RPT 246

AAPM REPORT NO.
246
Estimating Patient Organ Dose

with Computed Tomography:
A Review of Present Methodology and
Required DICOM Information
A Joint Report of AAPM Task Group 246 and the European

Federation of Organizations for Medical Physics (EFOMP)
August 2019
DISCLAIMER: This publication is based on sources

and information believed to be reliable, but the
AAPM, the authors, and the editors disclaim any war-
ranty or liability based on or relating to the contents of
this publication.
The AAPM does not endorse any products, manufac-

turers, or suppliers. Nothing in this publication should
be interpreted as implying such endorsement.
© 2019 by American Association of Physicists in Medicine

This page intentionally left blank.
Estimating Patient Organ Dose
with Computed Tomography:
A Review of Present Methodology
and Required DICOM Information
A Joint Report of
AAPM Task Group 246 and the European Federation
of Organizations for Medical Physics (EFOMP)
Jonas Andersson, Chair1, William Pavlicek, Co-Chair2, Rani Al-Senan3, Wesley Bolch4,
Hilde Bosmans5, Dianna Cody6, Robert Dixon7, Paola Colombo8, Frank Dong9, Sue Edyvean10,
Jan Jansen10, Kalpana Kanal11, Shuai Leng12, Qing Liang13, Cynthia McCollough12,
Ed McDonagh14, Michael McNitt-Gray15, Robert Paden2, Madan Rehani16, Ehsan Samei17,
Ioannis Sechopoulos18, Mark Supanich19, Christine Theodorakou20, Xiaoyu Tian17,
Alberto Torresin21, Annalisa Trianni22, David Zamora11, and Federica Zanca23
1
Umeå University, Umeå, SE
2
Mayo Clinic, Phoenix, AZ, USA
3
Columbia University, New York, NY, USA
4
University of Florida, Gainesville, FL, USA
5
University of Leuven, Leuven, BE
6
MD Anderson Cancer Hospital, Houston, TX, USA
7
Wake Forest University School of Medicine, NC, USA
8
Niguarda Ca’Granda Hospital, Milano, IT
9
Cleveland Clinic, Cleveland, OH, USA
10
Computing Center, HPA, Oxfordshire, UK
11
University of Washington Medical Center, WA, USA
12
Mayo Clinic, Rochester, MN, USA
13
Mercy Health System, Janesville, WI, USA
14
Royal Marsden Hospital, London, UK
15
David Geffen School of Medicine, UCLA, CA, USA
16
European Society of Radiology, Vienna, AU
17
Duke University, Durham, NC, USA
18
Radboud University Medical Center, Nijmegen, NL
19
Rush Presbyterian Hospital, Chicago, IL, USA
20
Christie Medical Physics, Wilmslow, Manchester, UK
21
Niguarda Ca' Granda Hospital, Milano, IT
22
AZ University Hospital, Udine, IT
23
GE Healthcare, Paris, FR
THE REPORT OF AAPM TASK GROUP 246:
Estimating Patient Organ Dose with Computed Tomography: A Review of Present Methodology and Required DICOM Information
Task Group Consultants
Erin Angel Canon Medical Systems, USA

Nicholas Bevins Henry Ford Medical Systems, Detroit, MI, USA
Kevin Buckley Boston Children’s Hospital, Boston, MA, USA
Gregory Couch Radimetrics, Toronto, CA
Thaddeus Flood Director, MITA, Washington, DC, USA
Dustin Gress MD Anderson Cancer Hospital, Houston, TX, USA
Sofia Kottou University of Athens, Athens, GR
Mary Sue Kulpins GE Healthcare, USA
Richard Morin Mayo Clinic, Jacksonville, FL, USA
William O’Connell GE Healthcare, USA
Renato Padovani University Hospital, Undine, IT
Anthony Seibert University of California, Davis Medical Center, Irvine, CA, USA
Stanley Stern FDA, USA
Orhan Suleiman FDA, USA (ret)
Makoto Suzuki Japan Medical Imaging and Radiological Systems Industries Association, JP
Virginia Tsapaki Konstantopoulio General Hospital, Athens, GR
Stephen Vastagh General Secretary, DICOM, USA
4
DISCLAIMER: This publication is based on sources and information believed to be reliable,
but the AAPM, the authors, and the publisher disclaim any warranty or liability
based on or relating to the contents of this publication.
The AAPM does not endorse any products, manufacturers, or suppliers. Nothing in this
publication should be interpreted as implying such endorsement.
ISBN: 978-1-936366-72-9
ISSN: 0271-7344
© 2019 by American Association of Physicists in Medicine
All rights reserved
Published by
American Association of Physicists in Medicine

1631 Prince Street
Alexandria, VA 22314
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.1 Purpose and Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.2 Out of Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2. Current Dosimetry Metrics and Associated DICOM Information . . . . . . . . . . . 8
2.1 Overview of the CTDI and its Derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.1.1 CTDI100, CTDIw, CTDIvol and CTDIfree-in-air . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.1.2 Tolerance Levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.1.3 Special Considerations for Nominal Beam Collimations >40 mm. . . . . . . . . . . . . . . . . . . . . . . 10
2.1.4 CTDIvol for Examination Protocols Without Table Translation . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2 Wide Beam Dosimetry and Equilibrium Dose: AAPM Report 111 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.3 The Dose Length Product (DLP). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.4 The CT Localizer Radiograph . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.5 Scanner Parameters Connected to CTDIvol and DLP in DICOM Image Header Tags . . . . . . . . . . . . . 13
2.6 Size-Specific Dose Estimates (SSDE) and Water-Equivalent Diameter (WED) . . . . . . . . . . . . . . . . . . . 14
2.6.1 SSDE and Longitudinal Variations in Patient Dimension . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.6.2 Calculation of WED . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3. Current and Emerging Methods to Estimate Organ Dose . . . . . . . . . . . . . . . . 15
3.1 Studies on the Accuracy and Utility of SSDE in Estimating Organ Doses . . . . . . . . . . . . . . . . . . . . . . . 16
3.1.1 Typical Head and Body Examinations (Contiguous Axial and Helical). . . . . . . . . . . . . . . . . . . . 16
3.1.2 Perfusion Examinations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.2 Organ Dose Estimates with Regional CTDIvol Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.3 Application of the Convolution Method for Organ Dose Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.4 Monte Carlo Simulation Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.4.1 Monte Carlo Engines and Computational Phantoms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.4.2 Modeling of the X-ray Source. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.4.3 Monte Carlo and Normalization of Reported Organ Dose . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.4.4 Generic Axial Dose Libraries vs. Helical Protocol-Specific Dose Libraries . . . . . . . . . . . . . . . 24
3.4.5 Considerations of Starting Angle and Overranging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.4.6 Monte Carlo and Modeling of TCM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.5 Monte Carlo Simulation with Stylized Anthropomorphic Phantoms . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.6 Monte Carlo Simulation to Model the Dose Deposited by a CT Scanner . . . . . . . . . . . . . . . . . . . . . . 26
3.7 Benchmarking and Validation of Monte Carlo Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4. Sources of Uncertainty in Estimating Organ Dose . . . . . . . . . . . . . . . . . . . . . . 27
4.1 Uncertainties in SSDE–based Calculations of Patient Organ Dose . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
4.1.1 Pencil Ionization Chamber Measurements and CTDIvol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
4.1.2 The SSDE. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
4.1.3 SSDE and TCM Examinations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4.1.4 Patient Not Centered in the Gantry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4.1.5 The SSDE and Partial Organ Irradiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4.2 Uncertainties in Monte Carlo Estimates of Organ Dose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4.2.1 Computational Phantoms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4.2.2 Scanner Irradiation Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.2.3 Patient Not Centered in the Gantry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.3 Contrast Media Used in CT Examinations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
4.4 Reporting Uncertainty with Estimates of Patient Organ Dose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
5. Summary of Report and Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
6. References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Appendix: DICOM—Present and Future for Dosimetry and Estimating Organ Dose. . . 40
6
1. Introduction
The radiation absorbed dose or ‘dose’ that a patient receives from a routine CT examination is consid-
ered to yield very low risk of harm when properly used to obtain a diagnostic benefit, i.e., when the
justification and optimization of a given examination have been taken into account.1 However, a gap
is recognized in the ability of the conventional CT radiation dosimetry metrics—the Computed
Tomography Dose Index (CTDI) and Dose Length Product (DLP)2 —to accurately represent individ-
ual patient organ doses. Organ dose is generally regarded as one of the best metrics to quantify indi-
vidual radiation burden.
1.1 Purpose and Overview
The purpose of this report is (1) to summarize the current state of the art in estimating organ doses
from CT examinations and (2) to outline a road map for standardized reporting of essential parameters
necessary for estimation of organ doses from CT imaging in the DICOM standard. To address these
purposes, the report includes a comprehensive discussion of (1) the various metrics, concepts, and
methods that may be used to achieve estimates of patient organ dose and (2) the DICOM standard for
CT.
This Joint Report of the American Association of Physicists in Medicine (AAPM) Task Group
246 and the European Federation of Organizations for Medical Physics (EFOMP) contains three
major sections and an appendix. Section 2 (with additional material in the appendix) provides a review
of basic CT dosimetry metrics, their uses and limitations in the context of organ dosimetry, and the
DICOM information currently associated with parameters that affect CT dose metrics and, conse-
quently, organ dose estimates. Section 3 provides an overview of present and emerging organ dose
estimation methods reported in the literature, e.g., for the lens of the eye, breast tissue, colon, and skin.
Finally, the report concludes with section 4, which provides a discussion on the sources and magni-
tudes of uncertainty for different organ dose estimation methods.
Ongoing efforts to facilitate routine standardized estimation of patient organ doses from CT are
dependent, in large part, on the availability of the DICOM Radiation Dose Structured Report (RDSR),
which provides a host of information pertinent to radiation dose calculations. This report, therefore,
includes detailed information on DICOM header content in CT images and how it can be used in
organ dose estimation. The RDSR markedly expands the abilities of the clinical medical physicist to
estimate doses at the patient, device, and protocol level.3,4
1.2 Out of Scope
Effective dose, while important from a general radiation protection viewpoint, is not addressed in this
report. This quantity is used to account for partial body exposures to ionizing radiation and to reflect
the associated radiation detriment for populations of workers and the general public. This quantity
provides an estimate of detriment by weighting the absorbed dose to organs and tissues according to
the sensitivity of the exposed tissues and organs to ionizing radiation, as well as according to the qual-
ity, or linear energy transfer, or the applied type of radiation.4 Macroscopic irradiation conditions are
commonly used to determine the effective dose, e.g., the DLP for CT examinations, which means that
characteristics of a specific individual that will influence organ dose and radiation detriment are not
taken into account. Effective dose is commonly used to compare doses delivered by different exposure
conditions, such as to compare the radiation detriment from medical imaging to that from the dose
received from naturally occurring background sources of radiation. Such comparisons are helpful in
discussing the relative risk and justification of a given examination. Present methodology for estimat-
ing individual patient organ dose is the subject of this report, and thus the effective dose concept is out
of scope.
7
2. Current Dosimetry Metrics and Associated DICOM Information

2.1 Overview of the CTDI and its Derivatives
The computed tomography dose index (CTDI) is a quantity developed to reflect the amount of radia-
tion produced by a CT scan in a standardized fashion.5,6 It is determined from a measurement of the
scanner’s radiation output using a single rotation of the x-ray source with a static table. The CTDI
metric is given in units of gray (Gy), usually mGy, i.e., absorbed dose or energy deposited per unit
mass (i.e., J/kg). CTDI and its derivatives are described in detail elsewhere, and only briefly reviewed
here.7–10
2.1.1 CTDI100, CTDIw, CTDIvol and CTDIfree-in-air
CTDI is formally defined2 as
1 
CTDI  
NT  D(z )dz . (1)
Depending on the calibration of the dosimetry system used, air kerma, K(z), may be directly measured
instead of absorbed dose, D(z), in the patient longitudinal direction, or z-axis. While the CTDI can be
correctly defined in either term (absorbed dose or air kerma), we shall use absorbed dose in this report.
The CTDI is defined using an infinite integral of the dose profile, D(z), along the z-axis. To produce
the CTDI, the dose integral is divided by the nominal beam collimation, d = NT. The nominal colli-
mation is the product of the number of physical data channels, N, and the width of one detector data
channel in the z-axis, T. It is important to note that the reconstructed image thickness and the detector
data channel width are commonly described by the same term, ‘slice’ (as in a 64-slice CT scanner or a
3-mm image slice). In this report, the term ‘slice’ only refers to the number of data channels N, each of
width T.
As measurements cannot be performed using infinity as integration limits, the conventional appli-
cation of the CTDI formalism involves performing measurements using a calibrated 100-mm-long
pencil ionization chamber and an electrometer,5,11
1 50 mm
CTDI100 
NT 50 mm D(z )dz , (2)
where the pencil ionization chamber reading represents a direct measurement of the integral in Equa-
tion (2). The physical interpretation of the CTDI100 is the dose at the center of a 100-mm scan length
for a table increment per rotation d = NT. 6 CTDI (described further in section 2.2) is the asymptotic
(equilibrium) dose reached at the central slice of a large scan length.12
The CTDI100 alternatively can be determined empirically by measuring the spatial distribution of
absorbed dose, D(z), along the z-axis in a CTDI phantom. This approach requires a dosimeter capable
of performing measurements with a high spatial resolution, e.g., a small-measurement-volume ioniza-
tion chamber or a solid-state detector. An example of the dose profile acquired using a small-volume
detector is given in Figure 1 for a nominal collimation of 5 mm. Because the pencil ionization cham-
ber is 10 cm long, the scattered radiation tails of the dose profile are not fully captured. Due to this
truncation, the measured CTDI100 underestimates CTDI (i.e., equilibrium dose, as discussed below)
by approximately 20% for nominal beam collimations of 40 mm.2 Wider nominal beam collimations
cause more extensive underrepresentation of the equilibrium dose by CTDI100.2,13 In spite of this
underestimation of the equilibrium dose, CTDI100 is a robust, standardized measurement of the radia-
tion output for purposes of annual and post-maintenance CT quality control (see section 2.1.2).
8
Figureof1.1.
Figure 1. Example a dose profile from a CT scan acquired with a nominal collimation of 5 mm. The dose profile
was measured along the central axis of a CTDI head phantom (16 cm diameter) with a liquid ionization chamber
calibrated for dose to water.14,15
The weighted CTDI, CTDIw, is the CTDI100 calculated from measurements made at the center and
periphery of standardized phantoms of a specified size and composition, which represent a simplified
“head” and “body” patient cross section. The head CTDI phantom is 16 cm in diameter, and the body
CTDI phantom is 32 cm in diameter. Both are made of polymethylmethacrylate, PMMA, e.g., acrylic,
(r = 1.19 0.01 g cm–3). The ‘w’ refers to a weighting of the measurements made at the edge and
periphery of the CTDI measurement phantom (1/3 of the center value and 2/3 of the periphery value).
CTDIw represents the average dose over the central plane of a 100-mm scan length.
The volume CTDI, CTDIvol, takes into account any overlaps or gaps between successive rotations
of the x-ray tube. For contiguous source rotations (table travel per rotation = nominal beam collima-
tion), CTDIw = CTDIvol. For noncontiguous source rotations,
NT (3)
CTDI vol  CTDI w ,
d
CTDIw is divided by the pitch value (in helical scanning) or the table increment divided by the nomi-
nal beam collimation (in axial scanning).
The CTDI free-in-air is based on the same concept as the CTDI measured in a phantom, but the
CTDIfree-in-air is measured in air, usually at isocenter, as opposed to at locations within the field of view
that correspond to standard positions within the CTDI phantom. This metric is often used for quantify-
ing the x-ray output from the scanner. Because it can be measured without the use of CTDI phantoms,
it can be easily used for quality assurance purposes. CTDIfree-in-air is also required for certain Monte
Carlo dose estimation data sets, for example those from the National Radiological Protection
Board.16,17
2.1.2 Tolerance Levels
A part of the uncertainty in organ dose estimates comes from uncertainties in the reported CTDIvol,
which is used in several present and emerging dosimetry methods, and derived from measured CTDIw
9
Table 1: Examples of typical tolerance levels for the CTDIw

according to five different CT manufacturers (A–E)
X-ray Tube Potential Typical or Expected Maximum

Manufacturer
(kV) CTDIw Tolerance Tolerance
A kV < 80 15% 40%
kV > 80 15% 30%

B 15% 40%
C 80–140 kV 25%
Narrow collimation 30%

(2  0.625 mm, 2  0.5 mm)
D 120 kV 20%
E 80–140 kV 20% 20%
values. CTDIw tolerance levels are determined by manufacturers and reported to regulatory agencies
at the time that they submit for product market clearance. They correspond to the allowable differ-
ences between measured and displayed CTDIvol values for contiguous scans, and they often vary
between different manufacturers and modes of scanner operation, e.g., x-ray tube potential and nomi-
nal collimation. Differences that fall within these tolerances are not considered by the manufacturer to
require any service action. In Table 1, examples of typical tolerance levels as determined by five dif-
ferent CT manufacturers are given (personal communication, Medical Imaging and Technology Alli-
ance (MITA)).
From a dosimetry perspective, CT scanners typically demonstrate a high degree of reproducibility
between different makes and models,18 with relative reproducibility errors under 5%, well below the
CTDIw tolerances stated by the manufacturers in Table 1. However, many factors may influence the
CTDIw values for individual scanners: age of x-ray tube, calibration of scanner, dosimetry system
used and its calibration, measurement setup, and technical parameters such as x-ray tube potential
(especially lower tube potential settings as described for manufacturer A in Table 1), tube current
(mA), and the nominal collimation. The tolerances are typically higher for narrow collimations (e.g.,
manufacturer C in Table 1) due to uncertainty in actual beam width and higher dose contributions
from beyond the imaged volume, i.e., penumbra.19
Medical physicists commonly compare their CTDI measurements to the manufacturer-reported
CTDIvol, which is based on the manufacturer-determined CTDIw. The CTDIvol, which is displayed on
the operator console, represents an average value of CTDIw values measured on many systems of the
same model and software version. Each manufacturer’s operation manual specifies the tolerance val-
ues for differences between their average CTDIw values and those measured by the user. The measure-
ment precision achieved by an individual user on a specific scanner is typically much smaller than the
tolerances specified by the manufacturer.18
2.1.3 Special Considerations for Nominal Beam Collimations >40 mm

Due to the availability of CT scanners with relatively wide nominal collimation values, the IEC has
adopted a modified definition2 of the CTDI100 for nominal beam collimations >40 mm, as described by
free -in -air
1 CTDI NT
( NT ) REF 
CTDI100  DREF ( z ) dz  free -in -a
air
. (4)
CTDI REF
10
Here, parameters with a subscript REF should be evaluated in a standard CTDI phantom and with a
value of (NT)REF 20 mm.
The CTDIfree-in-air can be measured by using either an ionization chamber with a small measure-
ment volume, such as the liquid ionization chamber, or a solid-state detector.14,15 The dosimeter is
translated (stepped) through the x-ray beam using either the patient table or a separate stepping motor
system to measure the entire dose profile. CTDIfree-in-air can also be measured with pencil ionization
chambers having a length suitable for a given nominal collimation. The modified CTDI100 values are
used to calculate the CTDIw and CTDIvol, as previously described. An in-depth discussion on the intro-
duction and justification of Equation (4) for interested readers is found in IEC 60601-2-44 Edition
3.2.2
A recent International Atomic Energy Agency (IAEA) report on the status of computed tomogra-
phy dosimetry for wide cone-beam CT scanners provides practical advice on measuring CTDIfree-in-air
without recourse to special chambers and using the standard-length (100 mm) pencil ionization cham-
ber.20 The Institute of Physics and Engineering in Medicine (IPEM) also provides practical advice on
implementing this approach.21
2.1.4 CTDIvol for Examination Protocols Without Table Translation
Scanner-reported CTDIvol values significantly overestimate the actual dose to the patient when there is
no table translation. This is because the reported CTDIvol is based on the CTDI100 , i.e., reflecting a
scan length and scatter volume of 100 mm, where the tails of each individual dose profile are inte-
grated to represent the average dose in the center of the scan volume resulting from multiple contigu-
ous rotations of the x-ray source. This is not equivalent to the average dose in the center of the scan for
a single, axial rotation of a narrow beam, where scatter tails from adjacent rotations of the source are
not additive to the dose in the center of the scan volume. This effect becomes more pronounced at nar-
rower beam collimations (NT <20 mm).2 The resulting CTDIvol value overestimates the peak dose to
the irradiated tissue by as much as 300% for certain perfusion studies and bolus tracking scans.22,23
The difference between the CT scanner-reported CTDIvol and the peak skin dose within the scan vol-
ume for scans without table translation can be estimated by dividing the peripheral CTDI100 by the
peak peripheral dose for a standard CTDI phantom, which can be determined using a small 0.6 cm3
Farmer-type chamber centered in the scan volume, or direct measurement of the radiation dose pro-
file.23 As an example, measurements on a GE LightSpeed VCT scanner with a beam width setting of
10 mm (120 kV) yielded a CTDI100 of 21.3 mGy and peak peripheral dose of 7.5 mGy for 100 mAs,
corresponding to an overestimation of 284% (personal communication; Jonas Andersson, Umeå Uni-
versity, Sweden).
2.2 Wide Beam Dosimetry and Equilibrium Dose: AAPM Report 111
A comprehensive methodology for the evaluation of radiation dose in CT applications is given in
AAPM Report 111.12 The report presents a measurement paradigm that can be applied to any scan
mode (axial or helical), fan- or cone-beam scanners, and to stationary scans or those that perform
translation of the table during scanning. A small-volume ionization chamber (e.g., 0.6 cm3 Farmer-
type chamber) and a long CTDI phantom (e.g., 300 mm) are used to measure the accumulated dose
distributions for different measurement lengths, L.
With increasing L, the cumulative dose increases due to additional contributions from the radia-
tion scatter tails of the dose profile. Eventually, the cumulative dose reaches an equilibrium dose, i.e.,
the contribution of the dose profile tails will eventually (for large values of L) be negligible. This
upper limiting value is called the equilibrium dose,12 and it is determined by an infinite integral of the
dose profile over the patient table translation that occurs for a particular scan. The equilibrium dose
for a table increment of d =NT (a pitch of unity) is given by CTDI in Equation (1).
11
The AAPM Task Group 200 report, CT Dosimetry Phantoms and the Implementation of AAPM
Report Number 111, describes a practical method for measuring the equilibrium dose that should
address some of the limitations of CTDI metrics.24
2.3 The Dose Length Product (DLP)
Using the CTDI formalism, an estimate of the integrated absorbed dose to a scanned volume of tissue
having a length L is given by the dose length product (DLP), which is defined for axial, Equation (5),
and helical scans, Equation (6), respectively, as
DLPaxial  CTDI vol  d  n , (5)
DLPhelical  CTDI vol  L . (6)

Here, L represents the total table travel distance (cm) during the irradiation event, n is the number of
rotations in an axial scan, and d the table translation per rotation. It is important to note that for heli-
cal scanning, L typically exceeds the scan range prescribed by the user in order to acquire needed pro-
jection data beyond the limits of the image reconstruction range, a concept referred to as over-ranging.
Again, L represents the length over which radiation was delivered, which is not necessarily equal to
the length over which images can be produced. Scanner calculations of DLP take this over-ranging of
the beam into account as the scanner knows the precise length of the irradiation event, a number that is
not reported to the user.
When a user calculates the DLP according to Equations (5) or (6), care must be taken, therefore, to
take over-ranging into account, as otherwise the user-calculated results may differ substantially from
those reported in the Dose Data Page or RDSR.
To reduce the extra radiation applied during over-ranging, multiple manufacturers have imple-
mented beam collimators that can be quickly opened and closed at the beginning and ending of the
scan, respectively, to greatly reduce the unwanted over-ranging radiation.25 Here also, the true irradia-
tion length is not reported to the user. However, as the adaptive collimators block the majority of the
additional radiation that is applied due to over-ranging, the agreement between the user-calculated and
the scanner-calculated values of DLP will be much better.
2.4 The CT Localizer Radiograph
The CTDI formalism was developed specifically for CT, where an x-ray source rotates around the
patient. However, manufacturers have applied this concept to irradiation with a fixed source position,
based on the premise that the x-ray beam undergoes the same attenuation by the cylindrically symmet-
ric CTDI phantom whether or not the gantry is rotating. The “localizer CTDIvol” is then calculated
from the CTDIw per tube load Q determined in the standard fashion from a rotating gantry, where Q is
the tube-current-time product (tube current × rotation time, i.e., mAs or total charge) used to produce
x-rays during the rotation. Using the ratio CTDIw /Q determined at the same tube potential and beam
collimation NT used for the localizer,
CTDIW NT
localizer
CTDI vol  I , (7)
Q v
where I is the CT localizer tube current and v the table velocity (mm s–1), DICOM tag (0018,9309).
However, special consideration is required if the CTDI localizer
vol is used to estimate patient organ dose
since the absorbed dose to individual organs from a planar projection is not accurately represented by
the spatially averaged dose in a CTDI phantom.
It is important to note that the use of CTDIvol in this manner has not been vetted nor endorsed by
the peer-review process, but was rather adopted by the manufacturers without support of the scientific
12
community. In order to accurately estimate organ or skin doses from a CT localizer radiograph, the
attenuation of the planar projection radiograph by the patient must be adequately accounted for, just as
one would do for standard radiography exposures. The manufacturers’ approach of using CTDIvol to
report doses from the CT localizer radiograph does not do this and is, in the opinion of AAPM Task
Group 246 and EFOMP, not a sound dosimetric approach. Thus, we recommend against using
CTDIvol in this manner. In particular, the CTDIvol associated with a CT localizer radiography should
NOT be used to estimate patient dose.
2.5 Scanner Parameters Connected to CTDIvol and DLP in DICOM Image
Header Tags
Many different physical scanner parameters must be taken into account when determining CTDI-
based metrics. Examples of such scanner parameters that are stored in various DICOM image header
tags are shown in Table 2.26
Table 2: Sample values for DICOM image header tags and a CT head examination26
DICOM Tag Sample Value Description Comments and Units

(0018,0050) 3 Slice Thickness The nominal slice thickness, in mm. a
(0018,0060) 120 kV The peak tube potential of the x-ray generator, in
kilovoltage.
(0018,1150) 500 t, Exposure Time The time of x-ray exposure in msec. b
(0018,1151) 187 I, X-ray Tube Current X-ray tube current in mA c
(0018,1152) 170 Q, Exposure The exposure expressed in mAs, which is calculated
as the product of the exposure time and x-ray tube
current. d
(0018,1160) FLAT Filter Type The label for the type of filter inserted into the x-ray
beam.
(0018,9306) 0.6 T, Single Collimation Width The width of a single row of acquired data (in mm).
(0018,9307) 12 NT, Total Collimation Width The width of the total collimation (in mm) over the
area of active x-ray detection.
(0018,9309) 13.2 v, Table Speed The table speed (in mm/s) during the gathering of data
that resulted in this frame. e
(0018,9310) 6.6 d, Table Feed per Rotation The motion of the table (in mm) during a complete
revolution of the x-ray source around the gantry orbit.
(0018,9311) 0.55 p, Spiral Pitch Factor The ratio of the table feed per rotation (0018,9310)
to the total collimation width (0018,9307).
(0018,9323) Z_EC Exposure Modulation Type A label describing the type of exposure modulation
used for the purpose of limiting the dose.
(0018,9324) 15.3 Estimated Dose Saving A percent value of dose saving due to the use of expo-
sure modulation type (0018,9323). A negative percent
value of dose savings reflects an increase of exposure. f
(0018,9345) 31.9 CTDIvol Computed Tomography Dose Index (CTDIvol), in mGy
according to IEC 60601-2-44, Ed. 2.1 (clause
29.1.103.4). The Volume CTDIvol describes the average
dose for this frame (image) for the selected CT
conditions of operation.
a
The reconstructed image thickness is the nomenclature used in this report instead of slice.
b
How this field is used is vendor-dependent and may not represent the time of one full tube rotation. (0018, 9305) is defined to be the time for one gantry
revolution, but is not always utilized.
c
How this field is used is vendor-dependent. For tube current modulated scans, the value reported here may be the average tube current over a full rotation,
but this may vary by vendor.
d
How this field is used is vendor-dependent. For some vendors, this may be used to represent the product of the exposure time and x-ray tube current; for
others it may be used to represent the “effective mAs” or “mAs/slice,” which is defined to be the product of rotation time and average x-ray tube current
divided by the spiral pitch factor.
e
Velocity is the nomenclature used in this report instead of speed.
f
It is not clear what the percentage value of dose saving is relative to. It is most likely vendor-dependent.
13
In the DICOM image header, CTDIvol is reported for each reconstructed image. An average
CTDIvol over an entire CT scan is not reported directly in the DICOM header data of any single image,
but can be determined by averaging the CTDIvol from all reconstructed images.
Currently, there is no complete description of tube current modulation (TCM) supplied by CT
manufacturers as DICOM data. However, the DICOM image header data do contain, in tag
(0018,1151), the tube current value averaged over all projections used in the corresponding recon-
structed image. These average tube current values per image, when plotted against the longitudinal
patient axis, describe the longitudinal (z-axis) TCM.
AAPM Report 220 strongly recommended that CT manufacturers provide the complete TCM pro-
file, specifically the instantaneous x-ray tube current as a function of projection angle (e.g., every 1 or
5 degrees) and longitudinal position.27 This will allow modeling of TCM in three dimensions and
would be appropriately stored in the DICOM RDSR rather than the DICOM header data of any partic-
ular image, as the TCM profile describes the data acquisition as a whole.
2.6 Size-Specific Dose Estimates (SSDE) and Water-Equivalent Diameter (WED)
AAPM Report 204 introduced a method for estimating the absorbed dose at the center of the scan vol-
ume to a patient of a specific size. This value, SSDE, makes use of the scanner-reported CTDIvol and
conversion factors that take patient size into account.28 For a given water-equivalent-diameter
(WED),27 SSDE is calculated using a scanner-reported CTDIvol as
SSDE  fWED  CTDI vol . (8)
Here, fWED represents a conversion factor reflecting patient size.27,28 Although AAPM Report 204 does
not specifically use the term WED, the effective diameter values used in Report 204 to determine fWED
were for water- or tissue-equivalent materials. AAPM Report 220 clarifies that the effective diameter
values used in AAPM Report 204 are, therefore, equivalent to the WED values used in AAPM Report
220.27 Report 220 further demonstrates that WED can be calculated using either the CT localizer
radiograph or axial images reconstructed using a full field of view.27
2.6.1 SSDE and Longitudinal Variations in Patient Dimension
Patient dimension and attenuation can vary considerably along the longitudinal axis. When TCM is
used, scanner output also varies along the longitudinal (z) axis of a patient, according to changes in
patient attenuation. For a cylindrical CTDI phantom, the CTDIw normalized to a constant tube-current
time product (CTDIw / mAs) is easily measured or can be determined from the scanner’s accompany-
ing documents. The instantaneous mAs at every z-axis table position, expressed as mAs(z), is deter-
mined by the scanner according to the specific TCM profile used by the scanner. By multiplying the
spatially invariant CTDIw / mAs by mAs(z) and dividing by pitch (for a helical scan), the scanner out-
put at each location can be calculated. This information can be recorded as a series of mean CTDIvol
values for each reconstructed image, which is referred to as CTDIvol (z). SSDE can be calculated at
each position z along the longitudinal direction as
SSDE ( z )  fWED ( z )  CTDI vol ( z ), (9)
where fWED (z) is the size-specific conversion factor from AAPM Report 204 or 220.27,28 The mean
SSDE over the entire scan range can be expressed as
 ZN1 SSDE ( z ) (10)

SSDE  ,
N
14
where N is the total number of images. This approach calculates WED(z) and SSDE(z) for each recon-
structed image, determining the mean SSDE over the scan range using the fWED(z) and CTDIvol(z) at
each longitudinal position.
Leng et al. showed that the mean SSDE over a scan range calculated using Equations (9) and (10)
correlated well with a mean SSDE calculated using the WED from a central image in the scan range
and the mean CTDIvol over the entire scan range.29 Thus, for the purpose of estimating a single SSDE
value for a given patient, the “shortcut” of using the scanner-reported mean CTDIvol and the WED
from a central image in the scan range appears acceptable. AAPM Task Group 246 and EFOMP rec-
ommend that WED(z) be provided for each reconstructed image in a standardized DICOM tag, which
would provide future opportunities to better estimate organ doses.30,31
Using the SSDE, a small patient is correctly attributed a relatively higher absorbed dose compared
to that received by a large patient when the same CTDIvol is applied. The SSDE concept takes the scat-
ter tails of the dose profile into account, as the conversion factors fWED were calculated or measured
for examinations of typical scan lengths, based on patient size. SSDE has been shown to estimate the
mean dose in the center of the scan volume for both constant tube current and modulated tube current
acquisitions.32–34
2.6.2 Calculation of WED
Calculation of SSDE requires knowledge of the patient WED, either at the center of the scan range, as
an average for a certain anatomical region (e.g., thorax, abdomen, or pelvis), as an average over an
organ-specific region (e.g., liver), or at the reconstructed image level.
Conceptually, the simplest approach to determining patient size uses the CT localizer radiographs
(AP and lateral views) to calculate the WED of the patient. The patient’s outer surface is always pres-
ent in the CT localizer views, while it is not always included in axial image reconstructions, depend-
ing on operator selection of anatomy to be reconstructed and the patient size. However, the pixel
values in CT localizer image(s) are not provided in absolute units, which means that attenuation and
pixel values may not correlate well between different CT manufacturers.35 Furthermore, CT localizer
image(s) are typically presented with significant edge enhancement, which can result in a nonlinear
relationship between pixel value and attenuation. Thus, the CT localizer radiograph is difficult for the
medical physics community to utilize in the estimation of WED. Manufacturers, however, have the
needed tools to estimate WED from the CT localizer radiograph and, in some cases, provide this infor-
mation in the DICOM image header or the RDSR.36 Finally, accurate centering of the patient in the
gantry is very important, since off-center CT localizer radiographs make the patient, as well as the
estimated attenuation, appear larger or smaller, respectively, depending on centering in relation to the
x-ray tube and detector.
Both ICRU Report 87 and Report of AAPM Task Group 220 provide detailed information on
methods to determine the WED from reconstructed image data.7,27 These methods use CT number
thresholds to segment the patient contour and then determine the average pixel density relative to
water.7,27
3. Current and Emerging Methods to Estimate Organ Dose

There are several different approaches that have been used to estimate organ dose from CT examina-
tions. The most straightforward and easily adoptable is the SSDE, which has been investigated exten-
sively in the literature, as described in section 3.1.
Monte Carlo simulations, which require knowledge of and access to special software, are often
used to create the reference standards for organ dose estimation. Using this approach, it is possible to
model both the individual patient (either via the actual CT images or mathematical anthropomorphic
15
phantoms), the actual x-ray beam (kV, collimation, etc.), and conditions of irradiation (axial, helical,
etc.). There are also more straightforward solutions, where Monte Carlo organ dose estimates are tab-
ulated for generalized patient types, and the user can select technique parameters and scanner models.
These methods are described in section 3.4.
An intermediate solution, between the SSDE and Monte Carlo simulation in complexity, is the
convolution method described in section 3.3. Here, dose profiles (e.g., Figure 1) are used together
with data on the constant or modulated tube current used for a given CT scan to model the x-ray beam
and account for the scatter radiation contribution throughout an entire examination. Since the convolu-
tion method employs dose profiles, either measured or simulated in cylindrical phantoms, it may be
adopted as an SSDE derivative, together with WED to account for patient size.
3.1 Studies on the Accuracy and Utility of SSDE in Estimating Organ Doses
3.1.1 Typical Head and Body Examinations (Contiguous Axial and Helical)
3.1.1.1 Pediatric
Moore et al. investigated the applicability of SSDE in pediatric CT examinations using physical
anthropomorphic phantoms (representing children 5 to 55 kg) and MOSFET dosimeters for direct
organ dose measurements in the chest and abdominopelvic regions.37 The SSDE was determined
according to AAPM Report 204 and compared to measured dose for 23 different organs, resulting in
organ-specific conversion factors that could be used to determine organ dose given a value of the
SSDE.
The authors used the organ-specific conversion factors in a retrospective estimation of organ
doses in chest and abdominopelvic pediatric CT examinations, and the results were compared to pre-
viously published organ doses on the same examinations from Monte Carlo simulations. The authors
found that, for both constant tube current and with TCM, the average agreement between SSDE and
absolute organ dose was within 10% for organs fully covered by the scan volume. However, the
organ dose for partially irradiated organs and tissues, e.g., the liver in chest examinations, had a poor
correlation with SSDE, resulting in underestimation of organ dose. This is consistent with the recom-
mendations of AAPM Report 204, which states that when the organ is fully contained in the scan vol-
ume, SSDE can be used as an estimate of organ dose.28
3.1.1.2 Head–Eye Lens and Brain Dose
McMillan et al. and Hardy et al. have extended the SSDE concept for body CT examinations to
include head CT examinations.38,39 Eight patient models (two pediatric patients, three adult males, and
three adult females) from the GSF family of voxelized phantoms were used in Monte Carlo simula-
tions of both axial and helical scan modes to achieve conversion coefficients for organ dose (lens of
the eye and the brain) from geometric and attenuation-based estimates of head size (effective diameter
and WED, respectively) and CTDIvol.40
The authors report on the strong relationship (axial and helical: R2 > 0.92) between dose to the
brain determined by Monte Carlo simulation, as normalized by CTDIvol, and all head SSDE values.38
They concluded that the SSDE concept might be extended to include brain dose estimations.38,41–43
The relationship was not as strong for dose to the lens of the eye (axial: R2 > 0.73, helical R2 > 0.84).
The authors hypothesized that the weaker relationship in results for the lens of the eye might reflect
that this small peripheral organ was subject to surface dose variations not included in the simulated
model.38
It is essential to take into account CT device features that use adaptive shielding and angular, or
organ-based, TCM to protect the patient’s eyes, or the use of bismuth-containing eye shields.44–46
16
3.1.1.3 Chest–Lung and Breast Dose

SSDE is expected to yield acceptable estimates of absorbed dose for fully irradiated organs in routine
thorax examinations provided that patient “size” was determined using WED, as recommended by
AAPM Report 220.27 When only geometric dimensions are considered, errors of up to 20% may
occur.28 This is larger for thoracic examinations than for abdominopelvic examinations due to the
presence of lung tissue. Numerous studies have shown that with use of WED, the agreement between
SSDE and absorbed dose to fully irradiated organs in the thoracic region is indeed below the 20% tol-
erance value noted in AAPM Report 204, and often much smaller.47,48
As in x-ray mammography, the glandular portion of the breast tissue is the focus of patient-spe-
cific breast dosimetry with CT imaging.49–59 Khatonabadi et al. studied the relationship between aver-
age glandular doses normalized by global, regional, or organ-specific CTDIvol values.60 Correlations
between patient size (effective diameter) and normalized breast glandular dose were observed when
the scanner-reported CTDIvol was adjusted to take into account TCM along the scan length by using
regional or organ-specific average tube current.60
Bostani et al. found a good correlation between reference doses determined by Monte Carlo simu-
lation and region-specific normalized average glandular dose when patient size was observed using
the effective diameter.30 The group found that region-specific CTDIvol predicted the breast glandular
dose for the cases they investigated to within an average of 9.8%, with minimum and maximum errors
ranging from 0.6% to 34%, when compared with Monte Carlo dose calculations.30
3.1.1.4 Abdomen and Pelvis
Measurements by Wang et al.61 and Mueller et al.62 have demonstrated that SSDE can provide accu-
rate estimates of dose to fully irradiated abdominopelvic organs in constant tube current examinations.
In the CT colonography study by Mueller et al., SSDE was compared to dose measured using TLDs
placed within the inferior colon in vivo during screening CT colonography in a pilot project of 10
patients.62 The SSDE was found to be on average within 8% of the measured value.62 This study con-
firmed the accuracy of SSDE predictions to fully irradiated abdominopelvic organs in general, and to
the colon in particular. The absorbed dose to the kidneys, liver, pancreas, and spleen can also be
appropriately represented by the SSDE.
3.1.1.5 Fetal Dose
SSDE can be used to approximate fetal dose, as demonstrated by Angel et al., where maternal perime-
ter at the level of the fetus was used as a metric of patient size.63 Complexity arises in estimating fetal
dose when the fetus is only partially irradiated. In such a case, the SSDE serves as an upper limit of
dose to the fetus.
The Monte Carlo-based study using actual CT image data found that patient circumference dis-
played a strong relation to fetal dose, thus giving further support to the robustness of a SSDE
approach.63 However, they also found that including the fetal depth (distance from skin surface) in a
model of fetal dose had a better agreement to the Monte Carlo results than a one-dimensional
approach using only a patient size metric.63 This indicates that the SSDE concept can be further
refined to a two-dimensional quantity in order to improve the accuracy of fetal dose estimates.
3.1.1.6 Thyroid Dose
The most radiosensitive organ in neck CT examinations is the thyroid.4 A method to estimate thyroid
organ dose, comparable to a regional SSDE methodology, has been proposed by Huda et al.64 The
method was applied to 11 adult patients, where a neuro radiologist was consulted to specify size and
location of the thyroid in each patient.65
17
The average CTDIvol used for the examinations was 26 6 mGy, while the TCM increased the
tube current by an average of 44% over the thyroid.65 Estimated thyroid doses ranged from 29 to 80
mGy, with an average of 55 19 mGy.65 These findings suggest that the CTDIvol (and consequently
also the SSDE) is a poor metric to estimate thyroid dose. Further refinement is needed to increase the
accuracy of (thyroid) organ dose estimates, e.g., by taking TCM into account. Also, for organ dose
estimates in the neck region, it is important to note which CTDI formalism is being used, i.e., 16- or
32-cm diameter phantom.
3.1.2 Perfusion Examinations

3.1.2.1 Brain Perfusion
Brain perfusion examinations differ greatly from routine imaging of the brain. With perfusion exam-
inations of the brain, the table is stationary or motion is limited to a small z-axis extent, and exposures
are intentionally obtained to image the same anatomy over time. The eyes may or may not be in the
primary beam.
Lens Dose
Dosimetry for perfusion also differs from that of a routine brain acquisition. The use of a regional
exposure geometry, i.e., 4 or 8 cm, for perfusion imaging, results in the reported CTDIvol that overesti-
mates exposure to the lens of the eye and skin dose due to a lack of scatter x-ray equilibrium that is
assumed with the CTDIvol metric. Zhang et al. have shown in simulations for which the eye lens was
located fully within the scanning FOV that the CTDIvol overestimates the lens of the eye dose by up to
a factor of 1.7.66
More recently, an experiment that looked at lens of the eye dose was performed by Lopez-Rendon
et al.67 With the help of a pathologist, 35 thermoluminescent dosimeters (TLDs) were carefully placed
at different places (brain white and gray matter, skin, thyroid, salivary glands, oral mucosa, bone sur-
face, and eyes lens) in a female adult cadaver head. The head was subsequently scanned using the
clinical protocol for brain perfusion. The measured lens dose was 216 mGy for the right eye receiving
direct exposure and 154 mGy for the left eye, which also was directly exposed but was angled differ-
ently in the gantry; the average of the two eye lenses was 185 mGy. With a reported CTDIvol of
260 mGy, an overestimation of the eye lens dose was 17% for one eye and 41% for the other eye. On
average, the direct exposure to the two eyes was approximately 29% less than the reported CTDIvol.
The results are similar to those obtained by Zhang et al. and confirm that CTDIvol is a conservative
measure of the eye lens dose when used for conventional perfusion-type acquisitions.68
Peak Skin Dose
The findings for dose to the lens of the eye from the above studies were similar to those for peak skin
dose, i.e., skin dose had approximately the same dependencies on CTDIvol for the scanner type used by
Lopez-Rendon et al.67 Bauhs et al. have shown a point dose method for measuring skin dose with head
perfusion examinations.22 Their results indicated that CTDIvol overestimates skin dose by approxi-
mately a factor of 2.
Importantly, these results do not apply to whole-head imaging or to CT-guided biopsy of the
brain. Whole-head imaging results in a widely different case of scatter equilibrium than does perfu-
sion imaging. For CT biopsy of the brain, the brain may be positioned lower in the gantry, resulting in
a portion of the brain being possibly shielded by the shape of the bow-tie filter. The lens of the eye and
a portion of the skin will likely receive a higher exposure level than with the head centered at isocen-
ter.
18
3.1.2.2 Body Perfusion

CT perfusion examinations of the body are usually associated with higher radiation dose (CTDIvol) but
are performed with little or no table travel compared to conventional CT imaging. Skin damage or loss
of hair from x-ray exposure could occur in extreme cases, such as patients having multiple CT perfu-
sion examinations or interventional fluoroscopy procedures within a few days, or if scan techniques
are set incorrectly.69 The patient table does not translate with most perfusion examinations, and the
skin dose distribution differs from that of helical scans.22
Recently de las Heras et al. demonstrated a method for converting CTDIvol (measured and device-
reported values) into actual peak skin dose using four different CT scanners.70 The authors concluded
that the CTDIvol for the investigated scanners and scan modes can be used to estimate patient-specific
skin dose.
Leng et al. measured peak skin doses in shuttle-mode, body perfusion examinations on the ante-
rior and lateral surfaces of tissue-equivalent body-shaped phantoms.71 Phantoms ranging from 20 to
50 cm wide were used to represent patients from a 10-year-old child to a large adult. For each scan,
console CTDIvol was recorded, and the ratio between measured surface dose and console CTDIvol was
calculated. Regression analysis was used to determine the dependence of skin dose on patient size.
Skin dose was exponentially related to patient size by
skin dose = a  e  bCTDIvol ( R 2  0.97).
The fitting coefficients depended on kV and scanner type, with slight differences between 100 and
120 kV, and large differences for 80 kV. Surface dose expressed as a percentage of CTDIvol at 120 kV
varied from 49% to 65% on the 64-slice scanner at 120.6 mm collimation and from 86% to 116% on
the 128-slice dual-source scanner at 321.2 mm collimation across patient sizes. The method used in
this study is similar to that of the SSDE approach, i.e., using the CTDIvol and metrics of patient size to
determine organ (skin) dose. While SSDE estimates the average dose at the middle of the scan range,
this study estimated skin dose from perfusion scans.
3.2 Organ Dose Estimates with Regional CTDIvol Values
For the task of estimating organ doses, Khatonabadi et al. compared the accuracy of regional SSDE
(i.e., the SSDE calculated using the CTDIvol averaged over a specific anatomic region) to that of the
scan-averaged SSDE (i.e., the SSDE calculated using the CTDIvol averaged over the entire scan vol-
ume).31 The authors used Monte Carlo simulations and actual helical CT image data, acquired using
TCM, to determine the reference absorbed dose in five different organs (liver, spleen, kidneys, lungs,
and glandular breast tissue for female patients) to benchmark the accuracy of the different SSDE
approaches for estimating organ dose.31 They found that the SSDE calculated using the scan-averaged
CTDIvol was in poor agreement with the Monte Carlo results. When the SSDE was determined using
the CTDIvol and an image-by-image approach, Equations (9) and (10), either for a region (e.g., abdo-
men) or a specific organ (e.g., liver), the correlation to the Monte Carlo results was stronger.31 The
best agreement between Monte Carlo results and SSDE organ dose estimates based on SSDE occurred
with the organ-specific approach, which is a logical finding. However, even the estimates based on
regional SSDE were found to be an improvement compared to the estimates made using the CTDIvol
averaged over the entire scan.
3.3 Application of the Convolution Method for Organ Dose Estimation
As described in section 2, CTDI100 does not adequately account for the entire dose delivered by scat-
tered radiation. Any derivative of CTDI100 inherits this limitation. Thus, CTDIvol underestimates the
equilibrium dose by an amount that depends on patient size, examination scan length, and x-ray beam
collimation width.
19
Dixon has presented a convolution method for dose estimation that takes into account the com-
plete scanner profile.72 In the simplest form (constant tube current), the method uses dose profiles f(z),
either calculated or measured with a small-volume ionization chamber, to characterize the x-ray beam
and scatter tails in a cylindrical dosimetry phantom. The accumulated dose distribution, D(z), from a
CT scan can then be determined by convolution:72
1 L /2 1
D( z ) 
d  L / 2
f ( z  z )dz  
d
f ( z )  ( z / L ). (11)
Here, L represents the total CT scan length and d the table advance per rotation. Note that by setting
z = 0 (center of scan length), L = 100 mm, and d = NT (a pitch of unity), Equation (11) reduces to
CTDI100. The formalism described by Equation (11) can be demonstrated graphically, as shown in
Figure 2.
The method described in Equation (11) takes the axial dose profile f(z) for a single rotation (blue
line) and convolves it with a box function ( z / L ) to yield the accumulated dose distribution D(z) in
a cylindrical dosimetry phantom (red dashed line). The box function tells us how the dose profile is
added to the accumulated dose distribution at each point z. Thus, Equation (11) gives a description of
a constant tube current CT scan of length L in a homogeneous cylinder of water. The user input to
describe the accumulated dose distribution requires three parameters: the dose profile for a single rev-
olution of the x-ray tube resulting from specific scan parameter settings (e.g., a specific nominal beam
collimation and x-ray tube potential), the table advance per rotation, and the total scan length. Thus,
two pieces of the information needed to complete the calculation described in Equation (11) can be
found in the DICOM metadata. The obvious limitation with application of the convolution method is
that dose profile information is not readily available to the medical physics community. Dixon and
Boone, however, have derived analytical equations for the dose profile f(z) for the 32-cm PMMA
(body) phantom.73 The convolution method is not limited to describing constant tube current scans. To
model a helical scan acquired using TCM, Equation (11) can be modified to use the x-ray tube current

Figure 2. A graphical representation of the convolution method, adapted from Dixon.72 Here a dose profile f(z) is
convolved with a box function to represent a constant tube current scan.
20
at each z-axis location, i(z'), such that the dose profile contribution for each tube current strength i(z)
is added,
1 L /2 1
D( z )   i ( z ) f ( z  z )dz   f ( z )  i ( z ) ( z / L ) , (12)
d  L / 2 d
where f(z) is the dose profile per unit value of the tube current. The physical interpretation of Equation
(12) is that the dose at z depends on the tube current i(z') at all locations over the entire scan length L.
The addition of a description on how the x-ray tube current varies in the z-axis over a CT scan in the
convolution, Equation (12), thus makes it possible to determine the accumulated dose distribution for
spiral CT scans with TCM. Dixon et al. has described the difference between estimating dose accord-
ing to the convolution method for constant tube current and TCM, which is shown in Figures 3 and 4,
respectively.74
Figures 3 and 4 describe a constant tube current scan and TCM CT scan, respectively; the scan
lengths were identical, and the average tube current for the TCM scan was equal to the constant tube
current scan. Since the IEC method for computing CTDIvol treats the average tube current over the
entire scan length as if it were a constant tube current, then both have identical values of scanner-
reported CTDIvol. The SSDE will differ between the two examples, however, because both i(z) and
f(z) are determined by the value of WED(z) in the case of TCM, whereas only f(z) varies with WED(z)
for a constant tube current scan.
Recently, Tian et al. evaluated the convolution method for use with TCM in 60 clinical patient
examinations.75 The results were benchmarked against Monte Carlo simulations to estimate patient
organ dose accuracy. For the Monte Carlo simulations, each patient was matched to a representative
Figure 3. An example of the convolution method Figure 4. An example of the convolution method
used for a constant tube current (mA) scan. Character- used for a TCM CT scan. Characterization of the accu-
ization of the accumulated dose profile was done for mulated dose profile was done for the central axis of a
the central axis of a body dosimetry phantom.74 The body dosimetry phantom.74 In this extreme example,
accumulated dose profile was calculated by convolution the dose profile for a single rotation, f(z), (including its
of the dose profile for a single rotation (the small col- long scatter tails) is scalable by the regional tube cur-
ored profiles) with a box function per Equation (11). rent (mA), whereas the accumulated dose distribution
is not due to scatter tail contributions from other dose
profiles over the scan length.
21
computational phantom.76 To model the impact of TCM in the Monte Carlo simulations, the tube cur-
rent profiles were estimated for each computational phantom.77 The convolution method was used
together with a Monte Carlo-modeled dose profile representative of the maximum x-ray beam colli-
mation of the scanner on the central axis in an infinitely long 32-cm-diameter CTDI phantom. The
dose profile was convolved with TCM profiles to generate accumulated dose distributions for 60 clin-
ical patients. Simulations of constant tube current were also performed for each patient to normalize
regional and organ dose to CTDIvol and compare results from the convolution method and Monte
Carlo simulations, respectively. The patient size was known for each subject through the computa-
tional phantom matching for Monte Carlo simulation employed by Tian et al.75 Exponential regression
models were used to determine organ dose as a function of patient body diameter. Since the distribu-
tion of organs was already estimated from matching to computational phantoms, organ dose estima-
tion by the convolution method was straightforward in this work. Tian et al. found that organ doses
estimated by the convolution method were generally within 10% of the Monte Carlo simulation
results.75 Larger deviations for the convolution method were found for the shoulder and pelvic
regions. Tian et al. also found that compared to a full Monte Carlo simulation, the estimation method
using a global CTDIvol based on the average tube current of the entire examination yielded a poor esti-
mate of organ dose, with a maximum error above 50%, similar to the results reported by Khatonabadi
et al.31,75 By comparison, the method based on the approximation that regional dose depends on
regional tube current significantly improved the accuracy of organ dose estimation, again similar to
the results reported by Khatonabadi et al.31 The convolution method provided the most accurate esti-
mation across patient models and TCM strengths, since it most accurately accounted for scatter ema-
nating from the entirety of the scan length.75
The convolution method cannot use the SSDE formalism discussed in AAPM Report 20428 since
dose profiles are used to describe the primary x-ray beam and scatter radiation instead of the CTDIvol.
However, similar correction factors for patient size, related to standard phantoms used for measuring
or simulating dose profiles, can be derived for the convolution method.
3.4 Monte Carlo Simulation Methods
There has been a substantial history of the use of Monte Carlo-based methods to estimate patient dose
from CT examinations.16,17 These methods require several components. The Monte Carlo simulation
code models the stochastic properties of x-ray interactions with tissues (e.g., photoelectric and Comp-
ton scatter interactions), which is adapted to estimating organ doses. Monte Carlo simulation requires:
(a) modeling the x-ray source’s spectra, filtration, and geometry;
(b) a model of the patient’s anatomy, and specifically the organs of interest;
(c) a description of the scan parameters, including start and stop locations, as well as elements
such as helical path; and
(d) a mechanism to transport photons through the geometry and tally doses to the organs of
interest.
These are described below.
3.4.1 Monte Carlo Engines and Computational Phantoms
A growing number of CT scanners are being modeled for use with Monte Carlo radiation transport
codes, where device-specific refinements in modeling exposure conditions improve estimates of
patient radiation dose through advanced modeling of the x-ray beam and irradiation geometry.78–86
A computational anatomic phantom is a computerized representation of human anatomy in a for-
mat amenable to coupling with Monte Carlo radiation transport codes. These codes are computer pro-
grams that use random sampling of probability distributions of radiation interaction type (scatter or
22
absorption), energy transfer (full or partial), and scatter/emission angles to follow individual x-ray
photons, and potentially their secondary electrons, from a simulated model of the CT x-ray source
through the simulated model of patient—the computational phantom.
During Monte Carlo radiation transport of x-ray photons within a computational phantom, energy
deposition is scored within spatial regions associated with individual organs. Organ absorbed dose is
thus reported as the ratio of the scored energy deposition and the phantom organ mass. The one gen-
eral exception is for estimates of absorbed dose to the active (or red) bone marrow and the bone end-
osteum. For these skeletal tissues, the microscopic structure of the bone trabeculae and individual
marrow cavities is generally too small to be properly represented in a computational phantom. One
solution is thus to score, not energy deposition, but energy-dependent photon fluence in these skeletal
regions, and then convolve those fluences with an energy-dependent and bone-specific fluence-to-
dose response function. Photon dose-response functions are presently available for adult phantoms in
a work by Johnson et al. and Annex D of ICRP Publication 116.87,88
Individual computational phantoms have two defining characteristics—their format type and their
morphometric category. Format types include stylized, voxel, or hybrid forms. Stylized (or mathemat-
ical) phantoms are composed of 3D geometric surface equations defining both internal organ anatomy
and the external body profile. Individual organs are composed of various geometric objects such as
spheres, ellipsoids, and cones, either as single entities or as geometric combinations. Voxel phantoms
are composed of a 3D array of voxels of labeled anatomy that are typically based upon the segmenta-
tion of medical images (CT, MR, etc.) of patients or cadavers. Hybrid models preserve the anatomical
realism of a voxel phantom and are typically modeled through either NURBS (nonuniform rational B-
spline) or polygon mesh surfaces.
Morphometric categories include: reference, patient-dependent, patient-sculpted, and patient-spe-
cific. A reference phantom is one in which the morphometry is defined formally, such as the reference
male and female phantoms established by the International Commission on Radiological Protec-
tion,89,90 which describe the 50th percentile individual in both height and weight at a given age and
gender. Patient-dependent phantoms are similar to reference phantoms, but with the removal of the
restriction of 50th percentile height/weight. This allows the creation of a phantom library, whereby a
specific member from that library may represent, for example, a female patient at 30th height percen-
tile and 65th weight percentile at a specified age. Patient-sculpted phantoms typically start from
patient-dependent phantoms and then reshape the outer body contour (typically using NURBs or
meshes) to more closely match the body shape of the individual patient for which CT organ doses are
to be assessed. Finally, patient-specific phantoms are those in which both internal organ anatomy and
outer body shape is uniquely matched to the patient for which the organ dose estimate is sought (such
as in voxel phantoms).
Once a particular patient phantom model is selected for CT organ dose assessment for a specific
patient, several steps are needed to complete the computational model of the imaging system. These
steps are briefly outlined below.
3.4.2 Modeling of the X-ray Source

In a Monte Carlo simulation of a CT imaging examination, one must specify the starting position,
emission direction, and energy of each simulated x-ray photon within the virtual imaging beam.
Depending upon whether axial or helical scanning is considered, Monte Carlo sampling occurs along
a simple circle at each exposed position on the phantom, or along a helical path of a selected pitch.
Emission direction is sampled within an angle defining the fan beam width, and the beam collimation
defines the emission direction with additional consideration of the beam penumbra (i.e., overbeam-
ing).
23
Early attempts at Monte Carlo modeling of CT imaging required explicit knowledge of the mate-
rial and geometric shape of the bow-tie filters to properly model the probability density function of
emitted x-ray energies. In 2009, Turner et al. published an experimental approach which circumvented
the need for this proprietary information.91
3.4.3 Monte Carlo and Normalization of Reported Organ Dose
Measurements are generally required for estimating organ doses to patients from Monte Carlo simula-
tion. The direct output of Monte Carlo simulations typically yields organ doses per simulated x-ray
photon (mGy/photon), while the organ dose is reported as dose per time-integrated tube current (mGy/
100 mAs). Thus, a normalization factor is required to provide the number of x-ray photons per
100 mAs for a given irradiation geometry, typically one single axial rotation. This normalization fac-
tor can be calculated as the ratio of measured and simulated CTDIw, respectively.92,93
3.4.4 Generic Axial Dose Libraries vs. Helical Protocol-Specific Dose Libraries
Under the basic approach of organ dose determination employing computational phantoms and Monte
Carlo simulation, two options exist for establishing a precomputed CT organ dose library. The first
option establishes an organ dose library based upon Monte Carlo simulation of an array of single-rota-
tion axial images at a given collimation beam width, tube potential, and a given selected computa-
tional phantom. This axial acquisition dose library provides the organ dose, typically normalized to
100 mAs, indexed to each z-position in the computational phantoms, and can in theory extend across
the entire craniocaudal length of the virtual patient. For a given clinical imaging examination, one
then linearly sums the z-level specific organ doses over all axial acquisitions that fall within the start-
ing and ending anatomy of the CT examination.
For modeling helical protocols, the summed organ doses are scaled by the inverse of the examina-
tion pitch. This approach thus permits explicit consideration of any combination of starting and ending
anatomical landmarks. In the second option, when CT imaging studies fall under a specific listing of
established clinical protocols, and where the anatomical landmarks of the scan length and the pitch of
the examination are well established, the CT organ dose library may be constructed for each explicit
imaging protocol through explicit modeling of the helical scan.
3.4.5 Considerations of Starting Angle and Overranging
A disadvantage of a protocol-specific organ dose library is that a fixed starting angle is typically
assumed during the Monte Carlo simulations of the helical path of the x-ray tube. Variations in CT
starting angle may influence doses to more superficial organs, such as breast, thyroid, and eye
lens.94,95 One solution is to consider a systematic variation in starting angle for the helical path proto-
col-specific dose library, and populate the organ dose library with the organ dose averaged over all
possible starting angles. For example, axial acquisition dose libraries consider all possible starting
angles and provide the average value. In either case, this dose value may differ greatly from the actual
organ dose to superficial organs for a given helical scan with an arbitrary starting angle. Overrang-
ing—which defines additional irradiated anatomy prior to and beyond the edges of the reconstructed
images—may be accounted for in the organ dose library by incorporating additional (whole or frac-
tional) simulated rotations in the final dose estimate.
3.4.6 Monte Carlo and Modeling of TCM
The incorporation of TCM information into Monte Carlo simulation methods has been described and
investigated, both for angular and longitudinal forms by Angel et al.96,97 The works by Angel et al.
used TCM functions from actual scans and, therefore, were only available retrospectively and were
specific to the patients being modeled. In addition, Khatonabadi et al. demonstrated that for a com-
bined angular and longitudinal modulation TCM scheme, modeling only the longitudinal modulation
24
(e.g., using values from the DICOM image header data) provided reasonable estimates for organ
doses.43
However, one significant challenge to reporting organ doses following CT imaging on a pre-com-
puted basis is the impact of TCM—in both its angular and longitudinal forms. While all TCM algo-
rithms are based on the measurement of patient attenuation and the adaptation of tube current to that
attenuation, the implementation of various manufacturers’ methods are quite different. Manufacturers
may use different control parameters (e.g., Quality Reference mAs or Noise Index), relationships
between the attenuation and tube current, as well as the ability to have users set minimum and maxi-
mum tube current limits, and these may all influence the actual TCM function. One approach to simu-
late and validate the TCM function from one manufacturer is described in McMillan et al.33
3.5 Monte Carlo Simulation with Stylized Anthropomorphic Phantoms
Organ dose estimates using stylized computational phantoms are available from different sources,
e.g., the ImPACT CT Patient Dosimetry Calculator, which uses tabulated Monte Carlo data sets from
the National Radiological Protection Bureau.16,17 Patient-specific attributes may be used to match a
stylized phantom to a patient. Stylized computational phantoms incorporate the Monte Carlo radiation
transport simulation of a representative CT examination. The Monte Carlo simulation depends on
basic output (kV, mA, rotation time, pitch, etc.) and also machine-specific factors.
An example for an examination of a standard adult modeled by the NRPB18+ stylized computa-
tional phantom for a given CT scanner using ImPACT is given in Figure 5.98 An average CTDIvol
value of 13.3 mGy was found for 2,434 patients receiving a routine abdomen examination (personal
Figure 5 Left: organFigure

dose 2.4.
calculation example from ImPACT CT Patient Dosimetry Calculator. Right: the
NRPB18+ stylized computational phantom.98
25
communication, William Pavlicek, Mayo Clinic, Scottsdale, AZ). The average size of patients among
the 2,434 examinations measured from the CT localizer radiographs was 28.6 cm in the A-P and
36.3 cm in the lateral direction. These measurements correspond to an effective patient diameter (geo-
metrical mean) of 32.2 cm.
Figure 5 shows the acquisition parameters used to achieve a CTDIvol of 13.3 mGy. The ImPACT
CT Patient Dosimetry Calculator gives estimates of absorbed dose for each organ and tissue (HT), as
well as an estimate of total effective dose, which have weighing factors (wT) defined in the ICRP 2007
recommendations.4
The organ dose estimates in Figure 5 may be compared to the SSDE; calculating the SSDE for the
average-sized patient (effective diameter 32.2 cm) in the example yields a value of 15.2 mGy.28 As
previously discussed, Mueller et al. found that the SSDE was a good estimate of dose to the colon in a
comparison with TLD measurements in vivo.62 The present colon dose estimate from the ImPACT
Patient Dosimetry Calculator is 18 mGy. The difference between these colon dose estimates is 16%,
which may arise in part from approximations in estimating patient size for SSDE. Using WED and
CTDIvol for each reconstructed image instead of patient size at a single point in the z-axis and the aver-
age CTDIvol for the examination may have improved the estimation accuracy. This underlines the
importance of uncertainty estimates in any methodology used to determine organ dose. Uncertainties
in organ dose estimates are discussed in section 4.
3.6 Monte Carlo Simulation to Model the Dose Deposited by a CT Scanner

It may be assumed that accuracy can be improved for patient-specific computational phantoms by
matching the actual high-resolution imaging data, at least for organs and tissues completely inside the
actual imaged volume. However, the segmentation of all organs and tissues necessary for organ dose
determination is very challenging, regardless if a manual or automatic approach is considered. Fur-
thermore, scatter, overbeaming, and overranging outside the imaged and modeled volumes (partially
or completely) may contribute substantial dose to these organs. Computational phantom models do
not have the problem of estimating organ doses outside the imaged volume, but they are limited in
their ability to mimic an individual patient. Finally, some small organs are varied in their position in
the body (e.g., the location of the ovaries varies within the pelvis). Hence, computational phantoms
may not accurately represent the location of such small organs for a given patient.
Recently, Kalender et al. used physical anthropomorphic phantoms—an adult Rando-Alderson
phantom (RANDO Phantoms Laboratory, New York, USA), 1-year-old, and 5-year-old child CIRS
phantoms (Computerized Imaging Reference Systems, VA, USA)—to explore the effects of scatter
and overranging on organ dose.99 The physical phantoms were imaged emulating thorax scans, and
organ doses to various sites were measured with the use of TLDs. Whole-body voxel models of the
physical phantoms were built from the CT images appended by data from computational phantom
models (ICRP and ORNL phantoms, respectively).
Figure 6 shows the dose distributions from a thorax scan achieved by Monte Carlo simulation of
the CT images and the whole-body volumes, respectively, for the adult Rando-Alderson phantom.99
The effect of scatter and overranging in the simulated spiral mode thorax scan is obvious in the
graphical representation of dose distribution in Figure 6. Organs outside the imaged volume, e.g., the
liver and thyroid are examples of organs with a too low of absorbed dose if calculations use only the
CT imaged volume. Furthermore, organs and tissues inside the imaged volume are also attributed less
absorbed dose due to neglect of scatter from the volume(s) on either side of the scan, as seen when
comparing Figure 2.5 (c) and (d), i.e., the CT image volume model assumes that the body ends where
the image data starts and stops, respectively.
26
Figure 2.5. dose distributions from a thorax scan for CT image (a, c) and whole-body (b, d)
Figure 6. Monte Carlo-simulated
volume, by Kalender et al.99
Kalender et al. compared the Monte Carlo results to TLD measurements and found that the CT
image model underestimated organ dose by 8% to 15% on average (up to 37%), while results from the
whole-body model were within 10% of the TLD measurements.99
While the work by Kalender et al. was performed on physical phantoms instead of patients, and
quite basic computational models were used to create the whole-body models, it shows the possibili-
ties for future dosimetry approaches.99 This refined modeling concept and strategy may be used for
automatic organ volume segmentation in advanced patient-specific models, incorporation of sophisti-
cated CT scanner models in Monte Carlo simulations, automatic reporting of input data into Monte
Carlo radiation transport code, and refined modeling with DICOM metadata for automatic input of
patient and CT scanner related parameters.
3.7 Benchmarking and Validation of Monte Carlo Simulations
Monte Carlo simulations are inherently based on models of CT scanners, patients, and examinations.
While the modeling can be quite detailed in many instances, a necessary step prior to reporting organ
doses should be the benchmarking and validation of the Monte Carlo simulation results under known
(e.g., phantom) conditions. There have been several approaches to this, including the creation and use
of reference datasets described in AAPM Report 195.100 In addition, there have been several efforts
describing the validation of Monte Carlo simulation results by comparing them to physical measure-
ments in phantoms and even in vivo measurements (e.g., TLDs used during CT colonography).41,42
AAPM Task Group 246 and EFOMP recommend that the steps used to benchmark and validate
Monte Carlo simulation results be reported together with organ dose estimates, according to publica-
tions by AAPM Task Groups 195 and 268.100,101
4. Sources of Uncertainty in Estimating Organ Dose

Among various dose metrics, organ dose is generally regarded as one of the best metrics to quantify
individual radiation burden. Since all methods of calculating organ dose estimates are inherently asso-
ciated with some amount of uncertainty, the factors contributing to the uncertainties associated with
27
different organ dose estimation methods need to be elucidated and understood. Variations of organ
dose estimates for the methods discussed in this report may range up to 50% from a single source.
While this uncertainty is small compared to that associated with the estimation of individual risk from
specified irradiations ( a factor of 3),102,103 a quantitative estimate of uncertainty provides a better
understanding of the limitations of current organ dose estimation methods and indicates areas for
which further research is needed.
AAPM Task Group 246 and EFOMP recommend the inclusion of percent uncertainty data for
estimates of organ dose, as well as any underlying assumptions included in the methodology that may
yield additional uncertainty. This uncertainty in dose should be determined in as rigorous a manner as
possible.104
4.1 Uncertainties in SSDE–based Calculations of Patient Organ Dose
Table 3 contains examples of sources of uncertainties in SSDE estimates of patient organ dose dis-
cussed in this report. Estimated uncertainties given as percentages are to be understood as indications
of the present knowledge. More work is needed to better understand the uncertainties associated with
organ dose estimates for CT examinations.
4.1.1 Pencil Ionization Chamber Measurements and CTDIvol
The pencil ionization chamber is an accepted and accurate dosimeter when calibrated, and as used in
the IEC methodology for quantifying the CTDIvol. The physicist can provide improved accuracy when
reporting patient organ dose that is based on an independent measure of CTDIvol, and consequently
also the SSDE, as compared to using the manufacturer’s reported values of CTDIvol (DICOM RDSR
or Dose Data Page) and their associated tolerances (Table 1).
4.1.2 The SSDE
The SSDE is an appropriate estimate of dose to internal organs fully located within the scanned
region. In general, for a constant tube current examination, a physicist can calculate an organ dose
estimate for fully irradiated organs for a patient who is centered in the gantry using the SSDE and
achieve an uncertainty within 20%.
Table 3: Summary of selected sources and level of uncertainties that can be

expected in SSDE estimates of patient organ dose
Source of Uncertainty Contributing Factors Estimated Uncertainty ()

CTDIvol measured by the Small differences in phantom dimensions, differences Up to 5% for measurements made by the same
physicist using a calibrated in the energy dependence and calibration reference individual, likely 10% to 20% across a range of
ionization chamber for dosimeters, reproducibility of measurements by measurements. The 15% to 40% tolerances used
individuals by the manufacturers likely overestimates the
uncertainty (see Table 1).
SSDE using geometric In the thorax, patient size is not an ideal surrogate of Up to 20%
measures of patient size patient attenuation due to the presence of large air
cavities
SSDE using attenuation-based This accounts for uncertainty in CTDIvol and patient Up to 10%
measures of patient size size
SSDE and TCM using the This accounts for uncertainty in CTDIvol and patient Up to 20%
concept of regional CTDIvol size
Assuming that the patient is Bow-tie filter modulates the x-ray beam(s) that Up to 50%, depending on the actual patient centering
at isocenter exposes the patient; if the CT localizer radiograph is achieved 105
used to estimate patient size for SSDE calculations,
the size may be over- or underestimated
28
4.1.3 SSDE and TCM Examinations

The uncertainty introduced into organ dose estimates from the SSDE by TCM along the z-axis can be
reduced by using regional values of the CTDIvol that best describe the irradiation over the organ of
interest. However, the SSDE will not properly account for organ dose variation at the patient periph-
ery due to non-uniform irradiation from TCM in the (rotational) xy-plane.
4.1.4 Patient Not Centered in the Gantry
The CTDIvol, and consequently the SSDE, may be a poor predictor of organ dose when a patient is
positioned either too high or low in the gantry, due to the impact of the bow-tie filter modulation on
the x-ray beam, and over- or under-estimation of patient size using a CT localizer radiograph. For rou-
tine examinations, the SSDE may under- or overestimate organ dose by as much as 50% when a
patient is not centered in the gantry.105
4.1.5 The SSDE and Partial Organ Irradiation
The SSDE does not yield accurate estimates of dose for partially irradiated organs. Monte Carlo esti-
mates of organ dose, which are currently considered the gold standard, should be used when tasked
with determining dose to partially irradiated organs.
4.2 Uncertainties in Monte Carlo Estimates of Organ Dose
Over the past decade, significant progress has been made in quantifying organ dose with various tech-
niques that make use of Monte Carlo methods.106–111 The estimation accuracy with such methodology
is critically dependent on how well the model represents the individual patient under consideration
and the actual examination exposure conditions. Table 4 offers a summary of the sources of uncer-
tainty in Monte Carlo estimates of organ dose. Estimated uncertainties given as percentages are to be
understood as indications of the present knowledge.
4.2.1 Computational Phantoms
The uncertainties associated with using computational phantoms have been previously reported in
several studies. Zhang et al. assessed the organ dose uncertainties associated with four different com-
putational phantoms (ICRP, CT-Expo, XCAT, and ImPACT) for 10 body and three neurological CT
Table 4: Summary of the sources and level of uncertainties with computational phantoms
and Monte Carlo organ dose estimations
Source of Uncertainty Contributing Factors Estimated Uncertainty ()

Computational phantoms Depends on how accurately different types of computational 3–66%
phantoms resemble the anatomical structure of the actual
patient
Patient matching Induced by geometry differences between a patient and the 10–15%
computational phantom that is used to represent that patient
Organ start/end location Induced by the heterogeneous dose pattern created across <10% for most organs
the body by the helical trajectory of the CT source 10–33% for the small surface organs
TCM simplification Induced by using simplified tube current profiles 20% depending on the method used to
(z-dimensional) to approximate organ dose under TCM model the dose field under TCM
Monte Carlo simulation Mainly caused by the underlying differences in the physical 5–10%
models used by different codes
Assuming that the patient is at Bow-tie filter spatially changes the assumed x-ray beam if the Up to 50% depending upon the actual
isocenter patient is not centered patient centering 105
Organs partially irradiated E.g., thyroid, bone marrow, and skin Depends on the extent of partial organ
irradiation
29
protocols.112 With one single dose estimation technique used across all phantoms, the average percent-
age differences were in the range of 3% to 38% for fully irradiated organs and 7% to 66% for partially
irradiated organs. Sizable differences were found for organs located near the scan boundary, e.g., tes-
tes for abdominopelvic examination and colon for chest examination. Furthermore, noticeable uncer-
tainties were found for organs with different spatial distribution across phantoms, e.g., breasts for
female phantoms. Liu et al. compared the organ dose differences between RPI and ICRP reference
phantoms for chest, abdominopelvic, and chest-abdomen-pelvis protocols and found that the ratio
between the organ doses for the two types of phantoms were within the range of 0.75 to 1.16 for the
majority of fully irradiated organs.113 However, significant differences were found for organs near the
scan start/end locations. In both studies, uncertainties were mainly introduced by variation in organ
location and spatial distribution.
Tian et al. assessed the uncertainties associated with patient matching to computational phantoms
for chest and abdominopelvic examinations.75 The organ dose differences between the matched
patient pairs were on average 11% and 15% for chest and abdominopelvic examinations, respectively.
The largest uncertainties were found for small organs near the scan start/end regions, e.g., testes for
abdominopelvic examinations and thyroid for chest examinations.
4.2.2 Scanner Irradiation Conditions

The uncertainty associated with scanner irradiation conditions refers to modeling of the scanner irradi-
ation mode, including geometry and physical properties of the CT scanner, scanning collimation, start
and end tube angle positions, overranging, table and head holder attenuation, and the TCM technique.
Mostly associated with spiral CT, the uncertainties associated with tube start/end locations are
mainly induced by the helical trajectory and the incorporation of ‘pitch’ (table movement) relative to
the CT source. This geometry creates a periodical dose pattern across the patient periphery (see Figure
6). Zhang et al. and Li et al. have studied the effect of tube start/end location under different condi-
tions (e.g., pitch, collimation) for different patient models (infant, small child, adult female, and preg-
nant patient) and found that the largest dose variations occur for eye lens, thyroid, breasts, and testes,
all of which are at or near the surface of the patient.95 The uncertainties were in the range of 10% to
33% across different phantoms for the small surface organs.
Another main source of organ dose uncertainty is the modeling of TCM, which requires effective
quantification of dose field distribution created by the changing tube current. Khatonabadi et al. and
Li et al. demonstrated the use of a regional CTDIvol estimated by averaging the tube current values
within the organ region to approximate organ dose under TCM.31,77 The uncertainties associated with
such techniques were found to be generally within 20% for most organs, but not organs located in the
pelvic and shoulder regions.
In addition to the uncertainty associated with geometry and irradiation conditions of the scan,
uncertainty is associated with the statistical fluctuations with any Monte Carlo simulation and the
underlying differences in the physical models used by different implementations. The latter uncertain-
ties are generally small and within 5% to 10%. As the organ dose is an average over a large volume of
tissue, the uncertainties generally exceed those associated with the statistical uncertainties, which are
normally in the range of 1% to 2%.100
4.2.3 Patient Not Centered in the Gantry

Monte Carlo methods typically use a computational phantom centered in the simulated CT scanner
gantry, while the individual patient may or may not be centered in the gantry. Organ dose estimates
from Monte Carlo methods are thus associated with the same additional uncertainties as the SSDE
when the patient is not positioned at isocenter. If the bow-tie filter modulation of the x-ray beam is
30
used in the Monte Carlo model of a CT scanner, and the patient location in the gantry during an exam-
ination is reported, it appears possible to correct for this variability.114
4.3 Contrast Media Used in CT Examinations
Recently, Sahbaee et al. assessed the dose increase due to the presence of contrast media in CT exam-
inations and found substantial increase of estimated local radiation dose.115 Furthermore, Tran et al.
found an even larger increase of estimated local radiation dose for a standard clinical contrast-
enhanced body CT examination compared with a non-contrast examination.116 To what extent those
increased radiation dose values corresponded to increased radiation burden to biological tissue, as
opposed to radiation energy being absorbed only to the contrast medium alone, requires further inves-
tigation.
4.4 Reporting Uncertainty with Estimates of Patient Organ Dose
AAPM Task Group 246 and EFOMP recognize the importance of including an estimate of the uncer-
tainty in the computation with any report of patient organ dose estimates. It is possible for a reported
patient organ dose estimate to indicate which methods, considerations, and underlying assumptions
were used in the computation. While the spectrum of patient physical representations, CT scanners,
and examination-specific conditions create complex variations in patient exposures, the physicist
should report the calculation methods for patient organ dose estimates and indicate the efforts under-
taken to improve their accuracy. AAPM Task Group 246 and EFOMP recommend that patient organ
dose estimates should be reported with a measure of the associated uncertainty.
5. Summary of Report and Recommendations

The goal of this Joint Report of AAPM Task Group 246 and EFOMP was to assist the clinical medical
physicist in estimating individual patient organ doses from CT examinations. We are greatly aided in
this endeavor by the continuing efforts of the scientific medical physics community, as well as the
emergence and greater availability of the DICOM Radiation Dose Structured Report (RDSR). Fur-
thermore, refined and new dosimetry systems continue to evolve and form the basis for demonstrating
accuracy. Information systems capable of handling big data are emerging for large-scale patient dose
monitoring and quality assurance tools.
The CTDIvol provides the fundamental radiation dose metric for CT, but it has limitations for esti-
mating patient organ dose because of the lack of specificity regarding patient size that is included in
the CTDIvol concept.28 The normalized dose coefficients detailed in the AAPM Task Group 204 report
for computing the SSDE from examination averages of CTDIvol and patient size contain information
that increases the robustness of the SSDE for estimating patient organ dose for a given sized patient.
AAPM Task Group 246 and EFOMP endorse the recommendations made in AAPM Report 220 on
how the WED should be estimated.27
The dose to a patient from CT localizer radiographs, while small compared to routine CT exam-
ination doses, may be of interest for specific protocols (e.g., pediatric imaging or very-low-dose
screening applications). Currently, some manufacturers provide a CTDIvol value for the CT localizer
radiograph, even though the CTDI concept is not applicable to planar projection radiography.
AAPM Task Group 246 and EFOMP recommend that:
1. Reported CTDIvol values from CT localizer radiographs should not be used to estimate patient
doses and should particularly not be used to estimate SSDE. Patient dose from CT localizer
radiographs, which irradiate the patient using a narrow fan beam, should be estimated by inte-
gration of the entrance skin exposure using a phantom and geometry similar to that used in
standard radiography dose measurements.
31
Despite the robustness of the SSDE concept, it still has limits from a patient radiation dosime-
try perspective. The SSDE is not well suited for estimating dose from partial organ irradia-
tions, and it does not communicate information on the spatial distribution of dose. Even if
organs are fully contained in the primary x-ray beam, the SSDE only provides the average
dose to a certain volume or region, since it is based on a spatially weighted average of CTDI100
measurements (CTDIw). Peripheral and central organs in the same irradiated region cannot be
assumed to receive the same dose due to attenuation, adding uncertainty to patient organ dose
estimates based on the SSDE.
Organ dose estimates can be calculated by Monte Carlo methods, which incorporate distinct
parameters for CT scanners and computational phantoms that more closely match the patients.
As discussed in this report, modeling patients and CT scanners requires significant effort, and
the most refined solutions are thus presently found in proprietary software and research initia-
tives. Accurate input data improve the accuracy obtained from both Monte Carlo methods and
the SSDE calculations. Use of actual CTDIw measurements help make patient organ dose esti-
mates reflect an actual CT scanner and not the typical make and model, which is reflected in
the tabulated values supplied by a manufacturer in DICOM. Use of actual measurements may
decrease the uncertainty of the radiation output (CTDIw and CTDIvol), and consequently also
the uncertainty of organ dose estimates from 40% to 5%.
2. Given the present refinement of methods for estimating patient organ dose, AAPM Task
Group 246 and EFOMP recommend that efforts be made to perform uncertainty analysis of
the respective methods, and that reports of patient organ dose should be accompanied by doc-
umented estimates of the uncertainty.
Patient organ dose estimates from Monte Carlo methods and the SSDE also share the need of
pertinent examination-specific information found in DICOM (both in image header data and
the RDSR), particularly to be able to perform automatic computations. While manual compu-
tation is certainly possible, it decreases the applicability and limits the possibilities of the clin-
ical medical physicist to perform required tasks in optimization of radiation dose and image
quality.
3. In the interest of improving adoption of the SSDE formalism, AAPM Task Group 246 and
EFOMP support the additional recommendations of AAPM report 220, which specifically
recommends that manufacturers of CT scanners report the WED for reconstructed images,
e.g., 5 mm slice thickness, as part of the RDSR.
4. Furthermore, for purposes of Monte Carlo modeling, it is recommended that manufacturers
report tube current (mA) both in the longitudinal (e.g., for 5-mm slice thickness) and rota-
tional plane (e.g., every 5 degrees) to improve the gold standard for benchmarking more sim-
ple methods of estimating organ dose.
5. AAPM Task Group 246 and EFOMP also recommend that the steps used to benchmark and
validate Monte Carlo simulation results be reported together with organ dose estimates,
according to publications by AAPM Task Groups 195 and 268.100,101
AAPM Task Group 246 and EFOMP have identified several changes to the presently avail-
able DICOM information that will enhance the ability to estimate patient organ dose. Modifi-
cations and additions to DICOM procedural information were also noted, which are not
immediately relevant for estimating dose, but which have high importance for quality assur-
32
ance and longitudinal dose monitoring. Further collaboration with MITA, FDA, IEC, manu-
facturers of CT scanners, and the DICOM community will be crucial to advancing the
calculation and reporting of organ dose.
6. AAPM Task Group 246 and EFOMP recommend that a medical physicist should verify CT
scanner RDSR information as part of commissioning and acceptance testing. Furthermore,
RDSR verification is also advisable following CT scanner software upgrades.
7. Furthermore, familiarity with RDSR information and other DICOM content, as discussed in
this report, may be considered as newly added competence requirements for medical physi-
cists involved in optimization of CT examination protocols and image quality assurance.
6. References
1. American Association of Physicists in Medicine. AAPM Position Statement on Radiation Risks
from Medical Imaging Procedures (Policy 25-A). American Association of Physicists in Medi-
cine. https://www.aapm.org/org/policies/details.asp?type=PP&id=318. Published 2011.
Accessed February 5, 2018.
2. International Electrotechnical Commission. IEC 6060-2-44 Edition 3.2. Medical electrical equip-
ment – Part 2-44: Particular requirements for the basic safety and essential performance of X-ray
equipment for computed tomography. Geneva, Switzerland: 2016.
3. American Association of Physicists in Medicine. AAPM Committee Tree: Computed Tomogra-
phy Subcommittee. American Association of Physicists in Medicine. https://www.aapm.org/org/
structure/default.asp?committee_code=CTSC. Published 2004. Accessed February 5, 2018.
4. International Commission on Radiological Protection. (2007) The 2007 Recommendations of the
International Commission on Radiological Protection. Ann. ICRP. 37:1–332.
5. Jucius, R. A. and G. X. Kambic. (1977). Radiation dosimetry in computed tomography. Appl.
Opt. Instrum. Eng. Med. Phys. 127:286–95.
6. Shope, T. B., R. M. Gagne, and G. C. Johnson. (1991). A method for describing the doses deliv-
ered by transmission x-ray computed tomography. Med. Phys. 8:488–95.
7. ICRU Report No. 87: Radiation dose and image-quality assessment in computed tomography.
(2012.) J. ICRU 12(1):1–149.
8. American Association of Physicists in Medicine. Report of AAPM Task Group 96. The measure-
ment, reporting, and management of radiation dose in CT. American Association of Physicists in
Medicine. https://www.aapm.org/pubs/reports/RPT_96.pdf. Published 2008. Accessed February
6, 2018.
9. McCollough, C.H., S. Leng, L. Yu, D. D. Cody, J. M. Boone, and M. F. McNitt-Gray. (2011). CT
dose index and patient dose: they are not the same thing. Radiology 259(2):311–16.
10. McNitt-Gray, M. F. (2002). AAPM/RSNA Physics Tutorial for Residents: Topics in CT. Radia-
tion dose in CT. Radiographics 22(6):1541–53.
11. European Commission. EUR 16262 EN: European guidelines on quality criteria for computed
tomography. http://www.drs.dk/guidelines/ct/quality/index.htm. Published 2000. Accessed
February 19, 2018.
12. American Association of Physicists in Medicine. Report of AAPM Task Group 111. Comprehen-
sive Methodology for the Evaluation of Radiation Dose in X-ray Computed Tomography. A New
Measurement Paradigm Based on a Unified Theory for Axial, Helical, Fan-Beam, and Cone-
Beam Scanning With or Without Longitudinal Translation of the Patient Table. American Asso-
ciation of Physicists in Medicine. https://www.aapm.org/pubs/reports/RPT_111.pdf. Published
2010. Accessed February 6, 2018.
33
13. Boone, J. M. (2009). Dose spread functions in computed tomography: a Monte Carlo study. Med.
Phys. 36(10):4547–54.
14. Bahar Gogani, J., P. Hagglund, and G. Wickman. (2005). Assessment of correlated dose and sen-
sitivity profiles on a multi-slice CT scanner. Radiat. Prot. Dosim. 114(1–3):332–36.
15. Wickman, G., B. Johansson, J. Bahar-Gogani, T. Holmstrom, and J. E. Grindborg. (1998). Liquid
ionization chambers for absorbed dose measurements in water at low dose rates and intermediate
photon energies. Med. Phys. 25(6):900–07.
16. Jones, D. G. and P. C. Shrimpton. Report NRPB-R250: Survey of CT practice in the UK Part 3:
normalized organ doses calculated using Monte Carlo techniques. National Radiological Protec-
tion Board: 1991.
17. Jones, D. G. and B. F. Wall. Report NRPB-R186: Organ doses from medical x-ray examinations
calculated using Monte Carlo techniques National Radiological Protection Board: 1985.
18. Mathieu, K. B., M. F. McNitt-Gray, D. Zhang, H. J. Kim, and D. D. Cody. (2010). Precision of
dosimetry-related measurements obtained on current multidetector computed tomography scan-
ners. Med. Phys. 37(8):4102–09.
19. McCollough, C. H. and F. E. Zink. (1999). Performance evaluation of a multi-slice CT system.
Med. Phys. 26(11):2223–30.
20. International Atomic Energy Agency. IAEA Human Health Reports No. 5. Status of Computed
Tomography Dosimetry for Wide Cone Beam Scanners. Vienna, Austria: 2011.
21. Platten, D. J., I. A. Castellano, C. L. Chapple, et al. (2013). Radiation dosimetry for wide-beam
CT scanners: recommendations of a working party of the Institute of Physics and Engineering in
Medicine. Br. J. Radiol. 86(1027):20130089.
22. Bauhs, J. A., T. J. Vrieze, A. N. Primak, M. R. Bruesewitz, and C. H. McCollough. (2008). CT
dosimetry: comparison of measurement techniques and devices. Radiographics 28(1):245–53.
23. Dixon, R. L. and J. M. Boone. (2014). Stationary table CT dosimetry and anomalous scanner-
reported values of CTDIvol. Med. Phys. 41(1):011907.
24. Bakalyar, D., E. Angel, J. Boone, et al. Making Proper Use of the ICRU/AAPM CT Dose Phan-
tom: Recommendations and Limitations. Paper presented at: Radiological Society of North
America Scientific Assembly and Annual Meeting 2015, Chicago, IL.
25. Christner, J. A., V. A. Zavaletta, C. D. Eusemann, A. I. Walz-Flannigan, and C. H. McCollough.
(2010). Dose reduction in helical CT: dynamically adjustable z-axis X-ray beam collimation.
AJR Am. J. Roentgenol. 194(1):W49–55.
26. Digital Imaging and Communications in Medicine (DICOM). The DICOM Standard. https://
www.dicomstandard.org/current/. Accessed February 19, 2018.
27. American Association of Physicists in Medicine. Report of AAPM Task Group 220. Use of
Water Equivalent Diameter for calculating patient size and Size-Specific Dose Estimates (SSDE)
in CT. American Association of Physicists in Medicine. https://www.aapm.org/pubs/reports/
RPT_220.pdf. Published 2012. Accessed February 5, 2018.
28. American Association of Physicists in Medicine. Report of AAPM Task Group 204. Size-Spe-
cific Dose Estimates (SSDE) in Pediatric and Adult Body CT Examinations. American Associa-
tion of Physicists in Medicine. https://www.aapm.org/pubs/reports/RPT_204.pdf. Published
2012. Accessed February 5, 2018.
29. Leng, S., M. Shiung, X. Duan, L. Yu, Y. Zhang, and C. H. McCollough. Size Specific Dose Esti-
mation in Abdominal CT: Impact of Longitudinal Variations in Patient Size. Paper presented at:
55th Annual Meeting & Exhibition of the AAPM 2013, Indianapolis, IN.
30. Bostani, M., K. McMillan, P. Lu, et al. (2015). Attenuation-based size metric for estimating
organ dose to patients undergoing tube current modulated CT exams. Med. Phys. 42(2):958–68.
34
31. Khatonabadi, M., H. J. Kim, P. Lu, et al. (2013). The feasibility of a regional CTDIvol to esti-
mate organ dose from tube current modulated CT exams. Med. Phys. 40(5):051903.
32. Hardy, A., M. Bostani, C. H. Cagnon, and M. McNitt-Gray. Comparison of Size-Specific Dose
Estimate Conversion Factors for Fixed Tube Current and Tube Current Modulated Computed
Tomography. Paper presented at: Radiological Society of North America Scientific Assembly
and Annual Meeting 2016, Chicago, IL.
33. McMillan, K. Estimating Radiation Dose Metrics for Patients Undergoing Tube Current Modula-
tion CT Scans. University of California, Los Angeles, 2015.
34. McMillan, K., M. Bostani, L. Yu, et al. Size-Specific Effective Dose Estimates in Abdominal and
Chest CT Exams Using Either Fixed Tube Current or Tube Current Modulation. Paper presented
at: Radiological Society of North America Scientific Assembly and Annual Meeting 2015, Chi-
cago, IL.
35. Wang, J., J. A. Christner, X. Duan, S. Leng, L. Yu, and C. H. McCollough. (2012). Attenuation-
based estimation of patient size for the purpose of size specific dose estimation in CT. Part II.
Implementation on abdomen and thorax phantoms using cross sectional CT images and scanned
projection radiograph images. Med. Phys. 39(11):6772–78.
36. McMillan, K., M. Bostani, C. H. Cagnon, et al. (2017). Estimating patient dose from CT exams
that use automatic exposure control: Development and validation of methods to accurately esti-
mate tube current values. Med. Phys. 44(8):4262–75.
37. Moore, B. M., S. L. Brady, A. E. Mirro, and R. A. Kaufman. (2014). Size-specific dose estimate
(SSDE) provides a simple method to calculate organ dose for pediatric CT examinations. Med.
Phys. 41(7):071917.
38. McMillan, K., M. Bostani, C. Cagnon, M. Zankl, A. R. Sepahdari, and M. McNitt-Gray. (2014).
Size-specific, scanner-independent organ dose estimates in contiguous axial and helical head CT
examinations. Med. Phys. 41(12):121909.
39. Hardy, A. J., M. Bostani, A. M. Hernandez, et al. (2019). Estimating a size-specific dose for heli-
cal head CT examinations using Monte Carlo simulation methods. Med. Phys. 46(2):902–12.
40. Petoussi-Henss, N., M. Zanki, U. Fill, and D. Regulla. (2002). The GSF family of voxel phan-
toms. Phys. Med. Biol. 47(1):89–106.
41. Bostani, M., K. McMillan, J. J. DeMarco, C. H. Cagnon, and M. F. McNitt-Gray. (2014). Valida-
tion of a Monte Carlo model used for simulating tube current modulation in computed tomogra-
phy over a wide range of phantom conditions/challenges. Med. Phys. 41(11):112101.
42. Bostani, M., J. W. Mueller, K. McMillan, et al. (2015). Accuracy of Monte Carlo simulations
compared to in-vivo MDCT dosimetry. Med. Phys. 42(2):1080–86.
43. Khatonabadi, M., D. Zhang, K. Mathieu, et al. (2012). A comparison of methods to estimate
organ doses in CT when utilizing approximations to the tube current modulation function. Med.
Phys. 39(8):5212–28.
44. McCollough, C. H., J. Wang, and L. L. Berland. (2011). Bismuth shields for CT dose reduction:
do they help or hurt? J. Am. Coll. Radiol. 8(12):878–79.
45. Wang, J., X. Duan, J. A. Christner, S. Leng, K. L. Grant, and C. H. McCollough. (2012). Bismuth
shielding, organ-based tube current modulation, and global reduction of tube current for dose
reduction to the eye at head CT. Radiology 262(1):191–98.
46. Wang, J., X. Duan, J. A. Christner, S. Leng, L. Yu, and C. H. McCollough. (2011). Radiation
dose reduction to the breast in thoracic CT: comparison of bismuth shielding, organ-based tube
current modulation, and use of a globally decreased tube current. Med. Phys. 38(11):6084–92.
47. Li, X., K. Yang, and B. Liu. (2016). A study of the midpoint dose to CTDIvol ratio: Implications
for CT dose evaluation. Med. Phys. 43(11):5878.
35
48. Sinclair, L., A. Mench, T. Griglock, S. Bidari, and M. Arreola. Validation of Size-Specific Dose
Estimates (SSDE) in Body CT Studies with Directly Measured Organ Doses in Adult Female
Cadaveric Subjects. Paper presented at: 55th Annual Meeting of the American Association of
Physicists in Medicine 2013, Indianapolis, IN.
49. Boone, J. M. (1999). Glandular breast dose for monoenergetic and high-energy X-ray beams:
Monte Carlo assessment. Radiology 213(1):23–37.
50. Boone, J. M. (2009). Normalized glandular dose (DgN) coefficients for arbitrary X-ray spectra in
mammography: computer-fit values of Monte Carlo derived data. Med. Phys. 29(5):869–75.
51. Boone, J. M., A. L. Kwan, J. A. Seibert, N. Shah, K. K. Lindfors, and T. R. Nelson. (2005). Tech-
nique factors and their relationship to radiation dose in pendant geometry breast CT. Med. Phys.
32(12):3767–76.
52. Dance, D. R. (1990). Monte Carlo calculation of conversion factors for the estimation of mean
glandular breast dose. Phys. Med. Biol. 35(9):1211–19.
53. Dance, D. R., C. L. Skinner, K. C. Young, J. R. Beckett, and C. J. Kotre. (2000). Additional fac-
tors for the estimation of mean glandular breast dose using the UK mammography dosimetry pro-
tocol. Phys. Med. Biol. 45(11):3225–40.
54. Dance, D. R., K. C. Young, and R. E. van Engen. (2009). Further factors for the estimation of
mean glandular dose using the United Kingdom, European and IAEA breast dosimetry protocols.
Phys. Med. Biol. 54(14):4361–72.
55. Dance, D. R., K. C. Young, and R. E. van Engen. (2011). Estimation of mean glandular dose for
breast tomosynthesis: factors for use with the UK, European and IAEA breast dosimetry proto-
cols. Phys. Med. Biol. 56(2):453–71.
56. Hammerstein, G. R., D. W. Miller, D. R. White, M. E. Masterson, H. Q. Woodard, and
J. S. Laughlin. (1979). Absorbed radiation dose in mammography. Radiology 130(2):485–91.
57. Sechopoulos, I. and C. J. D’Orsi. (2008). Glandular radiation dose in tomosynthesis of the breast
using tungsten targets. J. Appl. Clin. Med. Phys. 9(4):2887.
58. Sechopoulos, I., S. S. Feng, and C. J. D’Orsi. (2010). Dosimetric characterization of a dedicated
breast computed tomography clinical prototype. Med. Phys. 37(8):4110–20.
59. Sechopoulos, I., S. Suryanarayanan, S. Vedantham, C. D’Orsi, and A. Karellas. (2007). Compu-
tation of the glandular radiation dose in digital tomosynthesis of the breast. Med. Phys.
34(1):221–32.
60. Khatonabadi, M., D. Zhang, J. Yang, J. J. Demarco, C. C. Cagnon, and M. McNitt-Gray. (2012).
The relationship between organ dose and patient size in tube current modulated adult thoracic CT
scans. Proc. SPIE 8313:83131Q.
61. Wang, J., X. Duan, J. A. Christner, S. Leng, L. Yu, and C. H. McCollough. (2012). Attenuation-
based estimation of patient size for the purpose of size specific dose estimation in CT. Part I.
Development and validation of methods using the CT image. Med. Phys. 39(11):6764–71.
62. Mueller, J. W., D. J. Vining, A. K. Jones, et al. (2014). In vivo CT dosimetry during CT colonog-
raphy. AJR Am. J. Roentgenol. 202(4):703–10.
63. Angel, E., C. V. Wellnitz, M. M. Goodsitt, et al. (2008). Radiation dose to the fetus for pregnant
patients undergoing multidetector CT imaging: Monte Carlo simulations estimating fetal dose for
a range of gestational age and patient size. Radiology 249(1):220–27.
64. Huda, W., D. Magill, and M. V. Spampinato. (2011). Technical note: estimating absorbed doses
to the thyroid in CT. Med. Phys. 38(6):3108–13.
65. Huda, W., M. V. Spampinato, S. V. Tipnis, and D. Magill. (2013). Computation of thyroid doses
and carcinogenic radiation risks to patients undergoing neck CT examinations. Radiat. Prot.
Dosimetry 156(4):436–44.
36
66. Zhang, D., C. H. Cagnon, J. P. Villablanca, et al. (2012). Peak skin and eye lens radiation dose
from brain perfusion CT based on Monte Carlo simulation. AJR Am. J. Roentgenol. 198(2):412–
17.
67. Lopez-Rendon, X., A. Stratis, W. Coudyzer, W. Develter, H. Bosmans, and F. Zanca. Dose to the
Eye Lens and Skin in CT Perfusion Exams. Paper presented at: 102nd Radiological Society of
North America Scientific Assembly and Annual Meeting 2016, Chicago, IL.
68. Zhang, D., C. H. Cagnon, J. P. Villablanca, et al. (2013). Estimating peak skin and eye lens dose
from neuroperfusion examinations: use of Monte Carlo based simulations and comparisons to
CTDIvol, AAPM Report No. 111, and ImPACT dosimetry tool values. Med. Phys. 40(9):091901.
69. U.S. Food and Drug Administration. Radiation Dose Quality Assurance: Questions and Answers.
https://www.fda.gov/Radiation-EmittingProducts/RadiationSafety/RadiationDoseReduction/
ucm232550.htm. Updated 12/2/2017. Accessed February 7, 2018.
70. de las Heras, H., R. Minniti, S. Wilson, et al. (2013). Experimental estimates of peak skin dose
and its relationship to the CT dose index using the CTDI head phantom. Radiat. Prot. Dosimetry
157(4):536–42.
71. Leng, S., T. Vrieze, L. Yu, and C. McCollough. Skin Dose Estimation from CT Perfusion Stud-
ies: Influence of Patient Size, Beam Collimation and Scanner Type. Paper presented at: 96th Sci-
entific Assembly and Annual Meeting of the Radiological Society of North America 2010,
Chicago, IL.
72. Dixon, R. L. (2003). A new look at CT dose measurement: beyond CTDI. Med. Phys.
30(6):1272–80.
73. Dixon, R. L. and J. M. Boone. (2011). Analytical equations for CT dose profiles derived using a
scatter kernel of Monte Carlo parentage with broad applicability to CT dosimetry problems.
Med. Phys. 38(7):4251–64.
74. Dixon, R. L., J. M. Boone, and R. A. Kraft. (2014). Dose equations for shift-variant CT acquisi-
tion modes using variable pitch, tube current, and aperture, and the meaning of their associated
CTDI(vol). Med. Phys. 41(11):111906.
75. Tian, X., W. P. Segars, R. L. Dixon, and E. Samei. (2016). Convolution-based estimation of
organ dose in tube current modulated CT. Phys. Med. Biol. 61(10):3935–54.
76. Segars, W. P., J. Bond, J. Frush, et al. (2013). Population of anatomically variable 4D XCAT
adult phantoms for imaging research and optimization. Med. Phys. 40(4):043701.
77. Li, X., W. P. Segars, and E. Samei. (2014). The impact on CT dose of the variability in tube cur-
rent modulation technology: a theoretical investigation. Phys. Med. Biol. 59(16):4525–48.
78. Jansen, J. T. and P. C. Shrimpton. (2011). Calculation of Normalised Organ and Effective Doses
to Adult Reference Computational Phantoms from Contemporary Computed Tomography Scan-
ners. Prog. Nucl. Sci. Technol. 2:165–71.
79. Jansen, J. T. and P. C. Shrimpton. Normalized doses for various reference patients and a range of
computed tomography scanners. Paper presented at: Proceedings of CP 2013: 4th International
Workshop on Computational Phantoms for Radiation Protection, Imaging and Radiotherapy
2013.
80. Li, X., E. Samei, T. Yoshizumi, J. G. Colsher, R. P. Jones, and D. P. Frush. (2007). Experimental
benchmarking of a Monte Carlo dose simulation code for pediatric CT. Proc. SPIE
6510:65102A.
81. Li, X., D. Zhang, and B. Liu. (2012). Estimation of the weighted CTDI(infinity) for multislice
CT examinations. Med. Phys. 39(2):901–05.
82. Oono, T., F. Araki, S. Tsuduki, and K. Kawasaki. (2014). Monte Carlo calculation of patient
organ doses from computed tomography. Radiol. Phys. Technol. 7(1):176–82.
37
83. van Straten, M., P. Deak, P. C. Shrimpton, and W. A. Kalender. (2009). The effect of angular and
longitudinal tube current modulations on the estimation of organ and effective doses in x-ray
computed tomography. Med. Phys. 36(11):4881–89.
84. Zhang, Y., X. Li, W. P. Segars, and E. Samei. (2013). Comparative dosimetry of radiography,
tomosynthesis, and CT for chest imaging across 59 adult patients. Proc. SPIE 8668:866844.
85. Stepusin, E. J., D. J. Long, K. R. Ficarrotta, D. E. Hintenlang, and W. E. Bolch. (2017). Physical
validation of a Monte Carlo-based, phantom-derived approach to computed tomography organ
dosimetry under tube current modulation. Med. Phys. 44(10):5423–32.
86. Stepusin, E. J., D. J. Long, E. L. Marshall, and W. E. Bolch. (2017). Assessment of different
patient-to-phantom matching criteria applied in Monte Carlo-based computed tomography
dosimetry. Med. Phys. 44(10):5498–508.
87. International Commission on Radiological Protection. (2010). Publication No. 116. Conversion
coefficients for radiological protection quantities for external radiation exposures. Ann. ICRP
40:1–257.
88. Johnson, P. B., A. A. Bahadori, K. F. Eckerman, C. Lee, and W. E. Bolch. (2011). Response
functions for computing absorbed dose to skeletal tissues from photon irradiation—an update.
Phys. Med. Biol. 56(8):2347–65.
89. International Commission on Radiological Protection. (1975). Publication No. 23. Report of the
Task Group on Reference Man. Ann. ICRP 23:1–480.
90. International Commission on Radiological Protection. (2002). Publication No. 89. Basic Ana-
tomical and Physiological Data for Use in Radiological Protection Reference Values. Ann. ICRP
32(3–4):1–277.
91. Turner, A. C., D. Zhang, H. J. Kim, et al. (2009). A method to generate equivalent energy spectra
and filtration models based on measurement for multidetector CT Monte Carlo dosimetry simula-
tions. Med. Phys. 36(6):2154–64.
92. Caon, M., G. Bibbo, and J. Pattison. (1997). A comparison of radiation dose measured in CT
dosimetry phantoms with calculations using EGS4 and voxel-based computational models. Phys.
Med. Biol. 42(1):219–29.
93. Jarry, G., J. J. DeMarco, U. Beifuss, C. H. Cagnon, and M. F. McNitt-Gray. (2003). A Monte
Carlo-based method to estimate radiation dose from spiral CT: from phantom testing to patient-
specific models. Phys. Med. Biol. 48(16):2645–63.
94. Zhang, D., A. S. Savandi, J. J. Demarco, et al. (2009). Variability of surface and center position
radiation dose in MDCT: Monte Carlo simulations using CTDI and anthropomorphic phantoms.
Med. Phys. 36(3):1025–38.
95. Zhang, D., M. Zankl, J. J. DeMarco, et al. (2009). Reducing radiation dose to selected organs by
selecting the tube start angle in MDCT helical scans: a Monte Carlo based study. Med. Phys.
36(12):5654–64.
96. Angel, E., N. Yaghmai, C. M. Jude, et al. (2009). Dose to radiosensitive organs during routine
chest CT: effects of tube current modulation. AJR Am. J. Roentgenol. 193(5):1340–45.
97. Angel, E., N. Yaghmai, C. M. Jude, et al. (2009). Monte Carlo simulations to assess the effects of
tube current modulation on breast dose for multidetector CT. Phys. Med. Biol. 54(3):497–512.
98. Jansen, J. T. and P. C. Shrimpton. Calculation of Normalized Organ Doses for Pediatric Patients
Undergoing CT Examinations on Four Types of CT Scanner. Paper presented at: World Congress
on Medical Physics and Biomedical Engineering 2009, Munich, Germany.
99. Kalender, W. A., N. Saltybaeva, D. Kolditz, M. Hupfer, M. Beister, and B. Schmidt. (2014).
Generating and using patient-specific whole-body models for organ dose estimates in CT with
increased accuracy: feasibility and validation. Phys. Med. 30(8):925–33.
38
100. American Association of Physicists in Medicine. Electronic Resources of the TG 195 Report.
Monte Carlo Reference Data Sets for Imaging Research. American Association of Physicists in
Medicine. https://www.aapm.org/pubs/reports/report195.asp. Accessed February 7, 2018.
101. Sechopoulos, I., D. W. O. Rogers, M. Bazalova-Carter, et al. (2018). RECORDS: improved
Reporting of montE CarlO RaDiation transport Studies. Int. J. Radiat. Oncol. Biol. Phys.
101(4):792–93.
102. Martin, C. J. (2007). Effective dose: how should it be applied to medical exposures?
Br. J. Radiol. 80(956):639–47.
103. Martin, C. J. (2008). The application of effective dose to medical exposures. Radiat. Prot.
Dosimetry 128(1):1–4.
104. Joint Committee for Guides in Metrology (JCGM). JCGM 100:2008. Evaluation of measurement
data - Guide to the expression of uncertainty in measurement https://www.bipm.org/utils/com-
mon/documents/jcgm/JCGM_100_2008_E.pdf. Published 2008. Accessed February 19, 2018.
105. Zanca, F., A. Jacobs, W. Crijns, and W. De Wever. (2014). Comparison of measured and esti-
mated maximum skin doses during CT fluoroscopy lung biopsies. Med. Phys. 41(7):073901.
106. Ding, A., M. M. Mille, T. Liu, P. F. Caracappa, and X. G. Xu. (2012). Extension of RPI-adult
male and female computational phantoms to obese patients and a Monte Carlo study of the effect
on CT imaging dose. Phys. Med. Biol. 57(9):2441–59.
107. Geyer, A. M., S. O’Reilly, C. Lee, D. J. Long, and W. E. Bolch. (2014). The UF/NCI family of
hybrid computational phantoms representing the current US population of male and female chil-
dren, adolescents, and adults--application to CT dosimetry. Phys. Med. Biol. 59(18):5225–42.
108. Samei, E., X. Tian, and W. P. Segars. (2014). Determining organ dose: the holy grail. Pediatr.
Radiol. 44 Suppl 3:460–67.
109. Schlattl, H., M. Zankl, J. Becker, and C. Hoeschen. (2010) Dose conversion coefficients for CT
examinations of adults with automatic tube current modulation. Phys. Med. Biol. 55(20):6243–
61.
110. Tian, X., X. Li, W. P. Segars, D. P. Frush, E. K. Paulson, and E. Samei. (2013). Dose coefficients
in pediatric and adult abdominopelvic CT based on 100 patient models. Phys. Med. Biol.
58(24):8755–68.
111. Turner, A. C., D. Zhang, M. Khatonabadi, et al. (2011). The feasibility of patient size-corrected,
scanner-independent organ dose estimates for abdominal CT exams. Med. Phys. 38(2):820–29.
112. Zhang, Y., X. Li, W. P. Segars, and E. Samei. (2012). Organ doses, effective doses, and risk indi-
ces in adult CT: comparison of four types of reference phantoms across different examination
protocols. Med. Phys. 39(6):3404–23.
113. Liu, H., J. Gu, P. F. Caracappa, and X. G. Xu. (2010). Comparison of two types of adult phan-
toms in terms of organ doses from diagnostic CT procedures. Phys. Med. Biol. 55(5):1441–51.
114. Kallman, H. E., R. Holmberg, J. Andersson, L. Kull, E. Traneus, and A. Ahnesjo. (2016). Source
modeling for Monte Carlo dose calculation of CT examinations with a radiotherapy treatment
planning system. Med. Phys. 43(11):6118.
115. Sahbaee, P., W. P. Segars, and E. Samei. Multi-phase CT: Impact of Contrast Medium Propaga-
tion on Radiation Dose across a Population of Patient Models. Paper presented at: Radiological
Society of North America 2014 Scientific Assembly and Annual Meeting 2014, Chicago, IL.
116. Tran, H., C. Lee, V. Derderian, L. Folio, and E. Jones. Estimating the Role of Iodinated IV Con-
trast Media in Organ Radiation Dose: Effects of Vascular Phase and Tube Voltage in Multiphase
Body CT. Paper presented at: Radiological Society of North America 2014 Scientific Assembly
and Annual Meeting 2014, Chicago, IL.
39
Appendix:
DICOM—Present and Future for Dosimetry

and Estimating Organ Dose
Determination of dose estimates, with associated quality assurance efforts, are now an established
informatics endeavor, with radiation dose being fully represented as part of the medical imaging elec-
tronic environment.1 This is currently possible with the use of the RDSR, which compiles patient and
examination data, technical, geometric, and dosimetric data for each radiation procedure, plus total
dose data for the complete procedure. This DICOM object is gaining wide availability for CT.
DICOM is the set of standards managed by the National Electrical Manufacturer’s Association
(NEMA) and is the key method of transferring medical image data. The continuing aim of the
DICOM Standard is to achieve a higher level of compatibility between manufacturers and systems for
important information content, thereby enabling an enhanced quality of service with improved work-
flow efficiency between and among imaging systems and other information systems in healthcare
environments worldwide.2–9
One of the components of the Medical Imaging and Technology Alliance (MITA) standard XR-29
known as MITA Smart Dose is that the CT scanner must be capable of generating a DICOM RDSR.10
Each manufacturer has set up a vendor certification web portal on the MITA Smart Dose website
(http://www.medicalimaging.org/policy-and-positions/mita-smart-dose/). For CT scanners in the
United States, compliance with all components of MITA XR-29 was required for full reimbursement
of outpatient scans by the Center for Medicare and Medicaid Systems (CMS). Without compliance
(effective Jan. 1, 2016), CMS reimbursement is reduced by 5% for outpatient scans in 2016 and by
15% in 2017. Note that compliance with XR-29 is established based on CT scanner capabilities, not
on usage. When used appropriately, these features improve the safety of CT scans for the patients and
can improve the image quality.
At a high level, radiation dose and related information may be found in three locations: the
DICOM tags associated with the produced images’ header data, a CT Dose Data Page (known by
other names, including the Patient Protocol page, Patient Dose page, etc.), and more recently the
RDSR.11 Quality assurance dose-related information is found in the RDSR, but it also includes
DICOM data from the Study, Series, and even Image level tags.
Specifically related to the CT radiation dose, DICOM has a CT Exposure Functional Group with
attributes dedicated to capture the CT radiation dose. Examples of existing DICOM attributes in
images relating to CT irradiation are listed in Table 2. However, most of these DICOM attributes are
“optional” in the DICOM Standard and may not be populated by all CT manufacturers, or some man-
ufacturers may have implemented this information in private tags. Other important parameters useful
for patient dose estimates, such as anatomy location, are currently not available in DICOM.
A.1 CT Radiation Dose Reporting from DICOM Objects

This section describes the DICOM content from two different sources, the Dose Data Page and
DICOM image header data.
A.1.1 The Dose Data Page
CT manufacturers initially provided a summary of dose information in a Dose Data Page several years
prior to the introduction of the DICOM RDSR. The Dose Data Page is an image of text information,
automatically created and saved as an image (Secondary Capture), with or without associated dose
information in the DICOM image headers. Since manufacturers introduced the Dose Data Page at dif-
40
ferent times, clinical scanners have different degrees of compliance with optional fields and even
required fields. Therefore, the availability of CT dose-related information is totally dependent on the
scanner model and the software version currently installed.
For legacy scanners (defined in this context as those devices having only the Dose Data Page), the
dose information may only be available from the screen capture once the examination is completed.
The Dose Data Page may be sent to the PACS (picture archiving and communication system) as one
of the scan series and archived in the image database. The Dose Data Page lists all imaging scan series
performed in the examination (even if those image series are not sent to the PACS), and it always
includes the CTDIvol and DLP information for each scan series.
To extract the CT radiation dose-related information directly from Dose Data Page images, Opti-
cal Character Recognition (OCR) software can be used to read the numerical values of CTDIvol or
DLP. 12
A.1.2 DICOM Image Header Data
CT manufacturers support the use of DICOM standards and have populated the radiation dose-related
attributes at the series and image level. This information may be electronically extracted from DICOM
image header information using readily available DICOM software. However, there are some funda-
mental limitations to the dose extraction using image-based DICOM tags. For example, two image
sets reconstructed with different image thickness could be generated from the same scan series, there-
fore the radiation dose parameter extracted from one series is actually duplicated by the other series.
DICOM images have an attribute that specifies ‘Image Type’ (0008,0008). Images that are created
from source data are designated original. These original images were created by an x-ray irradiation in
CT, and differ from those that are derived from the original data, which are called reprocessed images,
e.g., with a different reconstructed image thickness. However, if the CT creates an image series with
another reconstruction kernel, these images would also be designated as original. If using the image
header data only, this second series could be interpreted as an added dose if the Image Type tag is
used to differentiate irradiation events. Furthermore, it is possible that the irradiation took place, but
the resulting images were not stored on the scanner or sent to the PACS by the operator. Such events
would be captured on the Dose Data Page, but it is difficult to automatically associate reconstructed
image series with irradiation events.
Since 2008, RDSR has been considered the preferred method for storing and extracting radiation
dose related information. The RDSR is discussed in the following sections. RDSR is one member of a
more extensive DICOM Structured Report family. RDSR is based on templates to “store” information
from different functional groups, such as the CT Exposure Functional Group and CT Acquisition
Details Group.
A.2 Radiation Dose Structured Report (RDSR)—The DICOM Object

An RDSR object is structured as a composite object (combination of various information entities),
similar to objects in DICOM images. An RDSR encompasses information to relate the object to the
right patient, the right examination, and to uniquely identify information generated during the exam-
ination. An RDSR object is identified by its DICOM SOP Class identifier
“1.2.840.10008.5.1.4.1.1.88.67” and a unique DICOM Instance identifier (UID). An RDSR object is
like an image, with the major difference that it does not contain pixel data; instead it contains struc-
tured information organized in a “content tree.” Current PACS solutions rarely have an RDSR reader,
and some legacy PACS cannot even store the RDSR. Use of legacy scanners requires configuration of
the CT scanner with a dedicated DICOM storage node to receive Dose Data Pages. A custom DICOM
storage can be created for this purpose, or radiation exposure management software can be used. In
41
principle for CT procedures, the contents can be divided into three parts: (1) DICOM header, (2) dose
accumulation container, and (3) container holding the information for each irradiation event. Contain-
ers are embedded in a root template container called “X-ray Radiation Dose Report.” Together, they
form the content tree. The basic module composition of the RDSR is described in PS3.3 Information
Object Definitions in section A.35.8 “X-ray Radiation Dose SR Information Object Definition
(IOD)”.13
A.2.1 DICOM Structured Report Templates
A Template Identifier (TID) uniquely specifies a Structured Report Template that contains the struc-
ture and the rules on filling in the content tree for a specific report. In DICOM, the TID is described in
PS3.16 Content Mapping Resource.14 The top-level template for the CT RDSR is TID 10011 CT
Radiation Dose. All TIDs have a simple structure, similar to a table with each line, or row, specifying
one content item of the RDSR. Templates are ordered in hierarchical structure and are used to group
related content or for repetition (e.g., “CT Irradiation Event Data” for irradiation event-related con-
tent, repeated one to many times as a number of acquisitions or CT series are performed). For exam-
ple, when a physician steps on the pedal for CT Fluoroscopy 34 times, 34 irradiation events are
captured, and this occurs for all activation of radiation, including the CT localizer radiograph(s). The
present CT RDSR template structure is shown in Figure A1.12
TID 1003
TID 10011 TID 1002
Person Observer
CT Radia on Dose Observer Context
Iden fying A ributes
TID 1004
Device Observer
Iden fying A ributes
TID 10012 TID 1021

Accumulated Dose Data Device Par cipant
TID 1002 TID 10013

Observer Context Irradia on Dose Event
TID 10014
Scanning Length
TID 10015 TID 1020

TID 1020 CT Dose Check Details Person Par cipant
Person Par cipant
TID 1020
Person Par cipant
TID 1021
Device Par cipant
Figure A1. Schema of TIDs used in the CT RDSR.12
42
A.2.2 The Patient RDSR

The RDSR content discussed above is limited to information about the irradiation system or informa-
tion the system can determine, e.g., radiation output, geometry, x-ray source, detector system, etc.
This content is missing pertinent information about the patient that can be used to estimate organ dose.
To address this deficiency, the Patient RDSR (P-RDSR) has been added to the DICOM standard. The
P-RDSR includes four TIDs, 10031 through 10034, which are detailed below.
TID 10031 Radiation Dose Estimate
This contains the dose information for all organs that were included in the dose estimation. Dose esti-
mation to the whole body or a phantom are possible options that may be used in place of an organ to
allow for simple estimations or quality control measurements. Only absorbed dose and equivalent
dose are allowed values to be included in the P-RDSR, and a single value for each organ is required.
The type of metric this value represents (e.g., maximum, minimum, mean, etc.) is required. There is
also an option to include an estimate of the uncertainty in the values if known.
TID 10032 Radiation Dose Estimate Representation
This allows the dose to be stored as a separate DICOM IOD to better visualize the distribution of dose
in the patient. Some examples of the representation are as a surface segmentation IOD for skin dose
maps, or storing a point cloud of individual dose estimations within a patient for use in more advanced
dosimetry.
TID 10033 Radiation Dose Estimate Methodology
This contains the information used in the dose estimation method. The first item to note is that an
RDSR that stores the radiation technique factors and output is required. If the modality does not pro-
vide an RDSR, one that includes the assumed values must be created. This was done so the P-RDSR
did not need to repeat the values already stored in the RDSR. The P-RDSR also provides the ability to
list the values within the RDSR that were used in the dose estimation. This is needed if the dose esti-
mation excluded some exposures, e.g., scout/localizer or setup exposures. TID 10033 also includes the
patient demographics used by the method (e.g., sex, age, weight, and height). These are only the val-
ues used in the method, and need not be the actual patient demographics. This was included to allow
storage of the actual values used by the method that may be needed for unique patient habitus, etc. The
TID also includes the type of model used to represent the patient (e.g., simple geometric, anthropo-
morphic, or actual patient segmentation models). The model can also be stored as a separate DICOM
IOD, similar to the dose representation. It is hoped that this may promote the development of radiation
transport and dosimetry models to be stored as DICOM IOD for standard storage and use. The regis-
tration of the patient model to the RDSR from the modality is required. Because the current RDSR
does not include a frame of reference, the registration may be a simple visual registration, or it can use
fiducial markers or the frame of reference from the images produced to accomplish the registration.
TID 10034 Radiation Dose Estimate Parameters
This is used to store any method-specific parameters used in the estimation. A listing of the parame-
ters that may be included are specified in the DICOM Standard, and the values and units for each must
be provided.
A.3 Using the DICOM RDSR

A.3.1 RDSR Reader
A CT RDSR is organized as a “content tree” with at least three major components: DICOM header
information, a container for CT accumulated dose data (based on TID 10012), and another container
43
Figure A.2. CT RDSR tree-like structure with containers based on TID 10012 and 10013.
for individual CT irradiation dose event (based on TID 10013). CT RDSR itself is an encoded tree-
like representation “layered” on top of traditional DICOM header data (see Figure A2).
An RDSR reader (or viewer) is typically necessary to parse these datasets into a human-readable
format. However, data can be extracted from RDSRs in a number of ways:
• Command line tools exist that can present the entire object as text in the form of a DICOM
dump. These tools are very useful for troubleshooting and quick checks, but the hierarchy, repe-
tition, and volume of information makes them unsuitable for general clinical use.
• Some RDSR readers can present the content in a formatted way, as shown in Figure A3. How-
ever, in order to avoid these reports being too long and unwieldy, the authors of the software
may have selected the fields that they consider to be pertinent. The reports present only the pre-
selected types of data.
• Some software products can compile a collection of RDSRs and present selected data in a
spreadsheet in per study/per event detail and summarize the data over all the studies. Similar to
text reports, the spreadsheet does not present all data.
• Radiation dose management software configures RDSRs into a database. The data can be
viewed and manipulated, commonly in a web browser, and usually can be exported to a spread-
sheet. The extent of the stored data and user access varies with the software.
All aforementioned software can be obtained as open source and from commercial providers.15
The NEMA DICOM Working Group 28 acts as a liaison body to facilitate including data relevant to
the medical physics community in the DICOM standard used by manufacturers of x-ray equipment
and software providers.
44
X-Ray Radiation Dose Report

Patient ID: Name:
Birth date: Age:
Sex: Ethnic group:
Exam Date: Exam No.:
Accession Number: Content Date:
Weight: Body Surface Area:
Height: Body Mass Index:
Cardiologist: Referring Physician:
Completion flag: COMPLETE Verification flag: UNVERIFIED
Procedure reported Computed Tomography X-Ray
Has Intent Diagnostic Intent
Observer Type Device
Device Observer UID 73657
Device Observer Name PX_CT02RAA031791
Device Observer Manufacturer SIEMENS
Device Observer Model Name SOMATOM Definition Flash
Device Observer Serial Number 73657
Device Observer Physical
Location During Observation Mayo Clinic PXMH CT02_1665
Start of X-Ray Irradiation 2013-07-24, 09:40:10.010988
End of X-Ray Irradiation 2013-07-24, 09:53:48.372994
Scope of Accumulation Study
Study Instance UID 1.2.124.113532.172.16.49.225.20130724.91741.2035726
CT Accumulated Dose Data
Total Number of Irradiation Events 8.0 {events}
CT Dose Length Product 5,346.0 mGycm
CT Acquisition
Acquisition Protocol Topogram
Target Region Head
CT Acquisition Type Constant Angle Acquisition
Procedure Context CT without contrast
Irradiation Event UID 1.3.12.2.1107.5.1.4.73657.30000013072404121171200000040
CT Acquisition Parameters
Exposure Time 4.3 s
Scanning Length 418.0 mm
Nominal Single Collimation Width 0.6 mm
Number of X-Ray Sources 1.0 {X-Ray sources}
Figure A.3. An example of the plain text format from a CT RDSR reader.
A.3.2 Verification of the DICOM RDSR

A physicist needs to verify that a CT scanner has the capability of generating a correct RDSR as part
of the CT acceptance test, or as part of a software upgrade for RDSR functionality. A healthcare sys-
tem uses these data as a record of individual patient exposure, and all data should thus be validated. If
a permanent RDSR storage solution is not available at the point of acceptance, the RSDR should be
exported from the scanner along with the acceptance images, for example by exporting to a travel
drive or an external hard drive.
The exported data should be inspected and the accuracy validated using any of the tools in section
A.3.1, along with the manufacturer DICOM conformance statement. Technical values such as CTDIvol
and DLP should be compared with the displayed acquisition parameters, as well as administrative
fields, such as the institution name, scanner details, and protocol names. Since these details are often
missing or set to defaults, and some technical parameters such as exposure time have been found to be
completely wrong, automated linking of the various identifiers with the necessary variables should be
a top priority of software upgrades.
Familiarity with the RDSR and other DICOM content may be considered as a newly added com-
petence requirement for medical physicists involved in optimization and image quality assurance.
45
Appendix References
1. Wang, J., X. Duan, J. A. Christner, S. Leng, L. Yu, and C. H. McCollough CH. (2012). Attenua-
tion-based estimation of patient size for the purpose of size specific dose estimation in CT. Part I.
Development and validation of methods using the CT image. Med. Phys. 39(11):6764–71.
2. Digital Imaging and Communications in Medicine (DICOM). Change Proposals (1077, 1114,
1151, 1160, 1201, 1223, 1254). http://www.dicomstandard.org/cps/. Accessed February 15,
2018.
3. O’Donnell, K. Correction Proposal (CP-1077): Add CR report type to Dose SR and relax content
conditions. 2012. ftp://medical.nema.org/medical/dicom/final/cp1077_ft3.pdf. Accessed Febru-
ary 15, 2018.
4. Correction Proposal (CP-1114): Correct UCUM multiplication. 2011. ftp://medical.nema.org/
medical/dicom/final/cp1114_ft.pdf. Accessed February 15, 2018.
5. Hoehn, H. and H. Solomon H. Correction Proposal (CP-1160): Degree sign in UCUM. 2012.
ftp://medical.nema.org/medical/dicom/final/cp1160_ft.pdf. Accessed February 15, 2018.
6. Revet, B. and H. Blendinger. Correction Proposal (CP-1151): Correct condition for “number of
pulses” in TID 10003. 2011. ftp://medical.nema.org/medical/dicom/final/cp1151_ft.pdf.
7. Nolte, B. and H. Blendinger. Correction Proposal (CP-1201): Correct calibration factor CI defini-
tion (TID 10002). 2013. ftp://medical.nema.org/medical/dicom/final/cp1201_ft.pdf. Accessed
February 15, 2018.
8. Revet, B. and H. Blendinger. Correction Proposal (CP-1223): Additional items for dose SR (by
IEC PT 61910-1). 2013. ftp://medical.nema.org/medical/dicom/final/cp1223_ft2.pdf. Accessed
February 15, 2018.
9. Revet B. and H. Blendinger. Correction Proposal (CP-1254): Correct definition of irradiation
duration. 2013. ftp://medical.nema.org/medical/dicom/final/cp1254_ft.pdf. Accessed February
15, 2018.
10. National Electrical Manufacturers Association. NEMA XR 29-2013: Standard Attributes on CT
Equipment Related to Dose Optimization and Management. In: The Association of Electrical
Equipment and Medical Imaging Manufacturers. 2013.
11. Digital Imaging and Communications in Medicine (DICOM). Supplement 127: CT radiation
dose reporting (Dose SR). 2007. ftp://medical.nema.org/medical/dicom/final/sup127_ft.pdf.
12. Clunie, D. Dose Utility Usage. http://www.dclunie.com/pixelmed/software/webstart/DoseUtili-
tyUsage.html. Accessed February 6, 2018.
13. Digital Imaging and Communications in Medicine (DICOM). DICOM PS3.3 2017e – Informa-
tion object definitions. 2017. http://dicom.nema.org/medical/dicom/current/output/chtml/part03/
PS3.3.html. Accessed February 15, 2018.
14. Digital Imaging and Communications in Medicine (DICOM). DICOM PS3.16 2017e – Content
mapping resource. 2017. http://dicom.nema.org/medical/dicom/current/output/html/part16.html.
15. McDonagh, E. OpenREM: Free and open source radiation exposure monitoring for the physicist.
http://openrem.org/. Accessed February 15, 2018.
46

RPT 246

Uploaded by

Copyright:

Available Formats

RPT 246

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

RPT 246

Uploaded by

Copyright:

Available Formats

AAPM REPORT NO.

Estimating Patient Organ Dose

A Joint Report of AAPM Task Group 246 and the European

DISCLAIMER: This publication is based on sources

The AAPM does not endorse any products, manufac-

© 2019 by American Association of Physicists in Medicine

Task Group Consultants

Erin Angel Canon Medical Systems, USA

© 2019 by American Association of Physicists in Medicine

All rights reserved

American Association of Physicists in Medicine

2. Current Dosimetry Metrics and Associated DICOM Information

Table 1: Examples of typical tolerance levels for the CTDIw

X-ray Tube Potential Typical or Expected Maximum

kV > 80 15% 30%

Narrow collimation 30%

2.1.3 Special Considerations for Nominal Beam Collimations >40 mm

DLPaxial  CTDI vol  d  n , (5)

DLPhelical  CTDI vol  L . (6)

DICOM Tag Sample Value Description Comments and Units

SSDE  fWED  CTDI vol . (8)

SSDE ( z )  fWED ( z )  CTDI vol ( z ), (9)

 ZN1 SSDE ( z ) (10)

3. Current and Emerging Methods to Estimate Organ Dose

3.1.1.3 Chest–Lung and Breast Dose

3.1.2 Perfusion Examinations

3.1.2.2 Body Perfusion

3.4.2 Modeling of the X-ray Source

Figure 5 Left: organFigure

3.6 Monte Carlo Simulation to Model the Dose Deposited by a CT Scanner

4. Sources of Uncertainty in Estimating Organ Dose

Table 3: Summary of selected sources and level of uncertainties that can be

Source of Uncertainty Contributing Factors Estimated Uncertainty ()

4.1.3 SSDE and TCM Examinations

Source of Uncertainty Contributing Factors Estimated Uncertainty ()

4.2.2 Scanner Irradiation Conditions

4.2.3 Patient Not Centered in the Gantry

5. Summary of Report and Recommendations

DICOM—Present and Future for Dosimetry

A.1 CT Radiation Dose Reporting from DICOM Objects

A.2 Radiation Dose Structured Report (RDSR)—The DICOM Object

TID 10012 TID 1021

TID 1002 TID 10013

TID 10015 TID 1020

Figure A1. Schema of TIDs used in the CT RDSR.12

A.2.2 The Patient RDSR

A.3 Using the DICOM RDSR

X-Ray Radiation Dose Report

A.3.2 Verification of the DICOM RDSR

You might also like