Strategies for Reducing Radiation Dose in CT Cynthia H. McCollough, PhD*, Andrew N. Primak, PhD , Natalie Braun, BS, James Kofler, PhD , LifengYu, PhD , Jodie Christner, PhD BACKGROUND AND SIGNIFICANCE High-end CT systems allow acquisition of isotropic volumetric data sets that permit high quality refor- matted images and images with less than 1 mm slice thickness. These capabilities have greatly ex- panded the usefulness of CT, and CT usage has correspondingly increased, replacing more and more radiographic examinations. In 1990, approx- imately 13 million CT scans were performed in the United States. 1 In 2000, the number of CT scans more than tripled to approximately 46 million. 1 The estimated number of CT scans for 2006 is 62 million. 1 With high-quality CT imaging being performed more frequently, patients can benefit from a quicker and more accurate diagnosis and precise anatomic information for planning thera- peutic procedures. Despite the tremendous con- tributions of CT to modern health care, however, some attention must also be given to the small health risk associated with the ionizing radiation received during a CT examination. CT scanners create cross-sectional images by measuring x-ray attenuation properties of the body from many different directions. Currently, the radiation dose associated with a typical CT scan (1–14 mSv depending on the examination) is comparable to the annual dose received from natural sources of radiation, such as radon and cosmic radiation (1–10 mSv), depending on where a person lives. 2 The health risk to an individual from exposure to radiation from a typical CT scan is thus comparable to background levels of radiation. Considering the growing population of people undergoing CT scans, however, the impli- cations of CT radiation dose on public health ef- fects may be significant, although considerable debate exists regarding this assumption. One study suggested that as much as 0.4% of all cur- rent cancers in the United States may be attribut- able to the radiation from CT studies based on CT usage data from 1991 to 1996. 1 When organ-spe- cific cancer risk was adjusted for current levels of CT usage, it was determined that 1.5% to 2% of cancers may eventually be caused by the ionizing radiation used in CT. 1 This study and a similar se- ries of previous articles 3,4 received considerable attention from the public media. One positive effect of the media attention was that the CT com- munity was obliged to review the amount of radia- tion prescribed for CT scans, especially for pediatric patients. This review ultimately resulted in an aggressive effort to minimize CT doses and optimize image quality. Concurrently, new tech- nologies, such as automatic exposure control (AEC), were in development, and were eventually made commercially available for all current CT systems. The use of AEC greatly enhances and simplifies efforts to decrease patient dose. The media reporting on the risk of CT radiation doses has had a negative effect on the public’s per- ception of CT imaging and radiation, influenced by the sensationalist and alarmist tone of the news stories. Phrases such as ‘‘.dangerous radiation from ‘super X-rays’.’’ 5 and comparisons to atomic bomb survivors 6 exploit the public’s general apprehension of radiation. Although informing the public of potential health risks—even small Department of Radiology, Mayo Clinic, 200 First Street SW, Rochester, MN 55905, USA * Corresponding author. E-mail address: [email protected] (C.H. McCollough). KEYWORDS CT Radiation dose Cardiac CT Dose reduction Automatic exposure control Effective dose Radiol Clin N Am 47 (2009) 27–40 doi:10.1016/j.rcl.2008.10.006 0033-8389/08/$ – see front matter ª 2009 Elsevier Inc. All rights reserved. radiologic.theclinics.com
Transcript
Strategies for ReducingRadiation Dose in CT
Cynthia H. McCollough, PhD*, Andrew N. Primak, PhD,Natalie Braun, BS, James Kofler, PhD, LifengYu, PhD,Jodie Christner, PhD
High-end CT systems allow acquisition of isotropicvolumetric data sets that permit high quality refor-matted images and images with less than 1 mmslice thickness. These capabilities have greatly ex-panded the usefulness of CT, and CT usage hascorrespondingly increased, replacing more andmore radiographic examinations. In 1990, approx-imately 13 million CT scans were performed in theUnited States.1 In 2000, the number of CT scansmore than tripled to approximately 46 million.1
The estimated number of CT scans for 2006 is62 million.1 With high-quality CT imaging beingperformed more frequently, patients can benefitfrom a quicker and more accurate diagnosis andprecise anatomic information for planning thera-peutic procedures. Despite the tremendous con-tributions of CT to modern health care, however,some attention must also be given to the smallhealth risk associated with the ionizing radiationreceived during a CT examination.
CT scanners create cross-sectional images bymeasuring x-ray attenuation properties of thebody from many different directions. Currently,the radiation dose associated with a typical CTscan (1–14 mSv depending on the examination)is comparable to the annual dose received fromnatural sources of radiation, such as radon andcosmic radiation (1–10 mSv), depending on wherea person lives.2 The health risk to an individualfrom exposure to radiation from a typical CTscan is thus comparable to background levels ofradiation. Considering the growing population of
Department of Radiology, Mayo Clinic, 200 First Street S* Corresponding author.E-mail address: [email protected] (C.H. McC
Radiol Clin N Am 47 (2009) 27–40doi:10.1016/j.rcl.2008.10.0060033-8389/08/$ – see front matter ª 2009 Elsevier Inc. All
people undergoing CT scans, however, the impli-cations of CT radiation dose on public health ef-fects may be significant, although considerabledebate exists regarding this assumption. Onestudy suggested that as much as 0.4% of all cur-rent cancers in the United States may be attribut-able to the radiation from CT studies based on CTusage data from 1991 to 1996.1 When organ-spe-cific cancer risk was adjusted for current levels ofCT usage, it was determined that 1.5% to 2% ofcancers may eventually be caused by the ionizingradiation used in CT.1 This study and a similar se-ries of previous articles3,4 received considerableattention from the public media. One positiveeffect of the media attention was that the CT com-munity was obliged to review the amount of radia-tion prescribed for CT scans, especially forpediatric patients. This review ultimately resultedin an aggressive effort to minimize CT doses andoptimize image quality. Concurrently, new tech-nologies, such as automatic exposure control(AEC), were in development, and were eventuallymade commercially available for all current CTsystems. The use of AEC greatly enhances andsimplifies efforts to decrease patient dose.
The media reporting on the risk of CT radiationdoses has had a negative effect on the public’s per-ception of CT imaging and radiation, influenced bythe sensationalist and alarmist tone of the newsstories. Phrases such as ‘‘.dangerous radiationfrom ‘super X-rays’.’’5 and comparisons toatomic bomb survivors6 exploit the public’s generalapprehension of radiation. Although informing thepublic of potential health risks—even small
risks—is not inappropriate, journalistic responsibil-ity should ensure that the data are presented so asnot to exaggerate or present the risk estimates ina manner that can be easily misinterpreted bya population that, in general, is not sufficientlyknowledgeable in radiation or radiobiology to accu-rately assess the information. Such alarmist articlesare a public disservice in that they cause unneces-sary stress to patients, or in some cases may per-suade a patient to decline a CT scan that couldhave a positive health impact. The latter casecannot be overstated because low-level radiationrisk estimates, which are derived primarily fromatomic bomb survivors and have considerableuncertainties at low doses (<100 mSv), give no con-sideration to the medical benefit of a CT scan.There is no question that the benefit of an appropri-ately indicated CT scan far exceeds the associatedestimated risk or that CT providers need to pre-scribe the minimal amount of radiation required toobtain images adequate for evaluating the patient’scondition. In general, the medical communityneeds to better educate the public to the risksand benefits associated with CT so that they canmake informed decisions regarding their healthcare.
BRIEF TUTORIAL ON THEMEASUREMENTOF RADIATION OUTPUT FOR CTScanner Output
The computed tomography dose index (CTDI) isthe primary metric used in CT to describe the radi-ation output from a scanner. It is a measure of theamount of radiation delivered from a series ofcontiguous irradiations to a pair of standardizedacrylic phantoms. It is, however, measured fromone axial CT scan (one rotation of the x-raytube).7–10 The CTDI was defined in the early daysof CT, when dose assessments were made usingthermoluminescent dosimeters and multiple axialscans, each one incremented from the previousscan by the nominal beam width. This procedurewas not only time consuming and laborious, butalso required many scans (exposures) for eachbeam width, phantom size, tube potential setting(kV), and position in the field of view (FOV, periph-ery or center). The resultant parameter was re-ferred to as the multiple-scan average dose(MSAD), which was typically a factor of two tothree times higher than the peak radiation dosefrom one axial scan. Shope and Gagne9 demon-strated the mathematical equivalence betweenthe scan intensive MSAD and the CTDI, which isable to be measured using only one scan (one gan-try rotation), when certain criteria are met with
regard to the length of the ionization chamberand the length of the clinical scan being assessed.
The theoretically perfect equivalence of MSADand CDTI is not achieved in many clinical scenar-ios, but because of the speed and ease of CTDImeasurements, the use of MSAD declined. CTDIis now used internationally following a standard-ized measurement technique. Several variants ofCTDI exist that describe specific steps in the mea-surement and calculation processes. These in-clude the CTDI100 and the weighted CTDI(CTDIw). These parameters are described in multi-ple publications.11–13
Volume CT dose indexThe CTDI variant that is currently of most rele-vance is the volume CTDI (CTDIvol). This parameteraccounts for gaps or overlaps between the x-raybeams from consecutive rotations of the x-raysource and variations in dose across the FOV.The CTDIvol provides a single parameter, basedon a directly and easily measured quantity, whichdescribes the radiation delivered to the scan vol-ume for a standardized (CTDI) phantom.13 The SIunits are milligray (mGy). CTDIvol is a useful indica-tor of the radiation output for a specific examina-tion protocol, because it takes into accountprotocol-specific information, such as pitch.CTDIvol is not a direct measurement of dose; it isa standardized measure of radiation output in theCT environment.14
Dose-length productTo better represent the overall energy delivered bya given scan protocol, the CTDIvol can be inte-grated along the scan length to compute thedose-length product (DLP),7 where the DLP (inmGy-cm) is equal to CTDIvol (in mGy) times scanlength (in cm). The DLP reflects the integrated ra-diation output (and thus the potential biologiceffect) attributable to the complete scan acquisi-tion. An abdomen-only CT examination mighthave the same CTDIvol as an abdomen/pelvis CTexamination, but the latter examination wouldhave a greater DLP, proportional to the greaterz-extent of the scan volume.
Effective Dose
Effective dose, E, is not a measurement of dose,but rather a concept that reflects the stochasticrisk (eg, cancer induction) from an exposure to ion-izing radiation.15,16 It is typically expressed in theunits of millisieverts. Effective dose reflects radia-tion detriment averaged over gender and ageand its use has several limitations when appliedto medical populations.15–18 In particular, it usesa mathematical model for a ‘‘standard’’ body in
Strategies for Reducing Radiation Dose in CT 29
its calculation19 and is hence not an appropriaterisk indicator for any one individual. It does, how-ever, facilitate the comparison of biologic effectbetween diagnostic examinations of differenttypes or having different acquisition parame-ters.15,16 By comparing patient effective dose tobackground radiation dose from natural sources,which in the United States averages 3 mSv peryear with a range across the United States from1 to 10 mSv,2 patients and their families are betterable to put the risk associated with medical dosesinto perspective.
The European Working Group for Guidelines onQuality Criteria in CT has proposed a generic esti-mation method for effective dose.7 A set of coeffi-cients k, where the values of k depend only on theregion of the body being scanned (Table 1),20–21
were determined in relation to the DLP. E (inmSv) can thus be estimated by multiplying theDLP value (in mGy-cm), which is reported onmost CT systems, by the region-specific k coeffi-cient (in mSv/(mGy-cm)).
The values of E predicted by DLP and the valuesof E estimated using more rigorous calculationmethods are remarkably consistent, with a maxi-mum deviation from the mean of approximately10% to 15%.22 Thus, the use of DLP to estimateE seems to be a reasonably robust method for esti-mating effective dose. However, effective dosealone does not give a complete picture of estimatedradiation risk to specific radiation-sensitive organsor patients of a specific age or gender. For a com-plete picture, specific organ doses and age, gender,and organ-specific risk estimates are needed.
Recently the International Commission on Radi-ation Protection (ICRP) altered their recommenda-tions regarding the relative radiation sensitivities ofvarious organs and tissues. For the same exact
Table1Normalized effective dose per dose-length product for adof various ages over various body regions
Region of Body 0-y-Old 1-y-Old
Head and neck 0.013 0.0085
Head 0.011 0.0067
Neck 0.017 0.012
Chest 0.039 0.026
Abdomen and pelvis 0.049 0.030
Trunk 0.044 0.028
Conversion factor for adult head and neck and pediatric patieother conversion factors assume use of the 32-cm diameter CTdose as defined in ICRP report #6015 and do not reflect the ne
scan performed on the same exact equipmentand resulting in the same predicted organ doses,the value for effective dose differs depending onwhether the recommendations from 1991 (ICRP60) or 2007 (ICRP 103) are used.15,18 The versionof E (eg, ICRP 60 or ICRP 103) must be providedwhen a value of effective dose is given. The datapresented in Table 1 make use of the organweighting factors from ICRP 60; the new valuesfrom ICRP103 have not yet been widely adopted.
JUSTIFICATION AND OPTIMIZATIONGeneral Principles of ‘‘as Low as ReasonablyAchievable’’
The guiding principles for radiation protection inmedicine are:
ults
k
nts asbod
wly re
Justification: The examination must be medi-cally indicated.
Optimization: The examination must beperformed using doses that are as low asreasonably achievable (ALARA), consistentwith the diagnostic task.
Limitation: Although dose levels to occupa-tionally exposed individuals (ie, the radiolo-gist or technologist) are limited to levelsrecommended by consensus organiza-tions, limits are not typical for medicallynecessary examinations or procedures.
As the growth in CT use increased, particularlyin pediatric patients, and concern over the popula-tion dose from CT was expressed in the scientificliterature and lay press,3,4,23,24 it became clearthat the responsible use of CT required adjustmentof technique factors based on patient size (attenu-ation characteristics).3,25,26 In response, the radi-ology community (radiologists, physicists, and
(standard physique) and pediatric patients
(mSvmGyL1cmL1)
5-y-Old 10-y-Old Adult
0.0057 0.0042 0.0031
0.0040 0.0032 0.0021
0.011 0.0079 0.0059
0.018 0.013 0.014
0.020 0.015 0.015
0.019 0.014 0.015
sumes use of the head CT dose phantom (16 cm). Ally phantom.20,21 These coefficients estimate effectivecommended tissue weighting values of ICRP #103.18
McCollough et al30
manufacturers) has worked to implement ALARAprinciples in CT imaging.23,27–33 The guiding prin-ciple for dose management in CT is that the rightdose for a CT examination takes into account thespecific patient attenuation and the specific diag-nostic task. For large patients, this means a doseincrease is consistent with ALARA principles.
Additionally, each CT examination must be ap-propriate for the individual patient. Justification isa shared responsibility between requesting clini-cians and radiologists. For medical exposures,the primary tasks of the imaging community areto work with ordering clinicians to direct patientsto the most appropriate imaging modality for therequired diagnostic task, and to ensure that alltechnical aspects of the examination are opti-mized, such that the required level of image qualitycan be obtained while keeping the doses as low aspossible. The American College of Radiology Ap-propriateness Criteria provides evidence-basedguidelines to help physicians in recommendingan appropriate imaging test.34 The EuropeanCommission guidelines and United Kingdom’sRoyal College of Radiologists document titled‘‘Referral Guidelines for Imaging’’ also providea detailed overview of clinical indications for imag-ing examinations, including CT.35 A CT examina-tion should thus be performed only when theradiation dose is deemed to be justified by thepotential clinical benefit to the patient.
DOSE REDUCTION STRATEGIESGeneral Strategies
All dose-reduction strategies are predicated onthe assumption that the CT scanner’s radiationdose levels and image quality fall within manufac-turer specifications and other general quality crite-ria. This goal can be accomplished througha quality control program that is designed andoverseen by a qualified medical physicist.
Fixed tube current (technique charts)Unlike traditional radiographic imaging, a CT im-age never looks ‘‘over-exposed’’ in the sense ofbeing too dark or too light; the normalized natureof CT data (ie, CT numbers represent a fixedamount of attenuation relative to water) ensuresthat the image always appears properly exposed.As a consequence, CT users are not technicallycompelled to decrease the tube current-timeproduct (mAs) for small patients, which may resultin excess radiation dose for these patients. It is,however, a fundamental responsibility of the CToperator to take patient size into account when se-lecting the parameters that affect radiation dose,the most basic of which is the mAs.27,30
As with radiographic and fluoroscopic imaging,the operator should be provided with appropriateguidelines for mAs selection as a function of pa-tient size. These are often referred to as techniquecharts. In CT, the tube current, exposure time andtube potential can all be altered to give the appro-priate exposure to the patient. Users most com-monly standardize the tube potential (kV) andgantry rotation time (s) for a given clinical applica-tion. The fastest rotation time should typically beused to minimize motion blurring and artifact,and the lowest kV consistent with the patientsize should be selected to maximize image con-trast.36–41 Tube current is thus the primary param-eter that is adapted to patient size.
Numerous investigators have shown that themanner in which mA should be adjusted as a func-tion of patient size should be related to the overallattenuation, or thickness, of the anatomy of inter-est as opposed to patient weight, which is corre-lated to patient girth, but not a perfect surrogatefor attenuation as a function of anatomic re-gion.25,26,42 The exception is for imaging of thehead, wherein attenuation is relatively well definedby age because the primary attenuation comesfrom the skull and the process of bone formationin the skull is age dependent.
Clinical evaluations of mA-adjusted imageshave demonstrated that radiologists do not findthe same noise level acceptable in small patientsas in larger patients.25 Because of the absenceof adipose tissue between organs and tissueplanes, and the smaller anatomic dimensions, ra-diologists tend to demand lower noise images inchildren and small adults relative to larger pa-tients.25,26,42,43 For CT imaging of the head, themAs reduction from an adult to a newborn of ap-proximately a factor of 2 to 2.5 is appropriate.For CT imaging of the body, typically a reductionin mAs of a factor of 4 to 5 from adult techniquesis acceptable in infants,42 whereas for obese pa-tients, an increase of a factor of 2 is appropriate.42
To achieve sufficient exposure levels for obese pa-tients, either the rotation time or the tube potentialmay also need to be increased.
Tube current (mA) modulationExtremely large variations in patient absorptionoccur with projection angle and anatomic regionand are not considered when using a fixed tubecurrent (Fig. 1). The projection with the most noiseprimarily determines the noise of the final image.Data acquired through body parts having less at-tenuation can thus be acquired with substantiallyless radiation without negatively affecting the finalimage noise.38,44–47 In addition, it is also possibleto reduce dose for projections of limited interest.
Fig. 1. Graph (top) of relative attenuation values asa function of table position and associated body region(bottom) shows almost three orders of magnitude ofvariation in attenuation, according to body regionand projection angle. (From McCollough CH, Bruesewitz MR, Kofler JM Jr. CT dose reduction and dosemanagement tools: overview of available options.Radiographics 2006;26(2):504; with permission.)
Fig. 2. Graph of relative tube current superimposed ona CT projection radiograph illustrates the concept oflongitudinal dose modulation. The prescribed tube cur-rent curve is determined by using attenuation datafrom the CT projection radiograph and a manufac-turer-specific algorithm (From McCollough CH,Bruesewitz MR, Kofler JM Jr. CT dose reduction anddose management tools: overview of availableoptions. Radiographics 2006;26(2):506; withpermission).
Strategies for Reducing Radiation Dose in CT 31
For example, in cardiac CT, decreasing mA duringsystole can significantly reduce dose to the pa-tient. Tube current modulation may occur angu-larly about the patient, along the long axis of thepatient, or incorporate both to adapt to attenuationdifferences within the patient.
Angular (x,y) mAmodulation Angular (x,y) mA mod-ulation addresses the variation in x-ray attenuationaround the patient by varying the mA as the x-raytube rotates about the patient (eg, in the anteropos-terior [AP] versus lateral direction). The operatorchooses the initial mA value, and the mA is modu-lated upward or downward from the initial valuewith a period of one gantry rotation. As the x-raytube rotates between the AP and lateral positions,the mA can be varied according to the attenuationinformation from the CT radiograph (ie, scout im-age) or in near–real time according to the measuredattenuation from the 180� previous projection.
Longitudinal (z) mA modulation Longitudinal (z) mAmodulation addresses the varying attenuation ofthe patient among anatomic regions by varyingthe mA along the z axis of the patient (eg, shouldersversus the abdomen), as shown in Fig. 2. The oper-ator must therefore provide as input to the algo-rithm the desired level of image quality, theparadigms for which are at present relatively man-ufacturer-specific.
Angular and longitudinal (x,y,z) mA modulationAngular and longitudinal (x,y,z) mA modulationcombines the previous two methods to vary themA during rotation and along the z axis of the pa-tient. The operator must still indicate the desiredlevel of image quality by one of the followingmethods. This results in the most comprehensiveapproach to CT dose reduction because thex-ray dose is adjusted according to the patientattenuation in all three dimensions.
Automatic exposure controlOverview It is technologically possible for CT sys-tems to adjust the x-ray tube current in real time inresponse to variations in x-ray intensity at the de-tector,36,47–49 much as fluoroscopic x-ray systemsadjust exposure automatically. The modulationmay be fully preprogrammed, occur in near–realtime by using a feedback mechanism, or incorpo-rate preprogramming and a feedback loop. Thesemethods of adapting the tube current to patient at-tenuation, known generically as AEC, are analo-gous to photo timing in general radiography andhave demonstrated reductions in dose of about20% to 40% when image quality is appropriatelyspecified. An exception to this trend occurs withobese patients. In large patients, the radiationdose is increased to ensure adequate image qual-ity. Much of the additional x-ray dose is absorbed
McCollough et al32
by excess adipose tissue. Thus, doses to internalorgans do not increase linearly with increases intube current settings.50,51 AEC is a broad termthat encompasses not only tube current modula-tion (to adapt to changes in patient attenuation)but also determining and delivering the rightdose for any patient (infant to obese) to achievethe diagnostic task.
Image quality selection paradigms for automaticexposure control systems Each manufacturer ofCT systems uses a different method of definingthe image quality in the user interface. GE usesa concept known as the Noise Index. The noise in-dex is referenced to the standard deviation of pixelvalues in a specific size water phantom and iscompared with patient attenuation measuredfrom the CT radiograph (scout) to maintain imagenoise. Toshiba allows two ways to prescribe imagequality in their Sure Exposure AEC algorithm:Standard Deviation and Image Quality Level. LikeGE’s Noise Index, Sure Exposure also comparesthe patient’s CT radiograph (Scanogram) data tothe standard deviation of a specific-attenuationwater phantom. Philips uses a Reference Imagefrom a satisfactory patient examination (ReferenceCase) stored in the system with which image qual-ity for future examinations is to be matched. Sie-mens uses a Quality Reference mAs to define theeffective mAs (5 mAs/pitch) required to producea specific image quality in an 80-kg patient(20 kg for pediatric cases) for a given protocol.For specific patients, the tube current is basedon the CT radiograph (Topogram) and fine-tunedby an online feedback system.
Future dose-reduction strategiesAdjusting kV based on patient size There have beenseveral physics and clinical studies on the use oflower tube potential (kV) in CT imaging to improveimage quality or reduce radiation dose. The princi-ple behind the benefit of lower kV in some clinicalapplications is this: The attenuation coefficient ofiodine increases as photon energy decreasestoward the k-edge energy of 33 kV. In many CT ex-aminations involving the use of iodinated contrastmedia, the superior enhancement of iodine atlower tube potentials improves the conspicuity ofhypervascular or hypovascular pathologies. Theimages obtained using lower tube potentials tendto be much noisier, however, mainly because ofthe higher absorption of low-energy photons bythe patient. A tradeoff between image noise andcontrast enhancement must therefore be made.
Fig. 3A shows the change of iodine CT numberwith tube potential for three different phantomsizes. The CT numbers of the iodine solution are
larger at lower kVs than at higher kVs. With an in-crease in phantom size, CT numbers decrease be-cause of beam hardening effects. Fig. 3B shows,for the same total radiation dose, image noise asthe tube potential and phantom size change. Forsmaller phantom sizes, images acquired at differ-ent kVs have almost the same noise levels, withslightly higher noise for images acquired at 80 kV.With the increase of the phantom size, however,lower-kV scans yield images with higher noiselevels compared with images attained at higherkVs. Fig. 3C combines the information fromFig. 3A and B to determine the change of con-trast-to-noise ratio (CNR) as a function of tube po-tential and phantom size, where contrast is definedas the CT number of the iodine solution minus theCT number of background, which in this case waswater (CT number z0 HU). The benefit of in-creased contrast enhancement at 80 kV is negatedby increased noise for the large phantom size.
When the patient size is below some threshold,the use of a lower tube potential can generate bet-ter image quality than the higher tube potential, forthe same radiation dose. Alternatively, dose canbe reduced while maintaining the same imagequality as a high tube potential image. Conse-quently, for a given patient size, an optimal tubepotential exists that yields the best image quality(in, eg, CNR, lesion detectability) or the lowest ra-diation dose. This optimal tube potential is highlydependent on the patient size and the specific di-agnostic task. For noncontrast CT examinations,the benefit of lower kV has not been establishedbecause soft tissue contrast is not highly depen-dent on the tube potential.
Iterative reconstruction Iterative reconstructiontechniques have demonstrated the potential forimproving image quality and reducing radiationdose in CT52–56 relative to the currently usedfiltered back projection techniques. The most no-ticeable benefit of iterative reconstruction is thatit is able to incorporate into the reconstruction pro-cess a physical model of the CT system that canaccurately characterize the data acquisition pro-cess, including noise, beam hardening, scatter,and so forth. This ability allows for dramatic im-provements in image quality, especially in thecase of low-dose CT scans, in which the propaga-tion of non-ideal data during the image reconstruc-tion becomes more significant than in routine CTscanning. This benefit has long been used bynuclear medicine imaging, where the photon num-bers are much smaller than in CT.
Iterative reconstruction is also superior to fil-tered back projection in handling insufficientdata. Recent advances in iterative reconstruction
Fig. 3. (A) Graph of the CT number of a 2% iodine solution for small, medium, and large phantoms at variousx-ray tube potentials. (B) Graph of noise (standard deviation of CT numbers within the water background) inimages of small, medium, and large phantoms at different tube potentials. (C) Graph of the contrast-to-noiseratio (CT number of iodine solution divided by the background noise level) in small, medium, and large phantomsat different tube potentials.
Strategies for Reducing Radiation Dose in CT 33
allow a significant reduction in the number of re-quired projection views, while still producingacceptable image quality. The use of iterativereconstruction techniques thus has the potentialto substantially reduce the radiation dose inCT.57–59 With computational power growingquickly, the clinical implementation of iterativereconstruction algorithms is within reach.56
Patient-Specific Dose Reduction Strategies
Pregnant patientsImaging the pregnant patient presents a uniquechallenge to the radiologist because of theconcern of radiation risk for the conceptus(embryo/fetus). Potential effects of radiation onthe conceptus include prenatal death, intrauterinegrowth restriction, small head size, severe mentalretardation, reduced intelligence quotient, organmalformation, and childhood cancer. The proba-bility of any effects depends on the radiationdose to the conceptus (Table 2).15,17,60–67
Common indications for CT scanning in a preg-nant patient include suspected appendicitis, pul-monary embolism, and urinary tract calculi. Tominimize radiation exposure to the fetus, it is
important to determine if the necessary diagnosticinformation can be obtained from an alternativenon–radiation-based imaging modality. For nona-cute symptoms, radiologists and physicians mustalso decide if immediate CT scanning is requiredor if CT scanning can be postponed until afterthe delivery.
For scanning body regions outside the abdomenand pelvis, such as chest CT for suspected pulmo-nary embolism, the dose to the fetus is extremelylow (<0.1 mGy) because the scattered radiationlevels fall off quickly away from the scan volume.For CT in a pregnant patient who has suspectedappendicitis, the scan volume should be restrictedto the necessary anatomy, and dual-pass (withand without contrast) studies should be avoided,if possible.68,69 In CT for renal calculi in a pregnantpatient, fetal dose can be reduced with use of lowmAs, high pitch, and a limited scan range withoutsubstantially compromising the study quality.70
For abdominal-pelvic CT, which directly irradiatesthe fetus, scan parameters (such as wider beamcollimation, higher pitch, and lower mAs, kV, andscan range) can be selected to reduce the fetaldose to approximately 23 mGy per scan phase.Even for a routine dose level, biphase CT
Table 2Probability of birthing healthy childrenwho have nomalformation or subsequent childhood cancerdevelopment for various radiation exposures during pregnancy
Dose to Conceptus(mGy)
Child has noMalformation (%)
Childwill notDevelop Cancer (%)
Childwill not DevelopCancer or haveaMalformation (%)
0 96 99.93 95.93
0.5 95.999 99.926 95.928
1.0 95.998 99.921 95.922
2.5 95.995 99.908 95.91
5.0 95.99 99.89 95.88
10.0 95.98 99.84 95.83
50.0 95.90 99.51 95.43
100.0 95.80 99.07 94.91
Data from Wagner LK, Lester RG, Saldana LR. Exposure of the pregnant patient to diagnostic radiations: a guide to med-ical management. Madison (WI): Medical Physics Publishing; 1997; with permission; and Wagner LK, Hayman LA. Preg-nancy and women radiologists. Radiology 1982;145(2):559–62.
McCollough et al34
examination of the abdomen and pelvis, the prob-ability of birthing a healthy baby decreases by only0.5% (see Table 2).
Pediatric patientsThe risk for cancer in children caused by radiationexposure is about two to three times higher thanadults because pediatric patients have a longer lifeexpectancy and their organs are more sensitive to ra-diationdamage.4,60–64 For newborns, the risk for can-cer induction is essentially the same as in the secondand third trimesters of pregnancy (see Table 2).
The best way to reduce the radiation dose to pe-diatric patients is to avoid unnecessary CT exam-inations and to look for alternative diagnosticimaging modalities with less or no exposure to ion-izing radiation. Pediatric protocols with scanningparameters specifically designed for childrenmust be used.71,72 These protocols usually includetube current modulation,39,47 a child-size bowtiefilter and scanning field of view (FOV), or a weight-or size-based technique chart that can determinethe appropriate kV or mAs for each patient.73,74
Automatic tube current modulation and manualtechnique charts are currently widely used. Addi-tionally, lower kV values may be used, dependingon patient size and clinical indication. For pediatricpatients, because of less attenuation in the body,the noise level does not increase significantlywith the decrease of kV for the same radiationdose. For iodine contrast-enhanced examinations,therefore, a lower kV can be used to improve thecontrast enhancement without increasing thenoise. Because of this, the benefit of image qualityimprovement or dose reduction is much moresignificant than in adult patients. Lower-kV
techniques have been actively investigated andare beginning to be widely used in pediatricCT.26,41,73,74
There are several factors that should be takeninto account when lower-kV techniques are usedin practice. First, because of the less efficientx-ray production of the tube at low kV values, themAs has to be increased to avoid excessive noiselevels. Second, for certain sized patients, a lowerkV may not be appropriate. To address these is-sues, a weight or size-based kV/mAs techniquechart should be used. Third, to avoid motion arti-facts and decrease scan times in pediatric pa-tients, a fast rotation time and a high helical pitchare desirable, which often limit the maximummAs that can be used because of tube current lim-itations. In this situation, a higher kV may be nec-essary to avoid compromising the examinationquality. Finally, although lower kV increases thecontrast of iodine, it may not increase the contrastof tissues, lesions, and other pathologic structureswithout iodine uptake. The use of lower kV has tobe carefully evaluated by radiologists and physi-cists for every particular type of pediatricexamination.
Other specific dose-reduction strategiesCardiac CT Dose in cardiac CT is a considerablymore complex issue compared with noncardiacCT applications. There are two major reasons forthis complexity. The first reason is that dose,noise, and pitch have different relationships in car-diac CT compared with noncardiac spiral CT.75
The second reason is that dose in cardiac CTcan depend on the patient’s heart rate (HR).
Strategies for Reducing Radiation Dose in CT 35
In noncardiac multidetector CT (MDCT), noisedepends on pitch. In cardiac spiral CT, however,noise is independent of pitch and depends onlyon the tube current-time product (mAs).75 Addi-tionally, because only a partial amount of the pro-jection data from one gantry rotation is used forimage reconstruction (to optimize the temporalresolution), relatively high mAs values are neededto provide an acceptable noise level for cardiacCT imaging, especially for cardiac CT angiography(CTA) examinations, which require the use of thinslices for better visualization of the coronaryarteries.
A combination of relatively high mAs values withthe low pitch values required in cardiac CT (dose isproportional to mAs/pitch) explains why cardiacCT examinations are associated with a higher radi-ation dose. Effective doses up to 21 mSv havebeen reported in the literature,76 and the recent re-view by Achenbach and colleagues77 claims thatmaximum organ dose values (based on MonteCarlo calculations) delivered during cardiac CT ex-aminations can be as high as 50 to 100 mGv whenno dose reduction measures are taken.
ECG-based tube current modulation is an im-portant dose-reduction tool in cardiac CT.38,78
The principle of ECG-based mA modulation is il-lustrated by Fig. 4. The percent dose reduction us-ing ECG-based tube current modulation is higherfor patients who have slow HRs compared withthose who have high HRs, because the maximummA time period is a smaller percentage of the timeinterval between successive R waves in the ECG(R-R interval) at low HR.
Additionally, the width of the maximum mA win-dow must be carefully chosen to make sure thatthe data for the best cardiac phase (the one withthe least motion artifact) is acquired with maxi-mum tube current (hence, best image quality). Fail-ure to properly set up the ECG-based tube currentmodulation parameters may compromise the
diagnostic quality of the cardiac examination. Ac-cording to the dual-source coronary CTA study byWeustink and colleagues,79 optimal windows forECG-based tube current modulation for low(HR%65 bpm), intermediate (65<HR<80 bpm),and high (HRR80 bpm) HRs were at 60% to76%, 30% to 77%, and 31% to 47% of the R-R in-terval, respectively.
Finally, for patients who have a very irregularHR, using ECG-based tube current modulationcan compromise the diagnostic quality of the im-ages if the optimal reconstruction window occursduring the reduced tube current interval. Somesystems address this concern by automatically in-creasing the tube current to the maximum levelwhen a statistical trend-analysis algorithm recog-nizes an R-R interval that is significantly differentfrom the previous rhythm.80
One of the earliest studies of ECG-based tubecurrent modulation reported dose savings of30% to 50% for 4-slice CT.78 A later study byHausleiter and colleagues81 reported effectivedose estimates for coronary CTA using 64-sliceCT of approximately 9.4 mSv with ECG-basedtube current modulation and 14.8 mSv withouttube current modulation.81 One study of dual-source coronary CTA reported mean effectivedose values of 7.8 to 8.8 mSv,82 and another re-ported effective doses of 6.8, 13.4, and 4.2 mSvfor low, intermediate, and high HR, respectively,when using the optimal windows for ECG-basedtube current modulation.79 The higher dose at in-termediate HR is attributable to the need fora wider maximum tube current window.
ECG modulation cannot be combined withsome other types of tube current modulation,such as angular modulation. Although it is possibleto use z-modulation in cardiac CT, this is notwidely available. The role of AEC in cardiacmode is typically limited to automatically adjustingthe mAs values based on patient size.
Fig. 4. ECG-based modulation ofthe tube current. The widthof the temporal window havingthe maximum tube current(Max mA) can be selected by theuser, whereas the temporalwidth of the image reconstruc-tion window is fixed (Recon).For fullquality images, the recon-struction window (darker graytime interval) should fall withinthe maximum mA window (ligh-ter gray time interval). (Courtesyof Suhny Abbara, MD, Boston,MA).
McCollough et al36
Alternatively, technique charts may be used tomanually choose the proper mAs settings basedon patient size, the type of cardiac examination(eg, coronary calcium scoring versus coronaryCTA), and the noise level accepted in the clinicalpractice. Appropriate use of lower kV values(80 or 100 kV) for coronary CTA examinations ofsmaller patients can further reduce radiationdose without compromising the image quality. Arecent study by Leschka and colleagues83 showedthat dual-source coronary CTA with 100 kV is fea-sible in patients of normal weight and can producea higher contrast-to-noise ratio at estimated effec-tive doses as low as 4.4 mSv. In a study by Achen-bach and colleagues,77 a 74-year-old patient whohad 63 kg body weight was scanned using 80-kVtube current and ECG-based tube current modula-tion, resulting in fully diagnostic image quality at anestimated effective dose of 3.0 mSv.
Additional dose reductions can be achieved us-ing prospective ECG-triggering (ie, ‘‘step andshoot’’ acquisitions). The lack of overlappingbeams at low spiral pitch values makes this modedose efficient. This mode is not as reliable as theconventional spiral, retrospectively-gated mode,however, and requires a careful selection of pa-tients having stable, low heart rates. Because thereare no data collected outside the narrow acquisi-tion window predicted to correspond to the bestphase of the cardiac cycle, any sudden change incardiac rhythm (eg, ectopic beat) can ruin imagequality for the portion of the heart included in thataxial scan. Nevertheless, a recent study by Scheffeland colleagues84 concluded that prospectivelytriggered dual source CT coronary angiography al-lows for the accurate diagnosis of significant coro-nary stenoses at a low radiation dose in patientswho have a regular HR. For 120 selected patientshaving an average heart rate of 59 � 6 bpm (range44–69 bpm), the reported mean effective dose was2.5 � 0.8 mSv (range 1.2–4.4 mSv).84
The susceptibility of prospectively triggered cor-onary CTA to artifacts caused by variations in car-diac cycle length or by the occurrence of ectopicbeats can be reduced by using adaptively trig-gered sequential scans with dynamic temporalwindows, where triggering for each cardiac cycleis adjusted according to the ECG trend and vari-ability. In a study in which cardiac CT scans weresimulated on the basis of 60 ECGs recorded dur-ing actual coronary CTA examinations, the adap-tively triggered sequential scans provided 68%dose reduction relative to spiral cardiac CT withconstant tube current, without compromising thereliability of image quality.85 The standard sequen-tial mode without adaptive triggering provided im-proved dose reductions (75%) but suffered from
inconsistent availability of the optimal cardiacphase, missing it in 18% of the cases.
Selective in-plane shielding Selective shielding of ra-diation-sensitive tissues and organs during CT scan-ning has been proposed and products to implementthis are commercially available. Their use is not gen-erally recommended, however, because the dosereduction they provide can be readily achieved bydecreasing x-ray tube current, which does not intro-duce noise or increase beam-hardening artifacts.
Shields made of thin sheets of flexible latex im-pregnated with bismuth and shaped to cover theeye lens, thyroid, or breasts can be used, respec-tively, during brain, cervical spine,orchest CT exam-inations. Dose savings to the superficially locatedtarget organ when using such shields have been re-ported to be 40% to 67% for adults86–90 and 30% to40% for children.91,92 Most of these studies reportedartifacts near the shields. Additionally, these studiesoverestimated organ dose reductions by assumingthat organ doses are equivalent to the measuredskin dose reductions.
A quantitative study by Geleijns andcolleagues93 assessed the tradeoff betweenabsorbed dose and image quality for the use ofselective shields. Using commercially availableshields and an anthropomorphic phantom, imagenoise was experimentally quantified by the stan-dard deviation of CT numbers in the target organ.Absorbed organ doses were also computed usinga validated Monte Carlo method. With use of theshields, organ dose reductions of 26%, 27%,and 30% were found for the thyroid, eye lens,and breast, respectively, contrary to the largersavings reported elsewhere. For each organ,dose reduction was accompanied by increasednoise and artifacts. This increase was mostmarked and most varied near the breast, wherethe noise increase ranged from a minimum of50% (8 HU increased to 12 HU) near the shieldcenter to a maximum of 100% (7.5 HU increasedto 15 HU) near the shield edge. They additionallyreport that the same 30% dose reduction couldbe achieved by decreasing the x-ray tube currentby 30%, yet with a smaller and more uniform noiseincrease of 20% to 30%; this is because while thedose shield attenuates the anterior x-ray beamand hence decreases anterior organ dose, it alsoattenuates x-rays coming from the posterior direc-tion that have already contributed to organ doseand contain important image information.
SUMMARY
In recent years, the media has focused on the po-tential danger of radiation exposure from CT, even
Strategies for Reducing Radiation Dose in CT 37
though the potential benefit of a medicallyindicated CT far outweighs the potential risks.This attention has reminded the radiology commu-nity that doses must be ALARA while maintainingdiagnostic image quality. To satisfy the ALARAprinciple, the dose reduction strategies describedin this article must be well understood and prop-erly used. The use of CT must also be justifiedfor the specific diagnostic task.
ACKNOWLEDGMENTS
The authors thank Kris Nunez and MeganJacobsen for their help with the manuscript prep-aration and submission.
REFERENCES
1. Brenner DJ, Hall EJ. Computed tomography—an
increasing source of radiation exposure. N Engl J
Med 2007;357(22):2277–84.
2. National Council on Radiation Protection and Mea-
surements. Ionizing radiation exposure of the popu-
lation of the United States. Report no. 93. Bethesda
(MD): National Council on Radiation Protection and
Measurements; 1987. NCRP Report No. 93.
3. Donnelly LF, Emery KH, Brody AS, et al. Minimizing
radiation dose for pediatric body applications of
single-detector helical CT: strategies at a large chil-
dren’s hospital. AJR Am J Roentgenol 2001;176:
303–6.
4. Brenner DJ, Elliston CD, Hall EJ, et al. Estimated
risks of radiation-induced fatal cancer from pediatric
CT. AJR Am J Roentgenol 2001;176:289–96.
5. The Associated Press. Report links increased can-
cer risk to CTscans. The New York Times. November
29, 2007.
6. Khamsi R. CT scan radiation can equal nuclear
bomb exposure. Available: NewScientist.com.
News Service. Accessed May 11, 2007.
7. Jessen KA, Shrimpton PC, Geleijns J, et al. Dosime-
try for optimisation of patient protection in computed
tomography. Appl Radiat Isot 1999;50:165–72.
8. Nagel HD. Radiation exposure in computed tomog-
raphy. 2nd edition. Frankfurt: COCIR; 2000.
9. Shope TB, Gagne RM, Johnson GC. A method for
describing the doses delivered by transmission
x-ray computed tomography. Med Phys 1981;8(4):
488–95.
10. American Association of Physicists in Medicine.
Standardized methods for measuring diagnostic
x-ray exposures. New York: AAPM; 1990. Report
no. 31.
11. McNitt-Gray MF. AAPM/RSNA physics tutorial for
residents: topics in CT. Radiation dose in CT. Radio-
graphics 2002;22(6):1541–53.
12. American Association of Physicists in Medicine. The
measurement, reporting and management of radia-
tion dose in CT 2008. College Park (MD) Report
No 96.
13. International Electrotechnical Commission. Medical
electrical equipment. Part 2-44: particular require-
ments for the safety of x-ray equipment for
computed tomography. IEC publication No. 60601-
2-44. Ed. 2.1. Geneva, Switzerland: International
Electrotechnical Commission (IEC) Central Office;
2002.
14. Boone JM. The trouble with CTDI100. Med Phys
2007;34(4):1364–71.
15. International Commission on Radiological Protec-
tion. 1990 recommendations of the international
commission on radiological protection (Report 60).
Ann ICRP 1991;21(1–3).
16. McCollough CH, Schueler BA. Calculation of effec-
tive dose. Med Phys 2000;27(5):828–37.
17. International Commission on Radiological Protec-
tion. Recommendations of the international commis-
sion on radiological protection (ICRP #26). Oxford:
The International Commission on Radiological
Protection; 1977. ICRP Publication 26.
18. International Commission on Radiological Protec-
tion. 2007 recommendations of the international
commission on radiological protection (ICRP Publi-
cation 103). Ann ICRP 2007;37:1–332.
19. Cristy M. Mathematical phantoms representing chil-
dren of various ages for use in estimates of internal
dose. Oak Ridge (TN): Oak Ridge National Labora-
tory; 1980. NUREG/CR-1159, ORNL/NUREG/TM-367.
20. Bongartz G, Golding S, Jurik A, et al. European
guidelines for multislice computed tomography.
Funded by the European Commission. Contract
number FIGM-CT2000-20078-CT-TIP. March 2004.
21. Shrimpton PC, Hillier MC, Lewis MA, et al. National
survey of doses from CT in the UK: 2003. Br J Radiol
2006;79(948):968–80.
22. McCollough CH. Patient dose in cardiac computed
tomography. Herz 2003;28(1):1–6.
23. Haaga JR. Radiation dose management: weighing
risk versus benefit. AJR Am J Roentgenol 2001;
177(2):289–91.
24. Nickoloff EL, Alderson PO. Radiation exposures to
patients from CT: reality, public perception, and
policy. AJR Am J Roentgenol 2001;177(2):285–7.
25. Wilting JE, Zwartkruis A, van Leeuwen MS, et al.
