This copy is for personal use only. To order printed copies, contact reprints@rsna.

org
n Medical Physics

U.S. Diagnostic Reference Levels

and Achievable Doses for 10
Adult CT Examinations1
Original Research

Kalpana M. Kanal, PhD

Purpose: To develop diagnostic reference levels (DRLs) and achiev-
Priscilla F. Butler, MS
able doses (ADs) for the 10 most common adult computed
Debapriya Sengupta, MBBS, MPH
tomographic (CT) examinations in the United States as
Mythreyi Bhargavan-Chatfield, PhD a function of patient size by using the CT Dose Index
Laura P. Coombs, PhD Registry.
Richard L. Morin, PhD
Materials and Data from the 10 most commonly performed adult CT
Methods: head, neck, and body examinations from 583 facilities
were analyzed. For head examinations, the lateral thick-
ness was used as an indicator of patient size; for neck and
body examinations, water-equivalent diameter was used.
Data from 1 310 727 examinations (analyzed by using SAS
9.3) provided median values, as well as means and 25th
and 75th (DRL) percentiles for volume CT dose index
(CTDIvol), dose-length product (DLP), and size-specific
dose estimate (SSDE). Applicable results were compared
with DRLs from eight countries.

Results: More than 46% of the facilities were community hospitals;

13% were academic facilities. More than 48% were in
metropolitan areas, 39% were suburban, and 13% were
rural. More than 50% of the facilities performed fewer
than 500 examinations per month. The abdomen and
pelvis was the most frequently performed examination in
the study (45%). For body examinations, DRLs (75th per-
centile) and ADs (median) for CTDIvol, SSDE, and DLP
increased consistently with the patient’s size (water-equiv-
alent diameter). The relationships between patient size
and DRLs and ADs were not as strong for head and neck
examinations. These results agree well with the data from
other countries.

Conclusion: DRLs and ADs as a function of patient size were devel-

oped for the 10 most common adult CT examinations per-
formed in the United States.

q
RSNA, 2017

1
From the Department of Radiology, University of Wash-
ington, Seattle, Wash (K.M.K.); Departments of Quality and
Safety (P.F.B., M.B.) and National Radiology Data Registries
(D.S., L.P.C.), American College of Radiology, 1891 Preston
White Dr, Reston, VA 20191; and Department of Radiology,
Mayo Clinic Florida, Jacksonville, Fla (R.L.M.). Received
September 23, 2016; revision requested October 26; re-
vision received November 21; accepted December 6; final
version accepted December 7. Address correspondence
to P.F.B. (e-mail: [email protected]).

q
RSNA, 2017

120 radiology.rsna.org n Radiology: Volume 284: Number 1—July 2017

MEDICAL PHYSICS: U.S. Diagnostic Reference Levels and Achievable Doses for 10 Adult CT Examinations Kanal et al

C
omputed tomography (CT) is crit- utilization and an increase in population relative to the size of the phantom
ical for screening, diagnosis, ther- exposure to ionizing radiation. used to report CTDIvol, the actual dose
apy, and the management of pa- Americans were exposed to more to the patient may be considerably dif-
tient care. In emergency departments than seven times as much ionizing radi- ferent (4,5). CTDIvol is primarily useful
alone, CT significantly impacts leading ation from medical procedures in 2006 as a quality assurance tool to compare
diagnosis, diagnostic confidence, and than in the early 1980s. Although CT doses from different protocols and to
admission decisions (1). However, scans represented only 12% of imaging compare scanner outputs from differ-
with these benefits come increased procedures, they contributed almost ent manufacturers.
50% of the total radiation dose to the More recently, the American As-
U.S. population from medical imaging sociation of Physicists in Medicine
Advances in Knowledge (2). This increase in population dose is (AAPM) developed a new CT param-
nn National diagnostic reference of concern because of the potential for eter, the size-specific dose estimate
levels (DRLs) and achievable radiation-induced malignancies. (SSDE), to more accurately estimate
doses (ADs) as a function of Until recently, a national collection dose at the center of the scanned re-
patient size are provided for the of patient-based dose estimates was gion of an individual patient by factor-
10 most common adult CT exam- not available in the United States. The ing in the patient’s size (6). SSDE is
inations (head and brain without Nationwide Evaluation of X-Ray Trends determined by applying a conversion
contrast material; neck with con- (3) program surveyed a representative factor based on linear dimensions of
trast material; cervical spine sample of U.S. CT facilities, but report- the transverse cross section of the pa-
without contrast material; chest ed only radiation exposure to a phan- tient to the CTDIvol. Although SSDE is
without contrast material; chest tom, not radiation exposure estimated not yet automatically reported by CT
with contrast material; chest and from individual patient scans. manufacturers, discussions are un-
pulmonary arteries with contrast Doses are routinely estimated by derway by the medical profession and
material; abdomen and pelvis using standard 16- or 32-cm diameter manufacturers to automatically acquire
without contrast material; ab- polymethylmethacrylate cylinder phan- the patient dimensions, apply them to
domen and pelvis with contrast toms representing “average” patients. the CTDIvol, and report SSDE for each
material; abdomen and pelvis for For CT, this parameter, the volume patient.
nephrolithiasis without contrast CT dose index (CTDIvol), approximates Diagnostic reference levels (DRLs)
material; and chest, abdomen, the average dose to a cross section of are benchmarks for radiation protection
and pelvis with contrast mate- the phantom (4). Dose-length product
rial) by using 2014 data. (DLP) is based on CTDIvol factors in
nn For the most common examina- the length of the scan. Presently, CT- https://doi.org/10.1148/radiol.2017161911
tion, abdomen and pelvis with DIvol and/or DLP are displayed on CT
units for each scan. Although these Content codes:
contrast material (25.8% of all
examinations in this study), the parameters are tagged to individual Radiology 2017; 284:120–133
DRLs for patients with water- examinations, they do not represent
Abbreviations:
equivalent diameters between 29 the patient’s dose but rather the dose
AAPM = American Association of Physicists in Medicine
and 33 cm were 15 mGy (volume to one of the standard phantoms. ACR = American College of Radiology
CT dose index [CTDIvol]), 18 Depending on the size of the patient AD = achievable dose
mGy (size-specific dose esti- CTDIvol = volume CT dose index
DIR = Dose Index Registry
mate), and 755 mGy-cm (dose- Implications for Patient Care
DLP = dose-length product
length product [DLP]). nn The results of this study will DRL = diagnostic reference level
nn For head and brain without con- enable facilities to compare their ICRP = International Commission on Radiological Protection
trast material examinations patient doses with national SSDE = size-specific dose estimate
(17.1% of all examinations in benchmarks. Author contributions:
this study), the DRLs for patients nn Because smaller patients require Guarantors of integrity of entire study, K.M.K., P.F.B., D.S.,
with lateral head thicknesses lower doses than larger ones to R.L.M.; study concepts/study design or data acquisition
between 14 and 16 cm were 56 or data analysis/interpretation, all authors; manuscript
yield adequate image quality, the
drafting or manuscript revision for important intellectual
mGy (CTDIvol) and 962 mGy-cm new size-specific DRLs and ADs content, all authors; manuscript final version approval,
(DLP). will enable facilities to more ef- all authors; agrees to ensure any questions related to the
nn The new DRLs show that exami- fectively optimize their CT proto- work are appropriately resolved, all authors; literature
nation exposures to the U.S. cols for the wide range of sizes research, P.F.B., D.S.; clinical studies, D.S.; statistical
adult population are generally of the patients they examine and analysis, P.F.B., D.S., M.B., L.P.C.; and manuscript editing,
all authors
not higher than those in other thus to appropriately reduce
countries. dose to patients. Conflicts of interest are listed at the end of this article.

Radiology: Volume 284: Number 1—July 2017 n radiology.rsna.org 121

MEDICAL PHYSICS: U.S. Diagnostic Reference Levels and Achievable Doses for 10 Adult CT Examinations Kanal et al

and optimization of patient imaging. Parameter for Diagnostic Reference study date), dose indexes, and study
They were first mentioned by the In- Levels and Achievable Doses in Med- descriptions were included. Examina-
ternational Commission on Radiological ical X-Ray Imaging (12) developed tions from facilities outside the United
Protection (ICRP) in 1990 (7) and were DRLs and ADs from data prior to States, multiscan examinations, and
clarified further in 1996 (8). The ICRP 2005 for only three adult examina- body examinations with missing water-
defines two key elements in medicine: tions (head, abdomen and pelvis, and equivalent diameters and head exami-
justification and optimization of radiol- chest). These DRLs and ADs are based nations with missing lateral thicknesses
ogy examinations. Justification implies on phantom data and apply only to a were excluded. Multiscan examinations
that the examination is indicated and the patient size that corresponds to the were identified as examinations with
patient’s benefit exceeds any potential size of the phantom. However, radia- and without contrast material studies
detriments. Optimization implies that tion dose must increase with patient or those in which more than one body
the radiation exposure is optimized for size (13) to maintain acceptable image part was included in the examination.
the clinical purpose of the examination. quality. Although some work develop- These were excluded to prevent overes-
An important optimization tool, ing size-specific DRLs for pediatric pa- timation of the radiation dose.
DRL is defined as an investigational tients has been done (14), guidance is Anteroposterior diameters and lat-
level that applies to an easily measured not available for small- and large-sized eral thicknesses were determined from
quantity using a standard phantom or adult patient populations. the localizer images (16) to determine
representative patient. It is intended The National Radiology Data Reg- patient size. For head examinations,
for use as a simple test for identifying istry (NRDR) is a data warehouse for only the lateral thickness was used as
situations where the levels of patient diagnostic imaging registries run by the the indicator of head size. Analysis
dose are unusually high (8). The ICRP ACR to collect examination data and of the anteroposterior diameters
emphasizes that DRLs “are not for results. The primary purpose of the showed an unexplained bimodal data
regulatory or commercial purposes, NRDR is to provide national and re- distribution, so they were considered
not a dose restraint and not linked to gional data to aid facilities in improving to be unreliable indicators of head size.
limits or constraints” (9). The use of patient care. The CT Dose Index Regis- The median lateral dimension (15 cm)
DRLs is endorsed by professional, ad- try (DIR) continuously collects, de-iden- was consistent with the mean published
visory, and regulatory organizations, tifies, and transmits dose indexes and by Huda et al (14.7 cm) (17).
including the ICRP, American College patient size information to the NRDR Water-equivalent diameter (18)
of Radiology (ACR), AAPM, United for storage and analysis (15), enabling was used for neck and body examina-
Kingdom Health Protection Agency, the development of benchmarks. tions and was calculated from the au-
International Atomic Energy Agency, The purpose of this study was to tomatically determined anteroposteri-
and European Commission. DRLs are use the power of the CT DIR to develop or diameter and lateral thickness (16),
typically set at the 75th percentile of DRLs and ADs for the 10 most common following the AAPM method. For body
the dose distribution from a survey adult CT examinations as a function of examinations, the water-equivalent
conducted across a broad user base patient size. diameter was used to determine the
(ie, large, small, public, private, hos- appropriate conversion factor to esti-
pital, and outpatient facilities) using a mate SSDE from CTDIvol normalized
specified dose-measurement protocol. Materials and Methods to a 32-cm phantom. SSDE conversion
They are established both regionally The 10 most common examinations in factors for head and neck examinations
and nationally, and considerable vari- the United States performed between are not available at this time from the
ations have been seen across both re- January and December 2014 in pa- AAPM (18) and will be integrated into
gions and countries (9,10). tients aged 19 years and older were the program at a later date.
The concept of achievable dose included in the study. These are head Results from the analysis were tab-
(AD) was introduced in 1999 by the and brain without contrast material; ulated alongside corresponding data
United Kingdom National Radiation neck with contrast material; cervical from other countries derived from the
Protection Board to further optimize spine without contrast material; chest literature. Descriptive comparisons
practice. In 2012, the National Council without contrast material; chest with were made; no statistical comparisons
on Radiation Protection and Measure- contrast material; chest and pulmonary were made because of the current vari-
ments, or NCRP, proposed that ADs be arteries with contrast material; abdo- ability in methods among countries and
set at the median (50th percentile) of men and pelvis without contrast mate- regions and inadequate data for statis-
a dose survey, on the basis that 50% rial; abdomen and pelvis with contrast tical comparisons.
of the facilities have already achieved material; abdomen and pelvis for neph-
doses at or below this value (11). rolithiasis without contrast material; Statistical Analysis
There are few current US recom- and chest, abdomen, and pelvis with A sensitivity analysis was performed
mendations for DRLs and ADs. For contrast material. Only examinations to determine whether examinations
example, the ACR-AAPM Practice with complete patient information (age, excluded from the study because of

122 radiology.rsna.org n Radiology: Volume 284: Number 1—July 2017

MEDICAL PHYSICS: U.S. Diagnostic Reference Levels and Achievable Doses for 10 Adult CT Examinations Kanal et al

Table 1 Table 2
Numbers of CT Examinations Included in Study Characteristics of Facilities and Examinations Included in the
Study
Body Part and Examination No. of Examinations Percentage
No. of
Head
No. of Facilities Examinations
CT of head and brain without contrast 223 908 17.1
Characteristic in DIR Percentage in Study Percentage
material
  Total 223 908 17.1 Facility category
Neck or cervical spine Academic 78 13.4 372 746 28.4
CT of neck with contrast material 33 740 2.6 Community hospital 271 46.5 794 839 60.6
CT of cervical spine without contrast 97 586 7.4 Multispecialty clinic 27 4.6 40 594 3.1
material Freestanding center 176 30.2 95 293 7.3
  Total 131 326 10.0 Children’s hospital 22 3.8 1278 0.1
Chest Other 9 1.5 5977 0.5
CT of chest without contrast material 159 909 12.2 Facility location
CT of chest with contrast material 111 898 8.5 Metropolitan 280 48.0 708 965 54.1
CT of chest pulmonary arteries with 58 986 4.5 Suburban 227 38.9 514 047 39.2
contrast material Rural 76 13.0 87 715 6.7
  Total 330 793 25.2 Census region
Abdomen and pelvis Northeast 168 28.8 555 612 42.4
CT of abdomen and pelvis without 201 754 15.4 Midwest 152 26.1 297 155 22.7
contrast material South 159 27.3 304 633 23.2
CT of abdomen and pelvis with contrast 338 056 25.8 West 104 17.8 153 327 11.7
material Trauma center level
CT of abdomen, pelvis, and kidney 47 748 3.6 I 87 14.9 351 883 26.9
without contrast material II 63 10.8 318 043 24.3
  Total 587 558 44.8 III 37 6.4 115 513 8.8
Chest, abdomen, and pelvis IV 12 2.1 18 007 1.4
CT of chest, abdomen, and pelvis with 37 142 2.8 Not a trauma center 384 65.9 507 281 38.7
contrast material Average no. of
  Total 37 142 2.8  examinations per
Grand total 1 310 727 month
0–500 330 56.6 181 656 13.9
501–1000 92 15.8 218 872 16.7
1001–2000 105 18.0 424 557 32.4
missing water-equivalent diameter or .2000 56 9.6 485 642 37.1
lateral thickness were inherently differ- Total in DIR 583 100 1 310 727 100
ent from those with non-missing values.
Distributions of lateral thickness
for head examinations and water-
equivalent diameter for body examina- Median values of CTDIvol, DLP, Table 3
tions were obtained by using univariate and SSDE were calculated for each
procedures. Head examinations were facility. Twenty-fifth percentiles, me- Demographic Distribution of Study
categorized into 2-cm lateral thick- dians, and 75th percentiles for these Population
ness bins because of the small range of median values were calculated for No. of Examinations
thicknesses. Neck and body examina- each of the examinations. Median Characteristic in Study Percentage
tions were categorized into 4-cm water- values were used for consistency with
Sex
equivalent–diameter bins because of proposed international recommen-
Female 726 485 55.4
the larger ranges of water-equivalent dations to enable comparison of our
Male 582 510 44.4
diameters. results with those of other countries Other/ 1732 0.1
Descriptive statistics were calcu- following these recommendations. unknown
lated for facility category, location, In early 2016, the ICRP published a Age group (y)
census region, and average volume draft of “Diagnostic Reference Levels 19–44 346 272 26.4
of examinations per month. One-way in Medical Imaging” for public com- 45–64 443 889 33.9
frequency tables were generated for ment. In that document, they say, 65 520 566 39.7
demographic distributions of the study “The Commission now recommends Total 1 310 727 100
population. that the median value (not the mean

Radiology: Volume 284: Number 1—July 2017 n radiology.rsna.org 123

MEDICAL PHYSICS: U.S. Diagnostic Reference Levels and Achievable Doses for 10 Adult CT Examinations Kanal et al

Table 4
Size-based ADs and DRLs for Head and Neck CT Examinations
CTDIvol (mGy) DLP (mGy-cm)
AD (50th DRL (75th AD (50th DRL (75th
Examination and Median Size (Thickness or Diameter) Size (cm) No. of Facilities No. of Patients Percentile) Percentile) Percentile) Percentile)

Head and brain without contrast material* 12–14 227 19 933 47 56 767 936
14–16 290 137 755 49 56 811 962
16–18 256 57 292 52 60 902 1020
18–20 160 5390 51 60 926 1069
All† 347† 223 908 49 57 849 1011
Neck with contrast material‡ 14–18 352 9458 14 18 377 509
18–22 350 8723 15 19 429 563
22–26 334 5717 15 19 423 560
26–30 307 5012 16 20 457 572
30–34 259 2655 17 23 494 656
All† 417† 33 740 15 20 431 572
Cervical spine without contrast material§ 13–17 350 22 739 18 24 362 495
17–21 375 36 711 20 28 421 562
21–25 346 18 600 21 28 438 575
25–29 326 11 640 22 29 450 609
29–33 265 5477 25 33 551 703
All† 434† 97 586 21 28 432 602

Note.—The AD is the 50th percentile of the distribution of median values (the 50th percentile) of all participating facilities; the DRL is the 75th percentile of the distribution of median values of all
participating facilities.
* Only lateral thickness (cm) was used. The median lateral thickness was 15 cm.
† “All” includes data beyond lowest- and highest-size bins; “No. of facilities” is the total number of facilities submitting data for any size patient.
‡ Water-equivalent diameter (cm) was used. The median diameter was 20 cm.
§ Water-equivalent diameter (cm) was used. The median diameter was 19 cm.

value) for the DRL quantity from each perspective and practical usefulness included in the study and the examina-
of the facilities in the survey should in mind. We considered collapsing the tions that were excluded from the study
be used. National DRLs should be set non–statistically significant bins into one (0%–0.13% difference in mean values of
as the 75th percentile of the median but realized that the resulting bins would CTDIvol and DLPs). We did not test for
values obtained in a sample of repre- be confusing and lose their usability. statistical significance because the dif-
sentative centers” (ICRP, unpublished All analyses were performed by us- ferences were too small to be clinically
document, 2016). ing SAS software, version 9.3, of the meaningful.
A multivariable mixed regression SAS System for Windows (2015, SAS Abdomen and pelvis was the most
analysis was performed to determine Institute, Chicago, Ill). common examination (45%), followed
whether dose indexes varied signifi- by chest (25%); head (17%); neck/cer-
cantly by water-equivalent diameter and vical spine (10%); and chest, abdomen,
lateral thickness. Facility was included Results and pelvis (2.8%). More than 46% of
as a random effect, and fixed effects The DIR collected 5 701 421 adult exami- participating facilities were community
included facility characteristics, age, nations between January and December hospitals; 13% were academic facilities;
and sex. An analysis was performed for 2014; 3 417 992 were in the top 10 most 280 facilities (48%) were in metropol-
multiple comparisons among size bins frequently performed examinations. Af- itan areas; 227 (39%) were in subur-
for each body part to determine if the ter the exclusion of multiphase examina- ban areas; and 76 (13%) were in rural
means of the dose indexes were signifi- tions, neck and body examinations with areas. Fewer than 500 examinations
cantly different from each other. missing water-equivalent diameters, and per month were performed at 56.6%
Size bins were constructed not by head examinations with missing lateral of facilities.
relying on statistical significance but by thicknesses, 1 310 727 examinations More than 55% of the examinations
using the distribution of the data—that were analyzed from 583 facilities (Tables were in female patients, and 60% were
is, the number of data points in each 1, 2). The sensitivity analysis indicated in patients between 19 and 64 years of
of the bins—and by keeping the clinical no difference between the examinations age (Table 3).

124 radiology.rsna.org n Radiology: Volume 284: Number 1—July 2017

MEDICAL PHYSICS: U.S. Diagnostic Reference Levels and Achievable Doses for 10 Adult CT Examinations Kanal et al

Figure 1

Figure 1: Graphs show head, neck, and cervical spine ADs and DRLs. (a) AD and DRL for head and brain without contrast material–CTDIvol. (b) AD and DRL for
head and brain without contrast material–DLP. (c) AD and DRL for neck with contrast material–CTDIvol. (d) AD and DRL for neck with contrast material–DLP. (e) AD
and DRL for cervical spine without contrast material–CTDIvol. (f) AD and DRL for cervical spine without contrast material–DLP.

Multivariate regression analysis significantly different from the next contrast examinations was 20 cm.
showed that water-equivalent diame- bin. There were 33 740 neck examinations,
ter and lateral thickness were signifi- Table 4 and Figure 1 show the of which 8723 (26%) fell in the 18–22-
cant predictors of dose indexes (after variation of dose indexes with lateral cm bin. The median water-equivalent
controlling for facility as random ef- thickness (head examinations) and diameter for examinations of the cer-
fect and facility characteristics, age, water-equivalent diameter (neck and vical spine without contrast material
and sex as fixed effects). The analysis cervical spine examinations). The me- was 19 cm. There were 97 586 cervical
for body examinations showed that all dian lateral thickness for examinations spine examinations, of which 36 711
bins were significantly different from of the head and brain without contrast (38%) fell in the 17–21-cm bin.
each other for CTDIvol. For chest ex- material was 15 cm. Among 223 908 Tables 5–7 show the variation of the
aminations, all but 21–25 cm and 25– head examinations, 137 755 (62%) fell dose indexes for chest, abdomen and
29 cm were different from each other in the 14–16-cm bin that included the pelvis, and chest, abdomen, and pel-
for DLP. For head and neck examina- median thickness. The median water- vis examinations with water-equivalent
tions, most bins were not statistically equivalent diameter for neck with diameter. Figures 2–4 show graphic

Radiology: Volume 284: Number 1—July 2017 n radiology.rsna.org 125

MEDICAL PHYSICS: U.S. Diagnostic Reference Levels and Achievable Doses for 10 Adult CT Examinations Kanal et al

Table 5
Size-based ADs and DRLs for Chest CT Examinations
CTDIvol (mGy) SSDE (mGy) DLP (mGy-cm)
Examination and Median AD (50th DRL (75th AD (50th DRL (75th AD (50th DRL (75th
Size (Diameter) Size (cm) No. of Facilities No. of Patients Percentile) Percentile) Percentile) Percentile) Percentile) Percentile)

Chest without contrast 21–25 389 10 863 5 7 8 12 186 270

material*
25–29 445 35 204 6 9 9 12 231 317
29–33 454 58 650 9 12 11 15 334 443
33–37 431 39 375 13 17 14 18 477 610
37–41 394 10 539 17 21 15 19 595 760
All† 485† 159 909 10 15 11 16 347 545
Chest with contrast 21–25 367 7507 5 8 9 12 205 299
material*
25–29 437 25 021 7 10 9 13 238 366
29–33 447 41 044 10 13 11 15 353 469
33–37 428 27 097 14 18 15 18 516 660
37–41 373 7581 18 23 16 20 663 855
All† 476† 111 898 10 16 12 17 374 596
Chest pulmonary 21–25 112 2578 6 9 9 14 203 282
arteries with
contrast material*
25–29 147 10 930 8 11 10 15 250 350
29–33 146 20 224 11 14 13 17 357 445
33–37 141 17 244 15 19 15 20 477 631
37–41 118 6683 19 25 18 23 611 838
All† 183† 58 986 11 18 13 19 357 557

Note.—The AD is the 50th percentile of the distribution of median values (the 50th percentile) of all participating facilities; the DRL is the 75th percentile of the distribution of median values of all
participating facilities.
* Water-equivalent diameter (cm) was used. The median diameter was 31 cm for all examinations.
†
“All” includes data beyond lowest- and highest-size bins; “No. of facilities” is the total no. of facilities submitting data for any size patient.

representations of the same data. The Table 9 shows that the U.S. DRLs size-based DRLs and ADs. Although
median water-equivalent diameter for are not markedly different from those the impact of patient size on radiation
all examinations was 31 cm. There in other countries (11–13,19–25). dose is well established (13,27), na-
were 330 793 chest examinations, of tional DRLs have previously provided
which 119 918 (36%) fell in the 29–33- only one value for each examination.
cm bin. There were 587 558 abdomen Discussion These are based on a standard-size
examinations, of which 187 860 (32%) This work establishes DRLs and ADs phantom representing an “average”
fell in the 29–33-cm bin. There were using data from the largest source in patient (11,12), a single patient size
37 142 chest, abdomen, and pelvis ex- the world of CT dose information from (19,20), or data averaged across all
aminations, of which 12 117 (33%) fell actual patient examinations. The DIR patient sizes (13,22). Size-based DRLs
in the 29–33 cm bin. The median (50th was launched in 2011 (26) and, as of will allow facilities to optimize proto-
percentile) and 75th percentile CTDIvol July 2016, has data on 30.3 million ex- cols so that the resultant dose is com-
and SSDE for these examinations in- aminations from 1524 facilities. This mensurate with the size of the patient,
creased with patient size, especially extensive participation and totally auto- thus avoiding unnecessary radiation
with the very large sizes. The median mated complete capture of all patient exposure to the patient.
DLPs also increased consistently from examinations enable the development SSDE (for body examinations) ad-
smaller to larger sizes. of robust, clinically based national justs the phantom-based CTDIvol for the
Table 8 and Figure 5 summarize DRLs and ADs. DRLs and ADs are pro- size of the patient and gives a more re-
the results for the trunk (chest; abdo- vided for CTDIvol, SSDE, and DLP for alistic estimation of patient dose. For
men and pelvis; chest, abdomen, and the 10 most common CT examinations. all body examinations, SSDE ADs and
pelvis) examinations for median-size One of the unique contributions DRLs were higher than CTDIvol values
patients. of this work is the development of for smaller patients; SSDE ADs and

126 radiology.rsna.org n Radiology: Volume 284: Number 1—July 2017

MEDICAL PHYSICS: U.S. Diagnostic Reference Levels and Achievable Doses for 10 Adult CT Examinations Kanal et al

Table 6
Size-based ADs and DRLs for Abdomen and Pelvis CT Examinations
CTDIvol (mGy) SSDE (mGy) DLP (mGy-cm)
Examination and Median AD (50th DRL (75th AD (50th DRL (75th AD (50th DRL (75th
Size (Diameter) Size (cm) No. of Facilities No. of Patients Percentile) Percentile) Percentile) Percentile) Percentile) Percentile)

Abdomen and pelvis 21–25 353 14 667 7 9 11 14 318 422

without contrast
material*
25–29 390 43 185 9 12 13 16 443 545
29–33 415 64 317 13 16 15 19 639 781
33–37 403 51 133 17 21 18 22 865 1048
37–41 365 21 901 21 25 19 22 1071 1306
All† 446† 201 754 13 20 15 19 657 1004
Abdomen and pelvis with 21–25 397 29 691 7 9 10 13 300 394
contrast material*
25–29 443 82 822 9 11 12 15 409 524
29–33 448 108 921 12 15 15 18 608 755
33–37 434 76 681 17 21 18 21 887 1056
37–41 392 30 640 21 24 19 22 1072 1266
All† 492† 338 056 13 19 15 19 615 995
Abdomen, pelvis, and 21–25 137 4173 7 9 10 15 291 408
kidney without contrast
material*
25–29 165 10 640 8 12 12 16 380 526
29–33 170 14 622 12 15 14 19 576 705
33–37 164 11 440 16 20 17 20 788 943
37–41 148 5111 19 22 17 20 901 1092
All† 202† 47 748 12 18 14 19 586 877

Note.—The AD is the 50th percentile of the distribution of median values (the 50th percentile) of all participating facilities; the DRL is the 75th percentile of the distribution of median values of all
participating facilities.
* Water-equivalent diameter (cm) was used. The median diameter was 31 cm for all examinations.
†
All” includes data beyond lowest- and highest-size bins; “No. of facilities” is the total no. of facilities submitting data for any size patient.

Table 7
Size-based ADs and DRLs for Chest, Abdomen, and Pelvis CT Examinations
CTDIvol (mGy) SSDE (mGy) DLP (mGy-cm)
Examination and Median Size AD (50th DRL (75th AD (50th DRL (75th AD (50th DRL (75th
(Diameter) Size (cm) No. of Facilities No. of Patients Percentile) Percentile) Percentile) Percentile) Percentile) Percentile)

Chest, abdomen, and 21–25 162 3465 7 10 11 15 407 591

pelvis with contrast
material*
25–29 197 9363 9 11 12 15 536 705
29–33 202 12 117 12 15 14 18 779 947
33–37 187 7817 17 21 17 21 1076 1348
37–41 147 3089 21 25 18 22 1328 1647
All† 250† 37 142 12 19 15 19 774 1193

Note.—The AD is the 50th percentile of the distribution of median values (the 50th percentile) of all participating facilities; the DRL is the 75th percentile of the distribution of median values of all
participating facilities.
* Water-equivalent diameter (cm) was used. The median diameter was 31 cm.
†
“All” includes data beyond lowest- and highest-size bins; “No. of facilities” is the total no. of facilities submitting data for any size patient.

Radiology: Volume 284: Number 1—July 2017 n radiology.rsna.org 127

MEDICAL PHYSICS: U.S. Diagnostic Reference Levels and Achievable Doses for 10 Adult CT Examinations Kanal et al

Figure 2

Figure 2: Graphs show chest ADs and DRLs. (a) AD and DRL for chest without contrast material—CTDIvol and SSDE. (b) AD and DRL for chest without contrast
material–DLP. (c) AD and DRL for chest with contrast material—CTDIvol and SSDE. (d) AD and DRL for chest with contrast material–DLP. (e) AD and DRL for chest
pulmonary arteries with contrast material—CTDIvol and SSDE. (f) AD and DRL for chest pulmonary arteries with contrast material–DLP.

DRLs were lower than CTDIvol values For example, in the United Kingdom, of a narrower dose distribution and a
for the largest patient sizes. the 2005 DRLs for radiography, fluoros- lower median dose.
DRLs for the size bin containing me- copy, and dental x-rays were approxi- DRLs should be used to determine
dian-size patients were similar to those mately 16% lower than those in 2000 if a facility’s dose indexes are unusu-
in other countries. As more modern and were approximately half of those ally high; they should not be used as
CT scanners with more dose-reduction in the mid-1980s (28). While improve- target doses. Both ADs and DRLs are
options become available, we anticipate ments in equipment dose efficiency may provided to encourage facilities to op-
a further reduction in radiation dose be reflected in these dose reductions, timize dose to a lower level than that
used for clinical examinations. The DIR investigations triggered when DRLs indicated by the DRL. Image quality
will continue to monitor this trend and are exceeded can often result in new, must be taken into consideration when
will revise the U.S. ADs and DRLs as lower-dose protocols that provide suf- using DRLs and ADs to evaluate CT
necessary. ficient image quality for the diagnostic protocols on each scanner to determine
The use of DRLs has been shown to task. Thus, data points above the 75th if protocols are optimized. Ideally, facil-
reduce the overall dose and the range percentile are, over time, moved below ities should analyze and compare their
of doses observed in clinical practice. the 75th percentile—with the net effect median, size-grouped dose indexes

128 radiology.rsna.org n Radiology: Volume 284: Number 1—July 2017

MEDICAL PHYSICS: U.S. Diagnostic Reference Levels and Achievable Doses for 10 Adult CT Examinations Kanal et al

Figure 3

Figure 3: Graphs show abdomen and pelvis ADs and DRLs. (a) AD and DRL for abdomen and pelvis without contrast material—CTDIvol and SSDE. (b) AD and
DRL for abdomen and pelvis without contrast material—DLP. (c) AD and DRL for abdomen and pelvis with contrast material—CTDIvol and SSDE. (d) AD and DRL for
abdomen and pelvis with contrast material—DLP. (e) AD and DRL for abdomen, pelvis, and kidney without contrast material—CTDIvol and SSDE. (f) AD and DRL for
abdomen, pelvis, and kidney without contrast material—DLP.

Figure 4

Figure 4: Graphs show chest, abdomen, and pelvis ADs and DRLs. (a) AD and DRL for chest, abdomen, and pelvis with contrast material—CTDIvol and SSDE. (b)
AD and DRL for chest, abdomen, and pelvis with contrast material—DLP.

Radiology: Volume 284: Number 1—July 2017 n radiology.rsna.org 129

MEDICAL PHYSICS: U.S. Diagnostic Reference Levels and Achievable Doses for 10 Adult CT Examinations Kanal et al

with the respective size-based ADs and

Table 8
DRLs. If size-grouped dose indexes are
AD and DRL Comparisons for Trunk Examinations not available, facilities should compare
their overall median indexes with the
CTDIvol (mGy) SSDE (mGy) DLP (mGy-cm)
“all” DRLs and ADs. Also, if the CTDI-
Examination AD DRL AD DRL AD DRL or SSDE for a protocol is below its
vol

Chest without contrast material 9 12 11 15 334 443 DRL but the DLP exceeds its DRL, the
Chest with contrast material 10 13 11 15 353 469 scan length should be reviewed. DRLs
Chest and pulmonary arteries with contrast material 11 14 13 17 357 445 and ADs are not intended to be used
Abdomen and pelvis without contrast material 13 16 15 19 639 781 for comparisons with dose indexes for
Abdomen and pelvis with contrast material 12 15 15 18 608 755 individual patients. Implementation of
Abdomen, pelvis, and kidney without contrast material 12 15 14 19 576 705 DRLs and ADs is most effective if the
Chest, abdomen, and pelvis with contrast material 12 15 14 18 779 947 facility has a system to automatically
monitor patient dose indexes so that
Note.—ADs and DRLs are based on the size bin containing median-size patients.
aggregate results may be evaluated.
One of the advantages of using
Figure 5 a dose index registry to determine
national ADs and DRLs is eliminat-
ing the need to manually collect data
from a small sample of facilities and
patients. Data from an enormous pa-
tient population and an all-inclusive
set of examinations are automatically
collected, resulting in fewer errors
and enabling frequent updates. Trans-
parency of DIR data encourages ongo-
ing data quality improvement at par-
ticipating facilities.
There were some limitations inher-
ent to any automated data-collection
process. The DIR is a voluntary registry
and is not a random sample of facilities,
examinations, or patients. However, the
DIR demographics show it has broad
participation from all types of facilities.
Participants in the DIR do not submit
clinical indication information, so ADs
and DRLs can be developed only based
on examination type. Also, the re-
ported values reflect the doses that are
currently used in practice rather than
the lowest doses that would provide
clinically adequate images (or are op-
timal in any other sense). In addition,
facilities do not submit clinical images
with their dose information, so image
quality at the participating sites is not
assessed. We have to assume that the
majority of the examinations submit-
ted to the DIR met the facilities’ image
quality standards because we assume
the vast majority were interpreted. An
independent assessment of image qual-
ity is addressed by other processes,
such as accreditation (29). Another
Figure 5: Graphs show AD and DRL comparisons for trunk examinations. limitation was the manual process for

130 radiology.rsna.org n Radiology: Volume 284: Number 1—July 2017

 Table 9
Domestic and International DRL Comparisons
DRLs
ACR DIR ACR-AAPM Australia Canada The Netherlands
Body Part, Examination, and Parameter (2016)* (2013)† NCRP (2012)‡ Japan (2015)§ EU (2014)|| UK (2014)# Ireland (2012)** (2011)†† (2016)‡‡ (2012)§§ Greece (2014)||||

Head
CT of head and brain without contrast

Radiology: Volume 284: Number 1—July 2017

material

n
  CTDIvol (mGy) 56 75 75 85 60 60 58 60 79 67
  DLP (mGy-cm) 962 1350 1000 970 940 1000 1302 1055
Neck/cervical spine
CT of neck with contrast material

radiology.rsna.org
  CTDIvol (mGy) 19 30
  DLP (mGy-cm) 563 500 600
CT of cervical spine with contrast material
  CTDIvol (mGy) 28 28 19
  DLP (mGy-cm) 562 400–600 600 420
Chest
CT of chest without contrast material
  CTDIvol (mGy) 12 21 21 15 10 12 9 15 14 14
  DLP (mGy-cm) 443 550 400 610 390 450 521 480
CT of chest with contrast material
  CTDIvol (mGy) 13 21 21 15 10 12 9 15 14 14
  DLP (mGy-cm) 469 550 400 610 390 450 521 480
CT of chest pulmonary arteries with
contrast material
  CTDIvol (mGy) 14 13 13 10
  DLP (mGy-cm) 445 440 430 350
MEDICAL PHYSICS: U.S. Diagnostic Reference Levels and Achievable Doses for 10 Adult CT Examinations

Abdomen and pelvis

CT of abdomen and pelvis without contrast
material
  CTDIvol (mGy) 16 25 25 20 25 15 12 15 18 15 16
  DLP (mGy-cm) 781 1000 800 745 600 700 874 700 760
CT of abdomen and pelvis with contrast
material
  CTDIvol (mGy) 15 25 25 20 25 15 12 15 18 15 16
  DLP (mGy-cm) 755 1000 800 745 600 700 874 700 760
CT of abdomen, pelvis, and kidney without
contrast material
  CTDIvol (mGy) 15 10
  DLP (mGy-cm) 705 460
Table 9 (continues)

131
Kanal et al
MEDICAL PHYSICS: U.S. Diagnostic Reference Levels and Achievable Doses for 10 Adult CT Examinations Kanal et al

examination code mapping, with its in-

herent and unavoidable inconsistencies.
Greece (2014)|||| Examinations not tagged accurately by
the facility may cause problems both

17
1020
by skewing the benchmark data and
by being compared with inappropriate
benchmark data. The DIR drives facil-
The Netherlands

ities to standardize procedure names

through the use of mapping tools and
(2012)§§

RadLex terminology (30). The use of

dose-reduction techniques, such as
iterative reconstruction, is not col-
(2016)‡‡
Canada

17
1269

lected in the DIR. The study did not

assess CT scanner type and detector
configuration.
Australia
(2011)††

30
1200

Another limitation was the use

of facility median dose indexes to de-
Ireland (2012)**

velop the DRL. This gives equal weight

to each facility, irrespective of its size
and volume. To address this concern,
ACR-AAPM Practice Parameter for Diagnostic Reference Levels and Achievable Doses in Medical X-Ray Imaging (12)—based on phantom of average-size patient.
13
12

we also analyzed the data so that each

patient examination was given equal
DRLs

UK (2014)#

weight and noted that the different

methods created only slight differences
1000

in the resulting DRLs. This is probably

because of the large number of exami-
EU (2014)||

nations submitted by all facilities. Irre-

spective of the limitations noted above,
the study’s enormous patient volume
Japan (2015)§

provides robust benchmarks for DRLs

and ADs.
National Council on Radiation Protection and Measurements (11)—based on phantom of average-size patient.
1300
18

This work provides DRLs and ADs

for the 10 most common CT adult exam-
inations performed in the United States
NCRP (2012)‡

using 2014 data from the DIR and rep-

resents the first time, to our knowledge,
that national adult DRLs and ADs have
European Union (20)—derived from “most common value” from surveyed countries.

been developed as a function of patient

ACR-AAPM

* ACR registry DRLs are based on the size bin containing median-size patients.

size. This will enable facilities to effec-

(2013)†

tively compare their patient doses with

national benchmarks and to optimize
Domestic and International DRL Comparisons

ACR DIR

their CT protocols, resulting in lower

(2016)*

United Kingdom (13)—no condition specified for patient size.

Canada (23)—based on average-sized (50–90-kg) patients.

** Ireland (21)—based on average-sized (60–80 kg) patients.
Japan (19)—based on standard-sized (50–60-kg) patients.

The Netherlands (24)—based on standard-sized patients.

15
947

doses at the appropriate image quality.

Australia (22)—no condition specified for patient size.

The DIR will update its ADs and

DRLs on a routine basis to capture
Greece (25)—based on a “typical” acquisition.
CT of chest, abdomen, and pelvis with

future trends in CT scanners and ra-

Body Part, Examination, and Parameter

diation dose. Future work will include

expanding the analysis to include
Chest, abdomen, and pelvis

high-dose studies and various scanner

contrast material
Table 9 (continued)

configurations.
  DLP (mGy-cm)
  CTDIvol (mGy)

Acknowledgments: The authors acknowledge

J.R. Wells, Y. Zhang, and E. Samei of the Duke
University Clinical Imaging Physics Group for al-
lowing the ACR to use their code to automati-
cally determine patient size from localizer im-
††

‡‡

§§

||||
†

#
||

ages in the ACR’s Dose Index Registry.

132 radiology.rsna.org n Radiology: Volume 284: Number 1—July 2017

MEDICAL PHYSICS: U.S. Diagnostic Reference Levels and Achievable Doses for 10 Adult CT Examinations Kanal et al

Disclosures of Conflicts of Interest: K.M.K. 11. National Council on Radiation Protection 2/2). EC website. https://ec.europa.eu/
disclosed no relevant relationships. P.F.B. dis- and Measurements (NCRP). Reference energy/sites/ener/files/documents/
closed no relevant relationships. D.S. disclosed levels and achievable doses in medical and RP180%20part2.pdf. Published 2014. Ac-
no relevant relationships. M.B. disclosed no dental imaging: recommendations for the cessed November 4, 2016.
relevant relationships. L.P.C. disclosed no rele- United States, Report No. 172. Bethesda,
vant relationships. R.L.M. disclosed no relevant 21. Foley SJ, McEntee MF, Rainford LA. Estab-
Md: NCRP, 2012.
relationships. lishment of CT diagnostic reference levels
12. American College of Radiology (ACR). ACR- in Ireland. Br J Radiol 2012;85(1018):1390–
AAPM practice parameter for diagnostic 1397.
References reference levels and achievable doses in
22. Australian Radiation Protection and Nuclear
medical x-ray imaging. ACR website. http://
1. Pandharipande PV, Reisner AT, Binder WD, Safety Agency (ARPNSA). Australian Na-
www.acr.org/~/media/ACR/Documents/
et al. CT in the emergency department: a re- tional Adult Diagnostic Reference Levels for
PGTS/guidelines/Reference_Levels_Diag-
al-time study of changes in physician decision MDCT. ARPNSA website. http://www.ar-
nostic_Xray.pdf. Published 2013. Amended
making. Radiology 2016;278(3):812–821. pansa.gov.au/services/ndrl/adult.cfm. Pub-
2014. Accessed November 4, 2016.
lished 2011. Accessed November 4, 2016.
2. National Council on Radiation Protection
13. Shrimpton PC, Hiller MC, Meeson S, Gold- 23. Health Canada. Canadian computed tomog-
and Measurements (NCRP). Ionizing radia-
ing SJ. Doses from computed tomography raphy survey – national diagnostic refer-
tion exposure of the population of the Unit-
(CT) examinations in the UK – 2011 review. ence levels. Health Canada website. http://
ed States. Report No. 160. Bethesda, Md:
Public Health England website. https:// www.healthycanadians.gc.ca/publications/
NCRP, 2009.
www.gov.uk/government/uploads/system/ security-securite/canadian-computed-to-
3. Conference of Radiation Control Program uploads/attachment_data/file/349188/PHE_ mography-survey-2016-sondage-canadien-
Directors (CRCPD). Nationwide Evalua- CRCE_013.pdf. Published 2014. Accessed tomodensitometrie/alt/cct-survey-sondage-
tion of X-Ray Trends (NEXT): tabulation November 4, 2016. ct-eng.pdf. Published May 2016. Accessed
and graphical summary of 2000 survey
14. Goske MJ, Strauss KJ, Coombs LP, et al. November 4, 2016.
of computed tomography, CRCPD Publi-
Diagnostic reference ranges for pediatric 24. Nederlandse Commissie voor Stralingsdo-
cation E-07-2. CRCPD website. http://c.
abdominal CT. Radiology 2013;268(1):208– simetrie (NCS). Diagnostische referentien-
ymcdn.com/sites/www.crcpd.org/resource/
218. iveaus in Nederland, Rapport 21. http://
collection/81C6DB13-25B1-4118-8600-
9615624818AA/NEXT2000-CT.pdf. Pub- 15. Bhargavan-Chatfield M, Morin RL. The ACR www.referentieniveau.nl/2012/07/bekijk-
lished 2007. Accessed November 4, 2016. Computed Tomography Dose Index Regis- hier-het-ncs-r-apport.html. Published June
try: the 5 million examination update. J Am 2012. Accessed November 4, 2016.
4. McCollough CH, Leng S, Yu L, Cody DD,
Coll Radiol 2013;10(12):980–983. 25. Simantirakis G, Hourdakis CJ, Economides
Boone JM, McNitt-Gray MF. CT dose index
and patient dose: they are not the same 16. Christianson O, Li X, Frush D, Samei E. S, et al. Diagnostic reference levels and pa-
thing. Radiology 2011;259(2):311–316. Automated size-specific CT dose monitoring tient doses in computed tomography exam-
program: assessing variability in CT dose. inations in Greece. Radiat Prot Dosimetry
5. Seibert JA, Boone JM, Wootton-Gorges SL, 2015;163(3):319–324.
Med Phys 2012;39(11):7131–7139.
Lamba R. Dose is not always what it seems:
where very misleading values can result from 17. Huda W, Lieberman KA, Chang J, Ros- 26. Morin RL, Coombs LP, Chatfield MB. ACR
volume CT dose index and dose length prod- kopf ML. Patient size and x-ray technique Dose Index Registry. J Am Coll Radiol
uct. J Am Coll Radiol 2014;11(3):233–237. factors in head computed tomography ex- 2011;8(4):288–291.
aminations. I. Radiation doses. Med Phys 27. Waszczuk LA, Guziński M, Czarnecka A,
6. American Association of Physicists in Med-
2004;31(3):588–594. Sąsiadek MJ. Size-specific dose estimates
icine (AAPM). Size-specific dose estimates
(SSDE) in pediatric and adult body CT ex- 18. American Association of Physicists in for evaluation of individual patient dose
aminations: the report of AAPM Task Group Medicine (AAPM). Use of water equivalent in CT protocol for renal colic. AJR Am J
204. AAPM website. http://www.aapm.org/ diameter for calculating patient size and size- Roentgenol 2015;205(1):100–105.
pubs/reports/RPT_204.pdf. Published 2011. specific dose estimates (SSDE) in CT: the 28. Hart D, Hillier MC, Wall BF. National ref-
Accessed November 4, 2016. report of AAPM Task Group 220. AAPM web- erence doses for common radiographic,
site. http://www.aapm.org/pubs/reports/ fluoroscopic and dental X-ray examina-
7. 1990 Recommendations of the International
RPT_220.pdf. Published 2014. Accessed tions in the UK. Br J Radiol 2009;82(973):
Commission on Radiological Protection.
November 4, 2016. 1–12.
Ann ICRP 1991;21(1-3):1–201.
19. Yonekura Y. Diagnostic reference levels 29. American College of Radiology (ACR). ACR
8. Radiological protection and safety in medi-
based on latest surveys in Japan – Japan Computed Tomography Accreditation Program
cine. A report of the International Commis-
DRLs 2015. Japanese Network for Research Requirements. ACR accreditation website.
sion on Radiological Protection. Ann ICRP
and Information on Medical Exposure. Med- http://www.acraccreditation.org/~/media/
1996;26(2):1–47. [Published correction ap-
ical exposure Research Information Net- ACRAccreditation/Documents/CT/Require-
pears in Ann ICRP 1997;27(2):61.]
work (J-RIME) website. http://www.radher. ments.pdf?la=en. Published 2002. Updated
9. Diagnostic reference levels in medical imag- jp/J-RIME/report/DRLhoukokusyoEng.pdf. May 3, 2016. Accessed November 4, 2016.
ing: review and additional advice. Ann ICRP Published 2015. Accessed November 4,
30. Radiological Society of North America
2001;31(4):33–52. 2016.
(RSNA). RadLex playbook. RSNA web-
10. Miller DL, Vano E, Rehani MM. Reducing 20. European Commission (EC). Radiation site. http://playbook.radlex.org/playbook/
radiation, revising reference levels. J Am Protection No. 180 – Diagnostic reference SearchRadlexAction. Published 2011. Updated
Coll Radiol 2015;12(3):214–216. levels in thirty-six European countries (Part July 15, 2016. Accessed November 4, 2016.

Radiology: Volume 284: Number 1—July 2017 n radiology.rsna.org 133