Cone Beam Computed Tomography

Cone Beam Computed Tomography Oral and Maxillofacial Diagnosis and Applications

Cone Beam Computed Tomography Oral and Maxillofacial Diagnosis and Applications Edited by David Sarment, DDS, MS

This edition first published 2014 © 2014 by John Wiley & Sons, Inc Editorial Offices 1606 Golden Aspen Drive, Suites 103 and 104, Ames, Iowa 50010, USA The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, UK 9600 Garsington Road, Oxford, OX4 2DQ, UK For details of our global editorial offices, for customer services and for information about how to apply for permission to reuse the copyright material in this book please see our website at www.wiley.com/wiley-blackwell. Authorization to photocopy items for internal or personal use, or the internal or personal use of specific clients, is granted by Blackwell Publishing, provided that the base fee is paid directly to the Copyright Clearance Center, 222 Rosewood Drive, Danvers, MA 01923. For those organizations that have been granted a photocopy license by CCC, a separate system of payments has been arranged. The fee codes for users of the Transactional Reporting Service are ISBN-13: 978-0-4709-6140-7/2014. Designations used by companies to distinguish their products are often claimed as trademarks. All brand names and product names used in this book are trade names, service marks, trademarks or registered trademarks of their respective owners. The publisher is not associated with any product or vendor mentioned in this book. The contents of this work are intended to further general scientific research, understanding, and discussion only and are not intended and should not be relied upon as recommending or promoting a specific method, diagnosis, or treatment by health science practitioners for any particular patient. The publisher and the author make no representations or warranties with respect to the accuracy or completeness of the contents of this work and specifically disclaim all warranties, including without limitation any implied warranties of fitness for a particular purpose. In view of ongoing research, equipment modifications, changes in governmental regulations, and the constant flow of information relating to the use of medicines, equipment, and devices, the reader is urged to review and evaluate the information provided in the package insert or instructions for each medicine, equipment, or device for, among other things, any changes in the instructions or indication of usage and for added warnings and precautions. Readers should consult with a specialist where appropriate. The fact that an organization or Website is referred to in this work as a citation and/or a potential source of further information does not mean that the author or the publisher endorses the information the organization or Website may provide or recommendations it may make. Further, readers should be aware that Internet Websites listed in this work may have changed or disappeared between when this work was written and when it is read. No warranty may be created or extended by any promotional statements for this work. Neither the publisher nor the author shall be liable for any damages arising herefrom. Library of Congress Cataloging-in-Publication Data Cone beam computed tomography : oral and maxillofacial diagnosis and applications / [edited by] David Sarment. p.; cm. Includes bibliographical references and index. ISBN 978-0-470-96140-7 (pbk. : alk. paper) – ISBN 978-1-118-76902-7 – ISBN 978-1-118-76906-5 (epub) – ISBN 978-1-118-76908-9 (mobi) – ISBN 978-1-118-76916-4 (ePdf) I. Sarment, David P., editor of compilation. [DNLM: 1. Stomatognathic Diseases–radiography. 2. Cone-Beam Computed Tomography–methods. WU 140] RK309 617.5′22075722–dc23 2013026841 A catalogue record for this book is available from the British Library. Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in electronic books. Cover design by Jen Miller Designs Set in 9.5/11.5pt Palatino by SPi Publisher Services, Pondicherry, India 1 2014

To my wife Sylvie To my children Lea, Myriam, and Nathanyel

Contents Contributors ix 7 Implant Planning Using Cone Beam 127 Preface xi Computed Tomography Acknowledgments xiii David Sarment 1 Technology and Principles of Cone 8 CAD/CAM Surgical Guidance Using Beam Computed Tomography Matthew W. Jacobson 3 Cone Beam Computed Tomography 147 George A. Mandelaris and Alan L. Rosenfeld 2 The Nature of Ionizing Radiation and 25 9 Assessment of the Airway and 197 the Risks from Maxillofacial Cone Supporting Structures Using Cone Beam Computed Tomography Beam Computed Tomography Sanjay M. Mallya and Stuart C. White David C. Hatcher 3 Diagnosis of Jaw Pathologies Using 43 10 Endodontics Using Cone Beam 211 Cone Beam Computed Tomography Computed Tomography Sharon L. Brooks Martin D. Levin 4 Diagnosis of Sinus Pathologies Using 65 11 Periodontal Disease Diagnosis Using 249 Cone Beam Computed Tomography Cone Beam Computed Tomography Aaron Miracle and Christian Güldner Bart Vandenberghe and David Sarment 5 Orthodontic and Orthognathic Planning 91 Index 271 Using Cone Beam Computed Tomography Lucia H. S. Cevidanes, Martin Styner, Beatriz Paniagua, and João Roberto Gonçalves 6 Three-Dimensional Planning in Maxillofacial Reconstruction of Large Defects Using Cone Beam Computed Tomography 109 Rutger Schepers, Gerry M. Raghoebar, Lars U. Lahoda, Harry Reintsema, Arjan Vissink, and Max J. Witjes vii

Contributors Sharon L. Brooks, DDS, MS Private practice Professor Emerita, Department of Periodontics Diagnostic Digital Imaging Sacramento, California, USA and Oral Medicine University of Michigan School of Dentistry Matthew W. Jacobson, MSc, PhD Ann Arbor, Michigan, USA Senior Research Scientist Xoran Technologies, Inc. Lucia H. S. Cevidanes, DDS, MS, PhD Ann Arbor, Michigan, USA Assistant Professor, Department of Orthodontics University of Michigan School of Dentistry Lars U. Lahoda, MD, PhD Ann Arbor, Michigan, USA Plastic surgeon, Department of Plastic Surgery University of Groningen and University Medical João Roberto Gonçalves, DDS, PhD Assistant Professor, Department of Pediatric Dentistry Center Groningen Faculdade de Odontologia Groningen, the Netherlands Universidade Estadual Paulista, Araraquara, Brazil Martin D. Levin, DMD Christian Güldner, MD Diplomate, American Board of Endodontics Specialist in ENT, Department of ENT, Head Chair, Dean’s Council and Adjunct Associate and Neck Surgery Professor of Endodontics University of Marburg University of Pennsylvania, School of Dental Germany Medicine David C. Hatcher, DDS, MSc, MRCD(c) Philadelphia, Pennsylvania, USA Adjunct Professor, Department of Orthodontics University of the Pacific School of Dentistry Private practice San Francisco, California, USA Chevy Chase, Maryland, USA Clinical Professor, Orofacial Sciences Sanjay M. Mallya, BDS, MDS, PhD University of California–San Francisco School Assistant Professor and Postgraduate Program of Dentistry Director San Francisco, California, USA Oral and Maxillofacial Radiology University of California–Los Angeles School of Clinical Professor Roseman University College of Dental Medicine Dentistry Henderson, Nevada, USA Los Angeles, California, USA ix

x Contributors George A. Mandelaris, DDS, MS David Sarment, DDS, MS Diplomate, American Board of Periodontology Diplomate, American Board of Periodontology Private practice Private practice Periodontics and Dental Implant Surgery Implantology and Periodontics Park Ridge and Oakbrook Terrace, Illinois, USA Alexandria, Virginia, USA Clinical Assistant Professor, Department of Oral Rutger Schepers, DDS, MD and Maxillofacial Surgery Maxillofacial Surgeon, Department of Oral and Louisiana State University School of Dentistry Maxillofacial Surgery New Orleans, Louisiana, USA University of Groningen and University Medical Aaron Miracle, MD Center Groningen Resident physician, Department of Radiology and Groningen, the Netherlands Biomedical Imaging Martin Styner, PhD University of California–San Francisco Associate Professor San Francisco, California, USA Department of Computer Science University of North Carolina Beatriz Paniagua, PhD Chapel Hill, North Carolina, USA Assistant Professor Department of Psychiatry Bart Vandenberghe, DDS, MSc, PhD Department of Computer Science Advimago, Center for Advanced Oral Imaging University of North Carolina Brussels, Belgium Chapel Hill, North Carolina, USA Prosthetics Section, Department of Oral Health Gerry M. Raghoebar, DDS, MD, PhD Sciences Professor, Oral and maxillofacial surgeon University of Groningen and University Medical KU Leuven, Belgium Center Groningen Arjan Vissink, DDS, MD, PhD Groningen, the Netherlands Professor, Oral and maxillofacial surgeon University of Groningen and University Medical Harry Reintsema, DDS Maxillofacial Prosthodontist, Department of Oral Center Groningen Groningen, the Netherlands and Maxillofacial Surgery University of Groningen and University Medical Stuart C. White, DDS, PhD Professor Emeritus, Oral and Maxillofacial Center Groningen Groningen, the Netherlands Radiology University of California–Los Angeles School Alan L. Rosenfeld, DDS, FACD Diplomate, American Board of Periodontology of Dentistry Los Angeles, California, USA Private practice Periodontics and Dental Implant Surgery Max J. Witjes, DDS, MD, PhD Park Ridge and Oakbrook Terrace, Illinois, USA Assistant Professor, Department of Oral and Clinical Professor, Department of Periodontology Maxillofacial Surgery University of Illinois College of Dentistry University of Groningen and University Medical Chicago, Illinois, USA Center Groningen Clinical Assistant Professor, Department of Oral Groningen, the Netherlands and Maxillofacial Surgery Louisiana State University School of Dentistry New Orleans, Louisiana, USA

Preface Technology surrounds our private and professional to practical use. It is only within the last fifteen lives, improving at ever-accelerating speeds. In years that CBCT machines could be built at afford- turn, medical imaging benefits from general enhance- able costs and reasonable sizes. Head and neck ments in computers, offering faster and more applications were an obvious choice. refined views of our patients’ anatomy and disease states. Although this Moore’s law progression Although the technology allows for outstand- appears to be exponential, it has actually been ing image quality and ease of use, we should not almost a century since mathematician Johann confuse information with education, data with Radon first laid the groundwork for reconstruction knowledge. Doctors treat disease with the ultimate of a three-dimensional object using a great number purpose to provide a good quality of life to patients. of two-dimensional projections. The first com- To do so, an in-depth knowledge of diagnosis and puted  tomography (CT) scanner was invented by treatment methods is necessary. This textbook aims Sir Godfrey Hounsfield, after he led a team to build at providing detailed understanding of CBCT tech- the first commercial computer at Electric and nology and its impact on oral and maxillofacial Musical Industries. The theoretical groundwork medicine. To achieve the goal of presenting a com- had been published a few years earlier by a particle prehensive text, world renowned engineers and physicist, Dr. Allan Cormack. In 1971, the first clinicians from industry, academic, and private human computed tomography of a brain tumor practice backgrounds came together to offer the was obtained. In 1979, the year Cormack and reader a broad spectrum of information. Hounsfield received the Nobel Prize for their con- tribution to medicine, more than a thousand hos- The clinician will want to jump in and utilize pitals had adopted the new technology. Several images for diagnostic and treatment purposes. generations of computed tomography scanners However, a basic understanding of CBCT properties were later developed, using more refined detec- is essential to better interpret the outcome. Trying to tors, faster rotations, and more complex movement comprehend electronics and formulas is daunting around the body. In parallel, starting in the mid- to most of us, but Dr. Jacobson manages, in the first 1960s, cone beam computed tomography (CBCT) chapter, to present the anatomy of the machine in prototypes were developed, initially for radio- an attractive and elegant way. Dr. Jacobson is the therapy and angiography. The first CBCT was built magician behind the scene who has been concerned in 1982 at the Mayo clinic. Yet, computers and for  many years with image quality, radiation, and detectors were not powerful enough to bring CBCT speed. In his chapter, he opens the hood and makes us marvel at the ingeniousness and creativity necessary to build a small CBCT scanner. xi

xii Preface The next three chapters are written by oral and have led since its inception. Dr. Hatcher, an early maxillofacial radiologists, as well as head and adopter and leader in dental radiology, is the expert neck radiologists. These two groups of specialists in three-dimensional airway measurement, which possess immense expertise in head and neck dis- he shares for the first time in a comprehensive eases and should be called upon whenever any chapter. Dr. Levine was first to measure the impact pathology might be present. In the second chap- of CBCT in endodontics, which he demonstrates ter, Doctors Mallya and White address the major in  his unique chapter. Finally, Dr. Vandenberghe issue of radiobiology risks. Their chapter allows shows us the way to use CBCT in periodontics, a us to  make sound and confident judgment, so new field with promising research he has in great that  X-ray emitting CBCT is only used when part produced. the  clinical benefits largely outweigh the risk. Dr. Brooks, a pioneer and mentor to us all, reviews At the turn of the century, some of us were asked major relevant pathologies and reminds us that by a small start-up company to estimate the num- findings can often be incidental. Drs. Miracle and ber of CBCT in dental offices in years to come. Our Christian’s unique chapter is a first: it introduces insight was critical to the business plan, and we the use of CBCT for pathologies usually studied anticipated the company could expect to sell about on medical CTs. fifteen units per year in the United States. Looking back, it is difficult to comprehend how we could The next chapters address clinical applications. have been so wrong! Immersed in existing options, Dr. Cevidanes and her team, who have pioneered we were unable to imagine how our practices could the study of orthognathic surgeries’ long-term be quickly transformed. We should also recall that, stability using three-dimensional imaging, review at the time, many other electronics now woven to the state of scientific knowledge in orthodontics. our personal lives were to be invented. So today, Next, Dr. Shepers and his colleagues share with us we wonder what comes next. This book is a detailed the most advanced surgical techniques they have testimony of our knowledge and a window to the invented while taking advantage of imaging. We near future. This time, we should attempt to use introduce the use of CBCT for everyday implanto- our imagination. We are clearly at the beginning of logy to make way to Drs. Mandelaris and Rosenfeld, an era where technological advances assist patient who present the most advanced use of CAD/CAM care. The thought leaders who wrote this book are surgical guidance for implantology, a field they showing us the road to our future.

Acknowledgments I would like to express my gratitude to the many colleagues. In particular, Professors William people who have helped bring this book together, Giannobile, Laurie McCauley, and Russel Taichman and to those who have developed the outstanding were immensely generous of their time, expertise, core technology around which it revolves. The and friendship while I struggled as a young faculty topic of this text embodies interdisciplinary inter- member. action at its best: clinical need, science, and engi- neering were intertwined for an outstanding Many engineers spend nights and weekends outcome. Behind each of these disciplines are ded- building, programming, and refining cone beam icated individuals and personal stories which machines. To them all, we must be thankful. I am I was blessed to often share. I hope to be forgiven particularly grateful to my friend Pedja Sukovic, by those who are not cited here. former CEO at Xoran Technologies in Ann Arbor, Michigan. We first met when he was a PhD student I am thankful to the editors at Wiley Blackwell, and I was a young faculty. He came to the dental who had the foresight many years ago to seek school as a patient, and casually asked if a three- and  support this project. In particular, Mr. Rick dimensional radiograph of the head would be Blanchette envisioned this book and encouraged of  interest to us. At the time, his mentor Neal me to dive into its conception. To Melissa Wahl, Clinthorne and he had built a bench prototype in a Nancy Turner, and their team, I am grateful for basement laboratory. It was only a matter of time their relentless “behind the scenes” editorial work. before it became one of the most sought-after machines in the world. I am forever indebted to the co-authors of the book. They are leaders of their respective fields, This work would simply have been unimagin- busy treating patients, discovering new solutions, able without the support of my family. I owe my or lecturing throughout the word. Yet, a short meet- grandmother Tosca Yulzari my graduate studies. ing, a phone call, or a letter was enough to have She saw the beginning of this book but will not see them on board with writing a chapter. They spent its completion. My father, long gone, taught me the countless hours refining their text, sacrificing meaning of being a doctor. My best mentor and precious moments with their families in order to friend is my wife Sylvie, who has supported me share their passion. As always, the work was much unconditionally during almost two decades. Finally, greater than initially anticipated, yet it was com- I thank my children Lea, Myriam, and little pleted to the finest detail and greatest quality. Nathanyel, for giving me such joy and purpose. At the University of Michigan, I received the David Sarment, DDS, MS unconditional support of several experienced xiii

Cone Beam Computed Tomography Oral and Maxillofacial Diagnosis and Applications

1 Technology and Principles of Cone Beam Computed Tomography Matthew W. Jacobson This chapter aims to convey a basic technical cells. The detector cells measure the amount of familiarity with compact Cone Beam Computed X-ray radiation penetrating the subject along dif- tomography (CBCT) systems, which have become ferent lines of response emanating from the source. prevalent since the late 1990s as enablers of in-office This process is called the acquisition of the X-ray CT imaging of the head and neck. The technical measurements. Once the X-ray measurements level of the chapter is designed to be accessible to are  acquired, they are transferred to a computer current or candidate end users of  this technology where they are processed to obtain a CT image and is organized as follows. In Section 1, a high-level volume. This process is called image reconstruction. overview of these systems is given, with a dis- Once image reconstruction has been performed, cussion of their basic hardware components and the computer components of the system make the their emergence as an alternative to conventional, CT image volume available for display in some sort hospital CT. Section 2 gives a treatment of imaging of image viewing software. The topics of image basics, including various aspects of how a CT image reconstruction and display will be discussed at is derived, manipulated, and evaluated for quality. greater length in Section 2. Section 1: Overview of compact Cone beam computed tomography refers to CT cone beam CT systems systems in which the beam projected by the X-ray source is in the shape of the cone wide enough to Computed tomography (CT) is an imaging tech- radiate either all or a significant part of the volume nique in which the internal structure of a subject is of interest. The shape of the beam is controlled by deduced from the way X-rays penetrate the subject the use of collimators, which block X-rays from from different source positions. In the most general being emitted into undesired regions of the scanner terms, a CT system consists of a gantry which field of view. Figure 1.1 depicts a CBCT system of moves an X-ray source to different positions around a compact variety suitable for use in small clinics. the subject and fires an X-ray beam of some shape In the particular system shown in the figure, the through the subject, toward an array of detector gantry rotates in a circular path about the subject firing a beam of X-rays that illuminates the entire desired field of view. This results in a series of Cone Beam Computed Tomography: Oral and Maxillofacial Diagnosis and Applications, First Edition. Edited by David Sarment. © 2014 John Wiley & Sons, Inc. Published 2014 by John Wiley & Sons, Inc. 3

4 Cone Beam Computed Tomography Rotating gantry DentoCAT Amorphous silicon (aSi:H) X-ray source detector array with rejection grid Computer running PWLS and DE PWLS reconstruction code and image display software Cone of X-rays Figure 1.1 The proposed design of DentoCAT. The patient is seated comfortably in chair (the chin-rest is not shown). DentoCAT features cone beam geometry, aSi:H detector array, PWLS and DE PWLS reconstruction methods. two-dimensional (2D) images of the X-ray shadow idea of fan beam geometries is that, as the source of the object that is recorded by a 2D array of moves along the length of the subject, the X-ray fan detector cells. Cone beam CT systems with this beam is used to scan one cross-sectional slice of the particular scan geometry will be the focus of this subject at a time, each of which can be reconstructed book, but it is important to realize that in the individually. There are several advantages to fan broader medical imaging industry, CT devices can beam geometries over cone beam geometries. First, vary considerably both in the shape of the X-ray since only one cross-section is being acquired beam and the trajectory of the source. at  a  time, only a 1-dimensional detector array is required, which lowers the size and cost of the Prior to the introduction of CBCT, it was common detector. Second, because a fan beam only irradi- for CT systems to use so-called fan beam scan geom- ates a small region of the object at a given time, the etries in which collimators are used to focus the occurrence of scattered X-rays is reduced. In cone X-ray beam into a flat fan shape. In a fan beam beam systems, conversely, there is a much larger geometry, the source must travel not only circu- component of scattered radiation, which has a cor- larly around the subject but also axially along the rupting effect on the scan (see “Common Image subject’s length in order to cover the entire volume Artifacts” section). Finally, in a fan beam geometry, of interest. A helical (spiral) source trajectory is the patient movement occurring during the scan will most traditional method used to accomplish this only degrade image quality in the small region of and is common to most hospital CT scanners. The

Technology and Principles of Cone Beam Computed Tomography 5 the subject being scanned when motion occurs. geometry increases the efficiency of X-ray use, Conversely, in cone beam systems, where larger leading to smaller and cheaper X-ray sources that regions of anatomy are irradiated at a given time, are easier to cool. Additionally, the imaging needs patient movement can have a much more perva- of dentomaxillofacial and otolaryngological sive effect on image quality. medical offices have generally been restricted to high-contrast differentiation between bone and The disadvantage of fan beam geometries, how- other tissues in nonmotion prone head and neck ever, is their inefficient use of X-ray output. Because anatomy. CT systems customized for such settings collimators screen away X-ray output from the can therefore operate both at lower X-ray exposure source except in the narrow fan region of the beam, levels and at slower scanning speeds (on the order much of the X-rays generated by the source go of 20–40 sec) than hospital systems. Not only does unused. Accordingly, the source must generate this further mitigate cooling needs of the X-ray more X-ray output than a cone beam geometry for source, it also leads to cheaper and smaller gantry the same region scanned, leading to problems control components. with source heating. Regulating the temperature of the  source in such systems requires fast rotating The emergence of compact CBCT was also facili- source components, accompanied by a consider- tated in part by recent progress in fast computer pro- able increase in mechanical size, complexity, and cessor technology and in X-ray detector technology. expense. As the desire for greater volume coverage The mathematical operations needed to reconstruct has grown in the CT industry, the difficulties with a CT image are computationally intensive and source heating have been found to outweigh the formerly achievable at clinically acceptable speeds advantages of fan beam scanning, and the CT only through expensive, special purpose elec- industry has been gradually moving to cone beam tronics.  With the advent of widely available fast scan geometries. Cone beam geometries have other computer processors, especially the massively par- advantages as well, which have further motivated allel programming now possible with common video this shift. The spatial resolution produced by cone game cards, the necessary computer hardware is beam CT scanners, when used in conjunction with cheaply available to CT manufacturers and hence flat panel X-ray detectors, tends to be more uni- also to small medical facilities. Improvements in form than fan beam–based systems. X-ray detector technology include the advent of flat panel X-ray detectors. Early work on compact CT Although the CT industry as a whole has been systems (circa 2000) proposed using X-ray detectors trending toward cone beam scanning, the hardware based on image intensifier technology, then common simplifications brought on by CBCT have played a to fluoroscopy and conventional radiography. particularly important role in the advent of com- However, flat panels have provided an alternative pact in-office CT systems, of the kind shown in that is both cheaply available and also offers X-ray Figure  1.1. Conventional hospital CT scanners are detection with less distortion, larger detector areas, bulky and expensive devices, not practical for in- and better dynamic range. office use. The reason for their large size is in part due to source cooling issues already mentioned The development of compact CBCT for the clinic and in part due to the fact that hospital CT systems has made CT imaging widely and quickly acces- need to be all-purpose, accommodating a compre- sible. Where once patients may have had to wait hensive range of CT imaging tasks. To accommo- weeks for a scan referred out to the hospital, they date cardiac imaging, for example, hospital CT may now be scanned and treated in the same office systems must be capable of very fast gantry rotation visit. The prompt availability of CT has also been (on the order of one revolution per second) to deal cited as a benefit to the learning process of physi- with the movement of the heart. This has further cians, allowing them to more quickly correlate CT exacerbated the mechanical power requirements, information with observed symptoms. Some con- and hence the size and expense of the system. troversy has sprung up around this technology, with questions including how best to regulate The evolution of compact CT came in part from X-ray dose to patients. The financial compensation recognizing how cone beam scanning and other that physicians receive when prescribing a CT scan system customizations can mitigate these issues. is argued to be a counterincentive to minimizing As discussed, the use of a cone beam scanning

6 Cone Beam Computed Tomography patient X-ray dose. In spite of the controversy, tendency of the anatomy at a particular location to CBCT has found its way into thousands of clinics obstruct the flight of X-ray photons. Because atten- over the last decade and is well on its way to uation is proportional to tissue density, a 3D map becoming standard of care. of attenuation can be used to observe spatial varia- tion in the tissue type of the subject anatomy (e.g., Section 2: Imaging basics for compact soft tissue versus bone). The attenuation applied to cone beam CT systems an X-ray photon at a certain location also depends on the photon energy. Ideally, when the X-ray This section describes the image processing soft- source emits photons of a single energy level only, ware components of compact CBCT systems that this energy dependence is of minor consequence. go into action once X-ray measurements have been In practice, however, an X-ray source will emit acquired. Tasks performed by these components photons of a spectrum of different energies, a fact include the derivation of a CT image volume from that introduces complications to be discussed later. the X-ray measurements (called image reconstruc- tion) and the subsequent display, manipulation, Once the X-ray measurements have been acquired, and analysis of this volume. In the subsections to the first processing step performed is to choose an follow, these topics will be covered in a largely imaging field of view (FOV), a region in space qualitative manner suited to practitioners, with a where the CT subject is to be imaged. For circularly minimum of mathematical detail. orbiting cone beam CT systems, this region will typically be a cylindrical region of points in space Overview of image processing and display that are all visible to the X-ray camera throughout its rotation and that cover the desired anatomy. A The volume image data obtained from a CT system process of image reconstruction is then performed is a 3D map of the attenuation of the CT subject in which the X-ray measurements are used to eval- at  different spatial locations. Attenuation, often uate the attenuation at various sample locations denoted μ, is a physical quantity measuring the within the FOV. The sample locations typically are part of a 3D rectangular lattice, or reconstruction grid, enclosing the FOV cylinder (see Figure  1.2). The sample locations are thought of as lying at the Field of view (FOV) Figure 1.2 The concept of a reconstruction grid and field of view.

Technology and Principles of Cone Beam Computed Tomography 7 center of small box-shaped cells, called voxels. The value of mwater is obtained in a system calibra- For image analysis and display purposes, the tion step by reconstructing a calibration phantom attenuation of the subject is approximated as consisting of water-equivalent material. CT num- being uniform over the region covered by a voxel. bers are measured in Hounsfield units (HU). In this Thus, when the reconstruction software assigns scale, water always has a CT number of zero, an attenuation value to a grid sample location, it while  for air (with mair = 0), the CT number is is in effect assigning it to the entire box-shaped –1000.  Expressing image intensity in HU instead region occupied by the voxel centered at that of  physical attenuation units provides a more location. sensitive scale for measuring fine attenuation dif- ferences. Additionally, it can help to cross-compare The following section will delve into image scans of the same object from different CT devices reconstruction in more detail. For now, we simply or using different X-ray source characteristics. The note that the selection of an FOV and reconstruc- effect of the different system characteristics on tion grid brings a number of design trade-offs into the  contrast between tissue types is more easily play, and must be optimized to the medical task at observed in the normalized Hounsfield scale, in hand. The selected FOV must first of all be large which waterlike soft tissue is always anchored at a enough to cover the anatomy to be viewed. In value near zero HU. addition, certain medical tasks will require the voxel sizes (equivalently, the spacing between Once the reconstructed 3D volume is converted sample points), to be chosen sufficiently small, to to Hounsfield units, it is made available for display achieve a needed resolution. On the other hand, in the system’s image viewing software. Typically, enlarging the FOV and/or increasing the sampling an image viewer will offer a number of standard fineness will, in turn, increase the number of voxels capabilities, among them a multiplanar rendering in the FOV that need reconstructing. For example, (MPR) feature that allows coronal, sagittal, or axial simply halving the voxel size in all three dimen- slices of the reconstructed object to be displayed sions while keeping the FOV size fixed translates (see Figure  1.3). The slices can be displayed as into an eight-fold increase in the number of FOV reconstructed, or one can set a range of neighboring voxels. This leads in turn to increased computa- slices to be averaged together. This averaging can tional burden during reconstruction and slows reduce noise and improve visibility of anatomy at reconstruction speed. Moreover, when sampling some expense in resolution. Other typical display fineness in 3D space is increased, the sampling functions include the ability to rotate the volume so fineness of the X-ray measurements must typically that MPR cross-sections at arbitrary angles can be be increased proportionately in order to reconstruct displayed, a tool to measure physical distances bet- accurate values. This leads to similar increases in ween points in the image, and a tool for plotting computational strain. Finally, as the FOV size is profiles of the voxel values across one-dimensional increased, there is a corresponding increase both in cross-sections. radiation dose to the patient, and also the presence of scattered radiation, which leads to a degrading CT display systems will also provide a drawing effect on the CT image (see “Common Image tool allowing regions of interest (ROIs) to be Artifacts” section). designated in the display. The drawing tool will typ- ically show the mean and standard deviation of the Attenuation is measured in absolute units of voxel values as well as the number of voxels within inverse length (mm–1 or cm–1). However, for pur- the ROI to be computed. For CT systems in the U.S. poses of analysis and display, it is standard market, this feature is in fact federally required throughout the CT industry to re-express recon- under 21 CFR 1020. Figure 1.4 illustrates a circular structed image intensities in CT numbers, a nor- ROI drawn in a commercial CT viewer, with the rel- malized quantity which measures reconstructed evant ROI statistics displayed. One function of this attenuation relative to the reconstructed attenua- tool is to verify certain performance specifications tion of water: that the CT manufacturer is federally required to provide in the system data sheets and user manual. CT No. = μ − μwater × 1000 These metrics will be discussed in greater detail in μwater the “Imaging Performance” section.

8 Cone Beam Computed Tomography Coronal Sagittal Axial Figure 1.3 Multiplanar rendering of a CT subject. Mean 24.90 stdev 53.02 660.52 mm2 Figure 1.4 Illustration of a region of interest drawing tool in the display of a reconstructed CT phantom. Another important display capability is the for  display purposes, the viewing software will ability to adjust the viewing contrast in the image. divide the available brightness levels among the Because there are a limited number of different CT numbers in a user-selected range, or window. brightness levels that can be assigned to a voxel Image voxels whose CT numbers fall between the

Technology and Principles of Cone Beam Computed Tomography 9 (A) (B) Figure 1.5 Axial slice of computer-generated phantom in (A) a high-contrast viewing window (L/W = 50/1200 HU), and (B) a low-contrast window (L/W = 30/90 HU). minimum and maximum values set by the window interior of the slice. At the time of this writing, are assigned a proportionate brightness level. If a however, low-contrast viewing windows are more voxel value falls below the minimum CT number commonly employed by users of compact CBCT in this range, it will be given zero brightness, systems. This is because certain limitations of the whereas if it lies above the maximum CT number, it cone beam geometry and of current flat panel will be assigned the maximum brightness. It is technology, to be elaborated upon later, render common to express a window setting in terms of a image quality poor when viewed in high-contrast level (L), meaning the CT number at the center of windows. The industry has therefore been limited the range, and a window width (W), meaning the to head and neck imaging where often only the difference between the maximum and minimum coarse differentiation between bone and soft CT number in the range. For example, a window tissue are needed. For these applications,  low- ranging between 400 HU and 500 HU would be contrast viewing windows, such as in Figure 1.5B, specified as L= 450 HU and W = 100 HU. tend to be sufficient. The terms soft tissue window and bone window are commonly used to distin- Narrowing the display window about a partic- guish between display range settings appropriate, ular intensity level allows for better contrast respectively, to soft tissue differentiation and between subtly different image intensities within coarse bone/soft tissue differentiation tasks. Soft the window. Figure  1.5 shows an axial slice of a tissue windows will use window levels of 30–50 HU computer-generated head phantom as displayed and window widths of one to several hundred in both a wide, high-contrast window (Figure 1.5A) HU. The bone window will use window levels of and a narrow, low-contrast window (Figure 1.5B). 50–500 HU and window widths of anywhere from Clearly, the narrower window offers better visi- several hundred to over a thousand Hounsfields. bility of the pattern of low-contrast discs in the

10 Cone Beam Computed Tomography The images in Figure  1.3 and Figure  1.5A are computationally but have certain limitations in the displayed at L/W = 50/1200 HU, a  setting repre- image quality they can produce. As computer pro- sentative of the bone window. Figure  1.5B is dis- cessor power has increased over time, however, played at L/W = 30/90 HU, a setting at the narrower and especially with the recent proliferation of end of different possible soft tissue windows. cheaply available parallel computing technology, the CT industry has begun to embrace more Image reconstruction powerful, if more computationally demanding, iterative reconstruction algorithms. The next sec- Image reconstruction is the process by which atten- tion will overview conventional filtered back uation values for each voxel in the CT image are projection reconstruction, which is still the most calculated from the X-ray measurements. This pro- prevalent approach. The section titled “Iterative cess tends to be the most computationally intensive Reconstruction” will then give a short introduction software task performed by a CBCT system. There to emerging iterative reconstruction methods and are tens of millions of voxels in a typical recon- some rudimentary demonstrations. struction grid and each computed voxel value derives information from X-ray measurements Conventional filtered back projection taken typically at hundreds of different gantry positions. A complete image reconstruction task To understand conventional image reconstruction, may hence require, at minimum, tens of billions one must first consider a particular line of X-ray of  arithmetic and memory transfer operations. photon flight, one that emanates from the X-ray CT  manufacturers therefore invest considerable focal spot (see Figure  1.6) to a particular pixel on development effort in making reconstructions the detector panel for some particular gantry posi- achievable within compute times acceptable in a tion. One then considers sample attenuation values clinical environment. Because of the computational of the CT subject along this line, with sample loca- hurdles associated with image reconstruction, com- tions spaced at a separation distance, d. If the sam- mercial systems have historically resorted to ples are weighted by this separation distance and filtered back projection algorithms. These are summed, then as the separation distance is taken among the simplest reconstruction approaches smaller and smaller (making the sampling more and more dense), this weighted sum approaches a d pi (m) Detector panel X-ray source Figure 1.6 The concept of a geometric projection.

Technology and Principles of Cone Beam Computed Tomography 11 limiting value, pi (μ), known as the geometric Once the geometric projections have been calcu- projection, or X-ray transform, of the attenuation lated, an inverse X-ray transform formula is applied. map, μ,  along the i-th measured X-ray path. Commonly, such formulas reduce to a filtering The  idea behind most conventional reconstruc- step, applied view-by-view to the geometric pro- tion techniques is to extract measurements of jections, followed by a so-called back projection step the  geometric projections from the raw physical in which the filtered projection values are smeared X-ray measurements and to then apply known back through the FOV. Algorithms that implement mathematical formulas for inverting the X-ray the reconstruction this way are thus called filtered transform. back projection (FBP) algorithms and are used in a range of tomographic systems, both in CT and The calculation of geometric projections from other modalities. The fine details of both the fil- raw X-ray measurements requires the knowledge tering step and the back projection step are some- of certain physical properties of the source-detector what dependent on the scanning geometry, that is, X-ray camera assembly. For example, it is necessary on the shape of the gantry orbit and the shape of to know the sensitivity of each detector pixel to the radiation beam. Generally speaking, however, X-rays fired in air, with no object present in the the filtering step will be an operation that sharpens field of view. It is also necessary to know the anatomical edges in the X-ray projections while detector offset values, which are nonzero signals dampening regions of slowly varying intensity. measured by the detector even when no X-rays are The smearing action of back projection, meanwhile, being fired from the source. The offset signals orig- will typically be along the measured X-ray paths inate from stray electrical currents in the photosen- connecting the X-ray source to the panel, in a sense sitive components of the detector. These properties undoing the forward projecting action of the radia- are measured in a calibration step performed at the tion source. For circular orbiting cone beam CT sys- time of scanner installation, by averaging together tems, our primary focus here, a well-known FBP many frames of an air scan and a blank scan (a scan algorithm is the Feldkamp Davis Kress (FDK) with no X-rays fired). The air scan and blank scan algorithm (Feldkamp and Davis, 1984). We will response will drift over time due to temperature focus on the FDK algorithm for the remainder of sensitivity of the X-ray detector and gradual X-ray this section. damage, and therefore they must be refreshed periodically, typically by recalibrating the device at Figure  1.7 illustrates the stages of FDK recon- least daily. struction up through filtering, including the data (A) (B) Figure 1.7 Illustration of the precorrection and filtering stages of the FDK algorithm for a CT subject. (A) One frame of precorrected geometric projection measurements. (B) The same frame after filtering.

12 Cone Beam Computed Tomography (B) (A) Figure 1.8 The back projection step of the FDK algorithm for progressively larger numbers of frames: (A) 1 frame. (B) 12 frames. precorrection step, for one frame of a cone beam As mentioned earlier, image reconstruction is CT scan. The edge sharpening effect of the filter computationally expensive compared to other pro- is  clear in Figure  1.7B. Because the sharpening cessing steps in a CT scan. For conventional fil- operation can also undesirably amplify sharp tered back projection, most of that expense tends intensity changes due to noise, the filtering opera- to be concentrated in the back projection step. For tion will also employ a user-chosen cutoff param- the filtering step, very efficient signal processing eter. Intensity changes that are “too sharp,” as algorithms exist so that filtering can be accom- determined by the cutoff, are interpreted by the plished in a few tens of operations per X-ray filter as noise, rather than actual anatomy, and are measurement. Conversely, in back projection, each therefore smoothed. Generally speaking, it is X-ray measurement contributes to hundreds of impossible to distinguish anatomical boundaries voxels lying along the corresponding X-ray path from noise with perfect reliability, and so applying and therefore results in hundreds of computations the cutoff always leads to some sacrifice in resolu- per data point. Perhaps even more troublesome tion in the final image. A judgment must be made is  that both the voxel array and the X-ray mea- by the system design engineers as to the best trade- surement array are too large to be held in com- off  between noise suppression and resolution puter cache memory. When naively implemented, preservation. a back projection operation can therefore result in very  time-consuming memory-access operations. Figure  1.8 shows the result of back projecting Accordingly, a great deal of research over the years progressively larger sets of X-ray frames. In has been devoted to acceleration of back projection Figure 1.8A, where only a single frame is back pro- operations. For example, a method for approxi- jected, one can see how smearing the projection mating a typical back projection with greatly intensities obtained at that particular gantry posi- reduced operations was proposed by Basu and tion back through the FOV results in a pattern Bresler (2001). Later, the same group proposed a demarcating the shape of the X-ray cone beam. In method that makes memory access patterns more Figure 1.8B, C, D, and E, as contributions of more efficient, resulting in strong acceleration over gantry positions are added, the true form of the CT previous methods (De Man and Basu, 2004). subject gradually coalesces.

Technology and Principles of Cone Beam Computed Tomography 13 (C) (D) (E) Figure 1.8 (Continued) (C) 40 frames. (D) 100 frames. (E) 600 frames. Much of the acceleration of image reconstruc- reconstruction operations (Wu, 1991). Since the cost tion  seen over the years has also been hardware- of developing such specialized chips can run into based.  For high-end CT systems, specialized millions of dollars, this route has generally been circuit  chips  known as application-specific available only to large CT manufacturers. Parallel integrated circuits (ASICs) have been used in computing technology has also often been used as place  of software to  implement time-consuming an approach to acceleration. Operations like back

14 Cone Beam Computed Tomography projection often consist of tasks that are indepen- computer-generated head phantom and its FDK dent and can be dispatched to several processors reconstruction from simulated cone beam CT mea- working in parallel. For example, the contribution surements. Comparing Figure 1.9B to Figure 1.9A, of each X-ray frame to the final image can be one can clearly see an erroneous drop-off in the computed independently of other frames. Similarly, image intensity values with distance from the plane different collections of slices in the reconstruction of the source, as well as the appearance of streaks grid can be reconstructed in parallel. and shading artifacts. These so-called cone beam artifacts become more pronounced where the axial Although parallel computing has become increas- cross-sections are less symmetric, for example, in ingly available to smaller manufacturers with the bony region of the sinuses. It is important to the  emergence of multicore CPUs, it has taken emphasize that artifacts such as these can arise a  particular significant leap forward in recent from a number of different causes in actual CT years with the advent of general purpose graphics scans, such as scatter and beam hardening (see processing units (GPGPUs). Essentially, it has been “Common Image Artifacts”). Here, however, the found that the massive parallel computing done by simulation has not included any such corrupting common video game graphics cards can be adapted effects. The artifacts we see here are therefore to a variety of scientific computing problems, assuredly and entirely due to the limitations of the including FDK back projection (Vaz, McLin, et al., circular scan geometry and the FDK algorithm. 2007; Zhao, Hu, et al., 2009). This advance has first of all led to a dramatic speed-up in reconstruction In spite of this fundamental weakness in circular time. Whereas five years ago a typical head CT cone beam scans, the circular scan geometry has reconstruction took on the order of several min- nevertheless been historically favored in the com- utes, it can now be performed in approximately pact CT device industry. This is in part because it 10  seconds. Additionally, the use of GPGPU has simplifies mechanical design. It is also because a greatly cut costs of both the relevant hardware and range of these artifacts are obscured when the software engineering work. In terms of hardware, phantom is viewed in a high-contrast bone window the only equipment required is a video card, costs (as illustrated in Figure 1.9C and Figure 1.9D), and for which may be as low as a few hundred dollars, bone window imaging has been an application of thanks to the size of the video gaming industry. predominant interest for compact CT. On the other The necessary software engineering work has been hand, this can also be seen as one reason why simplified by the emergence of GPGPU program- circular cone beam CT has had difficulty spreading ming languages, such as CUDA and OpenCL (Kirk in use from bone imaging to lower contrast imaging and Hwu, 2010). applications. In the next section, we discuss itera- tive reconstruction, which among other things While the FDK reconstruction algorithm is the offers possibilities for mitigating the problem of most common choice for circular-orbit cone beam cone beam artifacts. CT systems, there are limitations to a circular- orbiting CT scanner that appear when the FDK Iterative reconstruction algorithm is applied. Specifically, it is known that a circular-orbiting cone beam camera does not offer Although filtered back projection methods have complete enough coverage of the object to reliably been commercially implemented for many years, reconstruct all points in the FOV (or at least not by the science has continued to look for improve- an algorithm relying on the projection measure- ments using iterative reconstruction methods, ments alone). Conditions for a point in 3D space to both in CT  and in other kinds of tomography be recoverable in a given scan geometry are well (Shepp and Vardi, 1982; Lange and Carson, 1984; studied and are given, for example, in Tuy (1983). Erdoğan and Fessler, 1999a). With iterative recon- For circular-orbiting cameras, only points in the struction, instead of obtaining a single attenuation plane of the X-ray source satisfy these conditions. map from an explicit reconstruction formula, a Because of this, the accuracy and quality of the sequence of attenuation maps is generated that reconstructed image gradually deteriorate with converges to a final desired reconstructed map. distance from the source plane. This is illustrated While iterative methods are more computationally in  Figure  1.9, which shows sagittal views of a

Technology and Principles of Cone Beam Computed Tomography 15 (A) (B) (C) (D) Figure 1.9 Comparison of sagittal views of a computer-generated phantom and its FDK reconstruction in low- and high-contrast viewing window. The dashed line marks the position of the plane of the x-ray source. (A) True phantom, low-contrast window (L/W = 50/200 HU). (B) FDK reconstruction, low-contrast window (L/W = 50/200 HU). (C) True phantom, high-contrast window (L/W = 50/1200 HU). (D) FDK reconstruction, high-contrast window (L/W = 50/1200 HU). demanding than filtered back projection, they pro- algorithm, which affects image quality, and the vide a flexible framework for using better models computational expense of the algorithm, which of the CT system, leading to better image quality, affects reconstruction speed. The previous section sometimes at reduced dose levels. At this writing, overviewed traditional filtered back projection iterative methods have also begun to find their algorithms, which are among the simplest and fast- way into the  commercial CT device market. est reconstruction methods. An explicit formula is Notably, the larger medical device companies have used to obtain the reconstructed image, and only commercialized proprietary iterative methods one pass over the measured X-ray data is required. with claims of  reducing X-ray dose by several However, the amount of physical modeling factors without compromising image quality information used in filtered back projection is fairly (Freiherr, 2010). Iterative reconstruction software limited. As an example, filtered back projection is also marketed by private software vendors such ignores statistical variation in the X-ray measure- as InstaRecon, Inc., sample results of which are ments, leading to higher noise levels in the recon- shown subsequently. structed image (or alternatively higher radiation dose levels) than are actually necessary. FBP In the design of image reconstruction algorithms, also  ignores the fact that realistic X-ray beams there is a trade-off between the amount/accuracy consist of a multitude of X-ray photon energies, of physical modeling information included in an

16 Cone Beam Computed Tomography (B) (A) Figure 1.10 Reconstructions of a clinical helical CT scan of the abdomen using (A) filtered back projection and (B) a proprietary iterative algorithm developed by InstaRecon. approximating the beam instead as a monoener- years. Nevertheless, the advantages of iterative getic one. This leads to beam hardening artifacts, to reconstruction over filtered back projection are be discussed under “Common Image Artifacts.” readily demonstrated. Some relevant illustrations Finally, FBP only incorporates information avail- are provided in Figure  1.10, Figure  1.11, and able in the X-ray measurements, whereas more Figure 1.12. complicated iterative algorithms can also incorpo- rate a priori knowledge about the characteristics of Figure  1.10A and Figure  1.10B show a perfor- the patient anatomy. This has important implica- mance comparison of a proprietary iterative tions for circular-orbit CBCT systems, because for algorithm developed by InstaRecon with filtered this scanning geometry (see “Conventional Filtered back projection for a clinical abdominal scan. This Back Projection” section), the X-ray measurements particular scan was acquired using a conventional alone cannot provide enough information to accu- helical CT system, and so the filtered back projec- rately reconstruct the object at all points in the field tion algorithm used was not cone beam FDK. The of view. The FDK algorithm, a variation of FBP iterative algorithm achieves reduced image noise specific to circular-orbit systems, produces cone and hence more uniform images. Furthermore, beam artifacts, as a result. since image noise generally trades off with X-ray exposure, noise-reducing iterative algorithms such The desire to improve image quality has led as these also allow one to scan with reduced X-ray many researchers over the years to propose recon- dose, while achieving the same noise levels in the struction algorithms based on more detailed and reconstructed image as conventional filtered back complicated physical models of CT systems. These projection. Figure  1.11A and Figure  1.11B show a more complicated models lead to reconstruction similar comparison for simulated CT measure- equations that have no explicit solution. Instead, ments of a phantom commonly used to measure the solution must be obtained by iterative compu- low-contrast imaging performance. One sees how tation, in which a sequence of images is generated the iterative algorithm improves the detectability that gradually converges to the solution. Generally of low-contrast objects as compared to filtered back speaking, every iteration of an iterative reconstruc- projection. tion algorithm tends to have a computational cost comparable to an FBP reconstruction. This extra Figure  1.12A and Figure  1.12B show iterative computation puts a significant price tag on the reconstructions of the same computer-generated image quality improvements that iterative recon- CBCT phantom scan as in Figure 1.9. This recon- struction proposes to bring, a price tag that delayed struction algorithm incorporates prior informa- the clinical acceptability of these methods for many tion about the piece-wise smooth structure of the  patient anatomy. Reconstruction algorithms

Technology and Principles of Cone Beam Computed Tomography 17 (A) (B) Figure 1.11 Reconstructions of a simulated CBCT scan of a CIRS061 contrast phantom using (A) filtered back projection and (B) a proprietary iterative algorithm developed by InstaRecon. (A) (B) Figure 1.12 Sagittal views of a computer-generated phantom reconstructed using a rudimentary iterative algorithm in a low-contrast viewing window (L/W = 50/200 HU). (A) Result after 30 iterations. (B) Result after 300 iterations. that incorporate such information (Sukovic and that the intensity values in the region of the Clinthorne, 2000) are abundant in the medical sinuses are much closer to their true  value as imaging literature. The reconstruction algorithm compared to the FDK results in Figure 1.9B. This used here was more rudimentary than Insta- occurs because the addition of prior information Recon’s algorithm. Among other things, it has not about anatomical smoothness compensates for the been optimized for speed and it takes many more geometric incompleteness of the circular X-ray iterations to converge. However, it was sufficient camera orbit. to show how adding prior smoothness information can mitigate cone beam artifacts. Figure 1.12 shows Although the image quality benefits of iterative algorithms have been known for many years, it has

18 Cone Beam Computed Tomography only recently become possible to run at sufficient Measurement noise leads to sharp discontinu- speed to make them clinically acceptable for CT ities among the measured values of neighboring imaging. Computing hardware improvements over detector pixels. When the X-ray measurements the years, such as GPGPU discussed earlier, have are put through the image reconstruction pro- contributed to reducing computation time per cess,  the reconstructed CT volume will exhibit iteration. Additionally, much medical imaging correspondingly sharp discontinuities among neigh- research has been devoted to finding iterative boring voxel values that would otherwise be reconstruction algorithms requiring as few as uniform or gradually varying. This is the visual possible iterations to converge (Kamphuis and manifestation of image noise. A common way to Beekman, 1998; Erdoğan and Fessler, 1999b; Ahn, measure image noise is to compute the standard Fessler, et al. 2006). deviation of some region of voxels in a phantom of some uniform material (as in Figure  1.4, for Imaging performance example). As mentioned in the “Overview of Image Processing and Display,” most CT image This section discusses several quantitative mea- viewing software provides this capability. In sures of image quality that are commonly used to manuals for a CT device, the noise standard assess the performance of a CT device, namely deviation will often be reported as a fraction of noise performance, low-contrast detectability, and the attenuation of water. spatial resolution. CT manufacturers will typi- cally report such quality measurements in the user CT system engineers make design choices to manuals issued with their devices. Typically also, control noise but must take certain trade-offs into manufacturers provide customers equipment to account. Measurement noise can be reduced, for repeat these measurements and specify in the example, by increasing X-ray exposure to the user manual how reproducible the measurements patient, although health concerns place obvious should be. For CT manufacturers in the United limits on doing so. Certain types of detector panels States, providing this information is legally have better photon detection efficiency than others, required by the Code of Federal Regulations giving better resistance to noise. However, such (21 CFR 1020.33). detectors are also more expensive and lead to increased system cost. Other methods of reducing Image noise noise involve configuring the X-ray detection and image reconstruction process in a certain way, The term measurement noise refers to random var- although these methods entail trade-offs in image iations in CT measurements. Image noise refers resolution. For example, most detector panels to  the ensuing effect of these variations on the allow one to combine neighboring detector pixels reconstructed image. In a CT scan, there are sev- to form larger pixels. This “binning” of pixels effec- eral sources of measurement noise that make tively averages together the signal values that the measurements not precisely repeatable. When would be measured by the smaller pixels sepa- X-rays are fired through a patient along a certain rately and reduces noise. However, projection straight-line path, there is randomness in the sampling fineness, and hence resolution, are also number of photons that will penetrate through reduced as a trade-off. Similarly, the reconstruction the object to interact with the detector. There is software can be designed to include smoothing also randomness in the number of photons that, operations. As mentioned previously, filtered back after penetrating the object, will successfully projection methods include smoothing in the fil- interact with the X-ray detector panel to produce tering step, while iterative reconstruction methods a signal. Finally, there are also elements of can enforce image smoothness using a priori ana- random fluctuation in the detector electronics tomical information. These smoothing methods itself, independent of the object and the X-ray reduce noise but can also blur anatomical tissue source. borders as a side effect, and so resolution is again sacrificed. Reconstruction algorithms are often compared based on how favorably noise trades off with spatial resolution.

Technology and Principles of Cone Beam Computed Tomography 19 1.1 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 –0.1 Figure 1.13 Example of a slice sensitivity profile (SSP) illustrated with data from the xCAT-ENT, a commercial mobile cone beam CT scanner for sinus imaging. The profile is plotted on a horizontal axis in units of millimeters. Spatial resolution beam because X-rays are emitted from different points on the surface of the source, rather than Spatial resolution refers to how well small or from a single apex point. However, this effect is closely spaced objects are visualized in an image. commonly ignored by the reconstruction software, Spatial resolution in a cone beam CT system is at the expense of spatial resolution. partly limited by the size of the image voxels used for reconstruction. However, resolution is further In conventional helical fan beam CT systems, the limited by various sources of system blur. As amount of blur along the axis of the scanner has discussed in the previous section, certain sources historically been significantly different than the of  blur arise as a side effect of various engi- blur within an axial slice. This difference has led to neering  measures taken to reduce image noise. common practices, and in some cases regulations, Other sources of blur arise from the physics of the for CT manufacturers to report separate measure- X-ray detection process. Detector glare is an effect ments of axial and in-plane spatial resolution. With whereby X-ray photons striking the detector induce the advent of cone beam systems, the difference in a scattering event that causes a signal to be detected axial versus in-plane resolution has greatly dimin- in several neighboring pixels. This leads to a blur- ished, but laws designed for helical fan beam ring of the projection views and an ensuing blur in systems are so well established that they are still the reconstructed image. A similar effect is detector applied to CBCT. To measure spatial resolution lag, in which the signal detected in one X-ray shot axially, an object such as a wire or bead, whose fails to dissipate before the next X-ray shot is taken. cross-section along the scanner axis is narrow and This has the effect of blurring together adjacent pointlike, is imaged. Due to blur effects, the X-ray shots. Finally, imperfect modeling of the CT cross-section in the image will have a smeared, system geometry in the reconstruction process can lobelike profile, such as that shown in Figure 1.12. also blur the image. For example, no cone beam CT The amount of blur is reported on a slice sensitivity system produces a perfectly cone-shaped X-ray profile such as the one in Figure 1.13. The width of

20 Cone Beam Computed Tomography this profile at half its peak value is known as the average photon energies, obtained by lowering nominal tomographic section thickness. the  X-ray source voltage, attenuation differences among different materials generally increase, lead- To measure in-plane spatial resolution, it is tra- ing to better contrast. The engineering trade-off in ditional to report the modulation transfer function lowering source energy, however, is that the ability (MTF). An MTF is a graph showing how the imaged of X-ray photons to penetrate the CT subject is contrast of densely clustered objects decreases, as a reduced, leading to higher noise and photon result of system blur, with the clustering density. starvation artifacts. Contrast is also limited by As a result of blur effects, the intensity of small or certain features in the electronics of the X-ray narrow objects is diluted with background material detector. When detected X-rays are converted from in the image, thereby lowering their apparent con- analog to digital signals, information about tissue trast. Since objects must be of decreasing size to be contrast is somewhat degraded. This degradation clustered more densely, an accompanying decrease can be reduced by using A/D converters which in contrast with density is typically observed. This digitize signals more finely, but the trade-off in is illustrated in Figure 1.14A, which shows a series doing so is an increase in the cost of the detector of progressively denser line pair targets, with panel, and hence the overall system. the density expressed in line pairs per centimeter (lp/cm). One can see how not only the separation Common image artifacts between the more densely spaced line pairs dimin- ishes as a result of blur, but also their per- Image artifacts are visible patterns in an image aris- cent  contrast with the background medium. By ing from systematic errors in the reconstruction measuring the percent contrast of line pair phan- process. Common kinds of artifacts include streaks toms, one can plot contrast versus line pair density, and nonuniformity trends, such as in Figure 1.15A. which is how MTF plots are often expressed. MTFs For circular-orbiting CT systems, ring artifacts such can also be obtained more indirectly by measuring as in Figure 1.16A are also commonly encountered. an in-plane blur profile, similar to the slice sensi- Current use of compact CT systems is often tol- tivity profile (Boone, 2001). The MTF plots in erant to artifacts, since bone window viewing of CT Figure 1.14B were obtained in such a manner. They images is still very prevalent, and many artifacts show the MTFs for two imaging modes of a are obscured in the bone window. An under- commercial ear-nose-throat scanner. The temporal standing of artifacts and their causes can still be bone mode has a more slowly decreasing MTF, important, however, for several reasons. First, there indicative of less blur and higher spatial resolu- are exceptions where artifacts are severe enough to tion, than the sinus mode. This is typical, due to appear even in the bone window viewing applica- the higher resolution needs of temporal bone tions. When scanning very bony anatomy, for imaging tasks. example in dental or skull base imaging, very strong streak artifacts can be present. Artifacts can Low-contrast detectability also be a sign that a CT system is in need of mainte- nance. Strong ring artifacts can appear when the Low-contrast detectability is a performance param- system is in need of recalibration, for instance. eter of CT systems that measures its overall ability Finally, as practitioners expand their use of com- to resolve small differences in intensity between pact CT to low-contrast soft tissue imaging applica- objects. To test low-contrast detectability in a CT tions, the influence of artifacts becomes more system, phantoms such as that in Figure 1.11, con- noticeable in the less forgiving low-contrast view- taining low-contrast targets of a range of sizes, are ing windows. Means of suppressing artifacts will often used. be important to extending compact CT to these applications. As discussed in the previous section, system blur reduces the contrast of small objects. However, Causes of artifacts can be either advertent or inad- there are other contrast-limiting effects in a CBCT vertent. Inadvertent causes include inaccuracies in system that can affect the visibility of large objects the calibration of the CT system. When a CT system as well. One contrast-limiting effect in CT systems is the energy spectrum of the X-ray source. At lower

Technology and Principles of Cone Beam Computed Tomography 21 (A) 7 Ip/cm 6 Ip/cm 5 Ip/cm (B) Modulation transfer function (MTF) 100 90 Sinus Temporal bone 80 70 Percent contrast 60 50 40 30 20 10 0 0246 Spatial frequency (Ip/cm) Figure 1.14 Concepts of in-plane resolution measurement illustrated with data from the MiniCAT, a commercial cone beam CT scanner for sinus and temporal bone imaging. (A) Reconstructed image of a phantom containing line pair targets of different densities. lp/cm = line pairs per centimeter. (B) Modulation transfer function for the MiniCAT’s sinus and temporal bone scan protocols. is first installed, and possibly periodically there- some uncontrollable way. In the section after, certain physical properties of the system must “Conventional Filtered Back Projection,” for be measured through a calibration procedure. The example, it was discussed how certain detector physical properties to be calibrated are ones that pixel parameters must be calibrated periodically cannot be precisely controlled by the manufacturer, using air scans and blank scans. These kinds of or that may drift over the lifetime of the machine in calibrated quantities serve as input to the image

22 Cone Beam Computed Tomography (B) (A) Figure 1.15 (A) Illustration of streaks and nonuniformity artifacts in an axial slice of a low-contrast CBCT scan. (B) The same slice after a postcorrection method is applied. reconstruction process, which uses them to model treatment of other corrupting physical effects such system behavior. Inaccuracies in the calibration as beam hardening and scatter. Beam hardening is create disagreement between the true physical a physical effect whereby the average energy X-ray measurements and the mathematical model content of an X-ray beam gradually increases as the used by the reconstruction software, resulting photons in the beam pass through an object. This in  image artifacts. In circular-scanning CBCT occurs because lower energy X-ray photons have a systems, inaccuracies in pixel sensitivities and lower probability than higher energy photons of offsets are a typical cause of tree trunk–like ring passing through the object unattenuated, and are artifacts, like those shown in Figure  1.16A. progressively sifted out of the beam. Scatter is an Miscalibration of a given pixel will introduce effect whereby some X-ray photons traveling errors in how that pixel’s measurement is pro- through the CT subject are deflected from a cessed in every X-ray shot. The repetition of these straight-line path, due to interaction with matter, measurement errors throughout the circular and generate signal in the wrong detector pixels. orbit  of the X-ray camera leads to circularly When ignored by the reconstruction process, both symmetric artifact patterns in the image, thus beam hardening and scatter can contribute to showing as rings. coarse nonuniformity artifacts, such those as shown in Figure 1.15A. Moreover, when scanning Artifacts can also result from deliberate mathe- bony, asymmetric anatomy, beam hardening and matical errors and approximations made by the scatter can contribute to streak artifacts, also shown reconstruction algorithm to simplify computation. in the figure. Streaks result whenever certain As an example, in the “Conventional Filtered Back particular X-ray shots contain much more Projection” section, it was discussed how cone measurement errors than at other positions of the beam artifacts are an engineering trade-off to the X-ray camera. Beam hardening and scatter effects mechanical simplicity of a circular-orbiting CT are a common cause of such errors because their camera, as well as to the computational simplicity effect varies strongly with the thickness and density of the FDK reconstruction algorithm. Similar kinds of tissue through which the X-ray beam passes. of trade-offs have historically been made in the

Technology and Principles of Cone Beam Computed Tomography 23 (A) (B) Figure 1.16 (A) Illustration of ring artifacts in an axial slice of a low-contrast CBCT scan. (B) The same slice after a ring correction method is applied. For asymmetric patient anatomy, these in turn vary the resulting artifacts cannot be tolerated, com- strongly with the position of the X-ray camera mercial systems will often remove artifacts from relative to the patient. the reconstructed image using fast postcorrection methods. These methods are often proprietary, and Beam hardening and scatter have historically therefore it is hard to comment authoritatively on been computationally expensive to handle in the how they work for different CT vendors. However, image reconstruction process in a mathematically a variety of postcorrection methods have been pro- precise way, which means that in practice they are posed in public-domain scientific literature. It is either ignored or corrected using computationally likely that at least some methods used commer- cheaper compromises. One of the more mathe- cially are derived from these. The degree of matically rigorous ways of dealing with beam mathematical or physical modeling rigor on which hardening, for example, is to use an image recon- postcorrection methods are based can vary greatly. struction  algorithm that models the energy There is therefore much ongoing debate in scien- variation of the  beam (Elbakri and Fessler, 2002; tific literature over their limitations, as compared Elbakri and Fessler, 2003). However, reconstruction to  their more computationally expensive, mathe- algorithms with this level of modeling generally matically rigorous alternatives. However, postcor- require iterative methods, and only in recent years rection methods have certainly proven effective has computing technology become fast enough to enough to make them popular compromises. consider using such methods clinically. Similarly, Figure 1.15B, for example, demonstrates the reduc- scientific literature has proposed very accurate tion of streak and nonuniformity artifacts using a scatter modeling and correction approaches combination of postprocessing approach (Zbijewski (Zbijewski and Beekman, 2006). However, achiev- and Stayman, 2009; Hsieh, Molthen, et al., 2000). ing clinically viable computation time remains a Figure 1.16B demonstrates the reduction of ring arti- challenge with these methods. facts using a postcorrection method (Sijbers and Postnov, 2004). In situations where rigorous image reconstruc- tion is too expensive computationally, but where

24 Cone Beam Computed Tomography References subsets convex algorithm. IEEE Transactions on Medical Imaging 17(6): 1001–5. Ahn, S., Fessler, J.A., et al. (2006). Convergent incremental Kirk, D.B., and Hwu, W.W. (2010). Programming optimization transfer algorithms: Application to tomo- Massively  Parallel Processors: A Hands-on Approach. graphy. IEEE Transactions on Medical Imaging 25(3): Morgan Kaufman. 283–96. Lange, K., and Carson, R. (1984). EM reconstruction algo- rithms for emission and transmission tomography. Basu, S., and Bresler, Y. (2001). Error analysis and per- J Comp Assisted Tomo 8(2): 306–16. formance optimization of fast hierarchical backprojec- Shepp, L.A., and Vardi, Y. (1982). Maximum likelihood tion algorithms. IEEE Trans Im Proc 10(7): 1103–17. reconstruction for emission tomography. IEEE Trans Med Imag 1(2): 113–22. Boone, J.M. (2001). Determination of the presampled MTF Sijbers, J., and Postnov, A. (2004). Reduction of ring arte- in computed tomography. Med Phys 28(3): 356–60. facts in high resolution micro-CT reconstructions. Phys Med Biol 49(14): N247-54. De Man, B., and Basu, S. (2004). Distance-driven projec- Sukovic, P., and Clinthorne, N.H. (2000). Penalized tion and backprojection in three dimensions. Phys Med weighted least-squares as a metal streak artifacts Biol 49(11): 2463–75. removal technique in computed tomography. Proc IEEE Nuc Sci Symp Med Im Conf. Elbakri, I.A., and Fessler, J.A. (2002). Statistical image recon- Tuy, H.K. (1983). An inversion formula for cone-beam struction for polyenergetic X-ray computed tomog- reconstruction. SIAM J Appl Math 43(3): 546–52. raphy. IEEE Transactions on Medical Imaging 21: 89–99. Vaz, M.A., McLin, M., et al. (2007). Current and next generation GPUs for accelerating CT reconstruction: Elbakri, I.A., and Fessler, J.A. (2003). Segmentation-free Quality, performance, and tuning. Proc Intl Mtg on statistical image reconstruction for polyenergetic X-ray Fully 3D Image Recon in Rad and Nuc Med. computed tomography with experimental validation. Wu, M. A. (1991). ASIC applications in computed tomog- Phys Med Biol 48(15): 2543–78. raphy systems. Fourth Annual IEEE International ASIC Conference and Exhibit. Erdoğan, H., and Fessler, J.A. (1999a). Monotonic algo- Zbijewski, W., and Beekman, F.J. (2006). Efficient Monte rithms for transmission tomography. IEEE Transactions Carlo based scatter artifact reduction in cone-beam on Medical Imaging 18(9): 801–14. micro-CT. IEEE Trans Med Imag 25(7): 817–27. Zbijewski, W., and Stayman, J.W. (2009). Volumetric soft Erdoğan, H., and Fessler, J.A. (1999b). Ordered subsets tissue brain imaging on xCAT: A mobile flat-panel algorithms for transmission tomography. Phys Med x-ray CT system. Proc SPIE 7258, Medical Imaging 2009: Biol 44(11): 2835–51. Phys Med Im. Zhao, X., Hu, J.J., et al. (2009). GPU-based 3D cone-beam Feldkamp, L.A., and Davis, L.C. (1984). Practical cone- CT image reconstruction for large data volume. Int beam algorithm. J Opt Soc Amer 1: 612–19. J Biomed Imaging 2009: 149079. Freiherr, G. (2010). Iterative reconstruction cuts CT dose without harming image quality. Diagnostic Imaging 32(11). Available at www.diagnosticimaging.com. Hsieh, J., Molthen, R.C., et al. (2000). An iterative approach to the beam hardening correction in cone beam CT. Med Phys 27(1): 23–9. Kamphuis, C., and Beekman, F.J. (1998). Accelerated iter- ative transmission CT reconstruction using an ordered

2 The Nature of Ionizing Radiation and the Risks from Maxillofacial Cone Beam Computed Tomography Sanjay M. Mallya and Stuart C. White Configuration of matter sufficient energy, it can overcome this electrostatic attraction and move to a higher energy state. This Familiarity with the atomic structure is essential to energy is termed binding energy, and is specific for understanding production of X-rays and their an orbital and depends on the atomic number (Z) of interaction with matter. All matter is composed of the element. The higher the atomic number, the atoms. According to the classical view of the atom, more binding energy there is. For any given atom, as proposed by Niels Bohr, the atom is composed of the binding energy of the outer orbitals is lower a positively charged nucleus containing protons than that of the inner orbitals. Some radiations such and neutrons, with negatively charged electrons as ultraviolet light have sufficient energy to remove that revolve around the nucleus in well-defined outer electrons. Other radiations such as X- and orbits. The contemporary view of the atom is gamma rays have enough energy to displace inner described by the Standard Model. As with the electrons. In both these situations, the loss of an classical view, electrons are fundamental particles. electron causes an imbalance between the net But in contrast to the classical view, protons and charges of electrons and protons in the nucleus, and neutrons are not  considered fundamental units; thus results in ionization. These radiations are rather, they are composed of quarks. The contem- referred to as ionizing radiations. porary model of the atom also differs in its view of the relationship of electrons to the nucleus. Unlike Nature of ionizing radiation the classical view, which postulates that electrons revolve in a two-dimensional orbit, the modern Radiation is the propagation of energy through view considers that electrons are dispersed in three- space and matter. There are two types of radiation: dimension orbitals. Each orbital has a discrete particulate and electromagnetic (White and Pharoah, energy state. Within all atoms, the electrons occupy 2009). Particulate radiation is energy transmitted the lowest energy state first. The electrons are held by rapidly moving particles produced primarily in orbit by an electrostatic attraction to the posi- by disintegration of unstable atoms. The particles tively charged nucleus. If an electron absorbs Cone Beam Computed Tomography: Oral and Maxillofacial Diagnosis and Applications, First Edition. Edited by David Sarment. © 2014 John Wiley & Sons, Inc. Published 2014 by John Wiley & Sons, Inc. 25

26 Cone Beam Computed Tomography Figure 2.1 The electromagnetic spectrum, showing the relationship between photon energy and wavelength. Note that as photon energy decreases, wavelength increases. may be charged, for example, α- or β-particles, or Production of X-rays may be uncharged particles such as neutrons. Electromagnetic radiation is energy transmitted X-ray tube as  a combination of electric and magnetic fields. According to quantum theory, electromagnetic The X-ray tube is the heart of a radiographic imaging radiation is propagated in small bundles or system and is housed within the X-ray tube head packets of energy called photons. Photons have along with the essential electrical components that only energy and no mass are often described supply its power. The X-ray tube consists of a cathode in  terms of their energy (eV). Some aspects of and an anode within an evacuated glass tube. The electromagnetic radiation are better explained by process of X-ray production starts with the generation the wave theory, which assumes that these radia- of electrons at the cathode. The electrons are acceler- tions are transmitted as electric and magnetic ated toward the anode by providing a high potential fields that travel in a wavelike pattern. In this case difference between the anode and the cathode. As the radiation is better characterized by its wave- electrons travel from the cathode to the anode, they length. Photon energy is inversely proportional to accumulate kinetic energy. On striking the anode, this its wavelength. kinetic energy is converted into heat and X-rays. In cone beam computed tomography (CBCT) units, the The term electromagnetic radiation refers to a tube head is linked to the image detector (flat panel spectrum of radiations that differ in their or  image intensifier) by a C-arm. Control panels on energies but share some similar properties the CBCT unit allow the operator to regulate various (Figure 2.1). All electromagnetic radiations travel parameters of this process and thereby control the at the speed of light. The radiations within nature of the X-ray beam produced. Understanding this  spectrum have a broad range of energies the impact of these controls on X-ray beam produc- ranging from the low-energy (long-wavelength) tion is important. Selection of the optimal exposure radio waves to high-energy (short-wavelength) factors influences diagnostic quality of the images as gamma rays. High-energy electromagnetic radi- well as the radiation exposure to the patient. ations have sufficient energy to interact with and cause ionization of atoms. These radiations Cathode are  called ionizing radiations and include γ, X- and ultraviolet radiations. As described The cathode consists of a coil of metallic filament below,  ionizing radiations have the potential to (Figure 2.2). A low-voltage current is used to heat cause damage to biological molecules, including this coil. When the temperature of the filament is inducing cancer.

The Nature of Ionizing Radiation and the Risks from Maxillofacial Cone Beam Computed Tomography 27 (A) (B) Figure 2.2 Schematic diagram of the components of x-ray tubes with stationary anode (A) or rotating anode (B). high enough, electrons in the outer orbitals of the X-ray production tungsten atoms absorb sufficient energy to over- come their binding energy and are released from Electrons produced at the cathode are accelerated the filament. The focusing cup is negatively toward the anode by providing a high potential charged and thus electrostatically focuses the difference between the cathode and anode. As electrons to a small area of the anode. electrons strike the anode, the kinetic energy of the electrons is converted into heat and X-ray Anode photons. This accounts for more than 99% of the energy transfer from the striking electrons to the The anode is composed of a tungsten target tungsten atoms. The remainder 1% of energy is embedded into a block of copper (Figure 2.2). As converted in X-rays, primarily by bremsstrahlung the electrons strike the anode, their kinetic energy interactions. is converted into heat and X-rays. The production of X-rays is an inefficient process, with more than Bremsstrahlung photons 99% of the electron’s kinetic energy being con- verted into heat. The focal spot, the area of the target As the electrons course through the tungsten atoms struck by the electrons and from which X-rays are in the target, they may pass close to a nucleus. Due emitted, should be as small as possible. The to the electrostatic forces between the positively smaller the focal spot size, the sharper the final charged nucleus and the negatively charged elec- images. X-ray tubes have one of two designs. tron, the electron is deviated from its course and Some machines use a stationary anode (Figure 2.2A) loses some energy, which is converted into an X-ray like a conventional dental X-ray machine. Others photon (Figure  2.3). These photons are called use a  rotating anode (Figure  2.2B). In this design, bremsstrahlung photons. Bremsstrahlung photons the anode is a disc, with an angled surface that have a continuous spectrum of energies (Figure 2.4). serves as the target area. As the anode rotates, suc- The maximum energy of the bremsstrahlung pho- cessive electrons from the cathode strike sequential ton is determined by the potential difference regions of the target, and at any given time, the between the cathode and the anode. For example, area of the target producing X-rays, the focal spot, an X-ray machine set to operate at 100 kVp will is small. However, the heat is dissipated over the produce bremsstrahlung photons with a maximum larger  area of the entire disc. This design allows energy of 100 keV. Bremsstrahlung photons consti- production of X-rays from small focal spots even at tute the majority of the diagnostically useful X-ray high-energy outputs, or with prolonged exposure beam. From a diagnostic and radiation safety view- times. CBCT  units have either stationary or point, it is important to decrease the numbers of rotating anodes with focal spot sizes ranging from low-energy photons, which increase patient dose 0.15 mm to 0.7 mm.

28 Cone Beam Computed Tomography Figure 2.5 Increasing kVp (with constant mA and exposure time) results in more photons, with a higher mean energy and peak energy of the beam. Figure 2.3 Bremsstrahlung photons are produced when an influences both the image quality and the radiation electron is deviated from its path due to electrostatic dose to the patient. Thus, understanding these interaction with the nucleus. parameters is of importance to patient care. Figure 2.4 Spectrum of photons produced by an x-ray Tube voltage (kVp) tube operating at 100 kVp. The shaded area under the curve depicts the bremsstrahlung photons. The spike at Tube voltage refers to the potential difference approximately 69 keV represents characteristic radiation between the cathode and the anode and is con- from the tungsten atoms in the target. veyed as peak voltage (kVp). As kVp is increased, there is an increase in the number of photons gen- and decrease image quality. Manufacturers add fil- erated, a higher peak energy, and a higher mean ters to preferentially absorb low-energy photons. energy of the X-ray beam (Figure  2.5). Increasing the kVp increases the penetrating power of the Parameters of X-ray beams in CBCT units beam. Increasing kV increases the signal-to-noise ratio but also delivers a higher dose to the patient. The controls of X-ray units, including CBCT ma- Depending on the CBCT unit manufacturer, the chines, allow the operator to optimize various aspects kVp is fixed or adjustable. Few studies have exam- of  X-ray production. Altering these parameters ined the effect on kVp on optimization of image quality and patient dose. Tube current (mA) The tube current is the flow of electrons from the cathode to the anode and is expressed as milliam- peres (mA). It is a reflection of the power delivered to the tungsten filament in the cathode. When the mA setting is increased, the number of electrons liberated at the cathode is increased; this translates into a higher number of X-ray photons produced (Figure 2.6). However, the mean and peak energies of the beam remain same.

The Nature of Ionizing Radiation and the Risks from Maxillofacial Cone Beam Computed Tomography 29 Figure 2.6 Increasing the mA setting (with constant kVp and exposure time) results in more photons but no change in the mean and peak energies of the beam. Exposure time The exposure time is the total time during which Figure 2.7 Fields of view. Representation of the extent of X-ray production takes place during the CBCT anatomical coverage for small (limited), medium scan. During the CBCT scan, multiple projections (dentoalveolar) and large (craniofacial) fields of view. of the field of interest are obtained at varying angles. In most CBCT units, the exposure is pulsed, of anatomic coverage (Figure  2.7). In general, the so that X-ray production takes place only during scan collimations are as below: acquisition of the basis projections. In some units, the exposure is continuous—X-rays are produced a. Small field of view (also referred to as limited and expose the patient even when the detector is or focused fields of view): scan height and not recording images. Using a pulsed beam reduces width less than 5 cm. the radiation exposure to the patient. A second var- iable is the scan time or exposure time. For some b. Medium field of view (also referred to as den- units, the scan time is fixed and cannot be varied by toalveolar field of view): scan height 5–15 cm. the operator. Many contemporary units allow the operator to choose from a variety of scanning c. Large field of view (also referred to as craniofa- modes, such as “high speed” or “high resolution” cial field of view): scan height greater than modes. With high-speed modes, the number of 15 cm. basis projections is reduced, thereby decreasing scan times and thus radiation exposure to the Collimating the beam to as small a region as pos- patient. In high-resolution modes, the number of sible not only reduces patient exposure, it also basis projections is increased, and consequently enhances image quality due to decreased scatter scan time as well as patient radiation dose is radiation. It is of utmost importance to select the increased. optimal field of view for a particular diagnostic task. For example, when examining teeth for Field of view fractures, periapical lesions or accessory pulp canals, a limited field of view CBCT examination Many CBCT units allow the operator to restrict the acquired at a high resolution is necessary. Similarly, beam size to a predetermined area or field of view. smaller field of view scans have a better diagnostic Typically, the field of view is described as small (or efficacy for detection of temporomandibular joint limited), medium, or large depending on the extent erosions.

30 Cone Beam Computed Tomography Rotation angle Figure 2.8 Coherent scatter. The incident photon transfers its energy to the atoms, causing the electrons to momentarily Typically during a CBCT scan, the tube and vibrate. As the atom returns to the ground state, it emits a detector move around the patient acquiring mul- photon of the same energy as the incident photon. tiple projections during a 360-degree rotation. However, some contemporary units provide an Compton scatter acquisition mode where the tube and detector assembly rotate around the patient for 180 degrees, When an incident photon with moderate energy thereby reducing the patient exposure. These collides with an outer orbital electron, it transfers modes will use fewer basis projections and thus some of its energy to the electron, which overcomes typically yield images that are lower in resolution its binding energy and is ejected from its orbital, than a full 360-degree scan. Depending on the diag- causing ionization of the atom. The incident photon nostic task, the images may be of adequate diag- retains some of its energy and is scattered at nostic quality. The use of 180-degree scans has an  angle to its initial path (Figure  2.9). Compton implications not only in dose reduction but also in scatter has important implications in diagnostic situations where patient motion may be an issue. radiology. First, it causes ionization of biological Research comparing the diagnostic efficacies of molecules and thus, results in radiation-induced 360- and 180-degree scans is lacking. damage. Second, the photons are scattered in all directions. Some of the scattered photons may Interaction of X-rays with matter expose adjacent tissues outside the immediate field  of radiation, causing biological damage. X-ray photons that strike an object have different Scattered photons may also exit the patient and potential fates. Some photons pass through the strike the image receptor, resulting in reduced object without any loss of energy. Alternatively, image contrast. Manufacturers incorporate filters photons may transfer some or all of their energy to into the X-ray beam to preferentially decrease the object’s molecules. There are three mechanisms the  number of low-energy photons, thereby whereby diagnostic X-ray photons interact with decreasing Compton interactions. This added matter—coherent scatter, Compton scatter, and filtration reduces patient dose and also improves photoelectric effect. image quality. Importantly, during Compton inter- actions, photons may also be scattered at an angle Coherent scatter of 180 degrees—backscatter radiation—and could potentially expose the operator. This type of interaction occurs predominantly with X-ray photons with energies less than approxi- mately 10 keV. As a low-energy photon courses adjacent to an atom, it loses all of its energy and causes an outer orbital electron to become excited. As the excited electron returns to its steady state, it emits an X-ray photon, with the same energy as the initial incident photon (Figure  2.8). The scattered photon is typically at an angle to the incident photon. Importantly, coherent scatter does not cause ionization of the atom. At the photon energies used in CBCT imaging, coherent scatter accounts for only a minor proportion of the photon interac- tions and is of little importance in diagnostic imaging.

The Nature of Ionizing Radiation and the Risks from Maxillofacial Cone Beam Computed Tomography 31 Figure 2.9 Compton scatter. The incident photon transfers energy of the electron. The remainder of the energy some of its energy to an electron, resulting in ionization is converted to kinetic energy of this electron—a of the atom. Following this interaction, the photon is photoelectron—that is ejected from the atom. deviated from its path as a scattered photon. Thus, in a photoelectric interaction, the incident photon loses all of its energy and results in ioniza- tion of the atom. Atoms with a high atomic number absorb more photons than atoms with lower atomic numbers; this is the basis for radiographic image formation. Tissues with a higher effective atomic number such as enamel, dentin, and bone absorb more photons than soft tissue and thus are  depicted on the radiographic image as radi- opaque objects. Likewise, dental materials such as amalgam, gold, and titanium have high atomic numbers and are seen as radiopaque regions on a radiograph. Biological effects of ionizing radiation As X-ray photons interact with biological tissues they can cause ionization of atoms in biological tissues. Ionization of biological molecules may manifest as radiation-induced effects. The type and nature of these effects depends on the tissue type exposed as well as the dose. There are two principal types of radiation-induced effects: deterministic and stochastic. Figure 2.10 Photoelectric interaction. The incident photon Deterministic effects transfers all of its energy to the atom, resulting in ionization. Deterministic effects of radiation are caused when Photoelectric absorption the radiation exposure to an organ or tissue exceeds a particular threshold level. At doses Photoelectric absorption is an important interac- below the threshold, the effect does not occur. All tion and is the basis for formation of the radio- individuals exposed to doses above the threshold graphic image. In this interaction, the X-ray will develop deterministic effects. Importantly, photon interacts with an inner-orbital electron at  doses above the threshold, the severity of the (Figure 2.10). As the photon collides with the elec- effect is proportional to the dose. Deterministic tron, it loses all of its energy to the electron. A part effects are typically a result of radiation-induced of this energy is  used to overcome the binding cell killing. Examples of deterministic radiation- induced effects include cataract formation, skin burns, fibrosis, xerostomia, and mucosal ulcera- tions. All dentomaxillofacial radiographic exami- nations are designed so that we do not induce any deterministic effects. However, dentists may often encounter such effects in patients who have received radiation therapy.

32 Cone Beam Computed Tomography Stochastic effects tumor suppressor genes—they can deregulate cell growth and/or differentiation and ultimately lead Unlike deterministic effects, stochastic effects have to neoplastic development. no minimum threshold for causation. Thus, any dose of radiation has the potential to induce a sto- Current paradigms consider that carcinogenesis chastic effect. While the probability of causing a is a multistep process with accumulation of muta- stochastic effect increases as the radiation dose tions in multiple oncogenes and tumor suppressor is  increased, the severity of the effect itself is not genes. Several aspects of ionizing radiation–induced dependent on dose. Either you get it or you don’t. cancer can be explained in the context of these con- Stochastic effects are caused by radiation-induced temporary molecular genetic models. For example, damage to DNA. The most important stochastic in addition to ionizing radiations, spontaneously effect is radiation-induced cancer. The absence of a occurring DNA damage and genotoxic chemicals threshold implies that any amount of radiation also cause DNA mutations. Thus, radiation- carries with it a risk for causing cancer. Although induced neoplasms do not differ fundamentally the potential for causing this effect cannot be from chemical-induced or spontaneous neoplasms. entirely avoided, minimizing the radiation dose Radiation-induced tumors have no clinical or histo- can decrease the possibility of inducing this effect; logical signatures that allow us to differentiate them this is the basis of radiation protection. from sporadically occurring tumors. Second, there is a latent period between radiation exposure and Radiation-induced cancer the manifestation of the neoplasm. This is expec- ted  given the multistep nature of tumorigenesis. Cancer induction is the most important stochastic Depending on the tumor type, this may vary from a effect from diagnostic radiation. It is well estab- few years to decades. It is also important to empha- lished that exposure to ionizing radiation results in size that there is a wide variation in the risk— an increase in the incidence of malignancies. These young children are almost two to three times more data are largely derived from studies of human sensitive to radiation-induced cancer, compared populations that were exposed to ionizing radia- with middle-aged and older adults. Equally impor- tion, either intentionally or by accident. Examples tant is the fact that certain tissues are more sensitive of such populations include early radiation to the carcinogenic effects of radiation than others. workers, radium dial painters, uranium miners, In the maxillofacial region, these highly sensitive individuals irradiated for benign diseases, patients tissues include the bone marrow (leukemia) and the with tuberculosis who underwent repeated chest thyroid glands. These age- and tissue-dependent fluoroscopy, and survivors of the atomic bombings sensitivities are significant considerations for radia- and the radiation disaster at Chernobyl. Studies of tion safety and protection. these human populations, as well as animal studies, have provided an insight into the mechanistic basis Radiation-induced cancer is the principal risk of for radiation’s cancer-inducing effect. There is diagnostic radiography. When designing radiation strong evidence that radiation-induced carcinogen- protection policies, it is necessary to estimate the risk esis is a consequence of ionizing radiation–induced from a given dose of radiation. Currently, these risk DNA damage. Ionizing radiation causes several estimates are based on the linear nonthreshold (LNT) types of DNA damage, including damage to model. The LNT model assumes that cancer risk is individual bases, single strand breaks, double directly proportional to radiation dose at all dose strand breaks, and DNA–protein cross-links. levels. The LNT is a hypothesis and has not been sci- Misrepair of DNA damage results in mutations of entifically proven or disproven. Nevertheless, there the normal DNA sequence. Such mutations may is strong scientific justification that supports this occur as single base alterations, deletions or inser- hypothesis. As discussed above, radiation-induced tions of DNA segments, or chromosomal rear- cancer is a consequence of DNA damage. Even when rangements such as translocations and inversions. the dose of radiation is small, the possibility of When the mutations involve growth-regulating ionizing radiation–induced DNA damage and sub- genes—activation of oncogenes or inactivation of sequent DNA mutations exists, and this supports the nonthreshold assumption of this model. Second, cell culture studies have demonstrated that as radiation

The Nature of Ionizing Radiation and the Risks from Maxillofacial Cone Beam Computed Tomography 33 dose increases, the magnitude of DNA damage When prescribing and performing diagnostic radio- also  rises, and the probability of DNA muta- logical examinations, dentists should ensure that tions  increases. This finding provides justification both of these principles are satisfied. To maximize to the assumption of a linear relationship between diagnostic benefits, dentists must identify those radiation dose and risk. Most radiation prote- clinical situations where radiographic examina- ction  agencies around the world, including the tions  would provide additional information that is International Commission on Radiation Protection essential for diagnosis and management of the and the National Council on Radiation Protection patient’s condition. To minimize risks from radiation and Measurement, use the LNT to estimate radia- exposure, dentists must implement appropriate tion-induced risks. Nevertheless, the LNT model is dose-reduction procedures (White and Mallya, 2012). not universally accepted. The opponents of this Importantly, dentists must understand the magni- model argue that the assumptions do not take into tude of potential risks from radiographic examina- consideration cellular adaptive responses that may tions and convey this information in a manner that be effective at lower doses. Furthermore, the LNT can be easily comprehended by patients. model does not account for age at exposure, and assumes that sensitivity to radiation-induced cancer Sources of radiation for a particular organ is the same at all ages. Opponents of the LNT model argue that it overesti- Background radiation mates cancer risk from diagnostic radiation. All individuals are continuously exposed to radia- Risk from CBCT examinations tion from various natural and man-made sources (Figure  2.11). Natural radiation sources refer to The basic premise of diagnostic radiology is that the ubiquitous background radiation. The naturally diagnostic benefits from the radiographic examina- occurring radionuclides, in particular radon and tion far outweigh the risks from radiation exposure. thoron, contribute to a large part of this background Figure 2.11 Sources of radiation exposure in the United States. The average annual exposure to individuals in the U.S. is approximately 6.2 mSv. Half of this is from background sources and half from man-made sources. The relative contributions of the various sources are shown in the pie chart. Note that diagnostic imaging contributes a large proportion of the total exposure. Data derived from NCRP, 2009.

34 Cone Beam Computed Tomography radiation. Other natural sources include space radi- et  al., 2008; Loubele et  al., 2009; Loubele et  al., ation (cosmic rays and solar energetic particles), 2005;  Ludlow, 2011; Ludlow et  al., 2003; Ludlow terrestrial radiation from radioactive elements in et  al., 2006; Ludlow and Ivanovic, 2008; Okano rocks and soil, and internal radiation from radionu- et al., 2009; Pauwels et al., 2012; Roberts et al., 2009; clides that are ingested through food and water or Suomalainen et  al., 2009). Typically, these doses inhaled through air. The average annual effective are  determined using dosimeters placed at mul- dose from background radiation exposure in the tiple sites in a tissue-equivalent anthropomorphic United States is approximately 3.1 mSv (see “Units phantom to measure absorbed doses at specific of Radiation” section for definition of radiation organ sites. The measured absorbed doses are then dose units). Background radiation is often used as a used to calculate the effective dose from an exami- basis to convey the magnitude of radiation risks nation. Such studies provide an estimate of the from diagnostic radiological examinations. For dose that a patient is likely to receive from a specific example, an examination with an effective dose of CBCT examination. 0.31 mSv would result in an exposure equivalent to 36.5 days of background exposure. The striking point that emerges from these studies is that the effective dose, and thus radiation Man-made radiation risk, varies significantly between CBCT units from different manufacturers (Figure 2.12 and Table 2.1). The major contributor to this category of radiation Furthermore, different protocol settings of the same exposure is from diagnostic radiology and nuclear unit also result in markedly different radiation medicine. Consumer products, occupational expo- doses. This is particularly important because CBCT sure, and industrial sources account for a minor has often been publicized as a low-dose proce- component of this category. In the United States dure.  However, given the significant variability there has been a dramatic increase in medical radia- depending on manufacturer and selected imaging tion exposure. In 1980 medical radiation exposure protocol,  it is important that dentists fully was only one-sixth of natural background exposure. understand the radiation doses delivered by the In 2006, medical exposures equaled background specific CBCT exams that they prescribe and make. radiation, increasing the total annual effective dose Given the wide variation in the radiation dose bet- from all sources to 6.2 mSv. This increase is mainly ween manufacturers, dentists should give due due to exposures from computed tomography and consideration to this issue when purchasing a unit, reflects both an increase in the numbers of examina- or when referring a patient to an imaging facility. tions as well as the dose per examination. CT now Equally important is the increased radiation dose accounts for 24% of the annual total effective dose with some high-resolution imaging protocols. from all sources. Conventional radiography and Dentists must be familiar with the diagnostic situa- fluoroscopy account for 5% of the total dose. Dental tions that require such high-resolution protocols radiography accounts for approximately 2.5% of the and appropriately consider the balance between dose from conventional radiography. It should be diagnostic benefit and radiation risk. emphasized, however, that these data do not include exposure from CBCT, which is being increasingly Often patients who have been prescribed a CBCT used in dentistry. examination may inquire about the risks from these procedures. While dentists must be aware of Risk-estimates for CBCT examinations the estimated effective doses from such examina- The principal detriment from diagnostic X-radiation tions, it is often useful to convey these to patients, is radiation-induced neoplasia; the magnitude in the context of its equivalent of background of  this risk increases with radiation dose. Thus, exposure. Additionally, it is also useful to provide knowledge of the dose delivered by a diagnostic similar data for commonly used dental and med- radiographic examination is key for its risk-benefit ical radiographic procedures to allow the patient to analysis. Several studies have estimated effective place the dose to be received in proper perspective. doses that result from CBCT examinations (Hirsch Table 2.1 lists the effective dose from several CBCT, et  al., 2008; Librizzi et  al., 2011; Lofthag-Hansen multislice CT, and commonly used dentomaxillofa- cial radiographic examinations. Figure 2.12 shows these doses grouped by the size of the field of view.

The Nature of Ionizing Radiation and the Risks from Maxillofacial Cone Beam Computed Tomography 35 1000 Radiation dose (microSv)100 FMX PAN 10 Ceph BW 1 Limited Medium Large Intraoral Extraoral CBCT (FOV) Figure 2.12 Effective doses from dentomaxillofacial examinations. Note that doses are plotted on the y-axis on a logarithmic scale. Data are derived from sources listed in Table 2.1. Note also the striking overlap between limited, medium, and large field of view machines. Thus, in some situations a limited field of view machine can result in a larger effective dose than a large field of view machine from a different manufacturer. Methods to minimize radiation dose clinical findings, identifies those situations where from CBCT exams radiography is needed and prescribes the appro- priate radiographic examination that would provide While CBCT radiation doses are typically lower than the needed diagnostic examination. Selection criteria those from multislice maxillofacial CT examina- are an essential and often overlooked approach to tions, it should be remembered that the overarching minimizing patient radiation exposure. philosophy of radiation protection is minimizing the radiation dose to the patient while maintaining Guidelines have been established to help den- the diagnostic benefit. This philosophy is embodied tists select the appropriate radiographic examina- in  the principle and practice of ALARA—As Low tion. For example, the American Dental Association As  Reasonably Achievable. This principle aims to has developed guidelines that provide dentists reduce the radiation dose of exposed individuals with a framework to prescribe commonly used to as low levels as practically achievable. There are conventional radiographic modalities, including several means to satisfy this principle. intraoral, panoramic, and cephalometric imaging (ADA Council on Scientific Affairs, 2001). While Selection criteria these ADA guidelines do not include CBCT imaging, the principles underlying these guidelines apply to The basic premise of diagnostic radiography is that prescribing CBCT examinations. These basic prin- the diagnostic benefits of radiation far outweigh the ciples are clearly outlined in a position paper from risks from radiation exposure. Thus, a fundamental the American Academy of Oral and Maxillofacial requirement of all diagnostic radiological exams Radiology (White et  al., 2001) and in guidelines is  that they must have the potential to provide from the European Academy of Oral and information that is valuable for diagnosis and patient Maxillofacial Radiology (Horner et  al., 2009). management. It must be emphasized that any radio- Recently, the American Association of Endodon- graphic examination, including CBCT, be performed tists and the American Academy of Oral and after a complete history and clinical examination. Maxillofacial Radiology (2011) published a joint Judicious use of diagnostic radiation requires that position statement to provide guidance to the the dentist identify those clinical situations where the use  of CBCT imaging in endodontic treatment. radiological examination is likely to provide this These guidelines emphasize justification of radio- benefit. The term selection criteria refers to this process graphic examinations on an individual basis. CBCT where a dentist, based on the patient’s historical and has become increasing popular in orthodontic treatment planning. White and Pae (2009) have suggested guidelines for selection of orthodontic

36 Cone Beam Computed Tomography Table 2.1 Effective doses from selected CBCT and dentomaxillofacial radiographic examinations. Examination Effective Dose* (microSv) Equivalent Background Radiation (days)‡ CBCT small (limited) field of view 13–44 2–5 3D Accuitomo, 4 × 4 cm 19–40 2–5 Kodak 9000, 5 × 3.7 cm Pax-Uni 3D, 5 × 5 cm 44 5 CBCT medium field of view 3D Accuitomo170, 10 × 5 cm 54 6 CB Mercuray, 10 cm diameter 279 33 CB Mercuray, 15 cm diameter 548 65 iCAT next generation, 16 × 6 cm iCAT classic, 16 × 8 cm 45 5 iCAT classic, 16 × 8 cm, high-resolution 34–77 4–9 protocol 68–149 8–18 Kodak 9500, 15 × 8 cm NewTom3G, 10 cm diameter 76–166 9–20 NewTomVGi, 15 cm × 15 cm, 57 7 high-resolution protocol NewTomVGi, 12 cm × 8 cm, 194 23 high-resolution protocol Picasso Trio, 12 × 7, low dose 265 31 Picasso Trio, 12 × 7, high dose Prexion, 8 × 8 cm, low dose 81 10 Prexion, 8 × 8 cm, high dose 123 14 ProMax3D, 8 × 8 cm, low-dose protocol 189 22 Promax3D, 8 × 8 cm, high-dose protocol 389 46 Scanora 3D, 10 × 7.5 cm Veraviewepocs 3D, 8 × 8 cm 28 3 CBCT large field of view 122–652 14–77 CB Mercuray, 20 cm diameter Galileos Comfort, 15 × 15 cm 45 5 iCat Next generation, 16 × 13 cm 73 9 Illuma, 21 × 14 cm, low-resolution protocol 569–1073 67–126 Illuma, 21 × 14 cm, high-resolution 70–128 8–15 protocol 74–83 9–10 Kodak 9500, 20 × 18 cm 98 12 NewTom 3G, 19 cm diameter NewTom VG, 23 cm × 23 cm 368–498 43–59 Scanora 3D, 14.5 × 13.5 cm Skyview, 17 × 17 cm 93–260 11–31 30–68 4–8 83 10 68 8 87 10 (Continued )

The Nature of Ionizing Radiation and the Risks from Maxillofacial Cone Beam Computed Tomography 37 Table 2.1 (Continued) Examination Effective Dose* (microSv) Equivalent Background Radiation (days)‡ Multislice CT 860 101 534 63 Siemens Somatom (64-slice), 12 cm scan length 1500 177 180 21 Siemens Somatom (64-slice), 12 cm scan length, automatic exposure control 5 0.6 protocol 35 4 171 Siemens Sensation (16-slice), 22.7 cm 14–24 20 scan length 2–3 6 0.7 Siemens Sensation (16-slice), 22.7 cm scan length, low-dose protocol Intraoral radiographs Bitewings (PSP/F-speed, rectangular collimation) (PSP/F-speed, rectangular collimation) (PSP/F-speed, round collimation) Panoramic (digital, CCD-based) Lateral cephalomteric (digital, PSP-based) * Doses are rounded to the nearest whole number. Dose range is based on data derived from Hirsch et al., 2008; Librizzi et al., 2011; Lofthag-Hansen et al., 2008; Loubele et al., 2009; Loubele et al., 2005; Ludlow, 2011; Ludlow et al., 2003; Ludlow et al., 2006; Ludlow and Ivanovic, 2008; Okano et al., 2009; Pauwels et al., 2012; Roberts et al., 2009; and Suomalainen et al., 2009. ‡ Calculation of background equivalent days is based on an annual exposure of 3.1 milliSv. For doses above 10 microSv, the background equivalent days are rounded to the nearest whole number. patients who would likely benefit from this advantages, applications, and limitations of this imaging, emphasizing its value in assessing cranio- technology to ensure that patients selected for facial asymmetry, planning for orthognathic treat- these examinations will benefit from the diagnostic ment, evaluation of cleft palate patients, localizing information. Furthermore, these individuals must impacted and supernumerary teeth, and guiding be familiar with viewing and manipulation of mul- placement of orthodontic mini-implants. However, tiplanar CBCT images. This includes knowledge its routine use for all orthodontic patients is contro- of  dentomaxillofacial radiographic anatomy and versial and has not been substantiated by scientific appearances of pathological lesions on CBCT exami- evidence. nations. To maximize diagnostic yield and patient benefit, the entire CBCT volume must be inter- Operator training preted. This includes navigation through the multi- planar images outside of the region for which the Users of CBCT imaging at all levels should have examination was ordered and creating additional appropriate training in the use of this technology. reconstructions as appropriate. Where necessary, This is essential both to maximize the diagnostic dentists must consult with an oral and maxillofa- yield and to minimize the patient dose. The extent cial radiologist to report on the entire CBCT image of training will depend on the dentist’s role in volume. CBCT imaging. All dentists who use CBCT imaging for their patients’ care must be familiar with the Dentists who operate CBCT units in their clinics must have adequate training in the principles of CBCT production. All operators of CBCT units,

38 Cone Beam Computed Tomography Figure 2.13 Imaging protocol parameters. Control panel from the Accuitomo 170 demonstrating the various parameters to be selected for an imaging examination. These include the exposure factors (kVp and mA), the field of view, the rotational arc, and the scan mode. These parameters must be adjusted to optimize diagnostic quality and minimize radiation dose. including dentists and their technical staff, must alone will reduce radiation exposure to the thyroid understand the influence of exposure parameters gland and mandibular bone marrow and thus as well as any machine-specific parameters on significantly decrease effective dose. In addition to diagnostic quality and patient dose. These opera- the higher dose, a larger field of view results in tors must also receive appropriate training in more scattered radiation that compromises image quality assurance protocols and data storage and quality. To this end, it is important to recognize that transfer. Additionally, as with any other radio- a single CBCT unit may not be sufficient to provide graphic examination, these individuals must under- field of view sizes that encompass all diagnostic stand the principles of radiation protection and tasks, and this should be a consideration when implement the following methods to reduce dentists refer patients for CBCT examinations. patient dose. Librizzi et  al. (2011) showed that diagnostic effi- cacy to detect temporomandibular joint erosions Optimizing imaging protocols was significantly impacted by the field of view, with a higher diagnostic accuracy with smaller Although performing a CBCT examination appears field of view size. Thus, using a large field of view relatively simplistic, it is essential that operators of examination to examine the temporomandibular CBCT units optimize their imaging protocols to joints for osteoarthritic changes will not only ensure that the radiation dose to the patient is kept deliver a higher dose to the patient, it will also as low as reasonably achievable while maintaining result in a lower diagnostic benefit. It should also adequate diagnostic quality. There are several set- be emphasized that required diagnostic quality is tings in a CBCT unit that influence both the dose dependent on the diagnostic task. To this end, clini- and the image quality (Figure 2.13). cians who prescribe CBCT examinations must be familiar with the field of view and select the small- Field of view est that will provide an adequate view for each The smallest field of view needed for the diagnostic diagnostic task. task should be used. Typically, as the field of view increases, the volume of tissue irradiated increases Exposure factors and the radiation dose to the patient is higher The exposure settings should be optimized for the (Table  2.1). For example, when imaging the max- diagnostic task as well as considering individual illa, collimating the beam to the maxillary region patient size and anatomic site to be imaged. This is

The Nature of Ionizing Radiation and the Risks from Maxillofacial Cone Beam Computed Tomography 39 necessary to get diagnostic quality images and units and must be adequately trained to select the reduce retakes. Typically, this is accomplished by appropriate scan mode depending on the diagnostic reducing the mA to decrease the number of pho- task and the individual patient’s circumstances. tons and thus radiation dose. Such optimization is particularly important when imaging a child, due Angle of rotation to higher radiosensitivity of the bone marrow and Some CBCT units allow the operator to select thyroid gland. an  exposure mode where the rotation arc is 180 degrees instead of 360 degrees. In this mode, the One manufacturer, NewTom (Imageworks number of basis projections taken for image recon- Corporation), uses a patented “safebeam” tech- struction is lower; thus, radiation dose to the patient nology. In this technology, the amount of radiation is lower. Given the decreased number of basis received by the image sensor provides feedback projections, the resolution of the image is lower to  automatically adjust the exposure parameters, than when obtained with a full 360-degree rotation. thereby customizing exposure for every patient. This scan mode will reduce patient dose. However, Such automated adjustments provide an excellent the adequacy of the diagnostic information with approach to minimizing radiation exposure to this acquisition mode has not been well studied. patients. Protective thyroid collars and Scan modes protective aprons Some contemporary CBCT units allow the operator to select from a variety of scan modes. For example, The use of thyroid shields during maxillofacial some units provide the option of a “high resolu- CBCT reduces the absorbed dose to the thyroid tion” scan mode. These high-resolution modes gland and thus the patient effective dose. However, acquire images at a smaller voxel size. In order to it is important to ensure that the thyroid collar is increase the signal-to-noise ratio, these scan modes not in the path of the primary beam—this would use an increased mA or more basis projections, lead to significant artifacts that may compromise both of which increase patient dose (Table  2.1). the diagnostic quality of the image. When all other Prior to using this scan mode, the need for the high procedures are followed, it may not be necessary to resolution for the particular diagnostic task must use lead aprons during the CBCT exam. However, be evaluated. If the lower resolution mode pro- some states in the United States require the use of vides adequate diagnostic information, then the lead (or lead-equivalent) aprons for all dentomaxil- added radiation dose subjects the patient to addi- lofacial radiographic examinations. tional risk while not providing any additional benefit. For example, evaluation of dental and peri- Units of radiation apical structures and root fractures requires higher resolution, whereas evaluation of craniofacial asym- Exposure metry can be satisfactorily accomplished at lower resolutions. For some units, these higher resolution This unit of radiation conveys the dose of radiation scan modes also increase exposure time, and the in air. The traditional unit of exposure is Roentgen. clinician must take into consideration the possi- In the SI system, exposure is conveyed as cou- bility of patient motion, which could degrade lombs/kg. From a practical viewpoint, this unit is image quality and render the examination diagnos- used to measure the amount of radiation that exits tically inadequate. from the X-ray tube head, either at or at various distances from the tube head. These measurements Some manufacturers offer the option of a fast scan are used to calculate the need for protective shield- mode, where the number of basis projections is ing. This unit is also used to measure leakage of reduced, thereby decreasing scan time and lowering radiation from the tube head or denote the amount radiation dose. Such modes generally yield images of radiation at the skin surface. with a resolution lower than the standard scan mode. However, depending on the diagnostic task, this image quality may be sufficient. Operators of CBCT units must be familiar with these features of their

40 Cone Beam Computed Tomography Absorbed dose effects include the thyroid gland, active bone marrow, salivary glands, brain, and bone surface. As described above, x-radiation interacts with and Similar to equivalent dose, the units of effective transfers energy to the patient’s tissues. The unit of dose are Sieverts (Sv) or rems. Effective dose is absorbed dose is a measure of how much energy is mathematically denoted as: transferred to (absorbed by) the exposed tissues. In radiation protection, absorbed doses to the exposed & = ∑ 8T r)T tissues are measured as a first step in estimation of the overall dose from radiographic examinations. where E is effective dose, WT is the tissue-weighting In the SI system, absorbed dose is measured in factor, and HT is the equivalent dose. gray. One gray represents 1 joule of energy absorbed per kilogram of tissue. The tradition unit It is important to understand the concept of of absorbed dose is rad. effective dose. This is the unit that is used to convey the net detriment from a radiographic examina- Equivalent dose tion, and it is used to compare radiation risks between different modalities, specific imaging pro- The type of radiation influences the magnitude of tocols, and radiographic examinations that expose biological damage from the same absorbed radia- different regions of the body. For example, the risk tion dose. The unit of equivalent dose considers the of a maxillofacial CBCT examination with an effec- type of radiation that resulted in energy transfer. It tive dose of 100 μSv is ten times higher than the risk is a product of the absorbed dose and the radiation- from a panoramic radiographic examination with weighting factor, WR, and is mathematically sum- an effective dose of 10 μSv. marized as: References )T = ∑ 8R r%T ADA Council on Scientific Affairs. (2001). An update on where HT is the equivalent dose, WR is the radiation- radiographic practices: Information and recommenda- weighting factor, and DT is the absorbed dose. tions. Journal of the American Dental Association, 132(2): 234–8. For X-rays, the weighting factor is one; thus, absorbed dose is numerically equal to the equi- American Association of Endodontists and American valent dose. Equivalent dose is measured in Sieverts Academy of Oral and Maxillofacial Radiology. (2011). (Sv). The traditional unit of equivalent dose is Use of cone-beam computed tomography in endodon- the rem. tics: Joint Position Statement of the American Associa- tion of Endodontists and the American Academy of Effective dose Oral  and Maxillofacial Radiology. Oral Surgery, Oral Medicine, Oral Pathology, Oral Radiology and Endodontics, Different tissues have different sensitivities to 111(2): 234–7. radiation-induced stochastic effects. Thus, the total detriment per unit of equivalent dose varies depend- Hirsch, E., Wolf, U., Heinicke, F., et al. (2008). Dosimetry ing on the tissue types exposed. The unit of effective of the cone beam computed tomography Veraviewepocs dose accounts for this differential sensitivity. 3D compared with the 3D Accuitomo in different Depending on their sensitivity to radiation-induced fields of view. Dentomaxillofacial Radiology, 37(5): stochastic effects, tissues have been assigned a 268–73. weighting factor. This factor represents the relative contribution of injury to that organ or tissue to total Horner, K., Islam, M., Flygare, L., et  al. (2009). Basic risk of stochastic radiation effect. In the maxillofa- principles for use of dental cone beam computed cial region, tissues with increased risk of stochastic tomography: Consensus guidelines of the European Academy of Dental and Maxillofacial Radiology, Dentomaxillofacial Radiology, 38(4): 187–95. Librizzi, Z.T., Tadinada, A.S., Valiyaparambil, J.V., et  al. (2011). Cone-beam computed tomography to detect erosions of the temporomandibular joint: Effect of field of view and voxel size on diagnostic efficacy and