 2019 namely; the Advanced Intelligent Clear-IQ Engine
(AiCE) from Canon Medical Systems [24] and TrueFidelity™ from General Electric
(GE) Healthcare [19]. It is not within the scope of this chapter to describe the details
of each of these two DLIR algorithms and the reader is encouraged to refer to other
works [19, 24–26].
A generalized framework for the AiCE and TrueFidelity™ consists of 3 steps
(Fig. 4.6): deep learning algorithm development, training and optimization of the
algorithm, and performance or verification of the algorithm. While TrueFidelity™
training and optimization involve the use of low-dose (LD) and high-dose CT data
Fig. 4.6 A generalized framework for the AiCE and TrueFidelity™ deep learning image recon-
struction consists of 3 steps: deep learning algorithm development, training and optimization of the
algorithm, and performance or verification of the algorithm. (See text for further explanation)
reconstructed with the FBP algorithm; the AiCE uses iterative reconstruction (with
modeling), and the data are fed into the deep learning reconstruction engine which
performs the required operations to produce output images with high signal-to-
noise ratio. During the training process, the output images are compared with a
reference image using various parameters, including image noise, noise texture, low
contrast resolution, and low contrast detectability. Differences are reported by the
output image to the network via special DL network actions, which strengthens
some equations and weakens others, and tries again. This process repeats until the
output image is an accurate representation of the ground truth image. In deep learn-
ing, the term “ground truth” usually refers to correct labels prepared by experts [19].
During the final step, performance or verification, the algorithm is required to
reconstruct clinical and phantom cases it has never encountered, including rare
cases [19].
DLIR algorithms have been evaluated using phantom images as well a patient
images, on noise reduction, noise texture, contrast-to-noise ratio, and low contrast
detectability. The results show that these AI-based algorithms outperform iterative
reconstruction algorithms on these parameters as well as fast reconstruction speed
to meet the needs of clinical CT examinations [19, 24, 25, 27–29]. An example of a
comparison between an IR and DL reconstruction algorithm on image noise is
shown in Fig. 4.7.
4.2.5 Image Display/Storage/Communication
Image display/storage and communications represent the third system component

of the CT scanner. Images are displayed on a monitor for viewing and interpreta-
tion, and can be subject to post processing to suit the viewing needs of the observer.
Additionally, images are sent to the Picture Archiving and Communications System
(PACS) for storage and electronic communications. PACS is an organized system
that operates on computer networking and electronic communication infrastructure.
A major element of such infrastructure is the use of standards to communicate
images from the PACS to other remote monitors. This standard is the Digital
Imaging and Communication in Medicine (DICOM) standard. For further details of
DICOM the reader should refer to Bushberg et al. [4].
Fig. 4.7 A visual comparison of IR and DLIR algorithms on image noise. From Heinrich A,
Streckenbach F, Beller E, Groß J, Weber MA, Meinel FG (2021). Deep Learning-Based Image
Reconstruction for CT Angiography of the Aorta. Diagnostics (Basel). Nov 3;11 (11):2037.
(Reproduced under the Creative Commons Attribution License which permits unrestricted use,
distribution, and reproduction in any medium, provided the original work is properly cited)
4.3 Multi-slice CT Principles at a Glance
Multislice CT (MSCT) is state-of-the-art CT imaging system. MSCT is based on

continuous data acquisition while the patient moves through the CT gantry aperture
to collect a volume data set (multiple slices during a single rotation of the x-ray tube
and detectors) as opposed to single-slice CT (SSCT) which acquires a one slice dur-
ing a single rotation of the tube and detectors, as illustrated in Fig. 4.8. This strategy
requires the use of not only specialized equipment (slip rings and x-ray tubes with
higher output) but also two-dimensional (2-D) detector arrays, and continuous table
movement during scanning. These technical principles will not be elaborated in this
book, and therefore the reader should refer to the works of Seeram [1], Bushong [2],
Wolbarst et al. [3], and Bushberg et al. [4] for a more detailed description of funda-
mental physical principles.
For the purposes of this chapter, it is noteworthy to review briefly MSCT detec-
tors and selectable scan parameters since they play a significant role in radiation
dose to the patient. CT dosimetry and factors affecting CT dose will be described in
detail in Chap. 5, on CT Dose Optimization.
4.3 Multi-slice CT Principles at a Glance 51
Fig. 4.8 MSCT is based on continuous data acquisition while the patient moves through the CT
gantry aperture to collect a volume data set (multiple slices during a single rotation of the x-ray
tube and detectors) as opposed to single-slice CT (SSCT) which acquires a one slice during a sin-
gle rotation of the tube and detectors
4.3.1 MSCT Detectors: An Overview
The purpose of the detectors in CT is to capture x-ray photons transmitted through

the patient and convert them into electrical signals and subsequently digital data for
input into the CT host computer, which uses image reconstruction algorithms to
create the CT image.
Scintillation Detectors and Photon Counting Detectors
MSCT detectors fall into two categories, namely; scintillation detectors and photon
counting detectors. Furthermore, there are two types of scintillation detectors;
energy-integrating detectors and dual-layer detectors as illustrated in Fig. 4.9.
Scintillation crystals (such as cadmium tungstate; ceramic material made of high-
purity, rare-earth oxides based on doped rare-earth compounds such as yttria; and
gadolinium oxysulfide ultrafast ceramic) convert x-ray photons to light photons,
which then are converted to electrical signals by photodiodes. Detector electronics
called application-specific integrated circuits (ASIC) digitize the signals [30].
Furthermore, Philips Healthcare uses zinc selenide activated with tellurium in
their dual-layer scintillator detectors [30].
Photon-counting detectors are now used in CT scanners. The US Food and
Drug Administration (FDA) cleared the world’s first photon-counting computed
Fig. 4.9 MSCT detectors fall into two categories, namely; scintillation detectors and photon
counting detectors. (See text for further explanation)
tomography (CT) scanner, the Siemens Naeotom Alpha, in 2021. These detectors
use semiconductors such as cadmium telluride (CdTe) and cadmium zinc telluride
(CZT) [18] because they can convert x-ray photons directly into electron hole
pairs (electric charge). Studies which compared photon-counting detectors perfor-
mance with energy-integrating detectors for example shat that “…. CT images
obtained on a photon-counting detector CT scanner were rated as having superior
spatial resolution and better critical structure visualization than those obtained on
a conventional energy-integrating detector scanner, even with a substantial dose
reduction” [31, 32].
MSCT Detector Designs
MSCT scanners use 2-D detector arrays (Fig. 4.10) that acquire several slices per
single rotation of the x-ray tube and detectors during volume scanning, as compared
with 1-D detector arrays which acquire 1 slice per single rotation of the x-ray tube
and detectors. The detector design shown in Fig. 4.10 shows 8 detector rows, which
means that this detector will acquire 8 slices per rotation. The beam geometry
needed to cover this 2-D detector is referred to as a cone beam. Slice selection with
2-D detectors is based on the detector configuration used, and have been discussed
in the literature in detail [1, 2, 4],
4.4 Radiation Protection 53
Fig. 4.10 MSCT scanners use 2-D detector arrays that acquire several slices per single rotation of
the x-ray tube and detectors during volume scanning, as compared with 1-D detector arrays which
acquire 1 slice per single rotation of the x-ray tube and detectors. The detector in this figure shows
8 detector rows, which means that this detector will acquire 8 slices per rotation
4.3.2 Selectable Scan Parameters
There are several parameters affecting the dose in MSCT imaging, however, the
technologist is concerned with what has been referred to as selectable scan param-
eters. These include the scan mode, exposure factors (kV, mA, and scan time), gan-
try rotation time, pitch, scan length, collimation, and slice width. In particular, the
pitch determines image quality needed and is related to patient dose. The
International Electrotechnical Commission (IEC) [33] provides a universal defini-
tion of pitch for MSCT scanner technology as: the pitch (P) is equal to the distance
the table travels per rotation (d)/total collimation (W). The total collimation is equal
to the number of slices (M) times the collimated slice thickness (S). Algebraically,
the pitch is expressed as:
P d / W or P d / M S.
In addition, the pitch is related to dose and image quality. As the pitch increases
from 1 to 2, image quality is compromised (image becomes noisy), as shown illus-
trated in Fig. 4.11. The relationship between dose and pitch will be described in
Chap. 5 on dose optimization.
4.4 Radiation Protection
The need for radiation protection in CT is an important consideration based on the

fact that technological advances in CT have resulted in its increasing use in practice.
This fact provided the motivation to explore the impact of these changes on patient
dose. A number of studies show that patient doses in CT is higher relative to other
Fig. 4.11 As the pitch increases from 1 to 2, image quality is compromised (image becomes
noisy). Pitch is also related to dose to the patient (Chap. 5)
imaging modalities [34–36], and therefore the literature stresses the need to opti-
mize dose in CT, thus providing the best possible radiation protection strategies for
the patient. Dose optimization in CT will be described in detail in Chap. 5.
References
1. Seeram E (2022). Computed Tomography: Physical Principles, Patient Care, Clinical

Applications, and Quality Control. 5th Edition, Philadelphia, PA: Elsevier Inc.
2. Bushong S (2021). Radiologic Science for Technologists. 12th ed. St Louis, MO:
Mosby-Elsevier.
Imaging. 4th ed. Philadelphia, PA: Lippincott Williams & Wilkins.
5. Herman GT (1980). Image Reconstruction from Projections. New York, Academic Press.
References 55
6. Hounsfield GN (1973). Computerized transverse axial scanning (tomogra-

phy): Part 1. Description of system. Br J Radiol; 46:552, 1016–1022 https://doi.
org/10.1259/0007-1285-46-552-1016
7. Hounsfield, G. N. (1976). Picture quality of computed tomography. American journal of
Roentgenology, 127(1), 3–9.
8. Hsieh J (2008). Adaptive statistical iterative reconstruction. Whitepaper, GE Healthcare
9. Seibert JA (2014). Iterative reconstruction: how it works, how to apply it. Pediatr Radiol;
44(suppl 3):431–439. doi:https://doi.org/10.1007/s00247-014-3102-1.
10. Qiu D, Seeram E (2016). Does iterative reconstruction improve image quality and reduce
dose in computed tomography? Radiol Open J; 1(2):42–54. doi:https://doi.org/10.17140/
ROJ-1-108.
11. Boden MA (2018). Artificial Intelligence: A Very Short Introduction. Oxford University
Press; 2018.
12. Erickson BJ, Korfiatis P, Akkus Z, Kline TL (2017). Machine learning for medical imaging.
Radiogr; 37(2):505–515. doi:https://doi.org/10.1148/rg.2017160130
13. Lee J-G, Jun S, Cho Y-W, et al (2017). Deep learning in medical imaging: general overview.
Korean J Radiol;18(4):570–584. doi:https://doi.org/10.3348/kjr.2017.18.4.570
14. Chollet F (2018). Deep Learning with Python. Manning Publications Co.
15. Ertel W (2017). Introduction to Artificial Intelligence. 2nd ed. Springer International
Publishing.
16. Flasinski M (2017). Introduction to Artificial Intelligence. Springer International Publishing.
17. Wooldridge M (2018). Artificial Intelligence. Ladybird Books Ltd.
18. Wagner JB (2019). Artificial intelligence in medical imaging. Radiol Technol; 90(5):489–501.
19. Hsieh J, Liu E, Nett B, Tang J, Thibault J-B, Sahney S (2019). A new era of image recon-
struction: TrueFidelity. https://www.gehealthcare.ru/jssmedia/040dd213fa89463287155151
fdb01922.pdf. Published 2019. Accessed January 2, 2020.
20. Geyer LL, Schoepf UJ, Meinel FG, et al. State of the art: iterative CT reconstruction tech-
niques. Radiol. 2015;276(2):339–357. doi:https://doi.org/10.1148/radiol.2015132766
21. Do S, Song KD, Chung JW. Basics of Deep Learning: A radiologist’s guide to understanding
published radiology articles on deep learning. Korean J Radiol. 2020;21(1):33–41. doi:https://
doi.org/10.3348/kjr.2019.0312
22. Chartrand G, Cheng PM, Vorontsov E, et al. Deep learning: a primer for radiologists. Radiogr.
2017;37(7):2113–2131. doi:https://doi.org/10.1148/rg.2017170077
23. Seeram E (2020). Computed Tomography Image Reconstruction. Radiol Technol. 2020 Nov;
92(2):155CT-169CT. PMID: 33203780.
24. Boedeker K (2019). AiCE deep learning reconstruction: bringing the power of ultra-high reso-
lution ct to routine imaging. Canon Med Syst; 2:28-33
25. Zhang, M., Gu, S. & Shi, Y. (2022). The use of deep learning methods in low-dose com-
puted tomography image reconstruction: a systematic review. Complex Intell. Syst. https://doi.
org/10.1007/s40747-022-00724-7
26. Suzuki K. Overview of deep learning in medical imaging. Radiol Phys Technol.
2017;10(3):257–273. doi:https://doi.org/10.1007/s12194-017-0406-5
27. Shan H, Zhang Y, Yang Q, et al (2018). 3D convolutional encoder–decoder network for
low-dose CT via transfer learning from a 2D trained network. IEEE Trans Med Imaging;
37(6):1522–1534. doi:https://doi.org/10.1109/TMI.2018.2832217
28. Shan H, Padole A, Homavounieh F, et al (2019) Competitive performance of a modularized
deep neural network compared to commercial algorithms for low-dose CT image reconstruc-
tion. Nat Mach Intell; 1:269–276. doi:https://doi.org/10.1038/s42256-019-0057-9
29. Akagi M, Nakamura Y, Higaki T, et al (2019). Correction to: deep learning reconstruction
improves image quality of abdominal ultra-high-resolution CT. Eur Radiol; 29(8):4526–4527.
doi:https://doi.org/10.1007/s00330-019-06249-x
30. Shefer E, Altman A, Behling R, et al (2013). State of the art of CT detectors and sources: a lit-
erature review. Curr Radiol Rep; 1(1):76–91. doi:https://doi.org/10.1007/s40134-012-0006-4.
31. Zhou W, Lane JI, Carlson ML, Bruesewitz MR, Witte RJ, Koeller KK, Eckel LJ, Carter RE,
McCollough CH, Leng S (2018). Comparison of a Photon-Counting-Detector CT with an
Energy-Integrating-Detector CT for Temporal Bone Imaging: A Cadaveric Study. AJNR
Am J Neuroradiol;39(9):1733–1738. doi: https://doi.org/10.3174/ajnr.A5768. Epub. PMID:
30093479; PMCID: PMC6128765.
32. Benson JC, Rajendran K, Lane JI, Diehn FE, Weber NM, Thorne JE, Larson NB, Fletcher JG,
McCollough CH, Leng S. (2022) A New Frontier in Temporal Bone Imaging: Photon-Counting
Detector CT Demonstrates Superior Visualization of Critical Anatomic Structures at Reduced
Radiation Dose. AJNR Am J Neuroradiol;43(4):579–584. doi: https://doi.org/10.3174/ajnr.
A7452. Epub 2022 Mar 24. PMID: 35332019; PMCID: PMC8993187.
33. IEC (1999). Medical electrical equipment-60601 Part 2-44: particular requirements for
the safety of x-ray equipment for CT. Geneva, Switzerland: International Electrotechnical
Commission.
34. Berrington de Gonzalez A, Mahesh M, Kim KP, et al (2009). Projected cancer risks from
computed tomographic scans performed in the United States in 2007. Arch Intern Med;
169(22):2071–2077. doi:https://doi.org/10.1001/archinternmed.2009.440
dose in CT in The Netherlands. Insights Imaging; 4(3):383–390. doi:https://doi.org/10.1007/
s13244-013-0253-9.
Australians. BMJ; 346: f2360. doi:https://doi.org/10.1136/bmj.f2360.
Chapter 5
Dose Reduction and Optimization
Strategies in Computed Tomography
Keywords CT dose reduction · CT dose optimization · Radiation risks ·

Stochastic effects · Deterministic effects · Cancer risks from CT imaging ·
Selectable scan parameters · Tube current modulation · kV optimization · Image
reconstruction optimization · Deep learning image reconstruction · Diagnostic
reference levels
5.1 Introduction
The introduction of CT imaging in 1970s in medicine and specifically in diagnostic

imaging has provided numerous clinical benefits in the detection, evaluation and
diagnosis of human diseases. Hricak et al. [1] have shown that the growth and tech-
nical advances have resulted in more effective surgical treatments and have resulted
in eliminating the need for some invasive exploratory procedures. Furthermore,
these advances have helped decrease inpatient hospital stays; improve cancer,
stroke, cardiac, and trauma diagnosis and treatment; and enabled rapid diagnosis of
life-threatening vascular conditions [1]. Additionally, the there has been an increase
in the use of CT imaging in clinical medicine. For example, CT use in hospital
emergency departments has increased significantly from 2005 through 2013 [2], in
an effort to prioritize and reduce wait times in the care and management of
patients [3].
This increase in the use of CT has prompted research on that patient doses from
CT examinations, and the results have shown that these doses are high relative to
other radiography examinations [4–6]. Additionally, it has been reported that CT
doses typically range from 5 to 50 mGy to each organ within the image field [7]. As
of 2011, CT contributed the highest collective amount of medical radiation expo-
sure in the United States compared with any other medical imaging modality [1].

58 5 Dose Reduction and Optimization Strategies in Computed Tomography
5.2 Motivation for Dose Optimization in CT
The few studies cited in the introduction above, stress the need to optimize the doses
in CT imaging. The general motivation of dose optimization in CT and digital radi-
ography has been presented in Chap. 1 and include radiation risks which have been
studied extensively. The kinds of biological effects observed have been summarized
by Hendee and O’Connor [8], who drew attention to Radiation Effects Research
Foundation (RERF) data and models based on Japanese atomic bomb survivors that
form the basis of the Biological Effects of Ionizing Radiations VII (BEIR VII)
report [8].
These effects of radiation fall into 2 broad categories: stochastic effects and
deterministic effects [4, 9] and have been reviewed in Chap. 1. In summary, stochas-
tic effects are effects for which the probability of occurrence increases with increas-
ing dose, and for which there is no threshold dose (any amount of radiation, no
matter how small, has the potential to cause harm). Stochastic effects also are called
late effects, and examples of these effects include cancer and genetic damage [4, 9].
Stochastic effects are considered a risk from exposure to the low levels of radiation
used in medical imaging, including CT examinations [4, 9]. Deterministic effects
are those effects for which the severity of the effect (rather than the probability)
increases with radiation dose and for which there is a threshold dose. Examples of
deterministic effects include skin burns, hair loss, tissue damage, and organ dys-
function [4, 9].
An important and significant noteworthy point about these effects is that the
RERF data show that at high doses (100 mSv and higher), the evidence of increased
cancer is statistically significant [8]. In medical imaging using x-rays, the doses to
patients are much lower, and therefore in order to understand the cancer risks, dose
response models have been proposed to extrapolate cancer risks from a high-dose
situation to the risks of the low doses used in diagnostic imaging. These models
have been introduced in Chap. 1, and will not be described further in this Chapter.
In summary, there are tow categories of models, namely; the linear dose-response
models and nonlinear dose-response models [9]. Furthermore, there are two types
of linear dose-response models: a linear dose-response model without a threshold
(LNT) and a linear dose-response model with a threshold. While the former is used
to show that no amount of radiation is considered safe and that any dose, no matter
how small, carries some degree of risk, the latter proposes that no adverse effect
from a dose below a certain level, known as the threshold dose, is observed. An
effect occurs only when the threshold dose is reached [9].
The next obvious question is which model is used in diagnostic radiology. The
answer to this question is provided Hendee and O’Connor [8]. They argue that the
LNT model is more commonly used because it is a simpler and conservative
approach, which is more likely to overestimate cancer risk at low doses than to
underestimate risk of cancer induction. Furthermore, Hendee and O’Connor [8],
stress that because the doses in diagnostic radiology are so low, there is no evidence
that the LNT model is effective at estimating cancer induction risk [8]. This has led
5.2 Motivation for Dose Optimization in CT 59
to what is referred to as “the Pro-LNT/Anti-LNT Debate” in a commentary pre-

sented by Pamela J. Sykes, Flinders Centre for Innovation in Cancer, College of
Medicine and Public Health, Flinders University, Adelaide, South Australia,
Australia [10]. The author points out that “Until those on one side of the debate can
convince the other, it would be sensible to move forward toward a graded (risk-
based) approach to regulation, where the stringency of control is commensurate
with the risk, resulting hopefully in more sensible practical thresholds. This
approach is gradually being put forward by international radiation protection advi-
sory bodies” [10].
5.2.1 Literature on Cancer Risks from CT
The literature is replete with articles on cancer risks from CT, and studies have
emerged as a result of the relatively high doses from CT examinations [4, 6, 12],
particularly for pediatric patients. It is not within the scope of this Chapter to
describe details of these studies, however, a few examples of the results of selected
examples are noteworthy. For example, Pearce et al. [13] concluded that there is a
risk of cancer from diagnostic x-ray exposures and that the risk is very low, however
the risk is detectable, so statements that radiation risks associated with diagnostic
exposures are simply undetectable must be questioned. In one of the first large stud-
ies to derive direct estimates of cancer risk from CT instead of extrapolating risk
from high-dose exposures. Matthews et al. [7] found “an increased risk of cancer in
the children and adolescents who had CT scans from 1985 through 2005” but also
“also suggested that estimating risks from many studies has been difficult because
of relatively small study sizes and selection bias”.
Hall and Brenner [14] in an article entitled “Cancer risks from Diagnostic
Radiology” published in the British Journal of Radiology point out that “most of the
collective dose from diagnostic radiology comes from high-dose (in the radiological
context) procedures such as CT, interventional radiology and barium enemas; for
these procedures, the relevant organ doses are in the range for which there is now
direct credible epidemiological evidence of an excess risk of cancer, without the
need to extrapolate risks from higher doses. Even for high-dose radiological proce-
dures, the risk to the individual patient is small, so that the benefit/risk balance is
generally in the patients’ favour”. An interesting study by Shao et al. [15] showed
that “CT scans may be associated with an increased risk of thyroid cancer and leu-
kemia in adults and in those diagnosed with NHL at a younger age”.
Other studies followed and reported in the literature. Examples of two such stud-
ies are those of Moghadama [16] published in 2021, and Zewdea et al. [17] pub-
lished in 2022. While the purpose of the former study was “To estimate percentage
of patients undergoing multiple CT exams leading to cumulative effective dose
(CED) of more than 25, 50, 75 and 100 mSv in one year and assess per capita and
the collective effective dose”, the latter study was “to estimate cumulative organ
doses and age- and gender-stratified cancer mortality risks in patients undergoing
recurrent computed tomography (CT) exams.” The conclusions are as follows:
• “The alarming high CED received by large number of patients and with high
collective dose to population requires urgent actions by all stake holders in the
best interest of patient radiation safety” [16]
• “Organ doses over 100 mGy for most organs and for some organs ≥200 mGy
with unignorable associated lifetime attributable cancer mortality rates were
found” [17]
A final study to be cited in this Chapter is one by Hong et al. [18] entitled “Association
of Exposure to Diagnostic Low-Dose Ionizing Radiation with Risk of Cancer
Among Youths in South Korea” and published in JAMA Network Open. This origi-
nal investigation explored the following:
1. Cancer Risk Associated with CT Scanning
2. Risks of specific cancers
3. Cancer Risk According to Type of First Exposure, and the
4. Cancer Risk According to Number of Exposures
The investigators concluded that “the associations we found of diagnostic low-dose
ionizing radiation with increased incidence of cancer in youths suggest that there is
incentive to limit radiation doses to as low as reasonably achievable and to only scan
when justified. Medical professionals should weigh the benefits of diagnostic low-
dose ionizing radiation with the associated risks to justify each decision” [18].
5.2.2 Effects of Low-Dose Chest CT on Chromosomal DNA
An interesting and important study is one by Sakane et al. [19] where low-dose CT
examinations were compared to standard dose CT of the chest in exploring “the
effects on DNA double-strand breaks and chromosome aberrations (CAs) in periph-
eral blood lymphocytes” Sakane et al. [19] concluded that “no effect of low-dose
CT on human DNA was detected. In the same setting, DNA double-strand breaks
and chromosome aberrations increased after standard-dose CT”.
5.2.3 In Summary
The above are examples of studies on CT doses that are defined as high, standard,
and low on biological effects. These studies were cited to provide the motivation for
dose optimization in CT. In this regard, it is important to repeat here, the conclu-
sions drawn by Moghadama [16], which is “The alarming high CED (Cumulative
Effective Dose) received by large number of patients and with high collective dose
to population requires urgent actions by all stake holders in the best interest of
5.4 Elements of CT Dosimetry at a Glance 61
patient radiation safety” [16]. These “urgent actions” form the basis for the content
of the rest of this Chapter.
5.3 Radiation Protection Principles
Current standards of radiation protection include one of the fundamental guiding

principles of the ALARA (As Low as Reasonably Achievable) philosophy of the
International Commission on Radiological Protection (ICRP) [11]. This principle
identifies the notion of dose reduction and dose optimization, which are intended to
minimize stochastic effects and to prevent deterministic effects. The ICRP recom-
mends approaches to dose optimization associated with radiography equipment and
daily operations [20]. These were reviewed in Chap. 1 and involves at least four
elements, namely; the imaging equipment, the adequacy of the imaging equipment,
technical parameters of the examination, and diagnostic reference levels.
In CT, dose optimization requires that the user understands the elements of CT
dosimetry, reviewed in the next section.
5.4 Elements of CT Dosimetry at a Glance
CT dosimetry refers to the measurement of the dose to the patient. As shown in

Fig. 5.1, the typical dose distribution is a bell-shaped curve, given by the function
D(z), where D is the dose and z is the longitudinal axis of the patient. D(z) is
extremely important to the CT dose because this is the dose distribution, or dose
profile, that is measured. CT dose measurements are characterized by at least three
commonly used metrics (dose quantities): the Computed Tomography Dose Index
(CTDI), the Dose Length Product (DLP) and the Effective Dose (ED). While the
units of the CTDI and the DLP are expressed in milligrays (mGy), the ED is
expressed in millisieverts (mSv). In CT dose optimization studies, these dose quan-
tities are measured and reported. The ED relates the radiation exposure to risk and
Fig. 5.1 The typical dose D(z)

distribution in CT imaging
is a bell-shaped curve,
given by the function D(z), Area
where D is the dose and z
is the longitudinal axis of
the patient. D(z) is
extremely important to the
CT dose because this is the
dose distribution, or dose
profile, that is measured
is considered the best method available to estimate stochastic radiation risk [1].
Still, effective dose is an estimate only, based on weighting factors applied to the
body’s tissue or the organ being irradiated [9].
As discussed by McNitt-Gray [21], CTDI dosimetry is the current international
standard for estimating CT radiation dose. The CTDI has evolved along the techni-
cal advances made in CT during the years and readers interested in such evolution
should refer to a paper by Seeram [22]. The introduction of spiral/helical CT beam
geometry used for volume scanning led to the development of the CTDIvolune
(CTDIvol) expressed algebraically as:
CTDI vol = CTDI w / pitch
where CTDIw is the weighted CTDI, intended to addresses the average dose in the
x-y axis of the patient. When the pitch is 1, the CTDIvol equals the CTDIW. A signifi-
cant point to note is that the value of the CTDIvol is the same whether a 1-mm or
100-mm length of tissue of the patient is scanned. This problem is solved by the
introduction of the DLP in order to provide a much more accurate value of the dose
for a defined length of tissue. The DLP provides a measure of the total dose for a
CT, and it is proportional to the scan length (L). Algebraically, the DLP is
expressed as:
DLP CTDI vol L
For example, the data for 2 CT scans of the abdomen are: Abdomen 1-the scan
length is 16 cm, and the DLP is 160 mGy-cm. For abdomen 2, the scan length is
32 cm, and the DLP is 320 mGy-cm, that is, if the scan length doubles, the DLP
doubles.
CT dose reporting also include the ED in order to relate radiation exposure to
risk, and it takes into account that different tissues have different radiosensitivities.
Because only parts of the body (rather than the entire body) are exposed in CT
imaging, the risk of stochastic effects is proportional to the effective dose rather
than to the tissue dose [9]. The effective dose is expressed algebraically as:
ED WT H T
where HT is equal to the organ or tissue dose, and WT is the tissue weighting factor.
Using DLP values, the
ED k DLP
where k is a constant for different body parts. For more information on ED for dif-
ferent body parts and modalities, the reader should refer to McKnitt-Gray [21] and
Wolbarst et al. [23].
5.5 Factors Affecting the Dose in CT: An Overview 63
5.5 Factors Affecting the Dose in CT: An Overview
Understanding dose optimization in CT requires that users (operators) have a firm

grasp of note only the technical factors affecting the dose to the patient having CT
examinations, but also all image quality factors as well since dose and image quality
are closely related [24].
It is not within the scope of this Chapter to describe all image quality and dose
factors in detail and therefore, readers should refer to the Bushberg et al. [9], Seeram
[22], Wolbarst et al. [23] and Goo [24] for more detailed descriptions, however the
following are noteworthy and are briefly reviewed in the next subsections.
5.5.1 Image Quality Considerations
There are at least four image quality parameters that are related significantly to
dose. They parameters include image noise; contrast-to-noise resolution; spatial
resolution; and artifacts. Only the first three will be considered for the purposes of
this Chapter. According to Goo [24]:
• Noise is defined as a random variation of CT numbers, and is inversely propor-
tional to square root of radiation dose; inversely proportional to tube voltage;
inversely proportional to fourth power of spatial resolution; influenced by sec-
tion thickness and reconstruction algorithm.
• Contrast-to-noise resolution is defined as the ability to distinguish between dif-
ferent CT numbers. Not significantly changed at different tube voltages in most
materials except for some with high atomic numbers such as iodine (increased at
low tube voltage); low contrast resolution greatly influenced by image noise level.
• Spatial resolution is defined as the ability to distinguish small details of object,
and it is inversely proportional to focal spot size and detector collimation; influ-
enced by reconstruction algorithm.
5.5.2 Significant Dose Parameters
Important CT dose parameters have been identified by Goo [24], and include kV,
mAs, pitch, CTDIw, CTDIvol, DLP, and ED. The key factors that radiologic tech-
nologists can select to optimize dose are exposure technique factors (mAs and kV),
pitch, collimation and slices, Automatic Exposure Control (AEC), noise index,
overbeaming and overranging, and noise-reduction image reconstruction
algorithms.
The key relationships that are noteworthy in CT dose optimization however, are:
• The dose is directly proportional to the mAs, that is doubling the mAs doubles
the dose
• The dose is proportional to the square of the kilovolt peak (kV2). This exponen-
tial expression poses some difficulty for technologists in practice because this
power [2] can vary from 2.5 to 3.1 and depends on the patient’s size [25]. An
interesting study by Maldjian and Goldman [26], who showed that decreasing
the kilovolt peak from 140 to 120 kVp reduces the dose by 28% to 40% for a
typical phantom. The authors reported that further decreasing to 80 kVp reduces
the dose by approximately 65%.
• Dose (D) is inversely related to the pitch (P). If the pitch increases by 2, the dose
is reduced by 0.5 mGy. This relationship is an inverse one and is expressed:
Dα 1 / P
• When using the AEC, the tube current increases when the pitch is increased, and
therefore it is not best practice to reduce exposure with multislice CT scanners
• Generally, as the collimation width increases (wider beam = thicker section), the
dose decreases. When decreasing the section thickness (T, the reconstructed or
nominal thickness), the exposure must be increased to maintain the same signal-
to-noise ratio (SNR) as a thick section. For example, a 2.5-mm section requires
2 times more exposure than a 5-mm section [24]. The relationship for the noise
in the image is an inverse one expressed as follows:
Noise α 1 / T
• A technique referred to as adaptive or dynamic collimation is used to address

overranging and overbeaming in CT in an effort to reduce the dose to the patient
at the beginning and end of the scanning [27]. While overranging refers to the
use of additional rotations before and after the planned length of tissue so the first
and last images can be reconstructed, overbeaming is the excess dose beyond the
edge of the detector rows per rotation of a multisection. Furthermore, Christner
et al. [27] showed that dynamic collimation can reduce the dose by approxi-
mately 40%.
• The use of the AEC requires that a preselected image quality index also referred
to as a reference or target image quality, be established before scanning the
patient [24]. The index is stored in the CT scanner by the manufacturer before
shipping and is an operator-selectable parameter [24]. As the index increases, the
patient dose decreases but at the expense of a noisy image.
• Accurate patient positioning is an essential task of the technologist during CT
scanning, If the patient is not centered in the CT gantry isocenter (not centered
accurately) in the scan field-of-view, image noise and patient dose increase
because of poor bowtie filter performance. As demonstrated by Toth et al. [28]
miscentering the patient by as little as 3 cm increased surface dose by 18%, and
miscentering by 6 cm increased surface dose by as much as 41%
5.6 Dose Optimization Strategies in CT 65
• Iterative reconstruction algorithms and now used routinely in low-dose (LD) CT

imaging. These algorithms are intended to reduce image noise and minimize the
higher radiation dose [29–31]. A number of studies have demonstrated reduc-
tions in radiation dose using iterative reconstruction that vary from 30% to 50%
[26, 30]. These studies have included dose reductions in pediatric studies, CT
abdominal studies, and CT angiography.
5.6 Dose Optimization Strategies in CT
The idea of dose optimization was introduced in Chap. 1, and subsequently elabo-
rated in Chap. 3 for digital radiography. In Chap. 1, the distinction between dose
reduction and dose optimization was described. In review, operators working within
the ALARA philosophy of the ICRP balance the need for patient radiation protec-
tion with the need for acquiring high-quality diagnostic images. In doing so they are
guided by the meaning of two terms used in the literature, that is; reduction and
optimization. While reduction means to “reduce or diminish in size, amount, extent,
or number” the term optimization demands a more rigorous effort to be effective in
adhering to the ALARA philosophy. Optimization means “an act, process, or meth-
odology of marking something (as a design, system, or decision) as fully perfect,
functional, or effective as possible” [32]. Perhaps the key difference between these
two terms is that dose optimization takes into primary consideration that the image
quality must not be compromised as the dose is lowered. This is where the rigor of
an optimization study lies. For example, a CT dose optimization study would involve
an extensive component related to the methodology used in the study. Such method-
ology must use reliable and valid tools and procedures for the dosimetry, image
acquisition, and evaluation of image quality using human observers, keeping in
mind the nature of the detection task. The details within this definition of optimiza-
tion will be discussed further in Chap. 6.
CT dose reduction involves strategies to adjust and control the technical factors
affecting the dose with the goal of decreasing patient dose [33]. The effect of these
factors (eg, mAs, kV, pitch, scan field-of-view, beam collimation, AEC, overbeam-
ing and overranging, and iterative image reconstruction) on patient dose has been
researched by multiple authors [22–29, 34].
CT dose optimization strategies have been explored and the results published in
the literature. A few illustrative examples include:
1. Sohaib et al. [35] examined the effect of reducing milliampere seconds on image
quality and patient dose in sinus CT examinations. The authors reported a dose
reduction from 13.5 mGy at 200 mAs to 3.1 mGy at 50 mAs (P = .05) without
loss of image quality. The study used an observer performance method that
involved visual grading analysis.
2. Russell et al. [36] examined dose-image quality in neck volume CT and showed
that automatic tube current modulation reduced the CTDI by 20% with the noise
index set at 11.4 and by 34% with the noise index set at 20.2. The dose reduc-
tions were made without compromising image quality significantly.
5.7 Other Useful Tools for Dose Reduction

and Optimization in CT
There are several additional tools for dose reduction and optimization in CT and
these include the use of Diagnostic Reference levels (DRLs), artificial intelligence-
based (AI-Based) image reconstruction, and more recently the use of photon count-
ing detectors (PCD) in CT imaging.
5.7.1 The Use of DRLs
The ICRP and other Radiation Protection Authorities such as the National Council
on Radiation Protection and Measurements (NCRP) have established dose limits
not only for occupationally exposed individuals such as radiologists and technolo-
gists, pregnant workers, but for members of the public as well. There are no dose
limits however for patients having x-ray imaging procedures including diagnostic
radiographic and fluoroscopic as well as CT examinations, for the management of
their medical illness. This scenario was addressed by the ICRP, as early as 1990, and
in this regard the ICRP [37] introduced the notion of Diagnostic Reference Levels
(DRLs), and recommends the use of DRLs for patients. Dose limits and DRLs are
not the same thing. A thorough review of DRLs for technologist is one by Seeram
and Brennan [38] published in Radiologic Technology. Furthermore, a brief sum-
mary of the essential elements of the DRL is one by Seeram [39].
The DRL has been defined by several organizations, however, only two, the
ICRP and the American College of Radiology (ACR) definitions will be presented
here for the sake of brevity. The ICRP [37] states that DRLs “are a form of investi-
gation level, applied to an easily measured quantity, usually the absorbed dose in air,
or tissue-equivalent material at the surface of a simple standard phantom or a repre-
sentative patient.” The ACR [40], on the other hand, defines a DRL as “an investiga-
tion level to identify unusually high radiation dose or exposure levels for common
diagnostic medical x-ray procedures.”
These formal definitions imply that the DRL is a means to reduce and/or opti-
mize the radiation dose to the patient, and it “is to provide a benchmark for compari-
son, not to define a maximum or minimum exposure limit.” [37]. Additionally, the
ICRP states that DRLs “apply to medical exposure, not to occupational and public
exposure, thus they have no link to dose limits or dose constraints… The values
should be selected by professional medical bodies, and renewed at intervals that
represent a compromise between the necessary stability and long-term changes in
5.7 Other Useful Tools for Dose Reduction and Optimization in CT 67
the observed dose distributions. The selected value will be specific to a country or
region.” [37]
DRLs are tools that can be used by radiology departments to ensure that patients
are not exposed to unnecessary doses. If patients are exposed to higher DRL values
as established by radiation protection authorities in respective countries, then cor-
rective actions must be taken to ensure that proper radiation protection procedures
are implemented in these imaging facilities can use to measure and assess radiation
doses to patients for a defined set of procedures.
It is not within the scope of this Chapter to describe details of how DRLs are
established however, the following points summarize existing ICRP guidelines
for DRLs:
• The DRL is an advisory, not a regulatory, measure. It is not related to dose limits
established for radiation workers and members of the public.
• The DRL is intended to identify high levels of radiation dose to patients.
• The DRL applies to common examinations and specific equipment.
• The DRL selection is established by professional radiation protection organiza-
tions, using a percentile point on the observed distribution for patients, and spe-
cific to a country or region. One pragmatic method is to use what has been
referred to as a diagnostic reference range. While the upper level of the range is
established at 75th percentile, the lower level is set at 25th percentile of the com-
puted patient dose.
• Below a 25th percentile level, image quality may be compromised, above the
75th percentile may indicate excessive dose.
Figure 5.2 illustrates hypothetical data collected from a survey of 9 hospital CT
scanner rooms showing the doses (mGy) delivered for an AP lumbar spine projec-
tion. The dose values were arranged in increasing order. The next step is to establish
DRLs. In this regard the ICRP [37] recommends that “the median value (not the
mean value) for the DRL quantity from each of the facilities in the survey should be
used”, and that national DRL values should be set as the 75th percentile (Fig. 5.2).
This value becomes the reference level. It is clear from Fig. 5.2, that 2 hospitals are
above this level. This means that these 2 hospitals need to engage in radiation pro-
tection activities that would bring the dose levels below the 75th percentile. For a
thorough description of setting DRL values, the interested reader should refer to the
ICRP Report 135 [37].
For actual DRL values for various imaging examinations, the reader should con-
sult their respective national radiation protection organizations. For example, the
ACR, the American Association of Physicists in Medicine (AAPM), and the Society
for Pediatric Radiology (SPR) [40] have issued the following DRLs: for the adult
PA chest (23 cm patient thickness with grid), AP abdomen (22 cm patient thick-
ness), and the AP lumbosacral spine (22 cm patient thickness) to be 0.15, 3.4, and
4.2 mGy, respectively.
7 75th Percentile
Radiation Dose (mGy)
1 2 3 4 5 6 7 8 9
Hospitals
Fig. 5.2 Hypothetical data collected from a survey of 9 hospital CT scanner rooms showing the
doses (mGy) delivered for an AP lumbar spine projection. The dose values were arranged in
increasing order. (See text for further explanation)
5.7.2 The Use of AI-Based Image Reconstruction
The limitations of the FBP and IR image reconstruction algorithms in Low Dose
(LD) CT imaging provides the motivation to develop algorithms that overcome
these shortcomings as described in Chap. 4. In summary, for CT imaging using FBP
algorithms, image quality is adequate when the dose is high; however, image noise
and artifacts result when dose is lowered in an effort to reduce dose to patient and
work within the ALARA philosophy. This problem is solved using iterative recon-
struction (IR) algorithms, however, the literature for example has shown that in
low-dose CT imaging, the noise texture produced by IR algorithms do appear
blotchy, plastic-looking, or unnatural [41]. This shortcoming has provided the moti-
vation for the development and use of AI-based image reconstruction algorithms.
In the AI domain, deep learning is a subset of machine learning, which is a subset
of AI. For more details of the use of AI in medical imaging and in particular CT
image reconstruction, the reader should refer to a CT textbook by Seeram [22],
particularly intended for radiologic technologists. The use of deep learning image
reconstruction (DLIR) has gained acceptance in CT imaging since it provides
improved image quality, dose performance, and reconstruction speed compared
with iterative reconstruction methods, in low-dose CT imaging.
In 2019, the US Food and Drug Administration (FDA) approved two DLIR-
based, the Advanced Intelligent Clear-IQ Engine (AiCE) from Canon Medical
Systems and TrueFidelity from General Electric (GE) Healthcare [31, 42]. Figure 5.3
provides an illustration of a generalized framework for the AiCE and TrueFidelity
consisting of 3 steps, namely; deep learning algorithm development; training and
optimization of the algorithm; and performance or verification of the algorithm.
5.7 Other Useful Tools for Dose Reduction and Optimization in CT 69
Fig. 5.3 An illustration of a generalized framework for the AiCE and TrueFidelity AI-based image
reconstruction algorithms. There are essentially 3 steps, namely; deep learning algorithm develop-
ment; training and optimization of the algorithm; and performance or verification of the algorithm
Fig. 5.4 This figure illustrates a simple overview of the deep learning process; low-dose CT data
and high-dose CT data are fed into the deep learning reconstruction engine which performs the
required operations to produce output images with high signal-to-noise ratio
In Fig. 5.4 illustrates a simple overview of the process; low-dose CT data and
high-dose CT data are fed into the deep learning reconstruction engine which per-
forms the required operations to produce output images with high signal-to-
noise ratio.
The training process is complex, and therefore only a brief overview is given
here. Duing training, output images are compared with a reference image, referred
to as a ground truth image using various parameters, including image noise, noise
texture, low contrast resolution, and low contrast detectability. Differences in the
output image are sent to the network via backpropagation, which strengthens some
equations and weakens others, and tries again. This process repeats itself until the
output image is an accurate representation of the ground truth image [31].
The literature includes studies on assessing the performance of deep learning
reconstruction using both test phantom and clinical patient images on noise
reduction, noise texture, contrast-to-noise ratio, and low contrast detectability [31,
42–45]. These studies clearly demonstrate that deep learning reconstruction algo-
rithms outperform iterative reconstruction algorithms on these parameters.
Furthermore, they have fast reconstruction speed to meet the needs of sophisticated
clinical CT examinations [31, 42–45].
5.7.3 The Use of PCDs
The use of PCDs as a system component of a CT scanner has been approved by the
U.S. Food and Drug Administration (FDA) on September 30, 2021, for the Naeotom
Alpha is the first commercialized Photon-Counting CT (PCCT) scanner from
Siemens Healthineers. Major CT vendors are now developing PCCT Scanners.
The literature is replete with articles on PCCT scanners. For example, a Google
search on August 24, 2022, for photon counting CT generated 573,000 results. It is
not within the scope of this Chapter to describe the technical details of PCDs, how-
ever, the interested reader should refer to the works of Rajendran et al. [46], and
Danielsson et al. [47].
The major system differences between a PCD and an Energy Integrating Detector
(EID) were reviewed in Chap. 4. The major component differences are illustrated in
Fig. 5.5. One type of EID is illustrated in Fig. 5.5a. This type consists of scintillation
crystals (such as cadmium tungstate (CdWO4); ceramic material made of
Fig. 5.5 The major system differences between a Photon Counting Detector and an Energy
Integrating Detector (EID). See text for explanation
5.8 The Role of the Technologist in CT Dose Reduction and Optimization 71
high-purity, rare-earth oxides based on doped rare-earth compounds such as yttria;

and gadolinium oxysulfide ultrafast ceramic) convert x-ray photons to light pho-
tons, which then are converted to electrical signals by photodiodes. Detector elec-
tronics called application-specific integrated circuits (ASIC) digitize the signals.
Digital data are subsequently fed into the CT host computer. A PCD (Fig. 5.5b) on
the other hand use semiconductors such as cadmium telluride (CdTe) and cadmium
zinc telluride (CZT) to convert x-ray photons directly into electron hole pairs (elec-
tric signals). These signals are fed into the photon counter electronics which provide
respective energies (E1, E2….EN). Digital data are then fed into the CT host computer.
In terms of dose reduction, PCCT scanners provide images at lower dose than
conventional CT scanners, and provide images with higher contrast-to-noise ratio,
resulting in higher resolution. For example, two studies, one by Rajendran et al.
[48], and the other by Benson et al. [49] show that:
1. “The temporal bone patient images demonstrated a dose reduction of up to 85%
compared with clinical EID-CT” [48]
2. “Temporal bone CT images obtained on a photon-counting detector CT scanner
were rated as having superior spatial resolution and better critical structure visu-
alization than those obtained on a conventional energy-integrating detector scan-
ner, even with a substantial dose reduction” [49].
5.8 The Role of the Technologist in CT Dose Reduction

and Optimization
There are several types of imaging professionals working in CT imaging commu-

nity, each of whom play a role in responsibility in optimizing dose in CT. For exam-
ple, Strauss et al. [33] and Matthews and Brennan [50] identified radiologists,
physicians, medical physicists, technologists, and CT manufacturers in this regard.
Strauss et al. [33] have identified 10 steps to optimize image quality and reduce
radiation dose in CT. The first of these steps is to increase awareness and under-
standing of CT radiation dose issues among radiologic technologists.
As reviewed by Seeram [51], “the technologist is central to CT image acquisition
and generally is responsible for equipment start-up procedures and patient care and
communications throughout the entire examination, along with patient positioning,
communications with the radiologist regarding all clinical aspects of the examina-
tion, and overall radiation protection of the patient and other staff present during the
examination”. As a result, the technologist plays a significant role in patient dose
and image-quality optimization in CT. Seeram [51] identifies several factors that
technologists must be well-versed in order to be successful in CT dose optimization
strategies. These are listed in Table 5.1.
It is mandatory that in order to participate effectively in CT dose optimization
through research, technologists must understand the distinction between CT dose
Table 5.1 Several factors that technologists must be well-versed in order to be successful in CT
dose optimization strategies. Radiologic Technology 85 (4) 2014
1. The risks of radiation and CT dose in particular.
2. Current technical advances in CT.
3. CT dose metrics, particularly CTDIvol, the DLP and effective dose, and associated units.
4. CT image-quality metrics such as spatial resolution, contrast resolution, noise, and artifacts.
5. The technical factors affecting patient dose in CT, including exposure technique factors (kV
and mAs), Automatic Exposure Control, pitch, effective milliampere seconds, slice thickness,
scan field-of-view, beam collimation, noise-reducing algorithms (iterative reconstruction
algorithms), anatomical coverage, overbeaming and overranging, patient centering, and noise
index
6. Scan protocols and reviewing the protocols with the radiologist on an ongoing basis with the
goal of optimizing dose and image quality.
7. The prescan and postscan display of CT dose reports showing the CTDIvol, the DLP, and
effective dose.
8. Get involved with the development or implementation of a CT dose-monitoring or dose-
tracking system for the CT department. Monitoring and tracking should include items such as
dose capture, conversion of absorbed dose to effective dose, patient-specific storage, dose
analytics, dose communication, and data export
9. Participate in research on CT patient dose and image-quality optimization. This requires a
fundamental knowledge of CT equipment and dosimeter calibration, image acquisition details,
observer performance measures, and appropriate statistical tools.
10. Ensure continuous professional development through relevant continuing education
activities.
reduction and CT dose optimization and the application of ALARA. Dose optimiza-
tion research will be described in Chap. 6.
References
1. Hricak H, Brenner DJ, Adelstein SJ, Frush DP, et al (2011). Managing radiation use in medical
imaging: a multifaceted challenge. Radiology; 258(3):889–905. doi:https://doi.org/10.1148/
radiol.10101157.
2. Bellolio MF, Heien HC, Sangaralingham LR, Jeffery MM, Campbell RL, Cabrera D, Shah ND,
Hess EP (2017). Increased Computed Tomography Utilization in the Emergency Department
and Its Association with Hospital Admission. West J Emerg Med;18(5):835–845. doi: https://
doi.org/10.5811/westjem.2017.5.34152. Epub 2017 Jul 19. PMID: 28874935; PMCID:
PMC5576619.
3. Maxwell S, Ha NT, Bulsara MK, et al (2021). Increasing use of CT requested by emergency
department physicians in tertiary hospitals in Western Australia 2003–2015: an analysis of linked
administrative data. BMJ;11:e043315. doi: https://doi.org/10.1136/bmjopen-2020-043315
4. Brenner DJ, Hall EJ (2007). Computed tomography: an increasing source of radiation expo-
sure. N Engl J Med;357(22): 2277-2284.
5. Berrington de Gonzalez A, Mahesh M, Kim KP, et al (2009). Projected cancer risks
from computed tomographic scans performed in the United States in 2007. Arch Intern
Med;169(22):2071-2077. doi:https://doi.org/10.1001/archinternmed.2009.440.
References 73
dose in CT in The Netherlands. Insights Imaging;4(3):383–390. doi:https://doi.org/10.1007/
s13244-013-0253-9
Australians. BMJ; 346:1–18. doi:https://doi.org/10.1136/bmj.f2360.
8. Hendee WR, O’Connor MK (2013). Radiation risks in medical imaging: separating fact from
fantasy. Radiology;264(2), 312–320. doi:https://doi.org/10.1148/radiol.12112678
9. Bushberg JT, Seibert JA, Leidholdt EM, Boone JM (2021) The Essential Physics of Medical
Imaging. 3rd ed. Philadelphia, PA: Lippincott Williams & Wilkins; A Wolters Kluwer Business.
10. Sykes PJ (2020). Until There Is a Resolution of the Pro-LNT/Anti-LNT Debate, We Should Head
Toward a More Sensible Graded Approach for Protection from Low-Dose Ionizing Radiation.
Dose Response; 18(2):1559325820921651. doi: https://doi.org/10.1177/1559325820921651.
PMI: 32425725; PMCID: PMC7218310.
11. The optimisation of radiological protection: broadening the process. ICRP Publication 101.
Approved by the Commission in September 2005. Ann ICRP. 2006;36(3):69–87.
12. Miglioretti DL, Johnson E, Williams A, et al (2013). The use of computed tomography in
pediatrics and the associated radiation exposure and estimated cancer risk. JAMA Pediatr;
167(8):700–707. doi:https://doi.org/10.1001/jamapediatrics.2013.311
13. Pearce MS, Salotti JA, Little MP, et al (2012). Radiation exposure from CT scans in child-
hood and subsequent risk of leukemia and brain tumors: a retrospective cohort study. Lancet;
80(9840):499–505. doi:https://doi.org/10.1016/S0140-6736(12)60815-0
14. Hall EJ, Brenner DL (2014) Cancer risks from diagnostic radiology. British Journal of
Radiology Vol. 81, No. 965
15. Shao YH, Tsai K, Kim S, Wu YJ, Demissie K. (2020). Exposure to Tomographic Scans and
Cancer Risks. JNCI Cancer Spectr;4(1):pkz072. doi: https://doi.org/10.1093/jncics/pkz072.
PMID: 32337490; PMCID: PMC7050152.
16. Moghadama N, Rehani M M, Nassiri MA (2021) Assessment of patients’ cumulative
doses in one year and collective dose to population through CT examinations. European
Journal of Radiology, Volume 142, September 2021, 109871. doi:https://doi.org/10.1016/j.
ejrad.2021.109871
17. Zewdea N, Riab F, Rehani MM (2022) Organ doses and cancer risk assessment in patients
exposed to high doses from recurrent CT exams. European J of Radiol; 149: 110224.
18. Hong J-Y, Han K, Jung J-H Jung, Kim S (2019). Association of Exposure to Diagnostic Low-
Dose Ionizing Radiation with Risk of Cancer Among Youths in South Korea. JAMA Network
Open;2(9):e1910584. doi:https://doi.org/10.1001/jamanetworkopen.2019.10584
19. Sakane H, Ishida M, Shi L, et al (2020). Biological Effects of Low-Dose Chest CT
on Chromosomal DNA. Radiology;295(2):439-445. DOI: https://doi.org/10.1148/
radiol.2020190389. PMID: 32154776.
20. Radiological protection and safety in medicine (2007). Recommendations of the International
Commission on Radiological Protection. Publication No. 103. Ann ICRP. 2007; 37:2–4.
21. McNitt-Gray MF (2002). AAPM/RSNA Physics Tutorial for Residents: Topics in CT:
Radiation dose in CT. Radiographics;22(6):1541–1553.
22. Seeram E (2023). Computed Tomography: Physical Principles, Patient Care, Clinical
Applications, and Quality Control. St Louis, MO. Elsevier
Korean J Radiol;13(1):1–11. doi:https://doi.org/10.3348/kjr.2012.13.1.1.
25. Reid J, Gamberoni J, Dong F, Davros W (2010). Optimization of kVp and mAs for pediatric
low dose simulated abdominal CT: Is it best to base parameter selection on object circumfer-
ence? AJR Am J Roentgenol;195(4):1015–1020. doi:https://doi.org/10.2214/AJR.09.3862.
26. Maldjian PD, Goldman AR (2013). Reducing radiation dose in body CT: primer on dose met-
rics and key CT technical parameters. AJR Am J Roentgenol; 200(4):741–747. doi:https://doi.
org/10.2214/AJR.12.9768
27. Christner JA, Zavaletta VA, Eusemann CD, Walz-Flannigan AL, McCollough CH (2010).
Dose reduction in helical CT: dynamically adjustable z-axis x-ray beam collimation. AJR Am
J Roentgenol;194(1):W49–W55. doi:https://doi.org/10.2214/AJR.09.2878.
28. Toth T, Ge Z, Daly MP (2007). The influence of patient centering on CT dose and image noise.
Med Phys;34(7):3091–3101.
29. Beister M, Kolditz D, Kalender WA (2012). Iterative reconstruction methods in x ray CT. Phys
Med;28(2):94–108. doi:https://doi.org/10.1016/j.ejmp.2012.01.003.
30. Kaza RK, Platt JF, Goodsitt MM, et al (2014). Emerging techniques for dose optimization in
abdominal CT. Radiographics; 34(1):4–17. doi:https://doi.org/10.1148/rg.341135038.
31. Hsieh J, Nett B, Yu Z, Sauer K, Thibault J-B, Bouman CA (2013). Recent advances in CT
image reconstruction [published online January 16, 2013]. Curr Radiol Rep. 2013. doi:https://
doi.org/10.1007/s40134-012-0003-7.
32. “Reduction”, “Optimization”. Merriam-Webster Dictionary. Online English Language
Dictionary. www.merriam – webster.com/dictionary. Accessed August 2022.
33. Strauss KJ, Goske MJ, Kaste SC, et al (2010). Image gently: Ten steps you can take to optimize
image quality and lower CT dose for pediatric patients. AJR Am J Roentgenol;194(4):868–873.
doi:https://doi.org/10.2214/AJR.09.4091
34. Karami V, Gholami M (2018) Addressing as low as reasonably achievable (ALARA) in pediat-
ric computed tomography (CT) procedures, J Res Med Dent Sci; 6 (5):104–114
35. Sohaib SA, Peppercorn PD, Horrocks JA, Keene MH, Kenyon GS, Reznek RH (2001).
The effect of decreasing mAs on image quality and patient dose in sinus CT. Br J
Radiol;74(878):157–161.
36. Russell MT, Fink JR, Rebeles F, Kanal K, Ramos M, Anzal Y (2008). Balancing radiation dose
and image quality: clinical applications of neck volume CT. Am J Neuoradiol;29(4):727- 731.
doi:https://doi.org/10.3174/ajnr.A0891.
ICRP Volume 46, Issue 1, pages 1–144
38. Seeram, E. and Brennan, P (2006). Diagnostic Reference Levels in Radiology. Radiologic
Technology; Vol. 77/No. 5; 373–384
39. Seeram, E. (2020). Rad Tech’s Guide to Radiation Protection, 2e. Oxford: Wiley.
40. American College of Radiology ACR-AAPM-SPR (2020). Practice parameter for diagnostic
reference levels and achievable doses in medical x-ray imaging, 2018. https://www.acr.org/−/
media/ACR/Files/Practice-Parameters/Diag-Ref-Levels.pdf (accessed 30 August 2020)
41. Geyer LL, Schoepf UJ, Meinel FG, et al (2015). State of the art: iterative CT reconstruction
techniques. Radiol;276(2):339–357. doi:https://doi.org/10.1148/radiol.2015132766
42. Boedeker K (2019). AiCE deep learning reconstruction: bringing the power of ultra-high reso-
lution CT to routine imaging. Canon Med Syst; 2:28–33.
43. Shan H, Zhang Y, Yang Q, et al. 3D convolutional encoder–decoder network for low-dose CT via
transfer learning from a 2D trained network. IEEE Trans Med Imaging. 2018;37(6):1522–1534.
doi:https://doi.org/10.1109/TMI.2018.2832217
44. Shan H, Padole A, Homavounieh F, et al. Competitive performance of a modularized deep
neural network compared to commercial algorithms for low-dose CT image reconstruction.
Nat Mach Intell. 2019;1:269–276. doi:https://doi.org/10.1038/s42256-019-0057-9
45. Akagi M, Nakamura Y, Higaki T, et al. Correction to: deep learning reconstruction improves
image quality of abdominal ultra-high-resolution CT. Eur Radiol. 2019;29(8):4526–4527.
doi:https://doi.org/10.1007/s00330-019-06249-x
46. Rajendran K, Petersilka M, Henning A, Shanblatt ER, Schmidt B., Flohr TG, Ferrero A,
Baffour F, Diehn FE, Yu L, Rajiah P, Fletcher JG, Leng S., McCollough CH (2021). First
Clinical Photon-counting Detector CT System: Technical Evaluation. Radiology, Published
Online, doi:https://doi.org/10.1148/radiol.2125794
References 75
47. Danielsson M, Persson M, and Sjölin M (2021). Photon-counting x-ray detectors for CT. Phys.
Med. Biol. 66; 03TR01, doi:https://doi.org/10.1088/1361-6560/abc5a5
48. Rajendran K, Voss BA, Zhou W, Tao S, DeLone DR, Lane JI, Weaver JM, Carlson ML,
Fletcher JG, McCollough CH, Leng S. Dose Reduction for Sinus and Temporal Bone Imaging
Using Photon-Counting Detector CT With an Additional Tin Filter. Invest Radiol. 2020
Feb;55(2):91–100. doi: https://doi.org/10.1097/RLI.0000000000000614. PMID: 31770297;
PMCID: PMC8522262.
49. Benson JC, Rajendran K, Lane JI, Diehn FE, Weber NM, Thorne JE, Larson NB, Fletcher
JG, McCollough CH, Leng S. A New Frontier in Temporal Bone Imaging: Photon-Counting
Detector CT Demonstrates Superior Visualization of Critical Anatomic Structures at Reduced
Radiation Dose. AJNR Am J Neuroradiol. 2022 Apr;43(4):579–584. doi: https://doi.
org/10.3174/ajnr.A7452. Epub 2022 Mar 24. PMID: 35332019; PMCID: PMC8993187.
50. Matthews K, Brennan P. Optimization of x-ray examinations: general principles and an Irish
perspective. Radiography. 2009; 15:262–268.
51. Seeram E (2014). Computed Tomography Dose Optimization. Radiologic Technology, 85 (4),
p 655–671
Chapter 6
Optimization Research in Medical Imaging
Keywords Research · Scientific method · Dose/image quality optimization ·

Image quality descriptors · Image quality assessment tools · Observer · Observer
performance methods · Visual grading analysis · Receiver operating characteristics
6.1 Introduction
Research is based on science (a systematic and organized body of knowledge in any

area of inquiry) and is conducted using the “scientific method” [1]. There are at least
four primary tasks in research, based on the scientific method, namely; identifica-
tion of a problem, data collection to help answer the question posed by the problem,
data analysis, and last, provide a solution to the problem. These four tasks require a
thorough and defined set of steps that ensure a valid solution to the problem. The
major steps are usually outlined as follows:
• Problem identification
• Literature review
• Research purpose statement
• Research Objectives, Questions and Hypothesis
• Research methodology and research design
• Data Collection
• Data analysis and interpretation
• Research report and evaluation
Research in medical imaging, and in particular, on dose/image quality optimization
is complex and it is based on the scientific method [2]. A dose/image quality study
requires that images be obtained at doses ranging from low to high. Subsequently,
observers assess these images to determine the optimum dose, that is the lowest
dose without compromising the image quality. In particular, McCollough et al. [3],
point out that image quality assessment is a complex process and requires signifi-
cant attention to quantitative and qualitative measures. Furthermore, McCollough

78 6 Optimization Research in Medical Imaging
et al. [3] stress that quantitative metrics such as image noise and observer perfor-
mance, be considered.
6.1.1 Strategies Used in Dose/Image Quality

Optimization Research
The literature includes formal approaches used to guide dose/image quality optimi-
zation research [4]. One important and significant seminal work is a special issue of
the scientific journal Radiation Protection Dosimetry [5]. This issue is dedicated to
the nature and strategies for optimization research in medical imaging, including
radiography, fluoroscopy, mammography, and computed tomography (CT).
Several studies featured in this special issue have identified at least four essential
requirements that must be met for an effective and thorough study on dose/image
quality optimization research study [5] These requirements include:
1. Patient safety must be a primary consideration
2. The level of image quality for the diagnostic task must be determined
3. Use dose levels ranging from high to low keeping in mind that image quality
must not be compromised.
4. Valid research methodologies [6, 7] must be used for:
• Dosimetry measurements
• Image acquisition
• Evaluation of image quality obtained at all dose levels, by human observers
keeping in mind the
• Nature of the detection task
Additional Considerations
There are several additional requirements as described in a paper by Seeram [8]

published in Radiologic Technology, for dose/image quality optimization research
in CT, that are considered minimum essential elements. These include:
1. Calibration of the CT imaging system, to ensure that the system performance is
reliable and consistent during dose measurements
2. Calibration of the dosimeters used to measure doses during imaging acquisition
3. Appropriate anthropomorphic phantoms and image quality test tools must be
used for image acquisition
4. Reliable and approved methodology for dose measurements
5. Assessment of images should occur at least in two phases:
A. Expert human observers assess image quality on a defined and specific crite-
rion, for example, the appearance of image noise
6.1 Introduction 79
B. The assessment of images recorded with various levels of dose should be

done using the same set of observers using established, valid, and reliable
observer performance methods
C. The selection of the appropriate observer performance method may be based
on the nature of the detection task [9]
D. Visual Grading Analysis (VGA) is a useful tool if the detection task is based
on the visualization of anatomic structures. VGA is based on the assumption
that an observer’s ability to see and evaluate normal anatomy correlates to the
ability to evaluate pathology, or abnormal findings [9]
E. Receiver Operating Characteristic (ROC) method is a useful tool for the
assessment of pathology detection [9]
A brief overview of VGA and ROC will be provided later in the Chapter.
6.1.2 The Nature of Image Quality
Dose optimization research seeks to select the lowest possible dose to the patient
without compromising the image quality required to make a diagnosis by the human
observer. In this regard then, image quality characteristics are important and must
be considered, as an integral component in any research study.
Image Quality Descriptors
In a recent paper on “The assessment of image quality and diagnostic value in X-ray
images: a survey on radiographers’ reasons for rejecting image”, Kjelle and Chilanga
[10] stress the following: “The quality of a radiographic image influences diagnos-
tic accuracy and subsequent clinical management of the patient [11]. A radiographic
image is accepted as good quality when certain technical qualities are satisfied, and
the image considered of diagnostic value” [12]. Good image quality in diagnostic
imaging can be assessed mainly by at least three image quality factors, namely;
Image quality in digital imaging includes spatial resolution, contrast-to-noise reso-
lution, and image noise. These will be reviewed below.
Dose optimization not only address reduction of radiation dose according to the
ALARA principle, it focusses also on image quality. In CT imaging for example,
there are at least four essential image quality parameters that relate to radiation dose
[13]. These include spatial resolution, contrast-to-noise resolution, image noise, and
they have been described in detail in the literature by several authors [3, 4, 6–9]. The
first three have been outlined in Chap. 5. In review, Goo [13] provides a useful sum-
mary as follows:
• Image Noise is a “random variation of CT numbers” [13] and it is “inversely
proportional to the square root of radiation dose; inversely proportional to tube
voltage; inversely proportional to the fourth power of the spatial resolution;

influenced by section thickness and reconstruction algorithm” [13]
• Contrast-to-noise resolution is the “ability to distinguish between different CT
numbers” [13]. The relationship with dose is “not significantly changed at differ-
ent tube voltages in most materials except for some with high atomic numbers
such as iodine (increased at low tube voltages), low contrast resolution greatly
influenced by image noise level” [13].
• Spatial resolution is the “ability to distinguish small details of objects” [13]. Its
relationship with dose is that it is “inversely proportional to focal spot size,
detector collimation; and influenced by the reconstruction algorithm” [13].
Furthermore, Goo [13] points out that “the required image quality of CT differs
somewhat among different diagnostic tasks. Consequently, the required CT radia-
tion dose is also fairly diverse and thus should be tailored according to clinical
indications. For example, very low dose CT can be used for the identification of
high-contrast lesions, such as urinary stones, colonic polyps (virtual colonography),
or lung nodules.”
6.1.3 Image Quality Assessment Tools for Clinical CT Images:

An Overview
Chipiga et al. [14], Zarb et al. [15] and Seeram et al. [7], have described image qual-
ity assessment tools for optimization of CT images. In particular, two approaches
have been identified;
1. Objective physical measures
2. Observer and diagnostic performance tests
While the former is based on the evaluation of physical and technical parameters of
the image, the latter is based on assessment of clinical images of patients, and/or
anthropomorphic phantoms.
Objective Physical Measures
The major physical characteristics of a digital imaging system (radiographic or CT)

are those that relate to the performance of the imaging system. These characteristics
and performance measures include the modulation transfer function, the Signal-to-
Noise Ratio (SNR), the Wiener spectrum (also known as the noise power spectrum),
and the Detective Quantum Efficiency (DQE). While these are useful and signifi-
cant tools to evaluate image quality, they do not relate to all components of the
imaging chain [16–18]. Furthermore, Tapiovaara [19], in a review of the relation-
ships between physical measurements and human observer evaluation of image
6.1 Introduction 81
quality, emphasized that “exact physical measurements cannot supersede subjective

evaluation in judging the acceptability of clinical images.”
Clinical medical imaging involves subjective evaluation of patient images using
human observers, usually radiologists. The technologist however, is the first
observer to assess image quality before images are sent to radiologists for diagnos-
tic interpretation. In general, while the major task of the technologist is to evaluate
the visualization and reproduction of relevant anatomical structures, the major task
of the radiologist is detection of subtle lesions (detection of pathology). The proce-
dures used for these two purposes fall into the category of observer performance
methods.
Observer Performance Methods
Mansson [16], in describing various approaches to evaluating clinical image quality,

identifies 4 tasks that the observer must consider. These include:
1. The differences between varying states of diseases and health should be dis-
cerned (perceived or recognized)
2. Relevant anatomical structures and features should be described accurately
3. Abnormalities should be classified accurately, and
4. Relevant anatomy on images should be distinguished reliably
While the first three goals fall within the viewing and interpretation task of the radi-
ologist, the last one (visualization and inclusion of relevant anatomical structures)
belongs to the viewing task of the radiologic technologist. To measure these tasks,
requires the use of observer performance methods.
There are at least two classes of observer performance methods used to asses
clinical diagnostic image quality, not only for the visualization and reproduction of
normal anatomical structures but also for lesion detection. These include:
1. Visual Grading Analysis (VGA) used in the visualization and reproduction of
normal anatomical structures
2. Receiver operating characteristic (ROC) analysis is used in lesion detection tasks.
The major features of each of these will be described in the next sections of this
Chapter.
6.1.4 Receiver Operating Characteristics: A Brief Overview
The ROC analysis is a commonly used approach to measure the performance of

the human observer on the task of the detection of lesions in patient images in
medical imaging [20, 21]. ROC analysis is complex and requires a firm under-
standing of several integral elements on which the analysis is based. For example,
in order to identify lesions (abnormalities) in an image, calls for an understanding
of two terms, namely; sensitivity and specificity. While sensitivity refers to the
detection of abnormalities when present, specificity refers to ruling out abnormali-
ties that are absent (false positive rate = FPR or 1). These two factors are funda-
mental to ROC analysis.
If the sensitivity (on the y-axis) is plotted as a function of specificity (on the
x-axis), an ROC curve is obtained, which shows the relationship between sensitivity
and FPR. Zarb et al. [15], summarises the plot as follows: “The area under the ROC
curve is the most common measure of the accuracy of the test. A test with an ROC
curve area of 1 is perfectly accurate with a sensitivity of 1 and FPR of 0. On the
other hand, an ROC curve area of 0 is perfectly inaccurate where all patients are
incorrectly given positive results. In practice the lower limit for the RO C curve area
is given at 0.5 which is equivalent to guessing the patient’s result hence a 50%
chance of getting the correct interpretation. ROC curve areas above 0.5 have some
ability to discriminate between patients having a disease and those who do not”.
It is not within the scope of this Chapter to outline the details of ROC analysis,
and therefore the reader is encouraged to refer to McEntee [22], for a comprehen-
sive and further updates on ROC analysis, including its variants, such as for exam-
ple; the free-response ROC, localization ROC, and alternative free-response ROC,
and so forth. In terms of the shortcomings of ROC analysis, Zarb et al. [15], states
the following: “The data has to be divided in only 2 categories, normal or abnormal;
a large number of images with subtle pathology are required to produce statistically
significant results; restricted to one observer per report per image; ROC method
does not cater for multiple abnormalities within the same image; abnormality local-
ization is not considered and hence an image may be marked as abnormal for the
wrong reason”. Furthermore, an important consideration, however, and within the
context of this book, “concerns have been raised regarding their clinical relevance”
[16, 17]. Additionally, Bath [17] notes that “conducting these types of studies may
be cumbersome, because they are based on the establishment of truth for all cases
and normally require a large number of cases in order to produce statistically sig-
nificant results, which means that the reliability is relatively low. For these reasons,
ROC-related methods may not be the method of choice for the local optimization
task at a radiation department where the intention is to find the optimal image pro-
cessing setting or (dose level) for a given examination”.
6.1.5 Visual Grading Analysis: An Overview
A number of authors [4, 6, 9, 15–18, 23] have demonstrated that visual grading
using VGA is also another valid useful tool for evaluating image quality. VGA is
well-established where the observer’s overall goal is to subjectively assess image
quality, based on the visualization and reproduction of anatomical structures at dif-
ferent dose levels. Sund et al. [18] point out that a critical assumption of VGA is that
“the visibility of normal anatomy correlates to the detectability of pathological
6.1 Introduction 83
structures.” Furthermore, Bath [17] identifies several reasons for using visual grad-
ing in image quality assessment studies. These include:
1. It is assumed that the validity of VGA studies is high for assessing visualization
of relevant anatomical structures.
2. “Visual grading methods have in special cases been shown to agree both with
detection studies using human observers and with advanced calculations of the
physical image quality. This is important, and validates in some way the assump-
tion that the possibility to detect pathology correlates to the reproduction of
anatomy—the basic ideas of visual grading. Discrepancies between the methods
have been reported but have been explained with the different tasks for the meth-
ods rather than low validity for visual grading” [17].
3. VGA research is simple to carry out (compare to ROC studies). “How to perform
visual grading studies has been extensively described and the learning threshold
for conducting such studies is low” [17]
4. VGA studies do not demand too much time on the part of the observer.
The use of VGA however, has been subject to concerns that since subjectivity is one
of the tasks of the observer in assessing images, it could lead to bias. To address
with this bias problem, a set of formal criteria have been developed by a group of
expert radiologists and medical physicists. These criteria are the European
Guidelines on Quality Criteria for Diagnostic Radiographic Images and were pub-
lished by the Commission of European Communities in 1996 [24].
It is not within the scope of this Chapter to present details of such criteria there-
fore, the reader is encouraged to refer to reference 24 for full details.
VGA Methods
The literature identifies 2 common VGA approaches, and as noted earlier, VGA
requires that the observer assess image quality in a subjective manner using an
absolute or a relative rating scale. While the former requires the observer to rate
his or her opinion about the visibility of the anatomical structures, the latter
requires the observer to rate the visibility of each defined anatomical structure.
For example, with the absolute rating scale the structure in the image is: not visi-
ble; poorly reproduced; adequately reproduced; or very well reproduced. With the
relative rating scale, the reproduction of the structure in the image is ___; the
reproduction of the corresponding structure in the reference image. The blank
space would be filled in with one of the selected choices as follows: much worse
than (−2); worse than (−1); the same as (0); better than (+1); or much better
than (+2).
An important consideration in conducting a VGA study is that in using the rela-
tive approach, the reference image should always be presented side by side on the
same monitor used to display the test images to ensure that these images are dis-
played with the same monitor brightness and contrast [25, 26].
6.1.6 Example of a Dose Optimization Study

in Digital Radiography
A study that illustrates all the elements of a typical dose optimization study in medi-
cal imaging, based on the scientific method is one by Seeram et al. [27]. The study
used a descriptive correlational research design to investigate “the optimization of
a of the exposure indicator (EI) of a computed radiography imaging system as a
radiation dose management strategy”. A brief summary by Seeram et al. [2] is as
follows:
1. “The study compared the optimized EI with the manufacturer’s recommended
values for the anteroposterior (AP) pelvis and AP lumbar spine projections. One
of the study’s objectives was to correlate the milliampere seconds (mAs) and
associated EIs with the entrance skin dose using a range of mAs settings while
holding the kilovoltage peak (kVp) constant.
2. A noninterventional research design was used to relate or associate variables in
a predictable manner as opposed to an interventional approach, which is typical
in experiments. The noninterventional approach used a correlational research
method to measure the degree of association (or relation) between 2 or more
variables using statistical procedures. Entrance skin dose and EI were the depen-
dent variables, and mAs was the independent variable.
3. The measurement of the entrance skin dose for the AP pelvis and the AP lumbar
spine using exposure technique factors including mAs and kVp that the manu-
facturer suggested. Researchers obtained the entrance skin dose measurements
with a calibrated dosimeter (Unfors Instruments) free-in-air (ie, an air environ-
ment free of any absorbing or scattering objects) for the AP pelvis and the AP
lumbar spine. For all 9 of the mAs settings, researchers calculated the mean
milligray (mGy) per mAs setting for the AP pelvis and AP lumbar spine.
4. Descriptive statistics (e.g., such as sample size, mean, standard deviation, and
range) were computed for the dosimetry and EI data. In addition, the Pearson
correlation was applied to examine the relationship between the entrance skin
dose and the EI as well as the dose and mAs. Data was graphed and plotted the
mean dose for the AP pelvis and the AP lumbar spine as a function of mAs.
Furthermore, the inverse EI, the mAs, and the dose were compared. The inverse
EI shows a strong positive linear relationship (r = 0.999) for both the AP pelvis
and the AP lumbar spine.
5. At doses of 5 mGy, 10 mGy, and 20 mGy, the EIs were 400, 200, and 100,
respectively. When researchers plotted the inverse EI as a function of mAs and
plotted the dose as a function of the inverse EI, they saw a strong positive linear
relationship (r = 0.999) in both cases. Instead of displaying the S number on an
image (e.g., 100), the inverse S number (e.g., 0.01) should be displayed to better
aid technologists in understanding the relationship between the dose and the
patient (as opposed to the image plate) as well as the S number.
References 85
6. The results of this study show that technologists can optimize the dose and image
quality in computed radiography imaging using the mAs exposure technique
factor and associated EIs. Specifically, the dosimetry phase of this investigation
showed a strong positive linear relationship (r = 0.999) between mAs and dose,
mAs and the inverse EI, and the inverse EI and dose for the AP pelvis and AP
lumbar spine. Under the controlled conditions researchers used in this study for
dose optimization, the EI values were stable. The manufacturer’s reference val-
ues of 25 mAs (EI = 86) for the AP pelvis and 50 mAs (EI = 88) for the AP
lumbar spine were optimized to 16 mAs for the AP pelvis (EI = 136) and 32 mAs
for the AP lumbar spine (EI = 139). Seven expert observers, who have at least
10 years of experience teaching radiographic technique and positioning in class-
rooms, laboratories, and hospitals, performed an image quality assessment and
determined that the manufacturer’s recommended dose can be reduced by 36%
for the AP pelvis and AP lumbar spine without compromising image quality.”
In conclusion, dose optimization in digital radiography and computed tomography
is an important responsibility of all those who play an active role in the imaging of
patients. It simply means that doses should be kept as low as is reasonably achiev-
able so as not to compromise the image quality needed for diagnosis. This is the
ALARA principle of the ICRP and other international and national radiation protec-
tion organizations.
References
1. Bhattacherjee, A. (2012). Social science research: Principles, methods, and practices. In

Textbooks collection. 3. http://scholarcommons.usf.edu/oa_textbooks/3. University of
South Florida
2. Seeram E, Davidson R, England A, McEntee M (Eds) (2022) Research in Medical Imaging
and Radiation Sciences. Switzerland AG, Springer Nature.
3. McCollough C, Bruesewitz MR, Kofler JM Jr. (2006). CT dose reduction and dose manage-
ment tools: overview of available options. Radiographics; 26(2):503–512.
4. Mattsson S, ed (2005) Optimization strategies in medical x-ray imaging. Radiat Protect
Dosimetry; 114(1–3):1–3. doi:https://doi.org/10.1093/rpd/nch580.
5. Radiation Protection Dosimetry. Optimization Strategies in Medical X-Ray Imaging, Vol 114,
No 1–3; 2005
6. Zarb F, Rainford L, McEntee MF (2010). Image quality assessment tools for optimization of
CT images. Radiography; 16:147–153.
7. Seeram E, Davidson R, Bushong S, Swan H (2014). Image Quality Assessment Tools for
Radiation Dose Optimization in Digital Radiography: An Overview. Radiologic Technology;
Volume 85, Number 5, 555–562
8. Seeram E (2014) Computed Tomography Dose Optimization. Radiologic Technology; Vol 85,
No 6, 655–671
9. Tingberg A, Båth M, Håkansson M, et al (2005). Evaluation of image quality of lumbar spine
images: a comparison between FFE and VGA. Radiat Protect Dosimetry; 114(1–3):53–61.
10. Kjelle, E., Chilanga, C. The assessment of image quality and diagnostic value in X-ray images:
a survey on radiographers’ reasons for rejecting images. Insights Imaging 13, 36 (2022).
https://doi.org/10.1186/s13244-022-01169-9
11. Carmichael JHE, Maccia C, Moores BM et al (2000) European guidelines on quality criteria
for diagnostic radiographic images. EU Publication
12. Maccia C, Moores BM, Wall BF (1997) The 1991 CEC trial on quality criteria for diagnos-
tic radiographic images: detailed results and findings. EN Luxembourg Office for Official
Publication of the European Communities
Korean J Radiol;13(1):1–11.
14. Chipiga LA, Berkovich GV, Vodovatov AV, Trufanov GE (2021). Comparison of different
approaches for image quality assessment in computer tomography. American Institute of Physics
(AIP) Conference Proceedings 2356, 020005 (2021); https://doi.org/10.1063/5.0052960
15. Båth M, Månsson LG (2007). Visual grading characteristic (VGC) analysis: a non-parametric
rank invariant statistical method for image quality evaluation. Br J Radiol;80(951):169–176.
16. Mansson L (2005). Methods for the evaluation of image quality: a review. Radiat Prot
Dosimetry; 90(1e2):89e99
17. Båth M (2010) Evaluating imaging systems: practical applications. Radiat Prot Dosimetry;
139(1–3):26–36. doi:https://doi.org/10.1093/rpd/ncq007.
18. Sund P, Båth M, Kheddache S, Månsson LG (2004). Comparison of visual grading analysis
and determination of detective quantum efficiency for evaluating system performance in digi-
tal chest radiography. Euro Radiol;14(1):48–58.
19. Tapiovaara MJ (2008). Review of relationships between physical measurements and user
evaluation of image quality. Radiat Prot Dosimetry; 129(1–3):244–248. doi:https://doi.
org/10.1093/rpd/ncn009.
20. Metz C (1986). ROC methodology in radiologic imaging. Invest Radiol; 21:720e33.
21. Chakraborty D (2006) ROC curves predicted by a model of visual research. Phys Med Biol;
51:3463e82.
22. McEntee M (Chapter 6, In Seeram E, Davidson R, England A, and McEntee M (eds) (2022)
Research in Medical Imaging and Radiation Sciences. Springer
23. Precht H, Hansson J, Outzen C, Hogg P, Tingberg A (2019) Radiographers’ perspectives
on Visual Grading Analysis as a scientific method to evaluate image quality., Radiography;
Volume 25, Supplement 1, Pages S14–S18
24. Commission of the European Communities (1996). European guidelines on quality criteria for
diagnostic radiographic images. European Commission; Brussels, Belgium.
25. Tingberg AM (2000). Quantifying the Quality of Medical X-ray Images: An Evaluation Based
on Normal Anatomy for Lumbar Spine and Chest Radiography [dissertation]. Lund, Sweden:
Lund University, Department of Radiation Physics.
26. Gorham S, Brennan PC (2010). Impact of focal spot size on radiologic image quality: a visual
grading analysis. Radiography;16(4):304–313. doi:https://doi.org/10.1016/j.radi.2010.02.007
27. Seeram E, Davidson R, Bushong S, Swan H. Optimizing the exposure indicator as a dose man-
agement strategy in computed radiography. Radiol Technol. 2016;87(4):380–391.
Chapter 7
Review Questions
Answer the following questions to check your understanding of the materials

studied.
Chapter 1
1. In which of the following effects, a threshold dose exists?
A. Leukemia
B. Cancer
C. Cataracts
D. Hereditary effects
2. There is no threshold dose for:
A. Leukemia.
B. Cataracts
C. Skin burns
D. Tissue damage
3. Which of the following data sources show that there is statistically significant
evidence of increased cancers in Japanese survivors who received doses of
100 mSv and higher?
A. RERF data
B. ICRP data
C. NCRP data
D. LNT hypothesis
4. The ICRP suggests that justification addresses:
A. The concept of a good image
B. Low dose to the patient
C. Image quality optimization
D. The concept of net benefit

88 7 Review Questions
5. Which of the following radiation protection guideline relates to the ALARA

philosophy?
A. Optimization
B. Dose limitation
C. Justification
D. Shielding
6. Optimization takes into consideration all of the following except:
A. Imaging equipment and its adequacy
B. Technical parameters of the examination,
C. Dose limits for occupationally exposed individuals
D. Diagnostic reference levels
7. The phenomenon of an increase in the exposure over time as technologists use
these DR imaging systems is referred to as:
A. Dose creep or exposure creep
B. Dose limits
C. Deterministic effects
D. Stochastic effects
8. Which of the following imaging modality contributes to the highest collective
amount of radiation exposure in the United States, as of 2011?
A. Digital radiography
B. Computed tomography
C. Digital fluoroscopy
D. Digital mammography
9. The term used to refer to “an act, process, or methodology of marking some-
thing (as a design, system, or decision) as fully perfect, functional, or effective
as possible” is:
A. Reduction
B. ALARA philosophy
C. Radiation protection
D. Optimization
10. The exposure indicator (EI) and the deviation index (DI) are technical factors
used in:
A. Digital Fluoroscopy
B. Computed Radiography
C. Flat Panel Digital Radiography
D. B and C are correct
7 Review Questions 89
Chapter 2
11. A linear response of a digital detector to radiation exposure means that:
A. Digital image processing ensures that the density and contrast of the dis-
played image appear acceptable to the observer, whether the exposure is
low or high
B. A specific exposure is needed to produce an acceptable image
C. The digital detector has better spatial resolution that film-screen (F-S)
detectors
D. The contrast resolution of the digital detector is less that F-S detectors
12. The exposure indicator (EI) in digital radiography refers to:
A. An image with acceptable contrast and density
B. A number to identify the use of the correct detector
C. A number to indicate that the correct exposure technique factors have
been used
D. A repeat exposure in required
13. The current visual cue on a digital image that shows the use of the correct expo-
sure technique factors is the:
A. EI
B. Deviation Index (DI)
C. Sharpness of the image
D. Correct contrast of the image structures
14. A digital radiography imaging system based on the principle of photostimula-
ble luminescence is:
A. Flat-panel digital radiography
B. Image-intensifier based digital fluoroscopy
C. Computed radiography imaging system
D. All are correct
15. Which photostimulable phosphors are typically used in CR imaging plates?
A. Cesium bromide
B. Barium fluorobromide
C. Barium Fluoroiodide
D. All are correct
16. An indirect conversion flat-panel digital radiography (FPDR) detector uses:
A. A photocathode to convert x-rays to light
B. An analog-to-digital converter to convert light to electrical signal
C. Selenium to convert x-rays to electrical signal
D. A scintillator to convert x-rays to light
17. A direct conversion FPDR detector uses:

A. A photocathode to convert x-rays to light
B. An analog-to-digital converter to convert light to electrical signal
C. Selenium to convert x-rays to electrical signal
D. A scintillator to convert x-rays to light
18. The preprocessing feature characteristic of digital radiography imaging sys-
tems is mainly used to:
A. Identify, correct, and scale the raw image data prior to image display for the
purpose of viewing and interpretation
B. Enhance image contrast
C. Enhance density and sharpness of the raw data
D. Allow the technologist to use windowing to change image contrast and
brightness
19. Which of the following digital postprocessing operation changes the sharpness
of the displayed image?
A. Window width processing
B. Window level processing
C. Spatial frequency processing
D. All are correct
20. Which of the following refers to the expected value of the exposure index when
exposing the x-ray image receptor properly?
A. EI
B. Target EI (EIT)
C. DI
D. Volume of interest (VOI)
21. The number quantifying the deviation of the actual exposure index from a target
exposure index in digital radiography is the:
A. EI
B. Target EI (EIT)
C. DI
22. A measure of the detector response to radiation in the relevant image region of
an image acquired with a digital x-ray imaging system is referred to as the:
A. EI
B. Target EI (EIT)
C. DI
23. Which of the following DI numbers of a DR system is able to deliver the EIT
established by the department?
A. A DI = +3
B. A DI = −3
C. A DI value between +1 and −1
D. A DI value between +2 and −2
24. The dose in digital radiography is directly proportional to the:
A. kV3
B. mA2
C. mAs × kV
D. mAs
25. The dose in digital radiography is also directly proportional to the:
A. Square of the ratio of the kV
B. Square of the ratio of the mAs
C. Distance of the patient from the x-ray tube
D. mAs × kV
26. The American College of Radiology (ACR) defines the Diagnostic Reference
Level as:
A. An investigation level to identify unusually high radiation dose or exposure
levels for common diagnostic medical x-ray procedures
B. A reference dose level for chest x-ray examination
C. A dose limit for the patient
D. A radiation protection principle to keep all exposures as low as reasonably
achievable
27. The following percentile point used for DRLs indicate that image quality may
be compromised:
A. 100th percentile
B. 75th percentile
C. 25th percentile
D. None of the above
28. The ACR-AAPM-SPR for example, recommends a DRL for an adult PA chest
(23 cm thickness) with grid to be:
A. 0.15 mGy
B. 3.4 mGy
C. 4.2 mGy
D. 5.0 mGy
Chapter 3
29. According to the International Commission on Radiological Protection (ICRP),
the second part of optimization (the first part being observing the ALARA prin-
ciple) must include the:
A. Imaging equipment and adequacy of the equipment
B. Technical Parameters
C. DRLs
D. All of the above
30. The following are optimization strategies used in digital radiography except:
A. Optimization of the exposure technique factors
B. Optimization of the EI and DI
C. Optimization of image processing
D. Patient shielding
31. The major conclusion of research on the optimization of kV in digital radiogra-
phy showed that:
A. kV does not influence the magnitude of the dose to the patient
B. The patient dose cannot be controlled using kV
C. The kV can be reduced without compromising the image quality
D. The optimum kV for all digital radiography examinations is 120 kV
32. An optimization of the EI in digital radiography study by Seeram et al. [16],
essentially concluded that:
A. Operators can optimize the dose to the patient by manipulating the mAs and
associated EIs
B. The EI is the patient dose
C. An EI value of 300 indicates that the image quality is poor
D. All of the above
33. Research examining the use of clinical EI, Target EI (EIT) and DI as tools for
optimizing the patient dose concluded that:
A. Optimization is possible only if the EIT is periodically and properly updated
by the department
B. EIT is not related to image quality optimization required by departments
C. The kV and mAs are the only factors influencing optimization of dose in
D = digital radiography
D. The DI cannot be used as a dose reduction strategy
34. Research on the use of image processing as an optimization strategy in digital
radiography generally show that:
A. Preprocessing is can be used to optimize the dose
B. Multifrequency processing can allow a low-dose image with poor image
quality (e.g., high noise) to be processed using specialized software to
improve visibility of structures
C. Optimizing image processing parameters can significantly reduce the dose

without compromising image quality
D. B and C are correct
35. Which of the following is a useful tool for optimization of radiation protection
to identify high patient doses for various common examinations using specific
imaging equipment?
A. The DI and the EI
B. The EIT
C. DRLs
D. All of the above
Chapter 4
36. The first clinically useful CT scanner was invented by:
A. Allan Cormack
B. Godfrey Hounsfield
C. Siemens Medical Systems
D. Lambert and Beer
37. Radiation attenuation is defined as:
A. The calculation of CT numbers
B. The production of scattered radiation
C. A decrease in the intensity of x-rays as they pass through an object (patient).
D. An increase in the intensity of x-rays as they pass through an object
I I0 e x
38. The algebraic expression shown above is important to CT, and is know as:
A. Eulers law
B. Hounsfield’s law
C. Lambert-Beer’s law
D. Law of x-ray intensities
39. The CT number is:
A. A number calculated for each voxel (volume element) in the slice of tissue
to be imaged based on the attenuation coefficients of the tissues within
the voxel.
B. A number that denotes how many attenuation coefficients are in the tis-
sue voxel
C. A number indicating how many electrons are in the tissue voxel
D. A number used by the pioneer Hounsfield to describe the degree of radia-
tion transmission through the patient
40. Which of the following algebraic expressions is used to calculate CT numbers,

where the attenuation coefficients for tissue is μt; and the attenuation coefficient
for water is μw? The manufacturer’s contrast factor is K
A. CT = μw − μt/μw × K
B. CT = μt − μw/μw. × K
C. CT = μt − μw/μw × K.
D. CT = μw + μt × K
41. The primary purpose of a bow-tie filter in CT is to:
A. Shape the x-ray beam from the tube to make it more uniform at the detector
B. Filter the x-ray beam to reduce the dose to the patient
C. The ensure that the x-ray photons are filtered to get rid of high energy photons
D. All of the above
42. Which of the following algorithms uses a convolution filter (digital filter) to
solve the major problem of the first CT reconstruction algorithm?
A. Back projection algorithms
B. Filtered back projection (FBP) algorithms.
C. Iterative reconstruction (IR) algorithms
D. Artificial Intelligence (AI) based algorithms
43. Iterative algorithms were developed to:
A. Reduce the noise from low dose CT imaging
B. Allow technologists to scan the patient very quickly
C. Minimize radiation dose to patients
D. A and C are correct.
44. AI-based CT image reconstruction algorithms are intended to:
A. Overcome the problems (image noise) of the FBP algorithm in low-dose
CT imaging
B. Overcome the problems of noise texture of IR algorithms in low-dose
CT imaging
C. Improve image quality, dose performance, and reconstruction speed com-
pared with iterative reconstruction algorithms
D. All of the above
45. A generalized framework for an AI-based such as Deep Learning Image
Reconstruction (DLIR) algorithm include:
A. Development of the algorithm
B. Train and optimize the algorithm
C. Performance verification of the algorithm
D. All of the above
46. Multislice CT (MSCT) data acquisition involves:

A. Stop-and-go data acquisition
B. Slice-by-slice data acquisition
C. Move the table to scan a volume of tissue using a 1Dimensional (1D) detec-
tor array that rotates at the same time the table moves through the gantry
aperture
D. Move the table while the x-ray tube and detectors rotate simultaneously
around the patient using 2D detector array to collect multiple slices per
revolution.
47. Which of the following CT detectors convert x-ray photons directly into elec-
tron hole pairs (electric charge)?
A. Energy-integrating detectors
B. Photon counting detectors
C. Scintillation detectors
D. Dual layer detectors
48. The following are selectable scan parameters except:
A. Scan mode, and exposure factors (kV, mA, and scan time),
B. Gantry rotation time,
C. Pitch, scan length, collimation, and slice width.
D. Placement of the body part in the isocenter of the gantry aperture
49. The International Electrotechnical Commission (IEC) defines the pitch in a
multi-slice scanner as:
A. P = distance the table travels per rotation (d)/total collimation (W).
B. P = d + W
C. P = d × W
D. P = W/d
Chapter 5
50. Which of the following is the current international standard for estimating CT
radiation dose?
A. Thermo-luminescent dosimetry (TLD)
B. Computed Tomography Dose Index (CTDI) dosimetry
C. Pencil ionization chamber dosimetry
D. Dose-Area Product dosimetry
51. The CTDIvolume (CTDIvol) can be calculated using the following algebraic
expression:
A. CTDIvol = Weighted CTDI (CTDIw)/pitch
B. Pitch × CTDIw
C. Pitch × CTDIvol
D. CTDIvol × CTDIw × Pitch
52. In CT, the dose length product (DLP) is calculated using the following alge-
braic expression:
A. DLP = Pitch x Scan Length (L)
B. DLP = Pitch x CTDIvol
C. DLP = CTDIvol × L
D. DLP = CTDIw × Pitch
53. The quantity used in CT dose reporting to relate radiation exposure to risk is:
A. Effective Dose (ED)
B. Absorbed Dose
C. Entrance Surface Exposure (ESE)
D. CTDI
54. Which of the following image quality factors in CT addresses the random varia-
tion of CT numbers in the image?
A. Spatial resolution
B. Contrast-to-Noise resolution
C. Noise
D. Artifacts
55. Noise in CT is:
A. Inversely proportional to square root of radiation dose
B. Inversely proportional to tube voltage
C. Inversely proportional to fourth power of spatial resolution
D. All of the above
56. In CT, the ability to distinguish between different CT numbers is defined as the:
A. Contrast-to-noise resolution
B. Noise
C. Spatial resolution
D. Artifacts
57. Spatial resolution in CT is defined as:
A. The ability to distinguish small details of object
B. The ability to distinguish between different CT numbers.
C. A random variation of CT numbers
D. Directly proportional to focal spot size and detector collimation
58. The key relationships that are noteworthy in CT dose optimization are:
A. Dose is directly proportional to the mAs, that is, doubling the mAs doubles
the dose
B. The dose is proportional to the square of the kilovolt peak (kV2).
C. Dose is inversely related to the pitch (P).
D. All of the above
59. Which of the following is an effort to reduce the dose to the patient at the begin-
ning and end of the scanning?
A. Overbeaming
B. Overranging
C. Adaptive or dynamic collimation
D. Additional bow tie filtration of the primary beam from the x-ray tube
Chapter 6
60. Objective physical measures in dose optimization studies are used to assess
A. Modulation transfer function (MTF) of the imaging system
B. Signal-to-Noise Ratio (SNR), the Noise power spectrum (Wiener spectrum)
C. Detective Quantum Efficiency (DQE).
D. All of the above
61. In a dose optimization study, which approach is best suited to assess clinical
image quality?
A. Objective physical measures
B. Observer performance methods
C. Subjective impression of image quality parameters seen on clinical images
D. B and C of the above
62. Which of the following approaches is best suited for lesion detection in optimi-
zation research?
A. Visual Grading Analysis (VGA)
B. Receiver Operating Characteristics (ROC)
C. Increased intensity of the viewing monitor
D. Use of a smoothing digital filter
63. Which of the following approaches is best suited for visualization and repro-
duction of anatomical structures in optimization research?
A. Visual Grading Analysis (VGA)
B. Receiver Operating Characteristics (ROC)
C. Increased intensity of the viewing monitor
D. Use of a smoothing digital filter
Answers to Review Questions
1. C
2. A
3. A
4. D
5. A
6. C
7. A
8. B
9. D
10. D
11. A
12. C
13. B
14. C
15. D
16. D
17. C
18. A
19. C
20. B
21. C
22. A
23. C
24. D
25. A
26. A
27. C
28. A
29. D
30. D
31. C
32. A
33. A
34. D
35. C
36. B
37. C
38. C
39. A
40. C
41. A
42. B
43. D
44. D
45. D
46. D
47. B
48. D
49. A
50. B
51. A
52. C
53. A
54. C
55. D
56. A
57. A
58. D
59. C
60. D
61. B
62. B
63. A
Index
A Digital radiography (DR), 4–6, 9, 13–23,

Attenuation, 41–45, 93, 94 25–37, 41, 58, 65, 84–85, 88–92
Direct conversion digital radiography, 18–19
Dose and image quality optimization, 71
B Dose creep in digital radiography, 6–8
Biological effects of radiation exposure, 1–2 Dose/image quality optimization, 77–85
Dose in digital radiography, 22, 30, 91
Dose optimization, 4, 6, 19, 22, 23, 30, 31,
C 33–37, 41, 42, 47, 53, 54, 58–61, 63,
Cancer risks from CT imaging, 59, 60 65–66, 72, 79, 84–85, 97
Computed radiography (CR), 5, 13, 15–17, Dose optimization in digital radiography, 4,
19–21, 26–29, 84, 85, 88, 89 6–8, 22, 35, 85
Computed tomography (CT), 4–9, 37, 41–54, Doses in computed tomography (CT), 6–9, 41,
57–72, 78–81, 85, 88, 93–96 46, 47, 53, 57–72, 78, 80
CT dose optimization, 50, 61, 63, 65,
71, 72, 96
CT dose reduction, 65, 71–72 E
CT numbers, 43–44, 63, 79, 80, 93, 94, 96 Exposure indicator (EI), 6, 9, 15, 20–22,
25–31, 37, 84, 85, 88–90, 92, 93
Exposure technique factors, 4, 9, 15, 25–28,
D 37, 63, 84, 85, 89, 92
Data acquisition, 5, 16, 41, 42, 44–46,
50, 51, 95
Deep learning image reconstruction algorithms F
in CT, 48, 49 Flat-panel digital radiography (FPDR), 5, 13,
Deep learning image reconstruction (DLIR), 15, 17–21, 26, 89, 90
48–50, 68, 94
Detector response to radiation, 21, 90
Deterministic effects, 2, 3, 6, 58, 61, 88 I
Deviation index (DI), 9, 15, 20–22, 25, 30–31, Image display, 5, 13, 16, 19, 42, 49, 90
37, 88–93 Image post processing, 25, 35, 42
Diagnostic reference level (DRL), 4, 22–23, Image processing, 6, 9, 13, 14, 16, 17, 19–20,
25, 31, 36–37, 61, 66–67, 88, 91–93 22, 31–33, 44, 46, 82, 89, 92, 93
© The Editor(s) (if applicable) and The Author(s), under exclusive license to 101
Springer Nature Switzerland AG 2023
https://doi.org/10.1007/978-3-031-22871-1
102 Index
Image quality assessment tool, 33, 80–81 Optimization, 1–9, 22, 25–37, 48, 49, 65–72,
Image quality descriptors, 79–80 78, 80, 82, 84, 87, 88, 92, 93, 97
Image reconstruction algorithms, 46, 48–49, Optimization strategies, 4, 9, 19, 25–37,
51, 63, 68, 69, 94 57–72, 92
Image reconstruction optimization, 47
Image storage, 42, 49
Indirect conversion digital radiography, P
17–18 Photon counting detectors (PCD), 51–52, 66,
International Commission on Radiological 70–71, 95
Protection (ICRP), 2–4, 13, 22, 23, 25, Photostimulable luminescence, 16, 89
36, 61, 65–67, 85, 87, 92 Physical principles, 5, 15–19, 50
Iterative reconstruction algorithms, 46–47, 49, Pitch, 9, 53, 54, 62–65, 95, 96
65, 70, 94 Principles of radiation protection, 2–6
J R
Justification, 3, 87, 88 Radiation protection, 1–9, 22, 25, 36, 37, 41,
53–54, 59, 61, 65–67, 71, 78, 85,
88, 91, 93
K Radiation risks, 1, 6, 8, 58, 59
Kilovolt (kV), 6, 64, 96 Receiver operating characteristic (ROC), 79,
kV optimization, 26, 27, 37, 92 81–83, 97
Research, 1, 4, 9, 22, 26–28, 30, 33, 57, 58,
71, 72, 77–85, 92, 97
M
Milliampere-seconds (mAs), 6
Multifrequency processing, 32–35, S
37, 92 Scientific method, 77, 84
Multislice CT detectors, 41, 64, 95 Selectable scan parameters, 50, 53, 95
Multislice CT principles, 50–53 Standardized exposure indicator, 19–21
Stochastic effects, 2, 3, 6, 58, 61, 62, 88
N
Noise reduction algorithms, 9, 31–35, 37 T
Tube current modulation, 65
O
Observer, 13, 19, 29, 32, 49, 65, 77–83, V
85, 89 Visual grading analysis (VGA), 33, 65, 79,
Observer performance method, 65, 79, 81, 97 81–83, 97

Dose Optimization in Digital Radiography and Computed Tomography An Essential Guide 3031228707 9783031228704 - Compress

Uploaded by

Copyright:

Available Formats

Dose Optimization in Digital Radiography and Computed Tomography An Essential Guide 3031228707 9783031228704 - Compress

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

Dose Optimization in Digital Radiography and Computed Tomography An Essential Guide 3031228707 9783031228704 - Compress

Uploaded by

Copyright:

Available Formats

Euclid Seeram

Dose Optimization in Digital

ISBN 978-3-031-22870-4 ISBN 978-3-031-22871-1 (eBook)

There are seven chapters in this book as follows:

Burnaby, BC, Canada Euclid Seeram

personal communication with Dr. Hounsfield, and he graciously provided me with

Burnaby, BC, Canada Euclid Seeram

5.8 The Role of the Technologist in CT Dose Reduction

Keywords Biological effects of radiation exposure: International Commission on

1.1 Biological Effects of Radiation Exposure

Biological effects of radiation exposure are sometimes referred to as health effects,

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 1

1.2 Fundamental Principles of Radiation Protection

In order to protect patients undergoing medical imaging procedures using ionizing

1.2.1 The Principle of Justification

The principle of justification is described as follows by the ICRP:

1.2.2 The Principle of Optimization

Following justification, the principle of optimization is applied. The ICRP states

of images are sufficient to obtain the information needed for diagnosis or

 igital Radiography and Computed Tomography: System Components

Dose optimization in digital radiography and computed tomography (CT) requires

1.3 Motivation for Dose Optimization in Digital

1.3.2 Dose Creep in Digital Radiography

ICU CXR Exposures

ICU CXR Exposures

trend. This intervention demonstrated conclusively that dose creep is a challenge

1.3.3 Dose in Computed Tomography

1.4 Optimization Strategies: An Overview

In addressing optimization strategies, and operating within the ALARA principle,

5. ICRP (2013). Radiological protection in paediatric diagnostic and interventional radiology.

Keywords Digital radiography · Detector response to radiation · Computed

2.1 Digital Radiography: Two Notable Facts Related

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 13

LOW Radiation Exposure to the Detector HIGH

2.2 Digital Radiography Systems: A Technical Review

2.2.1 Computed Radiography: Physical Principles

CR imaging is based on the physical principle of photostimulable luminescence, a

2.2.2 Flat-Panel Digital Radiography: Physical Principles

Indirect Conversion Digital Radiography

Figure 2.6 illustrates the major components of a flat-panel digital radiography

Direct Conversion Digital Radiography

A schematic of the components of a direct conversion digital radiography system is

2.3 Other Considerations Related to CR and FPDR Systems

Dose optimization in DR requires other considerations such as image processing,

Digital image processing is an essential feature of digital radiography imaging sys-

2.3.2 The Standardized Exposure Indicator

of Physicists in Medicine (AAPM) [9] in collaboration with the Medical Imaging

The results of this calculation indicate the following [5, 9]:

2.4 Factors Affecting the Dose in Digital Radiography:

2.5 Diagnostic Reference Levels

2.5.1 Definitions and Major Guidelines

5. International Electrotechnical Commission (2008). IEC 62494-1 ed.1 Medical electrical

Keywords Dose optimization in digital radiography · Exposure technique factors

As reviewed in Chap. 1, optimization is a radiation protection principle, first pro-

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 25

3.2 Optimization of Exposure Technique Factors

The evolution and continued development of digital radiography imaging systems

3.2.1 Optimization of kV and mAs

One of the earliest reviews entitled “Optimization research exposure technique

Burnaby, BC, Canada Euclid Seeram

Burnaby, BC, Canada Euclid Seeram

5.8 The Role of the Technologist in CT Dose Reduction

1.1 Biological Effects of Radiation Exposure

1.2 Fundamental Principles of Radiation Protection

1.2.1 The Principle of Justification

1.2.2 The Principle of Optimization

igital Radiography and Computed Tomography: System Components

1.3 Motivation for Dose Optimization in Digital

1.3.2 Dose Creep in Digital Radiography

1.3.3 Dose in Computed Tomography

1.4 Optimization Strategies: An Overview

2.1 Digital Radiography: Two Notable Facts Related

2.2 Digital Radiography Systems: A Technical Review

2.2.1 Computed Radiography: Physical Principles

2.2.2 Flat-Panel Digital Radiography: Physical Principles

Indirect Conversion Digital Radiography

Direct Conversion Digital Radiography

2.3 Other Considerations Related to CR and FPDR Systems

2.3.2 The Standardized Exposure Indicator

2.4 Factors Affecting the Dose in Digital Radiography:

2.5 Diagnostic Reference Levels

2.5.1 Definitions and Major Guidelines

3.2 Optimization of Exposure Technique Factors

3.2.1 Optimization of kV and mAs

3.3 Optimization of the Exposure Indicator

3.4 Optimization of the Deviation Index

3.5 Optimization Using Image Postprocessing Algorithms

3.5.1 Optimization Studies Using Multifrequency Processing

3.6 Diagnostic Reference Levels: A Useful Tool

4.2 The CT Scanner: Fundamental Physics and Major

4.2.1 Attenuation Physics: An Essential Overview

4.2.2 Attenuation and CT Numbers

4.2.3 Data Acquisition: System Components and Principles

Iterative Reconstruction Algorithms

Artificial Intelligence in CT Image Reconstruction

Deep Learning CT Image Reconstruction Algorithms

4.3 Multi-slice CT Principles at a Glance

4.3.1 MSCT Detectors: An Overview

Scintillation Detectors and Photon Counting Detectors

MSCT Detector Designs

4.3.2 Selectable Scan Parameters

5.2 Motivation for Dose Optimization in CT

5.2.1 Literature on Cancer Risks from CT

5.2.2 Effects of Low-Dose Chest CT on Chromosomal DNA

5.3 Radiation Protection Principles

5.4 Elements of CT Dosimetry at a Glance

DLP CTDI vol L

5.5 Factors Affecting the Dose in CT: An Overview

5.5.1 Image Quality Considerations

5.5.2 Significant Dose Parameters

5.6 Dose Optimization Strategies in CT

5.7 Other Useful Tools for Dose Reduction

5.7.1 The Use of DRLs

5.7.2 The Use of AI-Based Image Reconstruction