Task Group No. 43-U1

Update of AAPM Task Group No.
43 Report: A revised AAPM protocol

for brachytherapy dose calculations
Mark J. Rivard
Department of Radiation Oncology, Tufts-New England Medical Center, Boston, Massachusetts 02111
Bert M. Coursey
Ionizing Radiation Division, National Institute of Standards and Technology, Gaithersburg, Maryland 20899
Larry A. DeWerd
Accredited Dosimetry and Calibration Laboratory, University of Wisconsin, Madison, Wisconsin 53706
William F. Hanson
Radiological Physics Center, M. D. Anderson Cancer Center, Houston, Texas 77030
M. Saiful Huq
Kimmel Cancer Center of Jefferson Medical College, Thomas Jefferson University, Philadelphia,
Pennsylvania 19107
Geoffrey S. Ibbott
Radiological Physics Center, M. D. Anderson Cancer Center, Houston, Texas 77030
Michael G. Mitch
Ionizing Radiation Division, National Institute of Standards and Technology, Gaithersburg, Maryland 20899
Ravinder Nath
Department of Therapeutic Radiology, Yale University, New Haven, Connecticut 06510
Jeffrey F. Williamson
Chair, Low-energy Interstitial Brachytherapy Dosimetry subcommittee of the Radiation Therapy Committee,
Department of Radiation Oncology, Virginia Commonwealth University, Richmond, Virginia 23298
~Received 22 August 2003; revised 11 December 2003; accepted for publication 16 December
2003; published 27 February 2004!
Since publication of the American Association of Physicists in Medicine ~AAPM! Task Group No.
43 Report in 1995 ~TG-43!, both the utilization of permanent source implantation and the number
of low-energy interstitial brachytherapy source models commercially available have dramatically
increased. In addition, the National Institute of Standards and Technology has introduced a new
primary standard of air-kerma strength, and the brachytherapy dosimetry literature has grown sub-
stantially, documenting both improved dosimetry methodologies and dosimetric characterization of
particular source models. In response to these advances, the AAPM Low-energy Interstitial Brachy-
therapy Dosimetry subcommittee ~LIBD! herein presents an update of the TG-43 protocol for
calculation of dose-rate distributions around photon-emitting brachytherapy sources. The updated
protocol ~TG-43U1! includes ~a! a revised definition of air-kerma strength; ~b! elimination of
apparent activity for specification of source strength; ~c! elimination of the anisotropy constant in
favor of the distance-dependent one-dimensional anisotropy function; ~d! guidance on extrapolating
tabulated TG-43 parameters to longer and shorter distances; and ~e! correction for minor inconsis-
tencies and omissions in the original protocol and its implementation. Among the corrections are
consistent guidelines for use of point- and line-source geometry functions. In addition, this report
recommends a unified approach to comparing reference dose distributions derived from different
investigators to develop a single critically evaluated consensus dataset as well as guidelines for
performing and describing future theoretical and experimental single-source dosimetry studies.
Finally, the report includes consensus datasets, in the form of dose-rate constants, radial dose
functions, and one-dimensional ~1D! and two-dimensional ~2D! anisotropy functions, for all low-
energy brachytherapy source models that met the AAPM dosimetric prerequisites @Med. Phys. 25,
2269 ~1998!# as of July 15, 2001. These include the following 125I sources: Amersham Health
models 6702 and 6711, Best Medical model 2301, North American Scientific Inc. ~NASI! model
MED3631-A/M, Bebig/Theragenics model I25.S06, and the Imagyn Medical Technologies Inc.
isostar model IS-12501. The 103Pd sources included are the Theragenics Corporation model 200
and NASI model MED3633. The AAPM recommends that the revised dose-calculation protocol
and revised source-specific dose-rate distributions be adopted by all end users for clinical treatment
planning of low energy brachytherapy interstitial sources. Depending upon the dose-calculation
protocol and parameters currently used by individual physicists, adoption of this protocol may
result in changes to patient dose calculations. These changes should be carefully evaluated and
633 Med. Phys. 31 „3…, March 2004 0094-2405Õ2004Õ31„3…Õ633Õ42Õ$22.00 © 2004 Am. Assoc. Phys. Med. 633
634 Rivard et al.: AAPM TG-43 update 634
reviewed with the radiation oncologist preceding implementation of the current protocol. © 2004
American Association of Physicists in Medicine. @DOI: 10.1118/1.1646040#
Key words: TG-43, brachytherapy dosimetry protocol, TLD dosimetry, Monte Carlo calculations,
125 103
I, Pd
TABLE OF CONTENTS Chamber. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 666

2. NIST 1999 standard using the Wide-Angle
I. INTRODUCTION. . . . . . . . . . . . . . . . . . . . . . . . . . . . 634 Free-Air Chamber. . . . . . . . . . . . . . . . . . . . . . 667
II. CLINICAL RATIONALE FOR ACCURATE APPENDIX C: EVALUATION OF DOSE RATE
DOSIMETRY. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 635 AT SMALL AND LARGE DISTANCES. . . . . . . . . . 669
III. TASK GROUP # 43 DOSIMETRY 1. 2D dose calculations. . . . . . . . . . . . . . . . . . . 669
FORMALISM. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 636 2. Collaborative Ocular Melanoma Study
dosimetry. . . . . . . . . . . . . . . . . . . . . . . . . . . . 669
A. General 2D formalism. . . . . . . . . . . . . . . . . . . . . 637
3. 1D dose calculations. . . . . . . . . . . . . . . . . . . 669
B. General 1D formalism. . . . . . . . . . . . . . . . . . . . . 639
APPENDIX D: ANISTOTROPY CONSTANT. . . . . . 670
IV. CONSENSUS DATASETS FOR CLINICAL
APPENDIX E: APPARENT ACTIVITY. . . . . . . . . . . 670
IMPLEMENTATION. . . . . . . . . . . . . . . . . . . . . . . . 640
A. Source geometry variations. . . . . . . . . . . . . . . . . 641
I. INTRODUCTION
B. General discussion of TG-43 dosimetry
parameters. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 642
In 1995, the American Association of Physicists in Medicine
C. Uncertainty analysis. . . . . . . . . . . . . . . . . . . . . . . 643
~AAPM! Task Group No. 43 published a protocol ~TG-43!1
V. RECOMMENDED METHODOLOGY TO
introducing a new brachytherapy dose calculation formalism
OBTAIN BRACHYTHERAPY DOSIMETRY
based largely on findings of the Interstitial Collaborative
PARAMETERS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 650
A. General recommendations. . . . . . . . . . . . . . . . . . 650 Working Group ~ICWG!.2 Previous calculation formalisms
B. Preparation of dosimetry parameters. . . . . . . . . . 650 were based upon apparent activity (A app), equivalent mass of
C. Reference data and conditions for radium, exposure-rate constants, and tissue-attenuation coef-
brachytherapy dosimetry. . . . . . . . . . . . . . . . . . . 652 ficients. These older formalisms did not account for source-
D. Methodological recommendations for to-source differences in encapsulation or internal construc-
experimental dosimetry. . . . . . . . . . . . . . . . . . . . 652 tion. Except for radium, the exposure-rate constants and
E. Methodological recommendations for Monte other input parameters to these algorithms depended only on
Carlo-based dosimetry. . . . . . . . . . . . . . . . . . . . . 654 the radionuclide.3 In contrast, TG-43 employed dose-rate
F. Publication of dosimetry results. . . . . . . . . . . . . 655 constants and other dosimetric parameters that depended on
VI. CLINICAL IMPLEMENTATION. . . . . . . . . . . . . . 656 the specific source design, and are directly measured or cal-
A. Dose-calculation formalism. . . . . . . . . . . . . . . . . 656 culated for each source design. Additionally, TG-43 pre-
B. Acceptance testing and commissioning. . . . . . . . 657 sented consensus dosimetry data, in terms of the recom-
C. Source calibrations. . . . . . . . . . . . . . . . . . . . . . . . 658 mended formalism, for the three low-energy photon emitting
GLOSSARY OF SYMBOLS AND TERMS. . . . . . . . 659 source models then available ~Theragenics Corporation
APPENDIX A: MODEL-SPECIFIC SOURCE model 200 103Pd source and Amersham Health models 6702
DOSIMETRY DATA. . . . . . . . . . . . . . . . . . . . . . . . . . 660 and 6711 125I sources!. ~Certain commercial equipment, in-
1. Amersham Health model 6702 and 6711
125 struments, and materials are identified in this work in order
I sources. . . . . . . . . . . . . . . . . . . . . . . . . . . 660
to specify adequately the experimental procedure. Such iden-
2. Best medical model 2301 125I source. . . . . . 661
tification does not imply recommendation nor endorsement
3. North American Scientific Inc. model
by either the AAPM or National Institute of Standards and
MED3631-A/M 125I source. . . . . . . . . . . . . . 662
4. Bebig/Theragenics Corporation model Technology ~NIST!, nor does it imply that the material or
I25.S06 125I source. . . . . . . . . . . . . . . . . . . . . 663 equipment identified is necessarily the best available for
5. Imagyn Medical Technologies Inc. isostar these purposes.! These data were based upon a critical re-
model IS-12501 125I source. . . . . . . . . . . . . . 664 view of ICWG measured dose-rate distributions using LiF
6. Theragenics Corporation model 200 103Pd TLD as well as other measurements and Monte Carlo calcu-
source. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 664 lations available in the literature. Overall, the TG-43 protocol
7. North American Scientific Inc. model has resulted in significant improvements in the standardiza-
MED3633 103Pd source. . . . . . . . . . . . . . . . . 665 tion of both dose-calculation methodologies as well as dose-
APPENDIX B: NIST AIR-KERMA STRENGTH rate distributions used for clinical implementation of brachy-
STANDARDS FOR LOW-ENERGY. . . . . . . . . . . . . . 666 therapy. For example, the differences between the previously
1. NIST 1985 standard using the Free-Air used dose-rate distributions and those recommended by
Medical Physics, Vol. 31, No. 3, March 2004

TG-43 were as large as 17% for some sources. These low-energy source models included in the original TG-43
changes have been exhaustively reviewed by the physics protocol. The present protocol ~TG-43U1! recommends a
community and are generally accepted. Most treatment plan- single, consensus dataset for each source model from which
ning software vendors have implemented the TG-43 formal- the 1D and 2D dose-rate distribution can be reconstructed.
ism and the recommended dosimetry parameters in their sys- @This protocol was prepared by the AAPM Low-energy In-
tems. LiF TLD dose measurements and Monte Carlo dose terstitial Brachytherapy Dosimetry subcommittee, now the
calculations have largely replaced the semi-empirical dose- Photon-Emitting Brachytherapy Dosimetry subcommittee
calculation models of the past. ~Chair, Jeffrey F. Williamson! of the AAPM Radiation
Since publication of the TG-43 protocol over nine years Therapy Committee. This protocol has been reviewed and
ago, significant advances have taken place in the field of approved by the AAPM Radiation Therapy Committee and
permanent source implantation and brachytherapy dosimetry. AAPM Science Council, and represents the current recom-
To accommodate these advances, the AAPM deemed it nec- mendations of the AAPM on this subject.# Finally, method-
essary to update this protocol for the following reasons: ological guidelines are presented for physicist-investigators
aiming to obtain dosimetry parameters for brachytherapy
~a! To eliminate minor inconsistencies and omissions in sources using calculative methods or experimental tech-
the original TG-43 formalism and its niques.
implementation.4 – 6 Although many of the principles and the changes in meth-
~b! To incorporate subsequent AAPM recommendations, odology might apply, beta- or neutron-emitting sources such
addressing requirements for acquisition of dosimetry as 90Sr, 32P or 252Cf are not considered in this protocol. A
data as well as clinical implementation.7 These recom- further update of this protocol is anticipated to provide con-
mendations, e.g., elimination of A app ~see Appendix E! sensus, single source dose distributions and dosimetry pa-
and description of minimum standards for dosimetric rameters for high-energy photon-emitting ~e.g. 192Ir and
characterization of low-energy photon-emitting brachy- 137
Cs) sources, and to generate consensus data for new low-
therapy sources,8,9 needed to be consolidated in one energy photon sources that are not included in this report, yet
convenient document. meet the AAPM prerequisites and are posted on the AAPM/
~c! To critically reassess published brachytherapy dosime- RPC Seed Registry website10 as of December 1, 2003:
try data for the 125I and 103Pd source models introduced
~1! Amersham Health, OncoSeed model 6733 125I,
both prior and subsequent to publication of the TG-43
~2! Best Medical model 2335 103Pd,
protocol in 1995, and to recommend consensus datasets
~3! Draximage Inc., BrachySeed model LS-1 125I,
where appropriate.
~4! IBt, Intersource-125 model 1251L 125I,
~d! To develop guidelines for the determination of
~5! IBt, Intersource-103 model 1031L 103Pd,
reference-quality dose distributions by both experimen-
~6! Implant Sciences Corp. I-Plant model 3500 125I,
tal and Monte Carlo methods, and to promote consis-
~7! IsoAid, Advantage model 1A1-125A 125I,
tency in derivation of parameters used in TG-43 for-
~8! Mills Biopharmaceuticals Inc., ProstaSeed model SL/
malism.
SH-125 125I,
Updated tables of TG-43 parameters are necessary and ~9! Nucletron Corp., selectSeed model 130.002 125I, and
timely to accommodate the ;20 new low-energy interstitial ~10! SourceTech Medical, 125Implant model STM1251 125I.
brachytherapy source models that have been introduced to As indicated in the Table of Contents, this protocol is
the market since publication of TG-43 in 1995. These com- divided into various sections. Clinical medical physicists
mercial developments are due mostly to the rapid increase in should pay special attention to Secs. III–VI due to dosimetry
utilization of permanent prostate brachytherapy. Some of formalism and clinical implementation recommendations
these new brachytherapy sources were introduced into clini- presented herein. Section II updates the clinical rationale for
cal practice without thorough scientific evaluation of the nec- accurate dosimetry. The origin of consensus datasets for
essary dosimetric parameters. The AAPM addressed this is- eight seed models is presented in Appendix A. Dosimetry
sue in 1998, recommending that at least one experimental investigators will find useful the detailed recommendations
and one Monte Carlo determination of the TG-43 dosimetry presented in Secs. IV and V. The description of the NIST
parameters be published in the peer-reviewed literature be- calibration scheme is presented in Appendix B. Manufactur-
fore using new low-energy photon-emitting sources ~those ers of brachytherapy treatment planning software will find
with average photon energies less than 50 keV! in routine new recommendations in Secs. II, IV, VI, and Appendixes
clinical practice.9 Thus, many source models are supported C–E.
by multiple dosimetry datasets based upon a variety of basic
dosimetry techniques. This confronts the clinical physicist
with the problem of critically evaluating and selecting an II. CLINICAL RATIONALE FOR ACCURATE
appropriate dataset for clinical use. To address this problem, DOSIMETRY
this protocol presents a critical review of dosimetry data for While low-energy, photon-emitting brachytherapy sources
eight 125I and 103Pd source models which satisfied the afore- have been used to treat cancers involving a variety of ana-
mentioned criteria as of July 15, 2001, including the three tomical sites, including eye plaque therapy for choroidal

melanoma and permanent lung implants,11,12 their most fre- 160 Gy, and both indicated a steep dependence of clinical
quent indication today is for the treatment of prostate outcome with dose in the range of 100 to 160 Gy. For ex-
cancer.13 Prostate cancer is the most frequent type of cancer ample, Stock reported freedom from biochemical failure
in men in the United States with approximately 180 000 new rates of 53%, 82%, 80%, 95%, and 89% for patients receiv-
cases incident per year and an annual death rate of about ing D 90,100 Gy, 100<D 90,120 Gy, 120<D 90,140 Gy,
37 000.14 While approximately 60% of new cases are con- 140<D 90,160 Gy and D 90> 160 Gy, respectively. The close
fined to the organ at time of diagnosis, only about 2.2% of correlation between D 90 and PSA–RFS, and a dose response
these new cases were treated with brachytherapy in 1995. in the clinical dose range of 100 to 160 Gy are strong justi-
Since that time, the percentage has increased to about 30% of fications for improved accuracy in the dosimetry for intersti-
all eligible patients receiving implants in current practice. tial brachytherapy, which is the focus of this work. The up-
This increase was largely due to improvements in diagnosis dated dosimetry formalism and changes in calibration
and case selection facilitated by introduction of the prostate standards recommended herein will result in changes to the
specific antigen ~PSA! screening test, and to improved clinical practice of brachytherapy. The clinical medical
ultrasound-guided delivery techniques. In the United States, physicist is advised that guidance on prescribed-to-
the pioneering work was performed by a group of investiga- administered dose ratios for 125I and 103Pd will be forthcom-
tors based in Seattle.15 The most widely used technique uti- ing in a subsequent report.
lizes transrectal ultrasound ~TRUS! guided implantation of
either 125I or 103Pd brachytherapy sources using a template-
guided needle delivery system to avoid open surgery re-
quired by the retropubic approach.16,17
Several studies have shown that clinical outcomes in pros- III. TASK GROUP # 43 DOSIMETRY FORMALISM
tate brachytherapy, both for the retropubic approach and the
TRUS-guided technique, correlate with dose coverage pa- As in the original TG-43 protocol, both 2D ~cylindrically
rameters. The extensive clinical experience of Memorial symmetric line source! and 1D ~point source! dose-
Sloan Kettering Institute ~1078 patients with retropubic ap- calculation formalisms are given. To correct small errors and
proach surgery! from 1970–1987 was reviewed by Zelefsky to better address implementation details neglected in the
and Whitmore.18 Multivariate-analysis revealed a D 90 im- original protocol, all quantities are defined. Throughout this
plant dose of 140 Gy to be an independent predictor of protocol, the following definitions are used:
recurrence-free local control at 5, 10, and 15 years ( p ~1! A source is defined as any encapsulated radioactive ma-
50.001). D 90 is defined as the dose delivered to 90% of the terial that may be used for brachytherapy. There are no
prostate volume as outlined using post-implant CT images. restrictions on the size or on its symmetry.
Similarly, a review of 110 implants at Yale University using ~2! A point source is a dosimetric approximation whereby
the retropubic implant approach from 1976 to 1986 reported radioactivity is assumed to subtend a dimensionless
a correlation ( p50.02) of recurrence-free local control after point with a dose distribution assumed to be spherically
10 years with V 100 ; V 100 is defined as the percentage of symmetric at a given radial distance r. The influence of
target volume receiving the prescribed dose of 160 Gy.19 inverse square law, for the purpose of interpolating be-
Two recent retrospective studies of the TRUS technique tween tabulated transverse-plane dose-rate values, can be
demonstrate that the clinical outcome depends on dose deliv- calculated using 1/r 2 .
ered and prostate volume coverage. Stock et al. reported on ~3! The transverse-plane of a cylindrically symmetric source
an experience of 134 prostate cancer patients implanted with is that plane which is perpendicular to the longitudinal
125
I and not treated with teletherapy or hormonal therapy.20 axis of the source and bisects the radioactivity distribu-
They assessed rates of freedom from biochemical failure as a tion.
function of the D 90 dose. A significant increase in freedom ~4! A line source is a dosimetric approximation whereby ra-
from biochemical failure ~92% vs 68% after 4 years! was dioactivity is assumed to be uniformly distributed along
observed (p50.02) for patients (n569) where D 90 a 1D line-segment with active length L. While not accu-
>140 Gy. Potters et al. recently reviewed the impact of vari- rately characterizing the radioactivity distribution within
ous dosimetry parameters on biochemical control for their an actual source, this approximation is useful in charac-
experience of 719 patients treated with permanent prostate terizing the influence of inverse square law on a source’s
brachytherapy.21 Many of these patients also received tele- dose distribution for the purposes of interpolating be-
therapy ~28%! or hormone therapy ~35%!. Furthermore, 84% tween or extrapolating beyond tabulated TG-43 param-
of the implants used 103Pd with the remainder using 125I. eter values within clinical brachytherapy treatment plan-
Their results indicated that patient age, radionuclide selec- ning systems.
tion, and use of teletherapy did not significantly affect bio- ~5! A seed is defined as a cylindrical brachytherapy source
chemical relapse-free survival ~PSA–RFS!. The only dose- with active length, L, or effective length, L eff ~described
specification index that was predictive of PSA–RFS was later in greater detail! less than or equal to 0.5 cm.
D 90 .
Like the other two studies, studies by Stock et al. and These parameters are utilized by the TG-43U1 formalism
Potters et al. were based on pre-TG-43 prescription doses of in the following sections.

A. General 2D formalism
The general, two-dimensional ~2D! dose-rate equation
from the 1995 TG-43 protocol is retained,
G L ~ r, u !
Ḋ ~ r, u ! 5S K •L• •g ~ r ! •F ~ r, u ! , ~1!
G L~ r 0 , u 0 ! L
where r denotes the distance ~in centimeters! from the center

of the active source to the point of interest, r 0 denotes the
reference distance which is specified to be 1 cm in this pro- FIG. 1. Coordinate system used for brachytherapy dosimetry calculations.
tocol, and u denotes the polar angle specifying the point-of-
interest, P(r, u ), relative to the source longitudinal axis. The
reference angle, u 0 , defines the source transverse plane, and source approximation used for the geometry function ~Sec.
is specified to be 90° or p/2 radians ~Fig. 1!. III A 3!. For evaluation of dose rates at small and large dis-
In clinical practice, source position and orientation are tances, the reader is referred to Appendix C.
identified by means of radio-opaque markers. Generally,
these markers are positioned symmetrically within the source
capsule such that the marker, the radioactivity distribution, 1. Air-kerma strength
and the capsule have the same geometric center on the sym-
metry axis of the source. Thus, determination of the location This protocol proposes minor revisions to the definition of
of the radioisotope distribution is based upon identification air-kerma strength, S K , which was first introduced by the
of the radio-opaque markers. All sources discussed in this AAPM TG-32 report in 1987.22 Air-kerma strength has units
document can be accurately represented by a capsule and of m Gy m2 h-1 and is numerically identical to the quantity
radio-opaque markers that are symmetric with respect to the Reference Air Kerma Rate recommended by ICRU 38 and
transverse plane, which by definition bisects the active ICRU 60.23,24 For convenience, these unit combinations
source and specifies the origin of the dose-calculation for- are denoted by the symbol U where 1 U51 m Gy m2 h21
malism. However, Eq. ~1! can accommodate sources that are 51 cGy cm2 h21 .
asymmetric with respect to the transverse plane. For sources Air-kerma strength, S K , is the air-kerma rate, K̇ d (d), in
that exhibit all of the following characteristics: ~i! the radio- vacuo and due to photons of energy greater than d, at dis-
activity distribution is clearly asymmetric with respect to the tance d, multiplied by the square of this distance, d 2 ,
planes bisecting the capsule or marker; ~ii! the extent of S K 5K̇ d ~ d ! d 2 . ~2!
asymmetry is known a priori or can be measured via imag-
ing; and ~iii! the source orientation can be determined under The quantity d is the distance from the source center to the
clinical implant circumstances ~e.g., via CT or radiography!, point of K̇ d (d) specification ~usually but not necessarily as-
then the source coordinate system origin should be posi- sociated with the point of measurement! which should be
tioned at the geometric center of the radionuclide distribution located on the transverse plane of the source. The distance d
~as determined using positioning information obtained from can be any distance that is large relative to the maximum
the markers!, not the geometric center of the exterior surface linear dimension of the radioactivity distribution so that S K is
of the capsule or marker. If radio-opaque markers do not independent of d. K̇ d (d) is usually inferred from transverse-
facilitate identification of source orientation and the asym- plane air-kerma rate measurements performed in a free-air
metrical distribution under clinical circumstances, then the geometry at distances large in relation to the maximum linear
geometric center of the source must be presumed to reside at dimensions of the detector and source, typically of the order
the radio-opaque marker centroid as is conventionally per- of 1 meter. The qualification ‘‘in vacuo’’ means that the mea-
formed. surements should be corrected for photon attenuation and
The quantities used in Eq. ~1! are defined and discussed scattering in air and any other medium interposed between
later. This formalism applies to sources with cylindrically the source and detector, as well as photon scattering from
symmetric dose distributions with respect to the source lon- any nearby objects including walls, floors, and ceilings. Of
gitudinal axis. In addition, the consensus datasets presented course, air-kerma rate may also be calculated to subvert
in Sec. IV B assume that dose distributions are symmetric some of the limitations imposed on practical
with respect to the transverse plane, i.e., that radioactivity measurements.25 The energy cutoff, d, is intended to exclude
distributions to either side of the transverse plane are mirror low-energy or contaminant photons ~e.g., characteristic
images of one another. However, this formalism is readily x-rays originating in the outer layers of steel or titanium
generalized to accommodate sources that are not symmetric source cladding! that increase K̇ d (d) without contributing
with respect to the transverse plane. significantly to dose at distances greater than 0.1 cm in tis-
Equation ~1! includes additional notation compared with sue. The value of d is typically 5 keV for low-energy photon-
the corresponding equation in the original TG-43 formalism, emitting brachytherapy sources, and is dependent on the ap-
namely the subscript ‘‘L’’ has been added to denote the line plication.

In summary, the present definition of S K differs in two strength standard ~e.g., P5‘‘N’’ for NIST, or P5‘‘T ’’ for the
important ways from the original 1987 AAPM definition. in-house calibration-standard of Theragenics Corporation!,
First, the original AAPM definition of S K did not allow for a ‘‘qq’’ denotes the year in which this source strength standard
low-energy cutoff. Subsequent experience using free-air was implemented, and the ‘‘S’’ subscript denotes the word
chambers as primary S K standards clearly indicates that fail- standard.7 For example, L 97D,N99S indicates a dose-rate con-
ure to exclude nonpenetrating radiations greatly increases stant determined from dosimetry measurements published in
measurement uncertainty and invalidates theoretical dosime- 1997 and normalized to an S K traceable to the 1999 NIST
try models. Second, the conditions that should prevail in an standard. Additional notation may also be utilized such as
6702
experimental determination of S K are now explicitly stated. TLD L 97D,N85S for the dose-rate constant for the model 6702
source published in 1997 using TLDs and the 1985 NIST
2. Dose-rate constant standard. These notations are useful for comparing results
The definition of the dose-rate constant in water, L, is from multiple investigators, and readily highlight features
unchanged from the original TG-43 protocol: it is the ratio of such as utilization of the calibration procedure and whether
dose rate at the reference position, P(r 0 , u 0 ), and S K . L has or not influence of titanium K-shell x rays is included.
units of cGy h21 U21 which reduces to cm22 ,
3. Geometry function
Ḋ ~ r 0 , u 0 !
L5 . ~3! Within the context of clinical brachytherapy dose calcula-
SK
tions, the purpose of the geometry function is to improve the
The dose-rate constant depends on both the radionuclide and accuracy with which dose rates can be estimated by interpo-
source model, and is influenced by both the source internal lation from data tabulated at discrete points. Physically, the
design and the experimental methodology used by the pri- geometry function neglects scattering and attenuation, and
mary standard to realize S K . provides an effective inverse square-law correction based
In 1999, a notation was introduced, L nnD, PqqS , to identify upon an approximate model of the spatial distribution of ra-
both the dose-rate measurements or calculations used to de- dioactivity within the source. Because the geometry function
termine Ḋ(r 0 , u 0 ) and the calibration standard to which this is used only to interpolate between tabulated dose-rate values
dose rate was normalized. The subscript ‘‘D’’ denotes refer- at defined points, highly simplistic approximations yield suf-
ence dose rate, ‘‘nn’’ denotes the year in which this refer- ficient accuracy for treatment planning. This protocol recom-
ence dose rate was published ~either measurement or calcu- mends use of point- and line-source models giving rise to the
lation!, ‘‘P’’ denotes the provider or origin of the source following geometry functions:
G P ~ r, u ! 5r 22 point-source approximation,
H
~4!
b
if u Þ0°
G L ~ r, u ! 5 Lr sin u line-source approximation,
~ r 2 2L 2 /4! 21 if u 50°
where b is the angle, in radians, subtended by the tips of the Use of such simple functions is warranted since their purpose
hypothetical line source with respect to the calculation point, is to facilitate interpolation between tabulated data entries for
P(r, u ). duplication of the original dosimetry results.
In principle, either the point-source or line-source models In the case where the radioactivity is distributed over a
may be consistently implemented in both the 1D and 2D right-cylindrical volume or annulus, this protocol recom-
versions of the TG-43 formalism. In this case, the word mends taking active length to be the length of this cylinder.
‘‘consistently’’ means that the geometry function used for For brachytherapy sources containing uniformly spaced mul-
derivation of dose rates from TG-43 parameters should be tiple radioactive components, L should be taken as the effec-
identical to that used to prepare the radial dose function and tive length, L eff , given by
2D anisotropy function data, including use of the same active
L eff5DS3N, ~5!
length, L, used in G(r, u ). Under these conditions, TG-43
dose calculations will reproduce exactly the measured or where N represents the number of discrete pellets contained
Monte Carlo-derived dose rates from which g(r) and F(r, u ) in the source with a nominal pellet center-to-center spacing
tables were derived. This protocol recommends consistent DS. If L eff is greater than the physical length of the source
use of the line-source geometry function for evaluation of 2D capsule ~usually ;4.5 mm), the maximum separation ~dis-
dose distributions, and use of either point- or line-source tance between proximal and distal aspects of the activity dis-
geometry functions for evaluations of 1D dose distributions. tribution! should be used as the active length, L. This tech-

nique avoids singularities in evaluating G(r, u ) for points of off the transverse plane typically decreases as ~i! r decreases,
interest in tissue which are located on the hypothetical line ~ii! as u approaches 0° or 180°, ~iii! as encapsulation thick-
source just beyond the tip and end of the physical source. ness increases, and ~iv! as photon energy decreases. How-
More complex forms of the geometry function have a role ever, F(r, u ) may exceed unity at u u 290° u .6arcsin(L/2r)
in accurately estimating dose at small distances outside the for right-cylinder sources coated with low-energy photon
tabulated data range, i.e., extrapolating g(r) and F(r, u ) to emitters due to screening of photons by the active element at
small distances.26,27 Use of such expressions is permitted. angles towards the transverse plane.
However, most commercial brachytherapy treatment plan- As stated earlier, the active length, L, used to evaluate
ning systems support only point- or line-source geometry G L (r, u ) in Eq. ~4! shall be the same L used to extract g L (r)
functions. Therefore, it is the responsibility of the physicist and F(r, u ) from dose distributions via Eqs. ~6! and ~8!,
to transform the tabulated TG-43 parameters given in this respectively. Otherwise, significant errors in dosimetry re-
protocol, which are based upon point- and line-source ap- sults at small distances may arise. For example, at r
proximations, to a format consistent with more complex ge- 50.5 cm, a change in L from 3 to 5 mm results in a 5%
ometry functions that may be available on their treatment change in G L (r, u 0 ).
planning systems.28 –30
4. Radial dose function
B. General 1D formalism
The radial dose function, g X (r), accounts for dose fall-off
on the transverse-plane due to photon scattering and attenu- While a 1D isotropic point-source approximation @Eq. ~9!#
ation, i.e., excluding fall-off included by the geometry func- only approximates the true complex 2D dose distribution, it
tion. g X (r) is defined by Eq. ~6!, and is equal to unity at r 0 simplifies source localization procedures by eliminating the
51 cm. need to determine the orientation of the source longitudinal
axis from imaging studies.
Ḋ ~ r, u 0 ! G X ~ r 0 u 0 !
g X~ r ! 5 . ~6!
Ḋ ~ r 0 , u 0 ! G X ~ r, u 0 ! G X ~ r, u 0 !
Ḋ ~ r ! 5S K •L• •g ~ r ! • f an~ r ! . ~9!
The revised dose-calculation formalism has added the sub- G X~ r 0 , u 0 ! X
script ‘‘X’’ to the radial dose function and geometry function Users should adopt one of the following implementations of
to indicate whether a point-source, ‘‘P,’’ or line-source, Eq. ~9!:
‘‘L,’’ geometry function was used in transforming the data.
Consequently, this protocol presents tables of both g P (r) and
g L (r) values. Ḋ ~ r ! 5S K •L• SDr0
r
2
•g P ~ r ! • f an~ r ! , ~10!
Equation ~7! corrects a typographical error in the original
TG-43 protocol.31 While table lookup via linear interpolation or
or any appropriate mathematical model fit to the data may be
G L ~ r, u 0 !
used to evaluate g X (r), some commercial treatment planning Ḋ ~ r ! 5S K •L• •g ~ r! • f an~ r ! . ~11!
systems currently accommodate a fifth-order polynomial fit G L~ r 0 , u 0 ! L
to the tabulated g(r) data. Since this type of polynomial fit
While most treatment planning systems use the implementa-
may produce erroneous results with large errors outside the
tions in Eq. ~10!, we recommend use of Eq. ~11! due to
radial range used to determine the fit, alternate fitting equa-
improved accuracy at small distances, e.g., r,1 cm. Linear
tions have been proposed which are less susceptible to this
interpolation may be used to match the grid spacing of g X (r)
effect,32
with the grid spacing of f an(r).
g X ~ r ! 5a 0 1a 1 r1a 2 r 2 1a 3 r 3 1a 4 r 4 1a 5 r 5 . ~7! These revised formulations require consistency between
the geometry function used for dose calculation and the ge-
Parameters a 0 through a 5 should be determined so that they
ometry function used for extracting g X (r) from the
fit the data within 62%. Also, the radial range over which
transverse-plane dose distribution. Furthermore, these re-
the fit meets this specification should be clearly specified.
vised formulations correct an inconsistency in Eq. ~11! of the
original TG-43 protocol that indirectly recommended the fol-
5. 2D anisotropy function lowing incorrect equation:
The 2D anisotropy function, F(r, u ), is defined as
G P ~ r, u 0 !
Ḋ ~ r, u ! G L ~ r, u 0 ! Ḋ ~ r ! 5S K L• •g ~ r ! • f an~ r !
F ~ r, u ! 5 . ~8! G P~ r 0 , u 0 ! L
Ḋ ~ r, u 0 ! G L ~ r, u !
~ not recommended! . ~12!
Other than inclusion of the subscript L, this definition is
identical to the original TG-43 definition.1 The 2D anisot- While use of the wrong g X (r) datasets will typically give
ropy function describes the variation in dose as a function of errors in the calculated dose rate of less than 2% at distances
polar angle relative to the transverse plane. While F(r, u ) on beyond 1 cm, average errors of 3%, 15%, and 74% arise at
the transverse plane is defined as unity, the value of F(r, u ) distances of 0.5, 0.25, and 0.1 cm, respectively. Clinical uti-

lization of the 1D dosimetry formalism presented in Eq. ~12!, for the dose-rate constant ( EXPL) were averaged. Sepa-
or other formalisms that inconsistently apply the geometry rately, L values obtained using Monte Carlo techniques
function, are not recommended. ( MCL) were averaged. The consensus value recom-
mended in this protocol ( CONL) is the equally
1. 1D anisotropy function weighted average of the separately averaged experi-
The 1D anisotropy function, f an(r), is identical to the mental and Monte Carlo L values. In cases where there
anisotropy factor defined by the original TG-43 protocol. At is only one experimental result and one Monte Carlo
a given radial distance, f an(r) is the ratio of the solid angle- result: CONL5 @ EXPL1 MCL # /2.
weighted dose rate, averaged over the entire 4 p steradian ~b! Each candidate dataset was examined separately and
space, to the dose rate at the same distance r on the trans- eliminated from consideration if it was determined to
verse plane, see Eq. ~13!, have a problem, e.g., data inconsistency. Corrections
* p0 Ḋ ~ r, u ! sin~ u ! d u for use of a nonliquid water measurement phantom
f an~ r ! 5 . ~13! were applied if not included in the original investiga-
2Ḋ ~ r, u 0 ! tors’ analysis.
Note that one should integrate dose rate, not the values of the ~c! For the 2D anisotropy function, F(r, u ), and the radial
2D anisotropy function to arrive at f an(r). dose function, g(r), all candidate datasets for a given
With consistent use of the geometry function, both Eqs. source model were transformed using identical line-
~10! and ~11! will exactly reproduce the solid-angle weighted source geometry functions to permit fair comparison.
dose rate at a given r. Of the two, Eq. ~11! is recommended The radial dose function was corrected for nonliquid
because the line-source geometry function will provide more water measurement medium if necessary. Assuming
accurate interpolation and extrapolation at small distances. that the different datasets agreed within experimental
The accuracy achievable using the 1D formalism for prostate uncertainties, the consensus data were defined as the
implants was reported by Lindsay et al.,33 and Corbett ideal candidate dataset having the highest resolution,
et al.34 covering the largest distance range, and having the
For brachytherapy treatment planning systems that do not
highest degree of smoothness. For most source models
permit entry of f an(r), Eqs. ~10! or ~11! can still be imple-
examined in this protocol, the consensus F(r, u ) and
mented by carefully modifying g X (r) to include f an(r) as
g(r) data, CONF(r, u ) and CONg(r), were taken from
shown in Eq. ~14!. These modified dosimetry parameters,
the transformed Monte Carlo dataset.
g 8 (r) and f̄ an
8 , are defined as
~d! A few entries in the tabulated consensus datasets were
g 8 ~ r ! 5g X ~ r ! • f an~ r ! , taken from the nonideal candidate dataset~s! to cover a
larger range of distances and angles. These data were
f̄ an
8 51. ~14! italicized to indicate that they were not directly con-
While TG-43 introduced the anisotropy constant, f̄ an , firmed by other measurements or calculations.
LIBD no longer recommends its use. This is discussed in ~e! The 1D anisotropy function, f an(r), was derived using
greater detail in Appendix D. numerical integration of the dose rate, as calculated
from CONF(r, u ) dataset, with respect to solid angle.
IV. CONSENSUS DATASETS FOR CLINICAL Use of the anisotropy constant, f̄ an , is discouraged as
IMPLEMENTATION discussed in Appendix D.
The 125I and 103Pd source models reviewed in this proto- ~f! When scientifically justified for a given source model,
col ~Fig. 2! satisfied the AAPM recommendations that com- exceptions or modifications to these rules were made,
prehensive ~2D! reference-quality dose-rate distribution data and are described later. For example, if the datasets
be accepted for publication by a peer-reviewed scientific were too noisy, they were rejected.
journal on or before July 15, 2001. Appropriate publications ~g! Following tabulation of g(r) and F(r, u ) for all eight
can report either Monte Carlo, or experimentally derived source models, overly dense datasets were down-
TG-43 dosimetry parameters. As many as 12 sets of indepen- sampled to permit reasonable comparisons. Removal of
dently published data per source model were evaluated dur- a dataset point was deemed reasonable if linear inter-
ing preparation for this report. For each source model, a polation using adjacent points resulted in a difference
single consensus dataset was derived from multiple pub- no larger than 62% of the dataset point in question.
lished datasets according to the following methodology.35 If Similarly, because the various authors used different
items essential to critical evaluation were omitted, the au- table grids, it was necessary to interpolate some of the
thors were contacted for information or clarification. data into the common mesh selected for presenting all
~a! The peer-reviewed literature was examined to identify eight datasets. Linear–linear interpolation was used for
candidate dose distributions for each source model de- F(r, u ) datasets, and log-linear interpolation was used
rived either from experimental measurements or Monte for g(r) datasets. Interpolated data are indicated by
Carlo simulations. Experimentally determined values boldface.

FIG. 2. Brachytherapy seeds examined in this report: ~a! Amersham model 6702 source, ~b! Amersham model 6711 source, ~c! Best model 2301 source, ~d!
NASI model MED3631-A/M or MED3633 source, ~e! Bebig/Theragenics Corp. model I25.S06 source, ~f! Imagyn model IS-12501 source, and ~g! Ther-
agenics Corp. model 200 source. The titanium capsule is 0.06 mm thick for the Amersham and Theragenics seeds, while each capsule of the Best seed is 0.04
mm thick. The capsule thickness of the remaining seeds is 0.05 mm.
The details used to evaluate dosimetry parameters for A. Source geometry variations
each source were the following:
Source geometry and internal construction are highly
~1! internal source geometry and a description of the source, manufacturer specific. Source models vary from one another
~2! review of the pertinent literature for the source, with regard to weld thickness and type, radioactivity carrier
~3! correction coefficients for 1999 anomaly in NIST air- construction, presence of radio-opaque material with sharp
kerma strength measurements ~if applicable!, or rounded edges, the presence of silver ~which produces
~4! solid water-to-liquid water corrections, characteristic x rays that modify the photon spectrum!, and
~5! experimental method used, TLD or diode, capsule wall thickness. All of these properties can affect the
~6! active length assumed for the geometry function line- dosimetric characteristics of the source. Radioactive carriers
source approximation, may consist of a radio-transparent matrix, a radio-opaque
~7! name and version of the Monte Carlo transport code, object coated with radioactivity, or a radio-transparent matrix
~8! cross-section library used by Monte Carlo simulation, with highly attenuating radioactive coating. For example, the
~9! Monte Carlo estimator used to score kerma or dose, and Amersham model 6702 and NASI model 3631-A/M sources
~10! agreement between Monte Carlo calculations and ex- utilize spherical resin carriers coated or impregnated with
perimental measurement. radioactivity. The number of spheres varies from 3 or more

per source. Other sources, such as the Amersham model vendors discussed in this protocol have agreed to accept
6711, utilize a silver rod carrier. The amount of silver, or the these same WAFAC measurements as the basis for their
length of silver rod, varies by the source model. Graphite stated source strengths. Subsequent periodic intercompari-
pellets are also used. For example, in the Theragenics Cor- sons between NIST, ADCL, and vendor calibrations will be
poration model 200 103Pd source, the pellets are coated with compared to this original reference calibration, but will not
a mixture of radioactive and nonradioactive palladium. be modified unless large deviations are noted. Table I gives
All 125I and 103Pd source models, except for the now- the NIST standard calibration date that is presently used by
obsolete model 6702 source, contain some type of radio- the ADCLs, NIST and the manufacturer for each source
opaque marker to facilitate radiographic localization. For ex- model along with the corresponding correction applied to
ample, the graphite pellets of the Theragenics Corporation CONL values to account for the 1999 WAFAC anomaly. The
source are separated by a cylindrical lead marker. Beside the L values of Table I have been corrected to the air-kerma
obvious dependence of photon spectrum on the radioisotope standard specified by the WAFAC measurement performed
used, the backing material ~e.g., the radio-opaque marker! on the listed date specified in the table. Generally, intercom-
may further perturb the spectrum. For the sources containing parisons have agreed within 62% of the source strengths
125
I deposited on silver, the resultant silver x rays signifi- derived from the WAFAC measurements listed in Table I.
cantly modify the effective photon spectrum. These source These standardization dates are subject to revision should
construction features influence the resultant dose rate distri- changes in manufacturing procedures, source geometry, or
bution and the TG-43 dosimetry parameters to varying de- the WAFAC standard itself occur that affect the accuracy of
grees. Accurate knowledge of internal source geometry and vendor or ADCL secondary standards. Future source model-
construction details is especially important for Monte Carlo specific revisions to the calibration standard could require
modeling. Individual sources are briefly described later. Ref- corresponding corrections to the recommended dose-rate
erences describing each source and the TG-43 parameters are constant. For this reason, regular calibration comparisons
given in each section. While Sec. III presented the dosimetry among NIST, ADCL, and vendors are required.9
formalism, its applicability to the derivation of consensus In summary, there were two possible situations regarding
datasets is given later. A detailed description for seed models the calibration of seeds at NIST using the WAFAC-based
is provided in Appendix A. air-kerma strength standard. First, seed calibrations per-
formed at NIST during the 1999 calendar year need correc-
B. General discussion of TG-43 dosimetry parameters tion due to a measurement anomaly present in 1999 only.
This correction was determined by another WAFAC mea-
1. Air-kerma strength standards surement for each seed model made at a designated date in
The NIST Wide-Angle Free-Air Chamber or WAFAC- 2000 or later. Second, WAFAC measurements made prior to
based primary standard became available in 1998, and was 1999 and after January 1, 2000 needed no correction. Since
used to standardize the 125I sources then available ~models the notation S K,N99 represents the NIST WAFAC-based air-
6702, 6711, and MED3631-A/M!. For a more detailed dis- kerma strength standard as officially introduced on January
cussion of the NIST air-kerma strength standards, including 1, 1999, this notation is used for all WAFAC measurements,
those based on the Ritz free-air chamber ~1985! and WAFAC regardless of the date of calibration. Thus, all measured
~1999!, see Appendix B. The WAFAC standard shifted for dose-rate constant data given in this protocol have been nor-
unknown reasons in 1999, and was corrected in the first half malized to the S K,N99 standard. Any measured dose-rate con-
of 2000. For those sources available in 1998, the 1998 and stants normalized to NIST calibrations performed in 1999
2000 WAFAC measurements agreed within estimated mea- have been appropriately corrected for WAFAC measurement
surement uncertainty. Following restoration of the WAFAC anomalies.
to its 1998 sensitivity in 2000, all sources initially standard-
ized against WAFAC measurements performed in 1999, and
2. Dose-rate constant
the model 3631-A/M source, which had renormalized its
stated strength against the WAFAC in 1999, had to be stan- Specifying the dose-rate constant as accurately as possible
dardized against the corrected WAFAC measurements. To is essential, as it is used to transform the relative TG-43 dose
implement these corrections, five sources of each type were distribution into absolute dose rates given the air-kerma
calibrated using the NIST WAFAC and then sent to both the strength of the sources deployed in the patient. As discussed
accredited dosimetry calibration laboratories ~ADCLs! and in more detail in Sec. V E, Monte Carlo simulations have a
the manufacturer for intercomparisons with their transfer freedom from detector positioning and response artifacts,
standards. The AAPM Calibration Laboratory Accreditation smaller estimated uncertainty, and can yield artifact-free
subcommittee, in conjunction with NIST, selected the NIST dose-rate estimates at distances shorter or longer than those
WAFAC calibration date as the reference date for each accessible by TLD measurement techniques. However, the
source model, converting stated source strengths to the NIST accuracy of Monte Carlo is inherently limited by the inves-
WAFAC 1999 standard as corrected in 2000. This date, as tigators’ ability to accurately delineate the source internal
described on ADCL calibration reports as the vendor trace- geometry. Few Monte Carlo studies have systematically
ability date, gives the date of the WAFAC calibration mea- evaluated the effects of geometric uncertainty, internal com-
surements to which the certified calibration is traceable. All ponent mobility, tolerances in the fabrication of sources, and

TABLE I. NIST standard WAFAC calibration dates for air kerma strength for each manufacturer, and dose rate constant values. Note that for a given source
type, the % change in L from the 1999 value is not necessarily equal to the average % change in air-kerma strength due the 1999 NIST WAFAC anomaly
because some of the L values were calculated based on air-kerma strength measurements of a single seed.
NIST date used by ADCL CONL % difference in L

Manufacturer and source type and NIST as standard @ cGy•h21 •U21 # from 1999 value
125
Amersham 6702 I April 15, 1998 1.036 N/A
125
Amersham 6711 I April 15, 1998 0.965 N/A
125
Best Industries 2301 I August 18, 2000 1.018 13.3%
125
NASI MED3631-A/M I June 30, 2001 1.036 11.0%
125
Bebig/Theragenics I25.S06 I January 27, 2001 1.012 12.2%
125
Imagyn IS-12501 I October 21, 2000 0.940 13.5%
103
Theragenics 200 Pd July 8, 2000 0.686 14.0%
103
NASI MED3633 Pd April 23, 2001 0.688 14.3%
small manufacturing changes on the uncertainty of calculated source geometry functions are presented in Tables II and III,
dose-rate distributions. Therefore, the use of Monte Carlo respectively. Details used in the determination of g(r) for
values without confirmation by experimental studies is each source model are provided in Appendix A.
highly undesirable. Drawbacks of TLD dosimetry include ~a!
4. 2D anisotropy function
limited precision of repeated readings and spatial resolution;
~b! a large and somewhat uncertain relative energy response Because Monte Carlo based datasets generally have supe-
correction; ~c! failure of most investigators to monitor or rior smoothness, spatial and angular resolution, and distance
control the composition of the measurement medium. For range, all anisotropy functions recommended in this protocol
these reasons, the LIBD recommends using an equally are derived from Monte Carlo results which have been vali-
weighted average of the average measured ~e.g., using dated by comparison to less complete experimental datasets.
TLDs! and average calculated ~e.g., Monte Carlo derived! A graphical comparison of datasets was performed, and the
values ~see Table I for each source! since the two recom- agreement between the Monte Carlo datasets and the experi-
mended dosimetry characterization techniques have comple- mental datasets was again expected to be 610%. For u
mentary strengths and limitations. .30°, observed differences between the datasets were typi-
The values in Table I are the average of experimental and cally ,5% with a maximum of about 9%. For u <30°, dif-
Monte Carlo results, e.g., CONL, for each source model. Ex- ferences were larger ~typically ;10% with maximum
perimental results normalized to the 1985 Loftus NIST stan- ;17%), and are attributed to volume averaging and the
dard have been corrected to agree with the NIST WAFAC high-dose-rate gradient near the source longitudinal-axis as
1999 standard as corrected in 2000.158 In those cases where well as uncertainties in the source geometry assumed by
the authors did not correct for differences between Solid Wa- Monte Carlo simulations. Tables IV–XI present the F(r, u )
ter™ and liquid water, corrections were applied based on and f an(r) data for the sources examined herein.
Williamson’s Monte Carlo calculations.37 Also, a number of
the cited experimental dosimetry papers published dose-rate C. Uncertainty analysis
constants are normalized to WAFAC measurements per-
Most of the experimental and computational investiga-
formed in 1999. In these cases, appropriate corrections were
tions, especially those published prior to 1999, failed to in-
made to the published dose-rate constant values.
clude a rigorous uncertainty analysis. Thus, the AAPM rec-
ommends that the generic uncertainty analysis described by
3. Radial dose function Table XII, based on the best estimate of uncertainty of the
For each source, Monte Carlo values of g(r) were graphi- measured dose rate constants used to compute the CONL val-
cally compared with experimental values. A comparison of ues recommended by this report, should be included hence-
the Monte Carlo and experimental g(r) results were ex- forth. In the future, the AAPM recommends that dosimetry
pected to show an average agreement of 610%. While the investigators include rigorous uncertainty analyses, specific
observed differences were typically ,5% for r<5 cm, sys- to their methodology employed, in their published articles.
tematic differences as large as 10% were observed due to use Table XII, based on the works of Gearheart et al.38 and Nath
of outdated Monte Carlo cross-section libraries. Experimen- and Yue,39 assigns a total 1s uncertainty of 8%–9% to TLD
tal values are difficult to measure at r,1 cm, but Monte measurements of dose-rate constant and an uncertainty of
Carlo calculation of dose rate values are often available at 5%–7% to measurements of relative quantities.
smaller distances. In each case, the most complete dataset Based on results of Monroe and Williamson,37,40 purely
~typically Monte Carlo values! was used since values were Monte Carlo estimates of the transverse-axis dose-rate per
more readily available over a larger range of distances ~es- unit air-kerma strength typically have uncertainties of
pecially at clinically significant distances closer than 1 cm! 2%–3% at 1 cm and 3%–5% at 5 cm, depending on the type
than provided by experimental measurements. The CONg(r) and magnitude of internal seed geometric uncertainties.
data for all 125I and 103Pd sources and for line- and point- Since relatively little has been published on estimation of

125
TABLE II. Consensus g(r) values for six I sources. Interpolated data are boldface, and italicized data are nonconsensus data obtained from candidate
datasets.
Line source approximation Point source approximation
Amersham Amersham Best NASI Bebig Imagyn

6702 6711 2301 MED3631-A/M I25.S06 IS12501 Amersham Amersham Best NASI Bebig Imagyn
r @cm# L53.0 mm L53.0 mm L54.0 mm L54.2 mm L53.5 mm L53.4 mm 6702 6711 2301 MED3631-A/M I25.S06 IS12501
0.10 1.020 1.055 1.033 1.010 1.022 0.673 0.696 0.579 0.613 0.631
0.15 1.022 1.078 1.029 1.018 1.058 0.809 0.853 0.725 0.760 0.799
0.25 1.024 1.082 1.027 0.998 1.030 1.093 0.929 0.982 0.878 0.842 0.908 0.969
0.50 1.030 1.071 1.028 1.025 1.030 1.080 1.008 1.048 0.991 0.985 1.001 1.051
0.75 1.020 1.042 1.030 1.019 1.020 1.048 1.014 1.036 1.020 1.008 1.012 1.040
1.00 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000
1.50 0.935 0.908 0.938 0.954 0.937 0.907 0.939 0.912 0.945 0.962 0.942 0.912
2.00 0.861 0.814 0.866 0.836 0.857 0.808 0.866 0.819 0.875 0.845 0.863 0.814
3.00 0.697 0.632 0.707 0.676 0.689 0.618 0.702 0.636 0.715 0.685 0.695 0.623
4.00 0.553 0.496 0.555 0.523 0.538 0.463 0.557 0.499 0.562 0.530 0.543 0.467
5.00 0.425 0.364 0.427 0.395 0.409 0.348 0.428 0.367 0.432 0.401 0.413 0.351
6.00 0.322 0.270 0.320 0.293 0.313 0.253 0.324 0.272 0.324 0.297 0.316 0.255
7.00 0.241 0.199 0.248 0.211 0.232 0.193 0.243 0.200 0.251 0.214 0.234 0.195
8.00 0.179 0.148 0.187 0.176 0.149 0.180 0.149 0.189 0.178 0.150
9.00 0.134 0.109 0.142 0.134 0.100 0.135 0.110 0.144 0.135 0.101
10.00 0.0979 0.0803 0.103 0.0957 0.075 0.0986 0.0809 0.104 0.0967 0.076
systematic ~type B! uncertainties of Monte Carlo-based dose y5 f ~ x 1 ,...,x N ! ,

estimation, the following sections apply the principles of un- ~15!
S D
certainty analysis, as outlined in NIST Technical Note N 2 N21 N
]f ]f ]f
1297,41 to estimation of total uncertainty of Monte Carlo s 2y 5 ( s x2i 12 ( ( s ,
dose-rate constants, MCL, Monte Carlo radial dose functions i51 ]xi i51 j5i11 ] x i ] x j x i ,x j
MCg(r), consensus dose-rate constants, CONL, and absolute
transverse-axis dose as evaluated by the dosimetric param- where s x i ,x j ~assumed zero here! represents the covariance
eters recommended by this report. of the two variables. For each dosimetric quantity,
NIST Report 1297 recommends using the Law of Propa- Y (L,g(r), etc.!, the total percent uncertainty, % s Y , is con-
gation of Uncertainty ~LPU! to estimate the uncertainty of a sidered to be composed of three sources: type B uncertainty
quantity y, that has a functional dependence on measured or due to uncertainty of the underlying cross sections, % s Y u m ;
estimated quantities x 1 ,...,x N , as follows: type B uncertainties arising from uncertainty of the seed geo-
TABLE III. Consensus g(r) values for two 103Pd sources. Interpolated data are boldface, and italicized data are
nonconsensus data obtained from candidate datasets.
Line source approximation Point source approximation
Theragenics 200 NASI MED3633

r @cm# L54.23 mm L54.2 mm Theragenics 200 NASI MED3633
0.10 0.911 0.494

0.15 1.21 0.831
0.25 1.37 1.331 1.154 1.123
0.30 1.38 1.322 1.220 1.170
0.40 1.36 1.286 1.269 1.201
0.50 1.30 1.243 1.248 1.194
0.75 1.15 1.125 1.137 1.113
1.00 1.000 1.000 1.000 1.000
1.50 0.749 0.770 0.755 0.776
2.00 0.555 0.583 0.561 0.589
2.50 0.410 0.438 0.415 0.443
3.00 0.302 0.325 0.306 0.329
3.50 0.223 0.241 0.226 0.244
4.00 0.163 0.177 0.165 0.179
5.00 0.0887 0.098 0.0900 0.099
6.00 0.0482 0.053 0.0489 0.054
7.00 0.0262 0.028 0.0266 0.028
10.00 0.00615 0.00624

TABLE IV. F(r, u ) for Amersham model 6702.
r @cm#
Polar angle
u ~degrees! 0.5 1 2 3 4 5
0 0.385 0.420 0.493 0.533 0.569 0.589

5 0.413 0.472 0.546 0.586 0.613 0.631
10 0.531 0.584 0.630 0.660 0.681 0.697
15 0.700 0.700 0.719 0.738 0.749 0.758
20 0.788 0.789 0.793 0.805 0.810 0.814
30 0.892 0.888 0.888 0.891 0.892 0.892
40 0.949 0.948 0.944 0.944 0.944 0.944
50 0.977 0.973 0.967 0.967 0.967 0.967
60 0.989 0.985 0.983 0.983 0.983 0.983
70 0.996 0.992 0.990 0.990 0.990 0.990
80 1.000 0.998 0.998 0.998 0.998 0.998
f an(r) 0.986 0.960 0.952 0.951 0.954 0.954
TABLE V. F(r, u ) for Amersham model 6711.
r @cm#
Polar angle
u ~degrees! 0.5 1 2 3 4 5
0 0.333 0.370 0.442 0.488 0.520 0.550

5 0.400 0.429 0.497 0.535 0.561 0.587
10 0.519 0.537 0.580 0.609 0.630 0.645
20 0.716 0.705 0.727 0.743 0.752 0.760
30 0.846 0.834 0.842 0.846 0.848 0.852
40 0.926 0.925 0.926 0.926 0.928 0.928
50 0.972 0.972 0.970 0.969 0.969 0.969
60 0.991 0.991 0.987 0.987 0.987 0.987
70 0.996 0.996 0.996 0.995 0.995 0.995
80 1.000 1.000 1.000 0.999 0.999 0.999
f an(r) 0.973 0.944 0.941 0.942 0.943 0.944
TABLE VI. F(r, u ) for Best model 2301.
r @cm#
Polar angle
u ~degrees! 1 2 3 4 5 6 7
0 0.367 0.454 0.922 0.902 0.894 0.893 0.858

5 0.724 0.720 0.726 0.738 0.753 0.771 0.800
10 0.653 0.671 0.699 0.727 0.732 0.764 0.782
20 0.785 0.794 0.809 0.814 0.825 0.852 0.821
30 0.900 0.890 0.885 0.892 0.899 0.915 0.873
40 0.982 0.954 0.947 0.939 0.943 0.976 0.937
50 1.014 0.992 0.985 0.991 0.997 0.989 0.961
60 1.030 1.010 1.009 1.007 1.010 1.019 1.002
70 1.036 1.026 1.016 1.023 1.011 1.035 1.010
80 1.010 1.030 1.019 1.017 1.010 1.020 1.005
f an(r) 0.945 0.987 0.968 0.971 0.969 0.991 0.969

TABLE VII. F(r, u ) for NASI model MED3631-A/M.
r @cm#
Polar angle
u ~degrees! 0.25 0.5 1 2 5 10
0 1.038 0.690 0.702 0.667 0.718 0.771

10 0.984 0.700 0.662 0.676 0.728 0.758
20 0.916 0.761 0.747 0.764 0.794 0.815
30 0.928 0.854 0.846 0.852 0.871 0.878
40 0.941 0.909 0.906 0.909 0.918 0.914
50 0.962 0.949 0.949 0.950 0.958 0.954
60 0.975 0.975 0.975 0.975 0.983 0.972
70 0.991 0.989 0.992 0.990 0.993 0.989
80 0.999 0.999 1.003 0.996 0.998 0.999
f an(r) 1.288 1.008 0.952 0.945 0.948 0.948
TABLE VIII. F(r, u ) for Bebig/Theragenics model I25.S06. Italicized data are nonconsensus data obtained from
candidate datasets.
r @cm#
Polar angle
u ~degrees! 0.25 0.5 1 2 3 4 5 7
0 0.302 0.429 0.512 0.579 0.610 0.631 0.649 0.684

5 0.352 0.436 0.509 0.576 0.610 0.635 0.651 0.689
10 0.440 0.476 0.557 0.622 0.651 0.672 0.689 0.721
20 0.746 0.686 0.721 0.757 0.771 0.785 0.790 0.807
30 0.886 0.820 0.828 0.846 0.857 0.862 0.867 0.874
40 0.943 0.897 0.898 0.907 0.908 0.913 0.918 0.912
50 0.969 0.946 0.942 0.947 0.944 0.947 0.949 0.946
60 0.984 0.974 0.970 0.974 0.967 0.966 0.967 0.976
70 0.994 0.989 0.988 0.990 0.984 0.985 0.987 0.994
80 0.998 0.998 0.998 1.000 0.994 1.000 0.993 0.999
f an(r) 1.122 0.968 0.939 0.939 0.938 0.940 0.941 0.949
TABLE IX. F(r, u ) for Imagyn model IS-12501. Italicized data are nonconsensus data obtained from candidate
datasets.
r @cm#
Polar angle
u ~degrees! 1 2 3 5 7
0 0.241 0.337 0.362 0.424 0.454

10 0.327 0.399 0.440 0.486 0.510
20 0.479 0.532 0.563 0.584 0.581
30 0.634 0.663 0.681 0.706 0.700
40 0.768 0.775 0.786 0.806 0.776
50 0.867 0.870 0.878 0.875 0.849
60 0.946 0.944 0.944 0.943 0.913
70 0.986 0.985 0.987 0.974 0.955
80 0.998 0.994 1.004 0.981 0.956
f an(r) 0.867 0.886 0.894 0.897 0.879

TABLE X. F(r, u ) for Theragenics Corp. model 200. Italicized data are nonconsensus data obtained from candidate datasets.
r ~cm!
Polar angle
u ~degrees! 0.25 0.5 0.75 1 2 3 4 5 7.5
0 0.619 0.694 0.601 0.541 0.526 0.504 0.497 0.513 0.547

1 0.617 0.689 0.597 0.549 0.492 0.505 0.513 0.533 0.580
2 0.618 0.674 0.574 0.534 0.514 0.517 0.524 0.538 0.568
3 0.620 0.642 0.577 0.538 0.506 0.509 0.519 0.532 0.570
5 0.617 0.600 0.540 0.510 0.499 0.508 0.514 0.531 0.571
7 0.579 0.553 0.519 0.498 0.498 0.509 0.521 0.532 0.568
10 0.284 0.496 0.495 0.487 0.504 0.519 0.530 0.544 0.590
12 0.191 0.466 0.486 0.487 0.512 0.529 0.544 0.555 0.614
15 0.289 0.446 0.482 0.490 0.523 0.540 0.556 0.567 0.614
20 0.496 0.442 0.486 0.501 0.547 0.568 0.585 0.605 0.642
25 0.655 0.497 0.524 0.537 0.582 0.603 0.621 0.640 0.684
30 0.775 0.586 0.585 0.593 0.633 0.654 0.667 0.683 0.719
40 0.917 0.734 0.726 0.727 0.750 0.766 0.778 0.784 0.820
50 0.945 0.837 0.831 0.834 0.853 0.869 0.881 0.886 0.912
60 0.976 0.906 0.907 0.912 0.931 0.942 0.960 0.964 0.974
70 0.981 0.929 0.954 0.964 0.989 1.001 1.008 1.004 1.011
75 0.947 0.938 0.961 0.978 1.006 1.021 1.029 1.024 1.033
80 0.992 0.955 0.959 0.972 1.017 1.035 1.046 1.037 1.043
85 1.007 0.973 0.960 0.982 0.998 1.030 1.041 1.036 1.043
f an(r) 1.130 0.880 0.859 0.855 0.870 0.884 0.895 0.897 0.918
metric model, % s Y u geo ; and the type A statistical uncertainty, mated here are standard uncertainties, having a coverage fac-
% s Y u s inherent to the Monte Carlo technique. Applying Eq. tor of unity, approximating a 68% level of confidence.
~15!, one obtains
1. L uncertainty
% s Y 5 A% s Y2 u m 1% s Y2 u geo1% s Y2 u s
The influence of cross-section uncertainty was derived
5 AS D%
]Y
]m
2
S
% s m2 1 %
]Y 2
] geo D
% s Y2 u geo1% s Y2 u s ,
from the Monte Carlo data published by Hedtjarn et al.42
This paper gives Monte Carlo estimates of L and g(r) cal-
culated for two different cross-section libraries, DLC-99
~16!
~circa 1983! and DLC-146 ~1995!. The photoelectric cross
where the relative uncertainty propagation factor is defined sections of the two libraries differ by about 2% between
as 1– 40 keV, corresponding to a 1.1% change in m for the mean
]Y x ]Y photon energy emitted by 125I. Using these data to numeri-
% [ . ~17! cally estimate the derivative in Eq. ~17!, one obtains
]x Y ]x
% ] L/ ] m 50.68. Assuming that % s m 52%, 43 then uncer-
The variable x denotes either the cross-section value, m, or tainty in L due only to cross-section uncertainty, % s L u m , is
geometric dimension, geo, of interest. The uncertainties esti- 1.4%.
TABLE XI. F(r, u ) for NASI model MED3633.
r @cm#
Polar angle
u ~degrees! 0.25 0.5 1 2 5 10
0 1.024 0.667 0.566 0.589 0.609 0.733

10 0.888 0.581 0.536 0.536 0.569 0.641
20 0.850 0.627 0.603 0.614 0.652 0.716
30 0.892 0.748 0.729 0.734 0.756 0.786
40 0.931 0.838 0.821 0.824 0.837 0.853
50 0.952 0.897 0.890 0.891 0.901 0.905
60 0.971 0.942 0.942 0.940 0.948 0.939
70 0.995 0.976 0.974 0.973 0.980 0.974
80 1.003 0.994 0.997 0.994 1.000 0.986
f an(r) 1.257 0.962 0.903 0.895 0.898 0.917

TABLE XII. Generic uncertainty assessment for experimental measurements using TLDs, and Monte Carlo
methods for radiation transport calculations. Type A and B uncertainties correspond to statistical and systematic
uncertainties, respectively. All values provided are for 1 s.
TLD uncertainties
Component Type A Type B
Repetitive measurements 4.5%

TLD dose calibration ~including linac calibration! 2.0%
LiF energy correction 5.0%
Measurement medium correction factor 3.0%
Seed/TLD positioning 4.0%
Quadrature sum 4.5% 7.3%

Total uncertainty 8.6%
ADCL S K uncertainty 1.5%
Total combined uncertainty in L 8.7%
Monte Carlo uncertainties

Component r51 cm r55 cm
Statistics 0.3% 1.0%

Photoionizationa 1.5% 4.5%
Cross-sections ~2.3%!
Seed geometry 2.0% 2.0%
Source energy spectruma 0.1% 0.3%
Quadrature sum 2.5% 5.0%

a
On the transverse plane.
Estimation of geometric uncertainty, % s L u G , is a com- % s L u geo and % s L u m values, discussed above, % s L varies
plex and poorly understood undertaking. Each source design from 2.5% to 3.7% for the eight seeds described in this re-
is characterized by numerous and unique geometric param- port. Thus, assuming a standard or generic % s L of 3% for
eters, most of which have unknown and potentially corre- all Monte Carlo studies seems reasonable.
lated probability distributions. However, a few papers in the
literature report parametric studies, in which the sensitivity
of dosimetric parameters to specified sources of geometric
variability is documented. For example, Williamson has 2. CONL uncertainty
shown that the distance between the two radioactive spheri- This report defines the consensus dose-rate constant as
cal pellets of the DraxImage 125I source varies from 3.50 to
3.77 mm.44 This leads to a source-orientation dependent CONL5 a • EXPL1 ~ 12 a ! • MCL,
variation of approximately 5% in calculated dose-rate con- where a 50.5. Applying the LPU law from Eq. ~15!, obtains
S D
stant. Rivard published a similar finding for the NASI model 2
MED3631-A/M 125I source.45 If this phenomenon is modeled EXPL
%s2 L5 a
2
%s2
EXPL
by a Type B rectangular distribution bounded by the mini- CON
CONL
S D
mum and maximum L values, the standard uncertainty is 2
MCL
given by 1 ~ 12 a ! 2 %s2 1~ % s B !2. ~19!
MCL
CONL
u L max2L minu
% s L u geo5100 . ~18! % s B is an additional component of uncertainty in CONL due
2L̄) to the possible bias in the average of the results of experi-
For the DraxImage source, Eq. ~18! yields a % s L u geo mental and Monte Carlo methods, and is modeled by a Type
51.4%. For the Theragenics Corporation Model 200 seed, B rectangular distribution, bounded by EXPL and MCL. 47 The
Williamson has shown that L is relatively insensitive to Pd bias B is assumed to be equal to zero, with standard uncer-
metal layer thickness or end weld configuration.46 Thus 2% tainty given by % s B 5100u EXPL2 MCL u /(2) CONL). For
seems to be a reasonable and conservative estimate of the various seed models presented in this protocol, % s B var-
% s L u geo . ies from 0.4% to 1.5%, depending on the magnitude of the
The reported statistical precision of Monte Carlo L esti- discrepancy between Monte Carlo and TLD results. Assum-
mates ranges from 0.5% for Williamson’s recent studies to ing % sEXPL 58.7% along with model-specific % sMCL and
3% for Rivard’s MED3631-A/M study.44,45 Thus for a typical % s B values, % sCONL varies from 4.6% to 5.0%. Thus for
Williamson study, one obtains a % s L of 2.5%. Using the the purposes of practical uncertainty assessment, a model
% s L u s reported by each investigator along with the standard independent % sCONL value of 4.8% is recommended.

As common in the field of metrology, future changes and a conservative assumption since correlation of statistical
improvements to the NIST WAFAC air-kerma strength mea- variance between two Monte Carlo-derived variables gener-
surement system and other calibration standards are ex- ally reduces the total variance of any algebraic combination
pected, and may somewhat impact dose rate constant values. of such variables.
For example, the international metrology system has recently Determination of EXPg(r) values using a diode is afflicted
revised the 60Co air-kerma standard for teletherapy beams. by measurement uncertainties similar to those in TLD mea-
Consequently, NIST has revised its 60 Co air-kerma standard surement of EXPL. In measurements by Li et al.,48 diode
effective July 1, 2003 by about 1% due to new, Monte Carlo readings were corrected for energy response variations
based wall corrections (k wall) for graphite-wall ionization through the range of measurement distances ~0.5 cm to 7 cm!
chambers. Changes in the NIST 60Co air-kerma strength as well as volume averaging of diode active detector
standard, which is the basis for AAPM TG-51 teletherapy element.49 Measurement uncertainties included ~i! a diode
beam calibrations, will only affect ~i! detectors calibrated detector positioning uncertainty of 0.1 mm, ~ii! electrometer
using either 60Co beams directly, or ~ii! detectors calibrated leakage current correction uncertainty, and ~iii! measurement
using high-energy photon beams ~e.g., 6 MV! calibrated with reproducibility. Inspection of these uncertainties indicates
ionization chambers which were themselves calibrated using that at short distances (r,1 cm), positioning uncertainties
the 60Co standard. As long as these changes are small in contributed the most to the overall uncertainties; whereas at
comparison to the aforementioned value of 8.7%, the clinical large distances the electrometer leakage current dominated.
medical physicist need not be immediately concerned. The overall measurement uncertainties were calculated by
combining all three standard uncertainty components in
quadrature, and normalizing to 1 cm as discussed above. In
3. g „ r … uncertainty
summary, the combined standard uncertainty of EXPg(r) was
For the sources considered in this report, except for the 5.5% at r50.5 cm, approximately 3% over the distance
NASI model MED3631-A/M 125I source, the Monte Carlo- range of 1.5 to 4 cm, and gradually rose to 15% at r
derived values, MCg(r), were adopted as the consensus 57 cm.
dataset for radial dose function, CONg(r). For this one seed,
the CONg(r) values were based on diode measurements by Li 4. Anisotropy function uncertainty
et al.48 Therefore, an uncertainty analysis of both MCg(r)
and EXPg(r) are presented separately. Currently, sufficient data are not available in the literature
Since MCg(r) is a relative quantity that is not combined to provide a similarly comprehensive uncertainty analysis for
with experimental results which are used only for validation, Monte Carlo-derived 1D and 2D anisotropy functions. More
it is therefore assumed that experimental data do not contrib- research is needed, particularly in the area of identifying geo-
ute to the uncertainty of CONg(r). Again, three sources of metric parameters to which F(r, u ) is sensitive, such as end
uncertainty are considered: statistical variations, cross- weld configuration and thickness, capsule wall thickness, in-
section uncertainty, and geometric uncertainties. Using the ternal component alignment uncertainties, and deviation of
methods from the preceding section, % s g(r) u m is 1.8%, 0.8%, internal structures with sharp edges from their idealized ge-
and 0% at 0.1, 0.5, and 1.0 cm, respectively. As distance ometries.
increases from 2 to 5 cm, % s g(r) u m progressively increases
from 0.2% to 4.6%, respectively. % s g(r) u geo is again esti- 5. Total uncertainty of transverse-plane dose-rate
mated from Williamson’s DraxImage and Rivard’s calculations
MED3631-A/M data assuming a rectangular distribution
As Eqs. ~10! and ~11! demonstrate, absolute dose rate is
bounded by the extreme values. For the geometric variations
proportional to air-kerma strength, consensus dose-rate con-
described above, these data show a relative g(r) range of
stant, consensus anisotropy function, and consensus radial
about 8% for r,0.25 cm, and 2% at 0.5 cm, corresponding
dose function. Thus, ignoring uncertainties in anisotropy
to a % s g(r) u geo of 2.3% and 0.6% for r,0.25 and 0.5 cm,
function,
respectively. Conservatively rounding these values to 3% and
1%, respectively, % s g(r) varies from 3.5% at 0.1 cm, 0% at % s ˙D(r) 5 A% s S2 1% s 2 1% s 2 . ~20!
K CONL CONg(r)
1 cm, and 4.6% at 5 cm.
In this analysis, the uncertainty is zero at r 0 , and follows Assuming that the unexpanded uncertainty, % s S K , of
from the definition of g(r) which specifies that g(1) is the vendor-supplied calibrations is probably no better than 3%,
ratio of the same two identical numbers. In the general un- Eq. ~20! leads to the conclusion that the total dose-
certainty propagation formula, this is equivalent to assuming calculation uncertainties are 6.7%, 5.7%, and 7.3% at 0.1, 1,
the correlation coefficient is equal to 21 when r51 cm. The and 5 cm, respectively. As the combined uncertainties asso-
correlation coefficient is the covariance divided by the prod- ciated with the consensus dosimetric parameters probably
uct of the standard deviations, so if one sets the correlation exceed 5%, clinical physicists should meticulously verify
coefficient equal to 21, then Cov(x,y)52 s x s y . Letting and assign the air-kerma strength values associated with
y5x, analogous to our case of Ḋ(r)5Ḋ(1), Cov(x,x)5 clinical seeds to ensure that % s S K does not dominate the
2 s 2x . Substitution in the propagation of uncertainties for- total uncertainty of the dose calculation. Note that this as-
mula yields s ˙D(r),D ˙ (1) 50 when rÞ1 cm. This appears to be sessment does not account for uncertainties in evaluating the

effects of anisotropy; the analysis presented herein is neither rise to dose-delivery errors. Therefore, it is necessary to rec-
complete nor rigorous: the AAPM supports further research ommend minimum spatial resolutions and ranges for which
in the area of brachytherapy dose-calculation uncertainties. these parameters should be specified.
1. Air-kerma strength
V. RECOMMENDED METHODOLOGY TO OBTAIN For experimental measurement of absolute dose rates to
BRACHYTHERAPY DOSIMETRY PARAMETERS water, at least one source should have direct traceability of
In this section, the AAPM recommends a list of method- S K to the 1999 NIST WAFAC calibration standard. Other
ological details that should be described in brachytherapy sources used in the experiment should have a precisely trans-
dosimetry publications based upon either experimental or ferred air-kerma strength using high-precision transfer de-
theoretical methods, along with more prescriptive guidelines vices such as well-characterized well-ionization chambers
on performing such studies. The list of key details docu- and secondary standards maintained by the investigator as
mented in this report for each study is reviewed later. To well as the manufacturer’s laboratories. The investigator us-
better appreciate results from a particular dosimetric mea- ing experimental techniques should state the NIST S K cali-
surement and its uncertainties, the reader is referred to a bration uncertainty in the evaluation of L. Use of another
listing of parameters needed to assess data for TLD source, even the same model, to cross-calibrate dosimeters
measurements.50 Unfortunately, this level of description was for the determination of L is highly discouraged since un-
not realized in many of the papers cited. When key data or certainties propagate and hidden errors may exist.
methodological details were missing from a published paper,
the author was contacted and asked to provide the missing 2. Dose-rate constant
information. The experimental investigator should rigorously control
A. General recommendations and try to minimize all detector response artifacts such as
dose-rate dependence, dose response nonlinearity, energy de-
Since publication of TG-43,1 the LIBD has published pendence, volumetric averaging, temporal stability of read-
guidelines on dosimetric prerequisites for low-energy ings and calibration coefficients, and accuracy of detector
photon-emitting interstitial brachytherapy sources.9 The aim positioning both in the source measurement setup and the
of those recommendations was to assure that multiple dosim- detector calibration setup. These issues should be discussed
etry studies, each subjected to the rigors of the peer-review in the measurement methodology section of the published
process, were available for each source model. However, that paper, and a rigorous uncertainty analysis should also be pro-
publication gave few technical guidelines to investigators vided.
publishing reference-quality dose-rate distributions derived Experimentally, L is evaluated by taking the ratio of the
from measurements or radiation transport calculations. Based absolute dose rate, Ḋ(r 0 , u 0 ) ~the only absolute dose rate
on the LIBD experience of analyzing dosimetry datasets,35 required to define TG-43 dosimetry parameters! and the mea-
more detailed recommendations on dosimetry methodology sured air-kerma strength of the source, decayed to the time of
and data analysis are presented in this section. These recom- dose-rate measurement. Typically 8 –10 sources are used,
mendations are intended to define minimum requirements for with at least one source having direct traceability to a NIST
future source dosimetry studies so that the accuracy and con-
calibration. At least 15 measurements of Ḋ(r 0 , u 0 ) are gen-
sistency of the consensus datasets may be improved.
erally performed. For example, multiple measurements of
Ḋ(r 0 , u 0 ) around a single NIST WAFAC-calibrated source
could be made by placing multiple TLDs in different quad-
B. Preparation of dosimetry parameters
rants of the transverse-plane.
Dosimetric parameters should be tabulated for both 1D Monte Carlo radiation transport codes commonly provide
and 2D dose-calculation models. This will require the inves- direct estimates of absorbed dose or collision kerma per
tigator to calculate the geometry function and the radial dose number of histories simulated ~or some other internal nor-
function using both point-source ~1D! and line-source ~2D! malization quantity, e.g., number of disintegrations, propor-
geometry functions ~see Sec. III A 3!. Consequently, the in- tional to the number of primary photons simulated!. Two
vestigator should always specify the active length used for simulations are necessary: one with the source model embed-
the 2D line-source geometry function. As previously stated ded in a phantom, yielding estimates of dose at specified
in Sec. III B, Eq. ~11! is the recommended formalism for the points, and a second simulation with the source model posi-
1D approximation. tioned within a vacuum or a large air sphere. The dose-rate
Specification of dosimetry parameters at a few distances constant can then be estimated using the following ratio @Eq.
or angles will not allow a sufficiently complete or accurate ~21!# since the numerator and denominator are similarly nor-
dose reconstruction of the 2D dose distribution adequate for malized, and the normalization constant used by the Monte
clinical implementation. In many instances, the underlying Carlo code is irrelevant,51
dose distribution will have high gradients. Inadequate spatial
resolution may result in inaccurate interpolation by brachy- ḋ ~ r 0 , u 0 !
L5 . ~21!
therapy treatment planning systems, unnecessarily giving sK

125 103
TABLE XIII. Recommended nuclear data for I and Pd for brachytherapy dosimetry.
125 103
I (half-life559.4060.01 days) Pd (half-life516.99160.019 days)
Photon energy ~keV! Photons per disintegration Photon energy ~keV! Photons per disintegration
27.202 0.406 20.074 0.224

27.472 0.757 20.216 0.423
30.98 0.202 22.72 0.104
31.71 0.0439 23.18 0.0194
35.492 0.0668 39.75 0.00068
294.98 0.00003
357.5 0.00022
497.1 0.00004
Weighted mean energy528.37 keV Total51.476 Weighted mean energy520.74 keV Total50.7714
125 2 21 21 103 2 21
I G 5 keV50.0355 m Gy•m •h •Bq Pd G 5 keV50.0361 m Gy•m •h •Bq21
The parameter ḋ(r 0 , u 0 ) is the dose rate per history esti- mum of six data points to fully constrain the fit. Because
mated using Monte Carlo methods at the reference position, these fits tend to perform poorly at small distances where the
and s K is the air-kerma strength per history estimated using dose rate is highest, care should be taken to assure the fit is in
Monte Carlo methods. Note the lower-case notation used to good agreement at these positions. Acceptable levels of
differentiate the normalized parameter, e.g., dose rate per his- agreement are outlined in the following section ~Sec. V!.
tory (cGy h21 history21 ) as compared to absolute dose rate
(cGy h21 ). Although Monte Carlo studies are potentially
free from experimental artifacts such as positioning uncer- 4. 2D anisotropy function
tainties, energy response corrections, and signal-to-noise ra-
tio limitations, such simulations require an accurate and When reporting 2D anisotropy function data, at a mini-
complete geometric model of the source, selection of an ap- mum, F(r, u ) should be tabulated at radial distances,
propriate cross-section library, and careful selection of dose r5 $ 0.5, 1, 2, 3, 5, and 7 cm% for 125I and $0.5, 1, 2, 3, and 5
tallying ~estimation! and variance-reduction strategies. As cm% for 103Pd, and from u5$0° to 90° with 10° increments%
with experimental studies, Monte Carlo-based dosimetry for all sources that are symmetric about the transverse plane.
studies should include a complete uncertainty analysis such To minimize extrapolation errors ~see Appendix C!, data
as that provided in Sec. IV C. should be determined over as wide a radial distance range as
reasonably possible given the experimental method or calcu-
lation technique. To minimize interpolation errors, data
3. Radial dose function should also be obtained such that bilinear interpolation be-
The minimum distance range over which transverse-plane tween various F(r, u ) data points produce errors less than
dose-rate data shall be measured or calculated should be 2%. When measuring 2D anisotropy function data with
from 0.5 to 7 cm for 125I and from 0.5 to 5 cm for 103Pd. sources that are cylindrically symmetric and symmetric
Special attention to accuracy and completeness should be about the transverse plane ~four identical quadrants!, it is
given to the 0.5 cm to 1.5 cm distance range, which is the recommended that investigators determine relative dose rates
most critical for accurately calculating dose for typical pros- a minimum of three times at each position analyzed. For
tate and other volume implants and for accurately relating source designs that are asymmetric about the transverse
absolute dose rates ~via the dose-rate constant! to the relative plane or exhibit internal component shifts that may result in
dose distribution. However, accurate dose characterization at asymmetric dose distributions, F(r, u ) should be similarly
shorter distances is critical for some procedures ~e.g., epis- tabulated, except with u5$0° to 180° with 10° increments%.
cleral eye plaque therapy! and for estimating hotspots in all Because of the increased sampling capabilities using Monte
implants. Accurate dose distributions at larger distances also Carlo methods in comparison to experimental methods, in-
contribute to overall dose-calculation accuracy and are im- vestigators using Monte Carlo methods should consider cal-
portant for assessing dose to organs at risk. Thus dose rates culating dose rates using much finer grids in high-gradient
should be estimated down to the smallest distance supported regions such as near the source ends ~e.g., u near 0° or
by the methodology used. Some investigators have reported 180°). Angular resolution of a few degrees near the ends
g(r) data down to 1 mm or less. Monte Carlo simulation can may be needed, with 10° resolution elsewhere. For those
easily estimate dose at submillimeter distances as well as source designs with anisotropic photon-fluence distributions
distances of 10 to 15 cm. In addition, some investigators near the transverse plane, measurements and calculations
have reported TLD or diode measurements at distances less having higher angular resolution are required to ensure that
than 1 mm. experimentally determined anisotropy functions are accu-
Since a fifth-order polynomial is frequently used for rately normalized and that air-kerma strength per simulated
smoothing transverse-plane data, one should present a mini- history for Monte Carlo simulations is accurately calculated.

TABLE XIV. Composition ~percent mass! of air as a function of relative humidity at a pressure of 101.325 kPa.
Relative humidity
~%! Hydrogen Carbon Nitrogen Oxygen Argon
0 0.0000 0.0124 75.5268 23.1781 1.2827

10 0.0181 0.0124 75.4048 23.2841 1.2806
40 0.0732 0.0123 75.0325 23.6077 1.2743
60 0.1101 0.0123 74.7837 23.8238 1.2701
100 0.1842 0.0122 74.2835 24.2585 1.2616
5. 1D anisotropy function density of 0.001 196 g cm23 . Since the composition of air
To derive 1D anisotropy function data, a solid-angle may change as a function of relative humidity, Table XIV is
weighted-average of the relative dose rates, uncorrected by provided to account for this effect.57,58 The proportion by
the geometry function, should be performed over all angles. weight of water in air of 100% relative humidity varies only
When examining small radii where u 50° or 180° would between 1% and 2%, for temperatures between 16 °C and
place the calculation point within the source, the weighting 26 °C and pressures between 735 mm Hg and 780 mm Hg.
should exclude the capsule/source volume and include only The change in mass density of saturated air is no more than
the volume outside the encapsulation. This is easily calcu- a 1% reduction with respect to that for dry air, over this
lated for radii, r, less than half the capsule length where range of temperatures and pressures. Thus, the mass density
r sin u.rcap , where r cap is the outer radius of the capsule. will be set at 0.001 20 g cm23 for both dry and moist air. For
Monte Carlo calculations, the recommended relative humid-
C. Reference data and conditions for brachytherapy ity is 40%, which corresponds to the relative humidity in an
dosimetry air-conditioned environment where measurements should be
carried out.
1. Radionuclide data
Since publication of the 1995 TG-43 protocol, the half- D. Methodological recommendations for experimental
lives, abundances and energies of photons emitted by unfil- dosimetry
tered 125I and 103Pd sources have been re-evaluated by
Compared to Monte Carlo theorists who may idealize re-
NIST.52–55 The currently recommended values are presented
ality by a theoretic construct, the experimental investigator
in Table XIII. These values should be used to interpret future
should address the variability that represents the clinical en-
experimental measurements and as source spectra in Monte
vironment. The experimental study should investigate a rea-
Carlo calculations. The recommended 125I half-life is un-
sonably large sample of sources received from multiple ship-
changed from the original TG-43 protocol. Differences be-
ments at different stages of the production stream from the
tween the recommended 103Pd half-life and that reported in
manufacturer.
TG-43 yield differences in the decay corrections exceeding
1% only for decay times .200 days. Of note is that the 125I 1. Detector choice
spectrum should now be described in terms of five different
LiF TLD remains the method of choice for the experimen-
photon energies ~previously three! with a 5% increase in the
tal determination of TG-43 dosimetry parameters for low-
number of photons per decay ~previously 1.40!. The 103Pd
energy photon-emitting brachytherapy sources.59– 63 While a
emission spectrum should now be described in terms of eight
variety of other experimental dosimeters such as diodes, dia-
discrete photon emissions ~previously two! with a 4% de-
mond detectors, miniature ionization chambers, plastic scin-
crease in the number of photons per decay ~previously 0.8!.
tillators, liquid ionization chambers, polymer gels, radio-
Although the relative number of high-energy photons emit-
graphic and radiochromic film, and chemical dosimeters
ted by 103Pd is low, their contribution to dose at distances
have been used for brachytherapy dosimetry,48,49,64 – 80 their
beyond 10 cm can be clinically relevant and should also be
validity for obtaining brachytherapy dosimetry parameters
considered for shielding calculations and exposure-control
has not yet been convincingly demonstrated for absolute
procedures.56
dose-rate measurements near low-energy photon-emitting
brachytherapy sources. For dosimetry parameters based on
2. Reference media relative measurements, some of these other dosimeters have
Water continues to be the recommended medium for ref- been successfully used. Diode detectors, in particular, are
erence dosimetry of interstitial brachytherapy sources. For well established for relative measurements.65,68,70 For 125I, Li
dosimetry calculations and measurements, it may be neces- et al. has shown that the relative energy-response correction,
sary to know the composition of various forms of water and although large, is independent of the point of measurement.49
air. Pure, degassed water is composed of two parts hydrogen However, validity of the results of absolute and relative do-
atoms and one part oxygen atoms, with a mass density of simetry parameters using these experimental dosimeters
0.998 g cm23 at 22 °C. Reference conditions for dry air are ~other than LiF TLDs and diodes! remains to be demon-
taken as 22 °C and 101.325 kPa ~760 mm Hg! with a mass strated through comparison of results with established Monte

Carlo and experimental techniques using well-characterized its composition, Solid Water™ and similar water substitutes
125
I or 103Pd sources ~such as those contained in this proto- require solid-to-liquid water conversion corrections ranging
col!. Multiple publications of results in peer-review journals from within 5% to 15% from unity in the 1–5 cm range.
by independent investigators ~see the second meaning of ‘‘in- Alternative materials need to be researched by future inves-
dependent studies’’ in Sec. V F! are desirable to demonstrate tigators. Because some of the low-Z media such as polysty-
independence and consistency. Therefore, use of these ex- rene, polymethylmethacrylate, or plastic water ~model
perimental dosimeters is an area of future research of signifi- PW2030 by Computerized Imaging Reference Systems, Inc.!
cant scientific value. For measuring brachytherapy dosimetry generally have more uniform and better-characterized com-
parameters, detectors should have the following properties: positions, these media may be possible candidates for future
low-energy photon-emitting brachytherapy dosimetry stud-
~a! Detectors should have a relatively small active volume
ies. However, values for their plastic-to-water conversion co-
such that effects of averaging of high-gradient dose
efficients, which are expected to be larger than correspond-
fields over this volume are negligible or are accurately
ing Solid Water™ corrections, need to be accurately
accounted for by correction coefficients.
determined for dosimetrically well-characterized source
~b! A well-characterized energy-response function such
models, such as those covered in this protocol, and validated
that differences between the calibration energy and ex-
by independent investigators in peer-reviewed publications.
perimentally measured energy are either negligible or
The relative energy response correction, E(r), is the larg-
may be quantitatively accounted for.
est single source of Type B ~systematic! uncertainty for TLD
~c! Sufficient precision and reproducibility to permit dose-
and other secondary dosimeters used in brachytherapy do-
rate estimation with 1s statistical ~Type A! uncertain-
simetry. It is defined as the ratio of TLD response per unit
ties <5%, and 1s systematic uncertainties ,7%. For
dose in water medium at position r in the brachytherapy
example, TLD statistical uncertainties may be im-
source geometry, to its response per unit dose in the calibra-
proved through repeated measurement at a given loca-
tion geometry, usually a calibrated 60Co or 6 MV x-ray
tion, and systematic uncertainties may be improved
beam.83 In general, E(r) depends on source-to-detector
through measuring chip-specific calibration coeffi-
distance,83 r, and may include corrections for volume aver-
cients. Typical statistical and systematic uncertainties
aging ~influence of dose gradients in the TLD volume!, de-
for 13131 mm3 TLD-100 chips are 4% and 7%, re-
tector self-absorption, medium displacement, and conversion
spectively, with total combined uncertainties of
from the measurement medium to liquid water. Most inves-
7–9 %.81 Therefore, 13131 mm3 TLD-100 chips are
tigators treat E(r) as a distance-independent constant, al-
considered a valid detector to perform the aforemen-
though when it includes volume-averaging and solid-to-
tioned absolute and relative measurements.
liquid water corrections, as is often the case for Monte Carlo
Because none of the experimental dosimeters satisfy the estimates, E(r) varies significantly with distance.59 This cor-
above prerequisites for absolute dose measurement, LiF ther- rection can be evaluated by irradiating TLD detectors to a
moluminescent dosimetry is currently the method of choice known dose in free space in a calibration low-energy x-ray
for experimental determination of the dose-rate constant and beam having a spectrum that matches the brachytherapy
is the most extensively ~but not only! validated methodology spectrum of interest. For TLD-100 and liquid-water measure-
for relative dose measurement. Several important issues in ment medium, values ranging from 1.39 to 1.44 for 125I,
TLD dosimetry are discussed in more detail in the following relative to 4 MV x rays or 60Co rays, have been
section. reported.84 – 86 For 125I, Meigooni et al. and Reft have shown
that E(r) values inferred from in-air measurements depend
on TLD size.82,86 Since free-air measurements relate TLD
2. Medium and energy response characterization reading to dose in a void left by removing the chip, a re-
It is necessary that the measurement medium should also placement correction ~2%–5%!, is needed to correct for the
be well characterized.82 While epoxy-based substitutes for phantom material displaced by the detector. However, pre-
water, such as Solid Water™ by Gammex-RMI or Virtual cise measurement of E(r) is difficult because ~i! photons
Water™ by MED-TEC Inc., have liquid–water conversion from the low-energy tail of the Bremsstrahlung spectrum
coefficients that differ from unity by less than 5% for high- bias the measurements to an unknown extent, ~ii! the limited
energy teletherapy beams, coefficients range from within 5% precision of TLD readout, and ~iii! the relatively large uncer-
to 15% from unity for low-energy photon-emitting sources. tainty of ion chamber dosimetry in this energy range. Recent
Recently, the measured calcium concentration of Solid Wa- authors have assigned an uncertainty of 5% to E(r). 86,87 An
ter™ was found to have deviated from the vendor’s specifi- alternative to the experimental approach is to calculate E(r)
cation by as much as 30%.59 Therefore, when Solid Water™ directly by Monte Carlo simulation.59,83 Although volume-
is used in experimental dosimetry, the atomic composition of averaging, displacement and detector self-attenuation correc-
the material used should be measured and correction coeffi- tions can be easily included, the method assumes that TLD
cients based on the measured composition of Solid Water™ response is proportional to energy imparted to the detector
should be used. Although Solid Water™ is the most widely ~intrinsic linearity!, an assumption which has been ques-
used material for TG-43 reference dosimetry, it has several tioned for some TLD phosphors and annealing and glow-
shortcomings. In addition to concerns over the constancy of curve analysis techniques.62 For the widely used TLD-100

chips using Cameron annealing and readout techniques, the ~9! NIST S K value and uncertainty used for the measured
evidence for intrinsic linearity is controversial. Das et al. source~s!, and
compared the Monte Carlo and the experimental free-air ~10! uncertainty analysis section assessing statistical and
x-ray beam approaches. Their measured relative responses systematic uncertainties and their cumulative impact.
~1.42–1.48! were in good agreement ~relative to stated 4%
experimental precision! with measurements reported by
other investigators and with their own Monte Carlo E. Methodological recommendations for Monte Carlo-
calculations.63 However, a recent paper by Davis et al. con- based dosimetry
cludes the opposite, that the measured TLD-100 energy re- Monte Carlo codes used to model photon transport for
sponse correction is underestimated by Monte Carlo calcula- brachytherapy dose calculation should be able to support de-
tions by 10% to 5% in the 24 to 47 keV energy range.88 Their tailed 3D modeling of source geometry and appropriate
measured E(r) values, which have stated uncertainty of dose-estimation techniques. In addition, they should be based
0.6%, are about 10% larger than previously reported upon modern cross-section libraries and a sufficiently com-
measurements,63,84 – 86 having values ranging from 1.58 to plete model of photon scattering, absorption, and secondary
1.61 in the 103Pd– 125I energy range. photon creation. Codes that have been widely used for inter-
In utilizing measured or Monte Carlo E(r) estimates pub- stitial brachytherapy dosimetry include EGS, MCNP, and
lished by others, LIBD recommends that TLD experimental- Williamson’s PTRAN code.89–91 These codes have been
ists confirm that the associated measurement methodology widely benchmarked against experimental measurements or
matches their dosimetry technique with regard to TLD detec- each other, so that their appropriate operating parameters and
tor type and size, annealing and readout technique, and limitations can be considered to be well understood.68 In
megavoltage beam calibration technique. The latter requires general, the AAPM recommends Monte Carlo investigators
accounting for differences in calibration phantom material utilize such well-benchmarked codes for brachytherapy do-
and dose-specification media used by the experimentalist and simetry studies intended to produce reference-quality dose-
assumed by the selected E(r) estimate. The experimentalist rate distributions for clinical use. However, regardless of the
should confirm the appropriate volume averaging, displace- transport code chosen and its pedigree, all investigators
ment, and self-absorption corrections regardless of whether should assure themselves that they are able to reproduce pre-
they are included in E(r) or applied separately. Finally, fur- viously published dose distributions for at least one widely
ther research is needed to resolve the discrepancy between used brachytherapy source model. This exercise should be
published E(r) values, to identify the appropriate role for repeated whenever new features of the code are explored,
transport calculations in TLD dosimetry, and to reduce the upon installing a new code version, or as part of orienting a
large uncertainty associated with relative energy-response new user. Other radiation transport codes, including Monte
corrections. Carlo codes not previously used in brachytherapy dosimetry,
should be more rigorously tested and documented in the
3. Specification of measurement methodology peer-reviewed literature before proposing to use their results
clinically. This is especially true for other types of transport
The experimental investigator should describe the follow-
equation solutions, including multigroup Monte Carlo, dis-
ing important features of the measurement materials and
crete ordinates methods,92 and integral transport solutions
methods to permit assessment of the results:
that have been proposed for brachytherapy dosimetry.93,94
~1! description of the external and internal source geometry, Due to the short range of the secondary electrons pro-
~2! brachytherapy source irradiation geometry, orientation, duced by interactions from photons emitted by the radionu-
and irradiation timeline, clides covered in this protocol, electron transport is not re-
~3! radiation detector calibration technique ~including proto- quired and collision kerma closely approximates absorbed
col from which the technique is derived! and energy re- dose. Since the investigator performing Monte Carlo analysis
sponse function, E(r), can control many features of the transport calculations, it is
~4! radiation detector ~dimensions, model No., and vendor! imperative that the salient details be described in publica-
and readout system ~e.g., electrometer unit model No. tions presenting Monte Carlo-derived brachytherapy dosim-
and settings, or TLD readout unit model No., vendor, etry data. For instance, the collisional physics model should
time-temperature profiles, and annealing program!, be described. The standard model used by experienced
~5! measurement phantom ~composition, mass density, di- Monte Carlo users includes incoherent scattering corrected
mensions, model No., and vendor!, for electron binding by means of the incoherent scattering
~6! phantom dimensions and use of backscatter ~at least 5 factor, coherent scattering derived by applying the atomic
cm backscatter is recommended for 125I and 103Pd dosim- form factor to the Thompson cross section, and explicit
etry measurements!, simulation of characteristic x-ray emission following photo-
~7! estimation of the impact of volume averaging on the electric absorption in medium- and high-atomic number me-
results at all detector positions, dia. For sources containing Ag or Pd, it is imperative that, if
~8! number of repeated readings at each position, the num- characteristic x-ray production is not explicitly simulated,
ber of different sources used, and the standard deviation the primary source spectrum be appropriately augmented to
of the repeated readings, include their presence.

1. Specification of Monte Carlo calculation be verified through the use of physical measurements,45
methodology transmission radiography,44 and autoradiography.98 Just
A list of key features that should be specified by the in- as the TLD experimentalist should measure an appropri-
vestigator in the publication follows: ate sample of sources, the Monte Carlo investigator
should quantify the geometric variations in a sample of
~1! radiation transport code name, version number, and ma- similar size.
jor options if any, ~5! The impact of volume-averaging artifacts should be lim-
~2! cross-section library name, version number, and cus- ited to ,1% through the appropriate choice of estima-
tomizations performed if any, tors ~tallies! and scoring voxels if used.
~3! radiation spectrum of the source ~consider Table XIII in ~6! Calculations of d(r, u ) to derive F(r, u ) should include
Sec. V C 1!, high-resolution sampling in high-gradient regions such
~4! manner in which dose-to-water and air-kerma strength as near the source ends or in regions where internal
are calculated: name of estimator or tally, whether or not source shielding causes abrupt changes in d(r, u ) and
transport was performed in air and how attenuation cor- subsequently F(r, u ).
rection coefficients were applied, and how suppression ~7! k(d) should be modeled as a function of polar angle for
of contaminant x-ray production for k̇ d (d) calculations s K simulation and L derivation. Williamson has shown
was performed to be compliant with the NIST S K,N99 that for some sources, detectors with large angular sam-
standard, pling volumes ~such as the NIST WAFAC! will have a
~5! source geometry, phantom geometry, and sampling space significantly different response than point-kerma detec-
within the phantom, tors positioned on the transverse-plane ~see Appendix
~6! composition and mass density of the materials used in B.2.2 for greater detail!. When the radioactivity is dis-
the brachytherapy source, persed within or on the surface of a high-density core
~7! composition and mass density of the phantom media, with sharp corners and edges, it may be necessary to
~8! physical distribution of the radioisotope within the simulate, if only approximately, the WAFAC geometry
source, and ~dimensions and composition! to permit investigators the
~9! uncertainty analysis section assessing statistical and sys- opportunity to directly compare Monte Carlo calcula-
tematic uncertainties and their cumulative impact. tions of L with NIST-based measurements of L.
~8! Point source modeling is unacceptable.95
2. Good practice for Monte Carlo calculations ~9! Mechanical mobility of the internal source structures,
which has the potential to significantly affect the dose
For calculating brachytherapy dosimetry parameters, the
distribution, should be considered by the Monte Carlo
following requirements should be adhered to:
investigator in developing both the geometric model of
~1! Primary dosimetry calculations should be performed in a the source and the uncertainty budget.44,45
30 cm diameter liquid water phantom, but calculations in
Solid Water™ may also be performed to supplement ex-
perimental results, e.g., calculation of E(r), performed F. Publication of dosimetry results
in Solid Water™ or other solid water substitutes. Typical Previous AAPM recommendations stated that dosimetry
calculations will produce dosimetry results extending out results should be published preceding clinical
to r;10 cm, with at least 5 cm of backscatter material implementation.9 However, the journal Medical Physics es-
for 125I and 103Pd dosimetry calculations. tablished a ‘‘seed policy’’ in 2001 that, in effect, limits print-
~2! Enough histories should be calculated to ensure that do- ing of articles to Technical Notes unless they contain signifi-
simetry results have a 1s (k51, 67% confidence index! cant new science. In order to comply with this restriction
<2% at r<5 cm, and that k̇ d (d) calculations for deri- imposed by the journal, the AAPM will accept technical
vation of s K , have 1 s <1% at the point of interest. notes with limited details as acceptable, provided the full
~3! Modern, post-1980 cross-section libraries should be details as listed above are available to the committee at the
used, preferably those equivalent to the current NIST time of evaluation. This policy in no way prevents publica-
XCOM database such as DLC-146 or EPDL97. Exclude tion of the article in other journals, as other scientific jour-
or appropriately modify older cross-section libraries nals of interest to medical physicists are appropriate venues
based on Storm and Israel data.96,97 Note that EGS4, for publication of these dosimetry parameters.
EGSnrc, and MCNP all currently require modification or In a 1998 report,9 the AAPM recommended that dosime-
replacement of their default photoionization cross sec- try results be published by independent investigators, but did
tions to meet this requirement. Furthermore, moist air not offer a strict definition of what this independence entails.
best-describes experimental conditions in comparison to The spirit of the initial recommendation was to prompt pub-
dry air ~see Sec. V C 2!, and mass-energy absorption co- lication of multiple studies to assess all the TG-43 brachy-
efficients for moist air are recommended to minimize therapy dosimetry parameters, e.g., L, g(r), F(r, u ), and
systematic uncertainties. f (r). Through determining the consensus datasets for the
~4! Manufacturer-reported source dimensions and composi- brachytherapy sources evaluated in this protocol, a rigorous
tions of encapsulation and internal components should definition of the ‘‘Independence Policy’’ was adopted. There

are two aspects of this policy, and both shall be met for full tems use the single-source dosimetry data in a tabular form
compliance. as input, whereas others represent the data by means of a
The first meaning of ‘‘independent studies’’ is that they mathematical formula that requires input of certain coeffi-
are performed, written, and published by investigators cients. Some use the TG-43 dose-calculation formalism and
who are affiliated with institutions independent of the others do not. In this section, procedures for clinical imple-
source vendor and who have no major conflicts-of- mentation of the updated dosimetry parameters recom-
interest with that vendor. mended above are presented.
The medical physicist is reminded that before adopting
The second meaning of ‘‘independent studies’’ is that the recommendations presented in this report, the physicist
they are scientifically independent of one another, i.e., should implement the dose-calculation data and technique
they represent independent and distinct estimations of recommended by this report on his/her treatment planning
the same quantities. In the case of two measurement- system and quantitatively assess the influence of this action
based studies, this will usually mean that two different on dose delivery. This is best done by comparing the dose
investigators have used their own methodologies for distribution for typical implants based on the revised dose-
measuring L and sampling the relative dose distribution, calculation procedure with those based upon the currently
as TLD dosimetry is highly technique and investigator implemented algorithm for the same seed locations, source
dependent. In the case of an empirical study and a Monte strengths, and dose-calculation grid. The potential impact of
Carlo study, if properly executed, they will yield scien- these dose-calculation modifications on dose delivery rela-
tifically independent estimates of the TG-43 parameters. tive to the current dose-calculation technique should be dis-
Thus, so long as the two studies are successfully scruti- cussed with the appropriate radiation oncologist before clini-
nized by the peer-review process and satisfy the AAPM cally implementing the recommendations of this report.
scientific requirements, the empirical and Monte Carlo Finally, the comparison of old and new dose-calculation al-
investigator author lists can overlap or even be identical. gorithms for the same seed input data, and the resultant de-
It is permissible to publish the Monte Carlo and mea- cisions that may impact clinical dose delivery, should be
sured estimates in the same paper so long as the two documented for future reference and for regulatory purposes.
datasets are independently tabulated. In this context,
‘‘Not independent’’ means that the one study is used to A. Dose-calculation formalism
modify the outcomes and methods of the other to im- For this work, RTP systems are divided into those that
prove agreement between the two datasets in a manner comply fully with the TG-43 formalism, and those that do
that is not scientifically justified. not. Full compliance is defined here as the use of Eq. ~1! or
one of its two approximations, given in Eqs. ~10! and ~11!.
When possible, the authors should cite previous publica- Full compliance also requires the use of the dose-rate con-
tions where the measurement system or techniques were first stant, appropriate values of the radial dose function, and the
described, and illustrate only the key features. It does not 1D or 2D anisotropy functions that are provided in this pro-
benefit either the reader or the journal in question to continu- tocol.
ally restate the definition of TG-43 parameters or their for- For RTP systems that use the TG-43 dose-calculation for-
malism. Simply citing this protocol or the original TG-43 malism and permit customer input of dosimetry parameters,
publication will suffice. one should enter ~or verify the correct entry of! the recom-
mended parameters, and check the accuracy of the dose cal-
VI. CLINICAL IMPLEMENTATION culation algorithm. The tabulated data provided in this pro-
Dose distributions in and around clinical interstitial im- tocol should be used with such systems. In other cases,
plants are calculated using computerized radiotherapy treat- coefficients in an equation, e.g., third- to fifth-order polyno-
ment planning ~RTP! systems. For sources with radio-opaque mial fits, describing the TG-43 parameters may be required.
markers, the 3D coordinates of the centers ~or the two ends! In these cases, the RTP-specific model or formula should be
of the markers in implanted sources are determined using fit to the data provided by this protocol. For example, for
multiple-view radiographs or CT scans. The dose-rate con- systems that require a third-order ~or higher! polynomial fit
tributions from each source at the points of interest are cal- of the radial dose function, the clinical physicist is respon-
culated using a one-dimensional or two-dimensional dose- sible for obtaining the best-fit coefficients by fitting the poly-
calculation algorithm. These contributions are then summed nomial equation to appropriate g X (r) data from this protocol.
to determine the total dose rate. This procedure assumes Before implementing the dose-calculation model, it is neces-
that there are no source-to-source shielding effects, that all sary to evaluate the quality of the parametric fit. Deviation of
tissues in and around the implant are water equivalent, and the fitted data from those recommended in this protocol
that the scattering volume within the patient is equivalent to should be less than 62%. This will yield a dose-rate calcu-
that used in the consensus datasets. The term equivalent in lation accuracy of 62% over the distance range of interest.
this context means at least 5 cm of water-equivalent material The medical physicist shall take responsibility for verifying
surrounds the point of calculation. Many RTP systems are the accuracy of dosimetry data, whether the manufacturer or
available commercially and use a variety of methods to cal- the user actually enters the data.
culate clinical dose-rate distributions. Some of the RTP sys- Some RTP systems do not use the TG-43 dose-calculation

TABLE XV. Dose rates (cGy•h21 •U21 ) as a function of distance for 8 brachytherapy sources using the 1D dosimetry formalism of Eq. ~11! with interpolation
for g L (r) and f an(r).
Amersham Amersham Best model NASI model Bebig model Imagyn model Theragenics NASI model
r ~cm! model 6702 model 6711 2301 MED3631-A/M I25.S06 IS-12501 model 200 MED3633
0.5 4.119 3.937 3.813 4.112 3.922 3.426 3.014 3.184

1.0 0.995 0.911 0.962 0.986 0.950 0.815 0.587 0.626
1.5 0.413 0.368 0.413 0.420 0.398 0.334 0.199 0.215
2.0 0.213 0.186 0.220 0.207 0.205 0.169 0.0837 0.0914
3.0 0.0768 0.0643 0.0783 0.0746 0.0733 0.0582 0.0206 0.0227
4.0 0.0344 0.0284 0.0347 0.0325 0.0323 0.0246 0.00634 0.00697
5.0 0.0169 0.0134 0.0171 0.0157 0.0157 0.0118 0.00221 0.00247
6.0 0.00890 0.00688 0.00908 0.00811 0.00840 0.00592 0.000846 0.000933
7.0 0.00490 0.00373 0.00506 0.00429 0.00459 0.00328 0.000342 0.000364
algorithm and perform calculations using methods not based The dose rates calculated by the RTP system from a single
upon Eqs. ~1!, ~10!, or ~11!. In most cases, one can devise a source should be compared with the dose-rate distribution
method to force the algorithm to generate the single-source derived from the tabulated parameters and equations pre-
dose-rate distributions recommended here by using modified sented in this protocol. The user shall compare the RTP sys-
values for the dosimetry parameters required by the RTP tem calculations with the dose-rate distributions derived
system. This conversion should be performed with care. As from the appropriate 1D or 2D data tables from this protocol.
with RTP systems based on the TG-43 dose calculation for- To facilitate this comparison, Table XV presents the
malism, one should assure that the RTP system is generating transverse-plane dose rates ~incorporating the 1D anisotropy
correct single-source dose-rate data by creating a single- function!, derived from Eq. ~11! using the recommended
source treatment plan with the modified parameters before data, as a function of distance for the brachytherapy sources
clinical use. Meigooni et al. have described an example of included in this protocol. As has been recommended previ-
this approach.99 The methods used to arrive at modified data, ously by the AAPM, this comparison should yield agreement
as well as records of the evaluation of the RTP system, within 62% over all angles and over the range of radial
should be documented carefully and retained for use follow- distances commissioned.100–102
ing installation of upgrades and for inspection by regulatory The user should determine the range of distances from the
authorities. Extreme caution should be exercised whenever source over which the RTP calculations meet this recom-
parameters should be entered or displayed that have units mended level of accuracy. Use of the RTP system should be
that do not match the units on documentation printed by the limited to this range. If deviations between calculated results
RTP system or displayed on its monitor. Procedures should and the published data ~such as that recommended in Table
be developed and documented to describe exactly how the XV! exceed 62%, the deficiencies of the RTP system
modified data and parameters are related to the non-TG-43 should be documented and further investigated by the user.
parameters assumed by the RTP system. These procedures This is especially important for RTP systems that fit a model
should address both clinical treatment planning practices and to published data, because such models are prone to large
chart-checking procedures. Ratios of the unconventional errors outside the range of the reference data. In the high-
units to the conventional units should be supplied, to facili- dose-gradient regions close to a source, particularly near the
tate review of the planning method. Because this approach is ends of a source, the acceptable error may need to be larger.
prone to errors in implementation or interpretation, this These deficiencies should be considered when evaluating
method should be used as the last resort. The AAPM recom- treatment plans, and when considering the purchase of an
mends using RTP systems that comply fully with the TG-43 RTP system.
formalism, whenever possible. To perform comparisons at the recommended level of pre-
cision, numerical point dose rates calculated by the RTP sys-
tem, rather than measured diameters of plotted isodose con-
B. Acceptance testing and commissioning
tours, should be used. The user should evaluate any
Before a new RTP system or a new source model on an deviations from the recommended data, taking into consider-
established RTP system is used for patient treatment plan- ation the uncertainty of calculations at very small distances.
ning, thorough acceptance testing and commissioning shall The calculation matrix should be adjusted to a value appro-
be carried out. The user should document the results of these priate for the high gradients near a source. For low-energy
tests both for later reference, and for compliance with appli- photon-emitting sources, grid spacing should not exceed 1
cable regulations. As a minimum, calculations of the dose- 3131 mm3 ; this size generally represents a reasonable
rate distribution shall be performed for a single source of compromise between calculation speed and accuracy.
each type to be used clinically. The recommendations of the Isodose curves should be generated as part of the commis-
AAPM ~TG-40, TG-53, and TG-56! should be sioning and continuing quality assurance procedures, but
followed.100–102 should be viewed as a test of the spatial accuracy of the

graphic dose display function of the RTP system, rather than chamber having an ADCL-assigned, directly traceable cali-
a definitive test of the underlying dose-calculation algorithm. bration coefficient determined for the source model to be
Because comparisons should include both point dose-rate used. TG-56 recommends returning the chamber to the
calculations and the placement of isodose lines, the user ADCL at intervals of two years for recalibration.
should also ensure that the RTP system and its graphical
output devices cause isodose curves to appear in the correct
locations relative to corresponding point calculations.
2. Secondarily traceable calibrations at the
institution by source intercomparison
C. Source calibrations
When NIST or an ADCL provides an air-kerma strength
For calibrating radioactive sources, the AAPM has previ- value for a given source, that source is then said to have a
ously recommended that users not rely on the manufacturer’s directly traceable calibration. The user can then use this
calibrations, but instead confirm the accuracy of source source to calibrate his/her transfer instrument, such as a well
strength certificates themselves by making independent mea- chamber. The well chamber in this situation is said to have a
surements of source-strength that are secondarily traceable to calibration coefficient bearing secondary traceability. To ac-
the primary standard maintained at NIST.100 For patient complish this, the user may obtain a source from a vendor,
treatments, AAPM further recommended that all clinically send it to NIST or an ADCL and obtain an air-kerma strength
used sources bear calibrations that are secondarily traceable value for that source. The user can then use this source to
to the primary standard. AAPM defines ‘‘direct traceability,’’ calibrate his or her well chamber. The well chamber can then
‘‘secondary traceability,’’ and ‘‘secondary traceability with be used to confirm the vendor-stated source strengths of
statistical inference’’ as follows:102 other sources of the same model before they are used clini-
cally.
‘‘Direct traceability is established when either a source If the chamber is calibrated by the ADCL, the calibration
or a transfer instrument ~e.g., well chamber! is calibrated coefficient is directly traceable. If the calibrated source
against a national standard at an ADCL or at NIST it- method is used, the user is responsible for transferring the
self.’’ calibration to his or her instrument. Regardless of which
‘‘Secondary traceability is established when the source method is used, the well chamber should be checked for
is calibrated by comparison with the same radionuclide constancy on a regular basis using a long-lived source such
and design that has a directly traceable calibration or by as 137Cs. The AAPM has recommended that a constancy
a transfer instrument that bears a directly traceable cali- check be performed at each use, and that the well chamber
bration.’’ exhibit constancy to within 62%. 100
The clinical user is cautioned that the secondary traceabil-
‘‘Secondary traceability by statistical inference is estab- ity calibration procedure should meet some minimum stan-
lished when a source is one of a group of sources of dard of quality, i.e., that for the intercomparisons method, the
which a suitable random sample has direct or secondary total uncertainty should be comparable to that achievable
traceability.’’ using an instrument with a directly traceable calibration. For
example, for the calibrated chamber or the calibrated source
This protocol, in accord with the previously published approach, the total expanded uncertainty (2 s 595% confi-
AAPM guidance ~TG-56, TG-40!, also recommends that all dence level!41 is obtained by adding in quadrature the uncer-
sources implanted into patients have measurements based tainties of instrument or source calibration at the ADCL
upon secondary traceability. Normally, vendors should be ex- ~typically 2.4% including the uncertainty at NIST!. It is em-
pected to provide calibration certificates that document sec- phasized that the precision of measurement in the well cham-
ondary traceability to NIST for their sources. Each institution bers is better than this ~generally within 60.5%). The typi-
should maintain a means for verifying vendor calibrations by cal total, expanded uncertainty at the user facility using a
air-kerma strength measurements with secondary traceability. source transfer to a chamber is typically 3.0%. This is the
Source sampling and instrument quality assurance guidelines maximum uncertainty that is relevant for secondarily trace-
are documented in TG-56. Source strength verification with able calibration. Any additional steps in the intercomparison
secondary traceability can be achieved either by means of a process will increase the total uncertainty and thus will not
chamber bearing a calibration that is directly traceable to the meet the minimum standard of quality recommended in this
appropriate national standard, or by comparison to a source protocol. The user is further warned that under no circum-
of the same model having a calibration that is directly trace- stances should a vendor or user calibration be used as a basis
able to the national standard. These methods are described in of institutional verification calibrations.
the following section. Finally, it is important not to confuse a source bearing a
calibration with direct or secondary traceability with a ‘‘cali-
1. Secondarily traceable calibrations at the
brated source’’ obtained from the source manufacturer. Use
institution using a transfer instrument
of a source calibrated by a manufacturer is not an acceptable
For brachytherapy sources, calibrating sources with sec- alternative for providing in-house calibrations with traceabil-
ondary traceability is best done with a well-type ionization ity to NIST or an ADCL.

GLOSSARY OF SYMBOLS AND TERMS sented as a point. Dimensionless units.

AAPM American Association of Physicists in CONg(r) Radial dose function derived from consen-
Medicine sus dataset. Dimensionless units.
ADCL AAPM-Accredited Dosimetry Calibration k̇ d (d) Air-kerma rate per history in vacuo esti-
Laboratory mated using Monte Carlo methods, due to
b Angle subtended by P(r, u ) and the two photons of energy greater than d.
ends of the active length. As used in the K̇ d (d) Air-kerma rate in vacuo due to photons of
line source approximation, b has units of energy greater than d, with units of
radians. cGy h21 .
COMS Collaborative Ocular Melanoma Study of L Dose-rate constant in water, with units of
episcleral eye plaque therapy versus m Gy h21 U21 . L is defined as the dose
enucleation trial. rate at P(r 0 , u 0 ) per unit S K .
d Distance to the point of measurement from L nnD, PqqS Notation identifying the dose-rate mea-
the source center in its transverse plane. surements or calculations used to deter-
Typically measured in-air or in-vacuo. mine Ḋ(r 0 , u 0 ) and the calibration stan-
Units of cm. dard to which this dose rate is normalized.
ḋ(r 0 , u 0 ) The dose rate per history estimated using For example, L 97D,N99S indicates a dose-
Monte Carlo methods at the reference po- rate constant determined from dosimetry
sition. measurements made in 1997 and having an
Ḋ(r, u ) Dose rate in water at P(r, u ). The dose rate S K traceable to the 1999 NIST standard.
is generally specified with units cGy h21 CONL Notation indicating that the reported value
and the reference dose rate, Ḋ(r 0 , u 0 ), is of L is the consensus value determined by
specified at P(r 0 , u 0 ) with units of the AAPM from published data, with units
cGy h21 . of cGy h21 U21 .
d Energy cutoff parameter used for air-kerma EXPL Notation indicating that the reported value
rate evaluation, which is 5 keV for this of L was determined by experimental mea-
protocol. surement.
FAC Ritz parallel-plate free-air chamber devel- MCL Notation indicating that the reported value
oped by Loftus of NIST. of L was determined using Monte Carlo
F(r, u ) 2D anisotropy function describing the ratio calculations.
of dose rate at radius r and angle u around L Active length of the source ~length of the
the source, relative to the dose rate at r 0 radioactive portion of the source! with
51 cm and u 0 590° when removing ge- units of cm.
ometry function effects. Dimensionless L eff Effective active length of the source, with
units. units cm.
G X (r, u ) Geometry function approximating the in- LIBD Low-energy Interstitial Brachytherapy Do-
fluence of the radionuclide physical distri- simetry subcommittee of the AAPM Ra-
bution on the dose distribution. G X (r, u ) diation Therapy Committee
may be calculated by Monte Carlo simula- NIST National Institute of Standards and Tech-
tion or by the following: nology
GP~r,u!5r22 point-source approximation, P(r, u )
H
Point-of-interest, positioned at distance r
b and angle u from the geometric center of
if u Þ0° the radionuclide distribution.
G L ~ r, u ! 5 Lr sin u
f an(r) The one-dimensional anisotropy function.
~ r 2 2L 2 /4! 21 if u 50° At any radial distance r, f an(r) is the ratio
line-source approximation, of dose rate averaged over 4 p steradian
with units of cm22 . integrated solid-angle to the dose rate at
g(r) Radial dose function describing the dose the same distance r on the transverse
rate at distance r from the source relative plane. Dimensionless units.
to the dose rate at r 0 51 cm. Dimension- RTP Radiotherapy planning system. In the con-
less units. text of this protocol, a treatment planning
g L (r) Radial dose function, determined under the system that can perform dose calculations
assumption that the source can be repre- for brachytherapy implants.
sented as a line segment. Dimensionless r The distance from the source center to
units. P(r, u ), with units of cm.
g P (r) Radial dose function, determined under the r0 The reference distance, which is 1 cm for
assumption that the source can be repre- this protocol

sK The air-kerma strength per history esti- tion. Because its dosimetric properties have been so exten-
mated using Monte Carlo methods. sively studied, it is often used for the basis of comparison for
SK Air-kerma strength: the product of the air- many new sources.
kerma rate K̇ d (d) and the square of the dis- The Amersham model 6711 source has been the most
tance d to the point of specification from widely used source for permanent implantation since its in-
the center of the source in its transverse troduction in 1983 @Fig. 2~b!#. It consists of a 4.5 mm welded
plane. S K is expressed in units of titanium capsule, 0.05 mm thick, with welded end caps. The
m Gy m2 h21 , a unit also identified by U. capsule contains a 3.0 mm long silver rod onto which 125I is
S K,N85 The 1985 NIST FAC air-kerma standard. adsorbed.
S K,N99 The 1999 NIST WAFAC air-kerma stan- There are many references for both Monte Carlo and ex-
dard. perimental dose determinations for these two sources. Many
u The polar angle between the longitudinal of the pertinent references are listed here,37,42,65,66,95,103–111
axis of the source and the ray from the ac- and have been summarized in more detail by Hedtjarn
tive source center to the calculation point, et al.42 Many of the Monte Carlo runs used older cross-
P(r, u ). section libraries and attempted to simulate the Loftus free-air
u0 The reference polar angle, which is 90° or chamber air-kerma strength standard. Only studies explicitly
p/2 radians. simulating the NIST 1999 standard were included in the
TLD Thermoluminescent dosimeter, generally Monte Carlo L average. Monte Carlo studies using older
composed of LiF. cross-section libraries incompatible with the current NIST
U The unit of air-kerma strength, equivalent XCOM database ~equivalent to DLC-146 distributed by
to m Gy m22 h21 or cGy cm2 h21 . RSICC!,112 were rejected as candidates for both relative and
WAFAC The wide-angle free-air chamber presently absolute TG-43 dosimetry quantities. LIBD attempted to se-
used at NIST to determine the air-kerma lect the most up-to-date and accurate data available. Because
strength of a low-energy photon-emitting the model 6711 silver rod has sharp corners, which call into
brachytherapy source. question use of the transverse-plane point detectors for simu-
lating air-kerma, Monte Carlo studies that include more de-
tailed modeling of the WAFAC have been emphasized for
ACKNOWLEDGMENTS this source.
The authors wish to thank Stephen M. Seltzer of NIST for Many references include all the TG-43 parameters. For
his comments and suggestions made during preparation of example, Hedtjarn et al. includes L, g(r), F(r, u ), and
this paper. Our gratitude is extended to Dr. Frank A. Ibbott f an(r). There are also experimental papers for g(r), and
who is acknowledged for creation of the artwork, and was anisotropy functions based upon TLD
supported through an AAPM RTC-sponsored project. Fur- measurements.65,95,103–105 Details on the references are given
thermore, we thank Dr. Dennis Duggan, Dr. Robert Kline, in the appropriate sections.
Dr. Zuofeng Li, Dr. Bruce Thomadsen, and Dr. Michael Her-
1.1. 6702 and 6711 L
man of the AAPM Radiation Therapy Committee for their
constructive comments towards review of this protocol. Ef- No experimental measurements of L for these sources
forts by Dr. Li to improve the uncertainty analysis are espe- have been published since the measurements of Chiu-Tsao
cially appreciated. Some of the authors ~M.J.R., G.S.I., R.N., et al., Nath et al., and Weaver et al. which date back to the
J.F.W.! have received research support to perform dosimetry late 1980s.66,106,107 Both TLDs and diodes were used for
studies for the sources included herein ~NASI, Imagyn, Best these measurements. These values from all three papers were
Medical and Theragenics Corporation, Amersham, Bebig, corrected for the Solid Water™ medium used, and for imple-
Best Medical, and Theragenics Corporation!, respectively. mentation of the NIST WAFAC 1999 standard. For both the
6702 and the 6711 sources, the experimental values for the
dose-rate constant were averaged after correction. For Monte
APPENDIX A: MODEL-SPECIFIC SOURCE Carlo evaluation, the studies of Hedtjarn et al.42 for the 6702
DOSIMETRY DATA source and Williamson46,108 for the 6702 and 6711 sources
The following sections summarize the dosimetry param- were used.
eters for each source. A description of the source and its In particular, Williamson included the WAFAC geometry
references are first provided. Afterwards each dosimetry pa- in model 6711 source air-kerma rate simulations.46,108 Each
rameter is discussed briefly. of these Monte Carlo simulations used the DLC-146 cross-
125
section library, and a bounded next flight point kerma esti-
1. Amersham Health model 6702 and 6711 I mator was used. This estimator resulted in an uncertainty
sources
~1s! between 0.2% ~near the source! to 2.5% ~far from the
The model 6702 source consists of a welded titanium cap- source!. These results suggest that the response of the
sule containing 125I adsorbed on anion exchange resin WAFAC is affected by internal geometry and has a 3%
spheres @Fig. 2~a!#. Although the 6702 source is no longer higher response that lowers the calculated L value. The air-
manufactured, consensus data are included in this compila- kerma strength values of the older references are based upon

the Loftus standard. Corrections for the new standard were tropy functions determined by Nath et al. and Capote et al.
made as were corrections for the Solid Water™ phantom to appear quite ‘‘noisy’’ and were therefore excluded from fur-
liquid water. The 6711 average EXPL is 0.980 cGy h21 U21 ther analysis. The best remaining dataset is by Weaver. These
and MCL50.950 cGy h21 U21 agree within 3%.46,108 Be- data are the most uniform and complete, and are recom-
cause the LIBD is convinced the 6711 sharp edges phenom- mended as CONF(r, u ) in ~Table IV!.
enon deserves further study, MCL is the average of William- The anisotropy functions F(r, u ) for the 6711 source from
son’s air-kerma point detector and full WAFAC geometry Sloboda and Menon,104 Furhang and Anderson, and Chiu-
simulations. Tsao et al., were compared with Monte Carlo calculations by
Similar methodology and results for the 6702 result in the Weaver.95 Other than Sloboda and Menon, and Furhang and
average of the experimental values from the candidate Anderson, there is good agreement. The 1,r,5 cm results
datasets of Nath et al., Chiu-Tsao et al., and Weaver et al. for all angles are within 10%, with the exception of F(1,0°).
being EXPL51.0557 cGy h21 U21 . 75–77 The average of the The most uniform and complete dataset seems to be Weaver,
Monte Carlo values from the candidate datasets of and therefore results by Weaver are recommended as the
Williamson46,108 and Hedtjarn et al.42 is MCL CONF(r, u ) in Table V for the model 6711 source.
21 21
51.0165 cGy h U . Therefore the average of these val-
ues, CONL, is that presented in Table I. 125
2. Best medical model 2301 I source
1.2. 6702 and 6711 g „ r … In 1992, a double walled encapsulated source of radioac-
tive 125I on a tungsten substrate was developed for interstitial
For the model 6702 and the 6711 sources, the measured
brachytherapy ~Best Medical International, Springfield, VA.
and Monte Carlo values for r.1 cm agree within the experi-
model 2300! as described by Rustgi.113 A sketch of this
mental uncertainties. The agreement is within 5% for the
source is shown in Fig. 2~c!. The double walled encapsula-
6702 source and within 7% for the 6711 source. Table II
tion design was intended to provide thinner walls at the ends
shows CONg(r) for both models ~6702 and 6711!, and for
of the source so that the corresponding angular distributions
line- and point-source approximations. The references for the
are more isotropic. In contrast to the model 6711 source,
consensus datasets are provided.
which uses a silver substrate that also serves as the radio-
For the 6702 source, measured results and Monte Carlo
graphic x-ray marker for source localization in the patient,
calculations for r>1 cm agree to within 5% for 1,r
the model 2300 uses a tungsten rod. 125I is distributed within
,4 cm and within 10% for distances greater than 4 cm. The
a low atomic number cylindrical annulus that surrounds the
Monte Carlo results of Hedtjarn et al.,42 Williamson,37 and
rod ~much like the Bebig source!. Because the tungsten
Mainegra et al.108 agree well with one another within the
K-shell binding energy exceeds the maximum energy emit-
combined uncertainties. Monte Carlo results of Hedtjarn
ted during 125I decay, no characteristic K-shell x-rays are
et al. are used since they are the most complete and are most
produced whereas L-shell x rays are readily absorbed in the
consistent with other data for the model 6702 source.
encapsulation.
Published data for the 6711 source indicate agreement be-
In 1993, Nath and Melillo reported the dosimetric charac-
tween the experimentally measured values and the Monte
teristics of the model 2300 source.114 Six years later in 1999,
Carlo calculations for distances greater than or equal to 1.0
the manufacturer introduced a commercial product based on
cm. Experimental results agree to within 7% for 1,r
the earlier design, which has been designated as the model
,8 cm. Monte Carlo results of Williamson and Mainegra
2301 source. The model 2301 source has a physical length of
et al. agree to within 3%.37,110 The Monte Carlo values agree
4.95 mm and outer diameter of 0.8 mm. The 125I radionu-
with experimental values to within 5%. Therefore, for 6711
clide was infused within the organic matrix that was coated
g(r), values from Williamson are used since the calculations
on a tungsten rod with an active length of 3.95 mm and a
cover a wider range, including r,1 cm. 37
diameter of 0.25 mm @Fig. 2~c!#. Also in 1999, NIST estab-
lished a WAFAC calibration standard for the air-kerma
1.3. 6702 and 6711 F „ r , u … strength of the model 2301 source.
Experimental and Monte Carlo results agree within 5% at Meigooni et al. measured the TG-43 dosimetric param-
larger angles for both source models. Tables IV and V eters for the model 2301 source and reported the values
present the model 6702 and 6711 F(r, u ) data, respectively. based upon the original WAFAC 1999 standard.115 Because
The measured anisotropy functions F(r, u ) for the 6702 of the 1999 NIST WAFAC anomaly, which was discovered
source from Nath et al.,103 Furhang and Anderson,105 Schell after the publication of Meigooni et al., the air-kerma
et al.,65 and Chiu-Tsao et al.,66 were compared with the strength was revised, the value determined as in Table I, and
Monte Carlo calculations of Weaver95 and Capote et al.111 In the TG-43 dosimetry parameters reported by Meigooni et al.
place of a realistic source geometry model used by other were corrected to this new value. In 2002, Nath and Yue
Monte Carlo investigators, Weaver used a simple line-source published independent determinations of TG-43 parameters
model for F(r, u ), in conjunction with a photon fluence an- of the model 2301 source based on TLD measurements.116
isotropy function measured in air at 100 cm for randomly Finally, Sowards and Meigooni published a TG-43 dosim-
selecting primary photon trajectories. Other than Furhang etry dataset obtained using Monte Carlo methods in both
and Anderson, all datasets agree fairly well. The 2D aniso- liquid water and Solid Water™.117

2.1. 2301 L ket in October 1998 following a brief appearance by the

For comparison purposes, Sowards and Meigooni pub- model MED3631-A/S source.120–122 The capsule is made of
lished a dose rate constant value of 0.98 titanium, with a 0.81 mm outer diameter and 0.05 mm wall
60.03 cGy h21 U21 in Solid Water™, and obtained MCL thickness, and a nominal length of 4.5 mm with spherical
51.0160.03 cGy h21 U21 in liquid water.117 end welds of thickness 0.05 to 0.15 mm. Inside the
In a Solid Water™ phantom, Nath and Yue used LiF TLD MED3631-A/M are four polystyrene ion exchange resin
detectors which were calibrated against a 6 MV x-ray beads, within which 125I is uniformly distributed. The four
beam.116 A relative energy-response correction factor of 1.41 beads are separated into two sets by two gold-copper radio-
was used.83 Nath’s published L value was increased by opaque markers. Both the beads and markers have a nominal
4.3%, based on 125I Monte Carlo simulations, to correct for diameter of 0.5 mm, and are free to move about within the
nonwater equivalence of the Solid Water™37 measurement capsule interior @Fig. 2~d!#.
medium, yielding L51.0260.07 cGy h21 U21 . Meigooni The only complete ~2D! experimental characterization of
et al. have also measured EXPL using LiF TLDs in a Solid brachytherapy dosimetry parameters was performed by Wal-
Water™ phantom using a 6 MV x-ray beam for calibration lace and Fan.120 They irradiated TLD-100 rods in tissue- and
and a relative energy response correction factor of 1.40. The water-equivalent plastic phantoms. Detectors were calibrated
authors applied 1.05 as the correction factor to account for using a 60Co teletherapy beam with tissue-equivalent phan-
the Solid Water™ measurement medium. After applying the tom corrections, C p (r), of C p (r50.5 cm)50.778 and
1999 WAFAC anomaly correction to the published value, a C p (r57 cm)51.053. Measurements were performed in
value of L51.0160.08 cGy h21 U21 was obtained based on 1998, so corrections for the 1999 WAFAC anomaly were not
the authors’ uncertainty analysis.115 A further correction was necessary since the 2000 NIST WAFAC measurements dif-
reported in a private communication that resulted in a final fered by ,1% compared to the 1998 calibration. Table I
value of L51.03 cGy h21 U21 for the Meigooni group,86 as shows the value used now based on a 2001 calibration. Be-
described in detail within the publication by Nath and Yue.116 cause the G(r, u ) used by all investigators was based on a
In this protocol, the final measured values of the candidate four point source model, all g(r) and F(r, u ) datasets were
datasets of Nath and Yue and of Meigooni et al. were aver- converted using an active length of 4.2 mm to adhere to the
aged to obtain a mean value of EXPL51.025 cGy h21 U21 . 2D formalism of this protocol.
This mean measured value was averaged with MCL Rivard published a complete, 2D TG-43 dosimetry dataset
51.01 cGy h21 U21 yielding CONL51.018 cGy h21 U21 . for the MED3631-A/M source using Monte Carlo methods
developed from previous studies.123–126 The effect of internal
component motion on dose distributions external to the cap-
2.2. 2301 g „ r …
sule was considered for the first time. The WAFAC was not
The measured radial dose function of Meigooni et al.115 simulated, and a 30 cm diameter liquid water spherical phan-
for the Best model 2301 source is slightly more-penetrating tom encompassed the source. Radii ranged from 0.25 to 10
than that of model 6711 125I source, and slightly less pen- cm, and the angular range was 0° to 180° with 1° incre-
etrating than that of the model 6702 125I source at distances ments. Air-kerma strength was determined in a 6 meter di-
beyond 2 cm. Monte Carlo results from Sowards and Mei- ameter sphere of dry air by multiplying the total air-kerma
gooni were chosen as the CONg(r); these values are pre- strength, integrated over all photon energies, by 1.049 to
sented in Table II. account for photon transmission in air at 1 meter, and by
0.897 to account for Ti K-shell x rays. A corrected value was
2.3. 2301 F „ r , u … later published, recognizing that the S K,N99 /S K,N85 factor
measured by NIST does not accurately model the influence
The anisotropy function of the Best 125I source ~model
of Ti x rays in the geometry used for Monte Carlo
2301! was measured at 2, 4, and 6 cm, and at different u
calculations.127 Statistical uncertainties ranged from 0.1% to
angles by Nath and Yue and at 2, 5, and 7 cm by Meigooni
2% for F(r, u ) on the transverse plane to the source ends,
et al. Monte Carlo calculations at distances of 1, 2, 3, 4, 5, 6,
respectively. Statistical uncertainties in g(r) and L were
and 7 cm were reported by Sowards and Meigooni. A com-
typically <1%, and ;3%, respectively.
parison of the measured and calculated values indicates good
agreement between the different datasets. Following the con- 3.1. MED3631-AÕM L
sensus procedure, CONF(r) was chosen based on results re-
Wallace and Fan reported L51.056 cGy h21 U21 , and Li
ported by Sowards and Meigooni with the exception of 6 cm
et al. reported L51.067 cGy h21 U21 . This average yields
data which exhibited larger statistical noise. Therefore, re- 21 21
sults from Sowards and Meigooni were used for CONF(r, u ), EXPL51.0615 cGy h U . Rivard calculated L
21 21
51.066 cGy h U , but this value was later corrected to
and are presented in Table VI.
1.011 cGy h21 U21 based on an inappropriate correction
methodology.127 Taking an equally weighted average of
3. North American Scientific Inc. model MED3631-AÕM
125 1.0615 and 1.011, Table I shows CONL
I source
51.036 cGy h21 U21 .
The North American Scientific Inc. ~NASI! model Li et al. performed measurements only on the transverse-
MED3631-A/M source45,48,118,119 was introduced to the mar- plane in 1999.48 Dosimetry measurements were made using

TLD-100 chips and a diode in a large water phantom. Irra- tors was good, with F(r, u ;0°);0.7260.05 for 0.25,r
diation in the water phantom was accomplished by taping the ,10 cm. Furthermore, agreement among f an(r) results re-
TLD chips onto the tip of the diode detector for 30 minutes. ported by Rivard and by Wallace and Fan was similarly good
No phantom material correction was employed, but the TLD (63%).
energy response function of Weaver et al. was used.107 Cal-
culation of L was based on the ratio of measured readings of 125
4. BebigÕTheragenics Corporation model I25.S06 I
MED3631-A/M and 6702 125I sources and using the 6702
source
source S K value. Effectively, the TLD’s were calibrated
against the model 6702 source based upon the TG-43 L 6702 The Bebig/Theragenics Corporation model I25.S06
value. Due to propagation of uncertainties from both 6711 source ~IsoSeed! was introduced to the market in July 1999
and MED3631-A/M measurements into the final result, this initially under the name Symmetra. On February 4, 2003,
protocol does not recommend the practice of cross calibra- Theragenics Corporation purchased the U.S. 125I prostate
tion, and these values were omitted from the analysis. How- brachytherapy business of BEBIG Isotopen und Medizin-
ever, the Li et al. measurements were later used as CONg(r). technik GmbH—including the manufacturing and distribu-
tion rights to the IsoSeed® 125I brachytherapy source distrib-
3.2. MED3631-AÕM g „ r … uted by Isotope Products Laboratories. Theragenics
Corporation distributes this source under the brand name
For determination of g(r), Li et al. used the same geom-
I-Seed. The source design for both IsoSeed® and I-Seed are
etry function as obtained above by Wallace and Fan. While
the same, and the model ~I25.S06! also remains the same. Its
the range of distances to the source covered by Rivard for
encapsulation consists of a 0.05 mm thick titanium capsule
determination of g(r) was larger and closer than either Wal-
with spherical end-welds ~0.44 –0.48 mm thick! similar to
lace and Fan or by Li et al., the impact of outdated default
those of the 6711 source family. The radioactive core con-
photon cross-section libraries in MCNP has become recently
sists of a 3.5 mm long cylindrical ceramic ~alumina,
apparent.96 While by definition all datasets agree at r 0 , dif-
2.88 g/cm3 ) annulus within which the radioactive iodine 125I
ferences between Rivard’s g(r) data and that of Li et al. and
is uniformly distributed. A gold marker, 0.17 mm diameter
Wallace and Fan gradually increased—reaching 25% at 7
and 3.5 mm long placed inside the ceramic core, permits
cm. Consequently, the difference between results obtained by
radiographic source localization. A schematic diagram is
Rivard and by Li et al. and by Wallace and Fan are not
shown in Fig. 2~e!.
readily resolvable. Therefore, the measured data of Li et al.
Hedtjarn et al.42 describes a complete Monte Carlo study
were chosen for the CONg(r) data as they demonstrate more
based mainly on the DLC-99 cross-section library. The esti-
consistent behavior than that of Wallace and Fan. Agreement
mator used was a bounded next-flight point-kerma estimator.
with the Wallace and Fan g(r) data was within 65% for r
Patel et al. described TLD measurements of L, g(r), and
,6 cm. Since the impact of differences as a function of dis-
F(r, u ) for a limited number of distances ~1–5 cm!.57 This
tance is independent of normalization, the impact of cross-
group subjected the widely used Solid Water™ material to
section library differences diminishes as the distance de-
chemical analysis and found its calcium content, which com-
creases. Therefore, g(r) data by Rivard are used for r
pensates for nonwater equivalence of the organic polymer
,1 cm, and are italicized in Table II.
composition of the basic material, to differ significantly from
the vendor’s specified atomic composition. Based on the
3.3. MED3631-AÕM F „ r , u … measured composition, Williamson’s PTRAN code was used
Since F(r, u ) data are by definition normalized to a given to calculate distance and TLD-size dependent energy re-
distance and the impact of outdated photon cross-section li- sponse correction factors ~1.3–1.5!. The measurement dis-
braries was assumed to be negligible, the Monte Carlo tance range was 1– 4 cm. Finally, Williamson44 compared
F(r, u ) results by Rivard are recommended as the consensus MCL with and without full simulation of the WAFAC geom-
dataset since they covered the largest angular and radial etry, using the DLC-146 cross-section library, and found that
ranges. While the dose distribution of this source model in the WAFAC and point-kerma detectors agreed within 1%.44
the longitudinal plane is highly nonsymmetric in close prox- In this case, the bounded next flight estimator for water
imity to the source, the F(r, u ) data were obtained using kerma was used, except for distances less than 3 mm, where
averaged dose-rate data above and below the transverse the once-more-collided point flux estimator was used. The
plane ~supplementary angles! to account for the asymmetric expected-value track-length estimator was used for the
geometric source model used by Rivard. These averaged re- WAFAC simulation. These two references, Hedtjarn et al.
sults using the line source approximation with L50.42 cm and Williamson,42,44 constitute the only published Monte
are presented, and the MED3631-A/M f an(r) results are pre- Carlo data available for this source model.
sented in Table VII. Using the same active length, results by
Rivard exhibited much less variation at u ;90° than Wallace 4.1. I25.S06 L
and Fan (,1% compared to 5%!. This was expected since Using Monte Carlo methods, Hedtjarn et al. estimated L
they used TLDs which were more susceptible to volume- using the DLC-146 cross-section library and by modeling the
averaging artifacts along the longitudinal axis. Agreement WAFAC geometry. Their MCL value, 0.991 cGy h21 U21 ,
among converted F(r, u ) results from both sets of investiga- was averaged with the EXPL value, 1.033 cGy h21 U21 , by

Patel et al. The two values agree within the experimental 5.2. IS-12501 g „ r …
uncertainties, and the CONL value is given in Table I. Gearheart et al. and Nath and Yue measured g(r) in Solid
4.2. I25.S06 g „ r … Water™. In both cases, g(r) was calculated using the line-
The g(r) data calculated by Hedtjarn et al. and measured source approximation method. Gearheart et al. also pub-
by Patel et al. are based upon a line source with L lished Monte Carlo calculation of g(r) in liquid water.
50.35 cm. These two datasets agree within experimental un- Analysis revealed the TLD measurements agreed with the
certainties ~5%! except for g(0.5). Due to its larger coverage Monte Carlo calculations within 28% to 16% with 1 stan-
of radial distance and closer coverage towards the source, the dard deviation of 14% ~1s!. Consequently, Gearheart et al.
Monte Carlo g(r) data of Hedtjarn et al. are recommended. Monte Carlo values in water are recommended as CONg(r),
and are listed in Table II.
4.3. I25.S06 F „ r , u …
Measured anisotropy functions by Patel et al., based on 5.3. IS-12501 F „ r , u …
an active length of 0.35 cm, were compared to Monte Carlo Nath and Yue did not measure F(r, u ). Gearheart et al.
data by Hedtjarn et al. and Williamson.42,44 For complete- published TLD measurements in Solid Water™ and also per-
ness, the anisotropy function derived from the 4 cm Monte formed Monte Carlo calculations in both Solid Water™ and
Carlo calculations, which was omitted from the published liquid water. The geometry function was modeled as a line
paper, was added. Agreement within 5% was usually ob- source with active length of 0.34 cm. Monte Carlo calcula-
served. The discrepancies were random and not indicative of tions were performed at 1, 2, 3, 5, and 7 cm to facilitate
different trends between the measured and computational ap- calculation of f an(r) at these distances. Comparison of mea-
proaches. The Hedtjarn et al. data are recommended and sured and calculated data demonstrate good agreement
given in Table VIII. within combined uncertainties of 10%. Consensus F(r, u )
and f an(r) data from Gearheart et al. are presented in Table
5. Imagyn Medical Technologies Inc. isostar model IX.
IS-12501 125I source
International Isotopes Inc. ~Denton, TX, now a division of 6. Theragenics Corporation model 200 103
Pd source
Imagyn Medical Technologies Inc.!, produced this source. It
is marketed by Imagyn Medical Technologies, Inc. under the The model 200 ~TheraSeed® ! source was introduced by
trade name ‘‘isostar IS-12501.’’ This source model first be- Theragenics Corporation in 1987, and remained the sole
came available for analysis in 1999, and was introduced to commercially available interstitial 103Pd source until 1999.
clinical sites later that year. The design consists of five 0.56 The encapsulation is a 0.056 mm thick Ti tube with a mea-
mm diameter silver spheres on which 125I silver iodide is sured external length of 4.50 mm and average measured
adhered @Fig. 2~f!#. The silver spheres are encapsulated in a outer diameter of 0.83 mm, respectively @Fig. 2~g!#. The tube
titanium tube whose ends are laser welded. ends are closed by means of inverted ‘‘end-cups’’ composed
There are four pertinent references for this of 0.040 mm thick Ti metal welded to the Ti tube. Using
source.38,39,128,129 Complete experimental and Monte Carlo transmission radiography and microscopic examination,
results are given in Gearheart et al. and experimental results Monroe and Williamson showed that the end-cup shape is
by Nath and Yue, respectively.38,39 Experimental measure- not well approximated by the right cylindrical shape used in
ments ~TLD in Solid Water™! of L, g(r) from 0.5 cm to 10 early, simplified Theragenics Corporation manufacturing
cm, and anisotropy function at 2 cm and 5 cm were first illustrations.40 The internal source components include two
reported by Gearheart et al. This work also contains Monte graphite pellets, modeled as right circular cylinders with 0.56
Carlo calculations of g(r) and F(r, u ), both in water and in mm diameters and 0.89 mm long, upon which a mixture of
Solid Water™ with ratios between each media. These ratios radioactive and nonradioactive palladium is added. The
were used to convert the TLD measurements of L in Solid graphite pellets are separated by a cylindrical lead marker.
Water™ to that in liquid water. The PTRAN Monte Carlo Dosimetric analysis is complicated by the fact that two
code was used, with the HUGO DLC-99 cross-section librar- manufacturing techniques were used to produce the 103Pd
ies. The bounded next-flight point-kerma estimator was used. used in these sources: lower specific activity reactor-
Nath and Yue presented TLD measurements of L and g(r) produced 103Pd until 1995 and higher specific activity
from 0.5 cm to 6 cm. Monte Carlo calculations were used to accelerator-produced palladium used thereafter. The older
relate L in water to the measurements in Solid Water™.37 reactor-produced heavy seeds had thicker layers of palladium
metal plated on the graphite pellets, ranging from 8 mm to 23
5.1. IS-12501 L mm ~200–570 mg/pellet!.7 The Monte Carlo study by Mon-
Ibbott and Nath explained that when the L value of Gear- roe and Williamson assumed a 10.5 mm thick ~260 mg/pellet!
heart et al. is corrected using the revised 1999 NIST calibra- layer.40 Monroe and Williamson also modeled the current
tion, agreement with Nath and Yue improved to 3%. Ibbott light seed configuration with a pellet coating of 2.2 mm for
and Nath published CONL50.940 cGy h21 U21 , based upon 57 mg of 103Pd per pellet.109 While pellet coating thickness
the methodology in this protocol, and this value is given in has been shown to minimally influence the source dosimetric
Table I. characteristics,95,109 most of the published dosimetry studies,

including the original TG-43 protocol dataset for the model another. Except for the 0.1 cm measurement point, very good
200 source, are based upon the obsolete heavy seed agreement with measured results by Chiu-Tsao and Anderson
geometry.7,66,106,107 at short distances is achieved. The light seed g(r) Monte
Until 1999, there was no air-kerma strength standard for Carlo data of Monroe and Williamson are recommended as
the model 200 source. Theragenics Corporation maintained consensus data ~Table III! since they include data from 0.1
an ‘‘apparent activity’’ standard based upon intercomparison cm to 12.5 cm. Williamson demonstrates using Monte Carlo
of photon fluence rates from model 200 103Pd sources and simulations that geometric differences between the light and
109
Cd activity standards using a NaI~Tl! detector. A detailed heavy seeds do not significantly influence the radial dose
history of these practices is given in LIBD’s recommenda- function.46
tions on 103Pd calibration and dose-prescription practices.7 6.3. 200 F „ r , u …
Briefly, the L value published in the original TG-43 protocol
was based upon the Theragenics Corporation A app,T88 stan- After conversion to a common active length of 0.423 cm,
dard which was initiated in 1988. TLD measurements of L the Yue and Nath F(r, u ) measured data were compared to
by Meigooni and Chiu-Tsao were normalized to S K,T88 esti- the Monroe and Williamson Monte Carlo data. Excellent
mates inferred from the vendor’s calibration using a nominal agreement of F(r, u ) between Monte Carlo results of Mon-
exposure-rate constant value. The vendor’s calibration ap- roe and Williamson and measured results by Yue and Nath is
peared to remain stable until 1997, when an 8% shift was observed, often within 5%. Since the Monroe and William-
reported.130 In 1999, the vendor’s A app,T97 calibration was son light seed data cover the distance range from 0.25 cm to
intercompared with the new 1999 NIST WAFAC standard, 7.5 cm and were similar to the heavy seed data, they are
and the 1999 NIST standard replaced the vendor’s calibra- recommended as the consensus dataset ~Table X!.
tion in March 2000. These initial NIST WAFAC calibrations
were affected by the 1999 WAFAC anomaly. On average, 7. North American Scientific Inc. model MED3633
103
WAFAC calibrations made after March 2000 were 5.3% less Pd source
than calibrations performed during 1999. On March 5, 2001, The model MED3633 source was introduced to the mar-
Theragenics Corporation adopted ~re: letter to users dated ket in February 1999, and is currently distributed by the
Feb 14, 2001! the corrected NIST calibrations. manufacturer, NASI. It has the same internal/external geom-
Six published papers were reviewed to determine the full etry as the MED3631-A/M @Fig. 2~d!# with L54.2 mm.
consensus dataset for the model 200 light seed.40,46,60,131–133 However, 103Pd is distributed only on the surface of the ion
Monroe and Williamson present comprehensive Monte exchange beads. During this time, the NIST 1999 WAFAC
Carlo-based dosimetry data for both reactor- ~heavy! and measurement anomaly impacted measurements of L by
cyclotron- ~light! produced sources.40 Their results indicate 13.2%. A letter to end-users was sent out in late 2000 dis-
that the heavy and light seed g(r) are nearly identical. Nath cussing the calibration change, and shift in L. There are only
et al.131 has published a measured L for the light seed while three peer-reviewed papers which assess the MED3633 2D
Yue and Nath published measured light seed anisotropy dosimetry parameters.
functions.133 Chiu-Tsao and Anderson, and Meigooni et al. Wallace and Fan measured 2D brachytherapy dosimetry
present measured g(r) for the heavy seed design.60,132 Re- parameters in a water-equivalent phantom using TLD rods
sults by Williamson and Monroe were based upon the PT- and an annealing procedure identical to that previously de-
RAN Monte Carlo code version 7.43 and used the DLC-146 scribed for the MED3631-A/M source.134 A phantom correc-
cross-section library and included a complete geometric tion factor of 1.0235 obtained by Luxton and Wallace was
model of the WAFAC. A bounded next flight point kerma used to calculate L,135,136 and a TLD energy response cor-
estimator was used. This estimator resulted in a 1s uncer- rection of 1.41 obtained by Meigooni et al. was also used.132
tainty between 0.2% ~near the source! to 2.5% ~far from the Radii on the transverse plane are listed to range from 0.17 to
source!. 7 cm, yet Table II of this paper contains an entry for r
6.1. 200 L 50.15 cm. Due to the small detector size (1 mm diameter
36 mm long LiF TLD-100 rods!, measurements at distances
For CONL, Nath’s measured value ~corrected by 1.05 to >0.5 cm may be considered free of volume-averaging arti-
reflect the 1999 WAFAC measurement anomaly! was aver- facts.
aged with Monroe and Williamson’s Monte Carlo estimate, Li et al. measured g(r) and determined L for the
yielding the value given in Table I. These two values agreed MED3633 source in a large water phantom using a
within 2%. Also, note that the Model 200 source L is nearly computer-controlled diode-electrometer system.137 Li et al.
independent of the thickness of the Pd metal matrix.46 used Williamson’s PTRAN Monte Carlo photon transport
code with the DLC-99 cross-section library, but did not de-
6.2. 200 g „ r … scribe the 103Pd photon energy spectrum, material densities
The g(r) results measured by Chiu-Tsao and Anderson and compositions, or the calculation geometry. The number
using TLDs were recalculated using a line-source approxi- of particle histories was such that 1s standard uncertainty
mation geometry function with L50.423 cm. The graphical about the mean were less than 2%. The authors state the
comparison indicates that the MC data agree better with each source geometry was explicitly modeled, including the 0.15
measured dataset than the two measurements agree with one mm thick semispherical end welds and 0.1 mm bead-marker

spacing. No information was given regarding the manner in et al. MCPT data, the Li et al. diode data varied by 67% for
which S K was determined. However, the bounded next flight 0.5,r,1.5 cm, and are not considered reliable. Therefore
point kerma estimator was employed. the CONg(r) is a combination of results by Rivard at close
Rivard calculated the 2D dosimetry parameters using distances and Li et al. for r.1 cm.
MCNP version 4B2 and the DLC-189 cross-section library
similar to the MED3631-A/M calculations ~accounting for 7.3. MED3633 F „ r , u …
motion of internal source components!.138 Results for g(r)
and F(r, u ) were evaluated over the 0.25 to 10 cm distance Rivard’s MED3633 F(r, u ) dataset covered the largest an-
range, and F(r, u ) was evaluated from 0°< u <180° in 1° gular and radial ranges, and its accuracy was expected not to
increments ~though only 10° increment data were reported!. be influenced by the outdated photon cross-section libraries
The Ti characteristic x-ray contributions were removed of- since data were normalized to a given distance. As for the
fline by binning kerma on energy and removing contributions MED3631-A/M source, the dose rate data above and below
,5 keV. The energy dependence of TG-43 dosimetry pa- the transverse plane were averaged to account for the asym-
rameters was analyzed by discretizing the polyenergetic metric geometric source model, and used to derive the con-
103
Pd spectra, and performing comparisons with results pre- sensus dataset 2D anisotropy function data ~Table XI!. These
sented by Chen and Nath, Luxton and Jozsef, and Carlsson averaged data were compared with the Monte Carlo data by
and Ahnjesö.109,139,140 Li et al. and the TLD results from Wallace and Fan at com-
mon radial distances of 1, 2, and 5 cm. Over these radii, the
7.1. MED3633 L Li et al. results agreed with Rivard’s data within 67% ~typi-
cally 14%). While differences as large as 20% were noted
Wallace and Fan reported a measured L value of 0.680 for small polar angles, these discrepancies may be attributed
60.033 cGy h21 U21 , yielding the value of 0.702 to different source models or the averaging technique used
60.034 cGy h21 U21 when corrected for the 1999 NIST for the F(r, u 50° – 180°) data. In comparison to the MCNP
WAFAC anomaly. Using a diode scanning system, a liquid results, the F(r, u ) dataset of Wallace and Fan exhibited un-
water phantom, and an in-house cross-calibration technique, expected irregularities (113% at r55 cm, u 580°, and
Li et al. reported two measured values ~0.714 and 118% at r51 cm, u 540°).
0.682 cGy h21 U21 ) before applying the 13.2% 1999 NIST
WAFAC anomaly correction ~Table I!. Though the corrected
average of these two diode readings would yield APPENDIX B: NIST AIR-KERMA STRENGTH
0.720 cGy h21 U21 , these measurements are not included in STANDARDS FOR LOW-ENERGY
this consensus since a cross-calibration method using a PHOTON-EMITTING SOURCES
source from a different manufacturer is discouraged. There- 1. NIST 1985 standard using the free-air chamber
fore, Wallace and Fan gave L50.702 cGy h21 U21 with
21 21 The National Institute of Standards and Technology
EXPL also equal to 0.702 cGy h U . Using MCPT, Li
21 21 ~NIST! maintains the U.S. primary air-kerma standards for x
et al. calculated 0.677 cGy h U , and Rivard calculated
rays in the energy range of 10 to 300 keV and for photon-
0.672 cGy h21 U21 using discretized photon energy fluence
emitting radionuclides such as 137Cs, 192Ir, 103Pd, and 125I.
estimators. Consequently, MCL50.6745 cGy h21 U21 was
The primary standard for 137Cs and 192Ir sources consists of
obtained. Combining results, CONL50.688 cGy h21 U21 is
Bragg–Gray cavity chambers.141 To provide similar trace-
shown in Table I.
ability for low-energy photon-emitting 125I sources, Loftus
developed a primary standard for 125I sources in 1985 based
7.2. MED3633 g „ r … on the Ritz parallel-plate free-air chamber ~FAC!, the na-
While the MCNP results of Rivard covered the largest tional primary x-ray standard for superficial therapy
radial distance range and came closest to the source, the beams.142,143 This chamber was used to measure the expo-
MCNP g(r) results could not be recommended. As was the sure rate in free-space on the transverse plane of model 6711
case for the MED3631-A/M source, Rivard used the default and 6702 sources. Because the Ritz FAC background current
MCNP cross-section library which is now known to cause was high relative to signal strength expected from a single
significant differences following radiation penetration source, this device was limited to a calibration arrangement
through multiple pathlengths due to obsolete photon cross- of a combination of 4 to 6 sources. These calibrations were
section data. Thus, the g(r) results of Li et al. generated then transferred to a spherical aluminum re-entrant ionization
using MCPT and updated cross-section data are recom- chamber which served as the secondary standard for routine
mended for CONg(r) data, with Rivard’s data recommended calibrations.144 Uncertainties (2 s 595% confidence level!
~italicized! only for r,1 cm ~Table III! where cross-section for the transferred measurements were 3% and 4% for the
data selection was less crucial. Note that the Rivard g(0.5) model 6702 and 6711 125I sources, respectively. Measure-
data exactly matched that of Li et al., i.e., g(0.5)51.243. ment uncertainties for subsequent source calibrations using
For r.4 cm the Rivard data differed from the Li et al. data the re-entrant chamber were estimated to be 5% and 6% for
by more than 10%, while Wallace and Fan g(r) data agreed the 6702 and 6711 sources, respectively.144 This Loftus cali-
with the Li et al. data ~except at r56 cm where the diode bration standard became available in 1985 and has been re-
signal was quite low! within 64%. Compared to the Li ferred to as the NIST 1985 air-kerma strength standard

(S K,N85) in recent AAPM guidance protocols.36,145 Soon after has a diameter of 8 cm, and is placed at a nominal distance of
introduction of this standard, Kubo called attention to the 30 cm from the source. This allows the measurement of ra-
influence on exposure measurements made in air by Ti diation in a cone with a half-angle of about 7.6°, rather than
K-shell characteristic x rays.146 These low-energy x rays the ;1° cone for the Ritz FAC measurements, resulting in a
(,5 keV) are clinically insignificant because they are 40-fold increase in solid angle; hence the ‘‘wide-angle’’ de-
largely absorbed by tissue or water within 1 mm of the scription. The effective or defined volume is ;704 cm3 , and
source. However, these x rays can affect air-kerma strength the collecting volume is ;2474 cm3 . This larger effective
measurements. Because of the extreme difficulty in using the volume increases WAFAC sensitivity 100-fold compared to
Ritz FAC for such measurements, NIST chose not to repeat the Ritz FAC. Moreover, the ratio of effective to collecting
this standardization process until a new instrument could be volumes is about 0.28 for the WAFAC compared to only
developed, precluding the addition of new low-energy source about 0.01 for the Ritz FAC, giving an improved signal-to-
models to the NIST system of standards or periodic inter- background ratio. While the Ritz FAC was limited to mea-
comparisons between the Ritz FAC and NIST re-entrant surements of multiple source arrangements, the WAFAC
chamber. Monte Carlo calculations by Williamson examined is able to measure S K with a precision of 0.01 m Gy m2 h21
the effects of x rays ,5 keV on Loftus’ FAC measurements for individual sources with strengths as low as
and data analysis procedures, and found that the air-kerma 1.00 m Gy m2 h21 .
rate was overestimated because the contributions by these x An automated version of the WAFAC was constructed for
rays were included in the measurements.51 computer-controlled, motor-driven, variable-length middle
electrode length, while holding fixed the positions of both
2. NIST 1999 standard using the wide-angle free-air the aperture plane and the center plane of the collecting vol-
chamber ume to remove detector measurement artifacts. Results from
2.1. The wide-angle free-air chamber both WAFACs have been compared for a large number of
sources of various designs, showing agreement to within
By the early 1990s, NIST was prompted to develop a new 0.5%. Determinations of air-kerma strength (S K ) using the
standard for brachytherapy sources because of deficiencies in WAFAC are referred to as S K,N99 in recent AAPM guidance
the existing primary standard, the need to develop a new documents,36,145 where N indicates the measurement was
instrument to directly measure air-kerma rate from individual performed at NIST and 99 indicates use of the January 1,
sources ~particularly those of low-energy photon-emitters 1999 WAFAC standard. WAFAC S K,N99 determinations have
such as 103Pd and 125I), and the need to calibrate sources of 1.4% Type-B ~systematic! uncertainties, and total uncertain-
newer design. In 1993, Loevinger developed the wide-angle ties of typically 2% when including Type-A ~statistical! un-
free-air chamber ~WAFAC!.147 Measurements were initiated certainties which depend on source strength.148
in 1998, and the new NIST WAFAC standard was formally
introduced on January 1, 1999. Numerous calibrations based
on this standard have been performed for the still-growing
number of new source designs. WAFAC results were com- 2.2. Impact of volume averaging by the NIST
pared to those of the Ritz FAC for four different NIST low- WAFAC
energy x-ray beam qualities. The average level of agreement Because of its large aperture, the WAFAC averages pho-
was 1.00360.003. ton fluence over a 7.6° half-angle cone centered about the
Figure 3 shows a schematic diagram of the WAFAC source transverse plane. Using Monte Carlo simulation tech-
chamber. It is a cylindrical chamber with circular symmetry niques, Williamson et al. have shown that WAFAC calibra-
about the beam-axis, and consists of: tions can deviate significantly from those based upon point
~a! a front, circular, aluminized-polyethylterephthalate detector air-kerma strength measurements for certain types
~PET! high voltage electrode, held at a potential V; of source geometries.25,40,46,149 Such source geometries con-
~b! a back, circular, aluminized-PET electrode that is di- tain internal components with sharp corners, e.g., right cir-
vided into a central circular area, defining the collect- cular cylinders, and are coated with radioactive, radio-
ing electrode, and an annular guard ring, both held at opaque materials. In the Theragenics Corporation model 200
ground potential; source ~cf. Appendix A.6. for more detail!, photons emitted
~c! a circular aluminum middle electrode separating the by the Pd-metal coatings on the planar end surfaces traverse
front and back electrodes, held at potential V/2 to shape moderate thicknesses of Pd metal and contribute to the dose
the electric field; and rate in condensed medium at short distances, but are much
~d! a mechanical support and auxiliary measurement in- more heavily self-absorbed at typical calibration distances
strumentation. and therefore, do not contribute significantly to transverse-
plane air-kerma rate measurements. This can result in
An aluminum filter, placed between the source- photon-fluence anisotropy factors as large as 1.25 over the
positioning device and the tungsten-alloy aperture, elimi- solid angle of which the WAFAC integrates.
nates the Ti K-shell x rays. Among other correction coeffi- While there is little direct experimental confirmation of
cients, attenuation of the penetrating components of source such distance- and angle-dependent radiation self-absorption
spectra by the aluminum filter is accounted for. The aperture effects and anisotropy near the transverse plane, the indirect

mended using (S K,N99)/(S K,N85)50.897 to convert between

the two standards for all applicable 125I source models ~Am-
ersham 6711 and 6702 and NASI models 3631 A/S and
A/M!.36 Because they were so tedious, NIST discontin-
ued periodic intercomparisons of Loftus re-entrant chamber
and WAFAC measurements in 1999. In preparation for the
introduction of the new S K,N99 standard, 6702 and 6711
sources were measured with the WAFAC and circulated to
the AAPM-Accredited Dosimetry Calibration Laboratories
~ADCLs!, enabling them to establish their secondary stan-
dards. Later, similar sources were circulated in Spring 1999
among the ADCLs in a successful proficiency test ~NIST was
not involved!. The new NIST standard based on the WAFAC
was formally introduced on January 1, 1999. At that time,
NIST also began the routine measurements of the new 103Pd
and 125I source models as well as the 6702 and 6711 sources.
Constancy checking of the WAFAC was performed by taking
readings with each source type using two commercial well-
ionization chambers: a Standard Imaging HDR 1000 Plus,
and a Capintec CRC-127R. During 1999 a new automated
WAFAC was tested, and agreement with the original
FIG. 3. The WAFAC measurement scheme, involving the subtraction of the
results of a second measurement using the small chamber length in order to WAFAC was repeatedly found to be at the 0.1–0.2 % level;
remove any possible effects due to the presence of the front and back the use of the new WAFAC was phased in. In December of
aluminized-PET electrodes. The middle electrode lengths shown are for the 1999, exceptionally large leakage and background currents
original WAFAC.
were noted. The new WAFAC was disassembled, cleaned,
reassembled, and compared with the original WAFAC—
again with agreement to within 0.1–0.2 %.
evidence is persuasive. In the case of the model 200
source,36,145 normalizing to point air-kerma strength calcula-
tions leads to dose-rate constant values that are sensitive to
2.3. 1999 WAFAC measurement anomaly
small changes in internal source geometry and differ from
experimental measurements by as much as 17%. Similar but In January 2000, NIST noticed a shift in well-chamber
smaller effects of ;5% occur for the models 6711 and calibration coefficients for a batch of Bebig 125I sources and
STM1251 125I sources.25,149 In contrast, Monte Carlo simu- a batch of IBt 103Pd sources. By February, shifts were found
lations incorporating the WAFAC measurement geometry re- also for the 6702 and for the 6711 models, and well-chamber
veal no such sensitivity, and result in dose-rate constant ~Sec. measurements by the ADCLs and manufacturers indicated
III A 2! values having close agreement with experimental similar shifts of 0–7 % between the 1999 and 2000 WAFAC
measurements.131 Furthermore, significant anisotropy implies calibrations ~as well as those done in 1998!.150,151 However,
that any air-kerma rate measurements based upon a point the new and original WAFACs were still in agreement. NIST
detector will have large uncertainties due to sensitivity to intensified its investigation of this anomaly, completely
source alignment and deviation of the actual source geom- checking all systems, mechanical, electrical, environmental,
etry from its idealized specifications. Despite its metrological software, etc., particularly any factor that could equally af-
impurity, the current WAFAC standard serves the clinical fect both WAFAC instruments. More calibrations in 2000
community well and has many advantages over previous confirmed the situation: an apparent downward shift in the
standards. By averaging air-kerma strength over regions of WAFAC air-kerma strength of approximately 3% for 125I
significant and possibly poorly reproducible anisotropy near sources, 5% for 125I-on-Ag sources, and 5% for 103Pd
the source transverse-plane, dosimetric uncertainties caused sources. However, a convincing explanation for this tempo-
by misalignment and source geometry specification uncer- rary shift has not been determined.
tainties are substantially mitigated. Exhaustive intercomparisons with 1998 brachytherapy
During the development and testing of the WAFAC in calibrations, Loftus transfer chamber readings, and low-
1997–1998, extensive intercomparisons were made between energy x-ray beams confirmed that post-January 2000
the WAFAC and the NIST re-entrant chamber ~the Loftus WAFAC S K measurements agreed with those performed in
transfer standard! for Models 6702 and 6711 sources. The 1998. A new comparison of the WAFAC ~original! with the
results from 10 sources established a ratio of the new NIST Ritz chamber on the NIST x-ray calibration range yielded
WAFAC standard (S K,N99) to the old Loftus standard agreement to ;0.8%. The original and newer automated-
(S K,N85) of 0.89860.014 for 6702 sources and 0.896 WAFACs continued to agree with each other. NIST com-
60.010 for 6711 sources, and 0.89760.011, combining the pleted a number of measurements of a 6702 and a 6711
two models. Based on this average, the AAPM recom- source, comparing the results in the WAFAC ~NIST 1999

standard! and in the re-entrant chamber ~NIST 1985 stan- mended for Eqs. ~6! and ~C1!, and approach zero at large
dard!. The results indicate a combined ratio for the sources distances. Where measured and calculated Monte Carlo data
of 0.89760.028, in very good agreement with the determi- are not available, such as for RTP systems that require
nation made in 1997–1998. This then confirmed the correct- F X (r, u ) and g X (r) values at r50, extrapolation using the
ness of the WAFAC measurements in 2000–2001 and in aforementioned approach is recommended. While a first-
1997–1998, and indicated that the problem was confined to order approach using linear extrapolation is also possible, the
measurements made in 1999 ~perhaps including late 1998!. merits of this approach have not been demonstrated.
The conclusion is that NIST WAFAC measurements up to Calculation of G X (r, u ) should employ Eq. ~4! down to
mid-1998 and after January 1 2000 are correct, and that cur- distances on the order of the interval between adjacent cal-
rent dose-rate constants for the 6702 and 6711 sources, based culation points. Daskalov et al. have found that the nearest
on the ratio NIST1999/NIST198550.897, are valid. How- neighbor method introduces uncertainties ,2% for 0.1 cm
ever, WAFAC air-kerma strengths measured in 1999 were along and away distances.154
too large by 2% to 7%, and required dose-rate constant mea-
surements normalized to NIST 1999 S K calibrations to be
revised accordingly.130 All measured dose-rate constant data
2. Collaborative Ocular Melanoma Study dosimetry
given in this protocol have been normalized to the S K,N99
standard as corrected in January 2000. In some specialized applications such as episcleral eye
plaque therapy, anisotropy corrections are prohibited when
following a protocol such as that published by the Collabo-
rative Ocular Melanoma Study ~COMS!.11,65,155–158 In eye
APPENDIX C: EVALUATION OF DOSE RATE AT plaque therapy, the sources are placed in a nearly spherical
SMALL AND LARGE DISTANCES cup and within the target volume most of the seeds contrib-
Care must be taken in evaluating dose rates at distances ute dose along their transverse directions. In this setup, the
falling outside the range of tabulated data, r min and r max , target volume receives very little dose in the longitudinal
especially at r,0.5 cm. While difficult to measure, modern directions of the sources. For practical reasons, the COMS
Monte Carlo techniques can easily calculate dose rates at protocol has recommended that for dose calculation within
distances as small as 0.1 cm from the center of the the target volume of eye plaque brachytherapy, the seeds can
source.152,153 This protocol recommends that g X (r) data be be identified as point sources provided one uses the dose
extracted and tabulated from Monte Carlo-derived dose rates rates on the transverse-plane alone. This procedure results in
for r>0.1 cm if possible. F X (r, u ) data are generally not the following expression for dose rate at different distances
available, even from most Monte Carlo evaluations, at r from the sources. For distances smaller than r min , the nearest
,0.25 cm. At shorter radii, points at small polar angles near neighbor approach should be used with the relevant r min
0° and 180° are located inside the source capsule, implying value used in the radial dose function,
that F X (r, u ) cannot be readily defined over the full angular
G X ~ r, u 0 !
interval @ 0°,90° # . Essentially, the TG-43 formalism using a Ḋ ~ r ! 5S K •L• •g ~ r ! for r,r min . ~C2!
polar coordinate system breaks down at r,L/2. Users are G X ~ r 0 , u 0 ! X min
warned that when working at small distances it is essential to
use the same geometry function and L for evaluating ab- 3. 1D dose calculations
sorbed dose as when used in the conversion of measured or
When anisotropy corrections are needed, and optimal ac-
Monte Carlo data in the TG-43 parameter ratios. Often,
curacy is desired at distances less than r min , the minimum
F X (r, u ) and g X (r) will exhibit nonphysical properties due to
distance for which f an(r) is tabulated, the following model
breakdown of the line-source G L (r, u ) very near the source.
can be used to extrapolate dose to short distances:
Three separate scenarios are described in the following
sections. G X ~ r, u 0 !
5
S K •L• •g ~ r ! • f an~ r ! , r>r min ,
1. 2D dose calculations G X~ r 0 , u 0 ! X
Evaluation of the 2D dose calculation formalism for dis-
tances r,r min or r.r max , should be made using the nearest
Ḋ ~ r ! 5
S K •L• S DS
r min
r
2
D
g L ~ r ! G X ~ r min , u 0 !
g L ~ r min! G X ~ r 0 , u 0 !
neighbor, zeroth-order approximation for selection of both
•g X ~ r min! • f an~ r min! , r,r min.
F X (r, u ) and g X (r), ~C3!
H
For distances r>r min , X5 P and Eq. ~C3! is equivalent to
interpolated value, r min<r<r max ,
the standard 1 formula, Eq. ~10!. The short-distance extrapo-
F X ~ r, u ! 5 F X ~ r min , u ! , r,r min , ~C1! lation model (r,r min) is based on applying Williamson’s
F X ~ r max , u ! , r.r max. approximation, r 22 • f̄ an'G(r, u 0 )• f an(r), to the short dis-
tance region, i.e., r 22 • f an(r min)'G(r,u0)•fan(r) for r
,r min .44 When X5 P, the short-distance extrapolation
g X (r) will follow the same extrapolation approach as recom- model simplifies as

Ḋ ~ r ! 5S K •L• S DS
r0
r
2
g L~ r !
g L ~ r min! D
•g P ~ r min! • f an~ r min! , Ḋ ~ r ! 5S K •L• SD
r0
r
2
•g P ~ r ! • f̄ an .
r,r min . ~C4! or
No matter how sophisticated an extrapolation model is cho- G L ~ r, u 0 !

D ~ r ! 5S K •L• •g ~ r ! • f̄ an . ~D3!
sen, users should realize that at r,0.25 cm where f an(r) G L~ r 0 , u 0 ! L
data may not be available due to lack of assessment or due to
positioning within the capsule, none of the 1D models de- APPENDIX E: APPARENT ACTIVITY
scribed above yield quantitatively meaningful estimates of
The quantity apparent activity, A app , is defined as the ac-
the solid angle-weighted average dose.
tivity of an unfiltered point source of a given radionuclide
that has the same air-kerma strength as that of the given
encapsulated source. It has been widely used by vendors and
APPENDIX D: ANISOTROPY CONSTANT customers alike to specify the strength of sealed radioactive
sources such as 125I, 103Pd and 198Au. Furthermore, A app is
The definition of the anisotropy constant, f̄ an , largely re- used in the treatment planning of permanent implants using
mains the same as that presented in TG-43, and is expanded 103
Pd and 125I interstitial sources. Using the apparent activity
upon. Use of f̄ an , as commonly practiced at this time, does as a method of source strength specification suffers from a
not exactly reproduce either the measured or Monte Carlo number of problems. For example, vendors using apparent
dosimetry data for r,1 cm. Therefore, it is strongly recom- activity to report source strength in essence convert a state-
mended that users utilize f an(r) and Eq. ~11!, or compensate ment of source output, e.g., S K , into A app by dividing S K by
for treatment planning inadequacies using the method out- an assumed value of the exposure rate constant (G d ) X . To
lined in Eq. ~14!, for implementing 1D dosimetry formalism. calculate absorbed dose in a medium around such sources,
For those users who insist on using f̄ an , the following users shall multiply the vendor supplied A app value by the
implementation will minimize dose-calculation errors at same (G d ) X value. Although (G d ) X is a clearly defined physi-
small distances, e.g., r<1 cm: cal concept, it has no meaningful physical role in the dosim-
SD 2
etry of output calibrated sources. Continued use of such
r0 dummy constants constitutes a significant potential source of
Ḋ ~ r ! 5S K •L• •g L ~ r ! • f̄ an . ~D1!
r dosimetric error since the user may choose the wrong (G d ) X
value. It is essential that users employ the same (G d ) X values
For this purpose, the inverse-square law weighted average of
as the manufacturer for dosimetric calculations rather than
f an(r) for r.1 cm should be used,
more physically accurate or definitive values taken from the
r recent literature.
max
( r>1 f r •r 22
cm an~ !
f̄ an5 r . ~D2! In 1999, NIST implemented a new primary standard for
max
( r>1 cm
r 22 air-kerma strength.8 The AAPM has consistently taken the
For instances where f an(r) data are not available over con- position that S K should be the quantity used for specifying
stant increments of r, linear interpolation of f an(r) may be brachytherapy source strength for the purpose of defining
calibration standards, documenting source strength on cali-
used for derivation of f̄ an . The constraint on r is needed
bration reports and for all aspects of dose calculation and
because dose distributions near typical brachytherapy
treatment prescription. Both users and vendors shall take ap-
sources generally take the shape of prolate ellipsoids due to
propriate steps to adopt this new source-strength standard
the dominating effects of solid angle. This is because F(r, u )
correctly. To facilitate an unambiguous conversion of source
excludes dose fall-off already included by the geometry
strength estimates and for transitional practice, the LIBD has
function, and f an(r) ~and subsequently f̄ an) do not. There- recommended a set of data and equations for the conversion
fore, f an(r) rapidly increases as r,2L, and can assume val- of (S K /A app) for 125I and 103Pd interstitial brachytherapy
ues much larger than unity. While nonintuitive, the use of sources. For all 125I and 103Pd sources, regardless of internal
g L (r) with the point-source geometry function in Eq. ~D1! construction, the values of (G d ) X recommended were 1.45
better approximates the average dose at small distances than and 1.476 R cm2 mCi21 h21 , respectively. When implement-
the superficially more consistent expression g X (r) ing the TG-43 formalism based upon apparent activity speci-
•G X (r, u 0 )• f̄ an•G X (r 0 , u 0 ) 21 again due to the ellipsoidal fication, the recommended (S K /A app) conversion coefficients
shape of the isodose surfaces. Thus r 22 • f̄ an'G(r, u 0 ) were 1.270 and 1.293 m Gy m2 h21 mCi21 for 125I and 103Pd,
• f an(r). For sources with L'0.3 cm, this approximation is respectively. The authors discussed the importance of using a
sufficiently good that errors introduced are often less than consistent set of values for the exposure rate constant (G d ) X
5% at distances ,0.3 cm. 150–152 However, Williamson dem- for the evaluation of (S K /A app). 8 For example, adopting
onstrated that Eq. ~D1! can produce much larger errors for S K -to-A app conversion coefficients derived from modern
low-energy sources with longer effective active lengths.44 (G d ) X values, different from the ones used by the authors;
Because of this short-distance behavior, the following forms would require all users and vendors to redefine the relation-
should not be used at r,0.5 cm: ship between S K and A app ; would require updating of dosi-

16
metric constants in all A app-based treatment planning sys- W. F. Whitmore, B. Hilaris, and H. Grabstald, ‘‘Retropubic implantation
of iodine-125 in the treatment of prostatic cancer,’’ J. Urol. ~Paris! 108,
tems; could cause significant confusion among clinical users; 918 –920 ~1972!.
would complicate future retrospective analyses of clinical 17
R. E. Peschel, C. R. King, and K. Roberts, ‘‘Pubic arch interference in
outcome data; and would not improve dosimetric accuracy. permanent prostate implant patients,’’ J. Brachytherapy. Intl. 14, 241–248
This further emphasizes the importance of using the same ~1998!.
18
M. J. Zelefsky and W. F. Whitmore, Jr., ‘‘Long-term results of retropubic
value of (G d ) X by the vendors and the users. Consequently, permanent 125Iodine implantation of the prostate for clinically localized
the AAPM continues to recommend that the quantity A app prostatic cancer,’’ J. Urol. ~Paris! 158, 23–30 ~1997!.
19
not be used for specification of brachytherapy source R. Nath, K. Roberts, M. Ng, R. Peschel, and Z. Chen, ‘‘Correlation of
strength. medical dosimetry quality indicators to the local tumor control in patients
with prostate cancer treated with iodine-125 interstitial implants,’’ Med.
Phys. 25, 2293–2307 ~1998!.
1 20
R. Nath, L. L. Anderson, G. Luxton, K. A. Weaver, J. F. Williamson, and R. G. Stock, N. N. Stone, A. Tabert, C. Iammuzzi, and J. K. DeWyngaert,
A. S. Meigooni, ‘‘Dosimetry of interstitial brachytherapy sources: Rec- ‘‘A dose-response study for I-125 prostate implants,’’ Int. J. Radiat. On-
ommendations of the AAPM Radiation Therapy Committee Task Group col., Biol., Phys. 41, 101–108 ~1998!.
21
No. 43,’’ Med. Phys. 22, 209–234 ~1995!. L. Potters, Y. Cao, E. Calugrau, T. Torre, P. Fearn, and X-H. Wang, ‘‘A
2
Interstitial Collaborative Working Group ~ICWG!, in Interstitial Brachy- comprehensive review of CT-based dosimetry parameters and biochemi-
therapy: Physical, Biological, and Clinical Considerations, edited by L. cal control in patients treated with permanent prostate brachytherapy,’’
L. Anderson, R. Nath, K. A. Weaver, D. Nori, T. L. Phillips, Y. H. Son, Int. J. Radiat. Oncol., Biol., Phys. 50, 605– 614 ~2001!.
22
S-T. Chiu-Tsao, A. S. Meigooni, J. A. Meli, and V. Smith ~Raven Press, R. Nath, L. Anderson, D. Jones, C. Ling, R. Loevinger, J. Williamson,
New York, 1990!. and W. Hanson, ‘‘Specification of brachytherapy source strength: A report
3
J. F. Williamson, ‘‘Physics of brachytherapy,’’ in Principles and Practice by Task Group 32 of the American Association of Physicists in Medi-
of Radiation Oncology, edited by C. A. Perez and L. W. Brady, 3rd ed. cine,’’ AAPM Report No. 21 ~American Institute of Physics, New York,
~Lippincott-Raven Press, Philadelphia, 1997! pp. 405– 467. 1987!.
4 23
A. S. Meigooni and R. Nath, ‘‘Response to ‘Comment on Dosimetry of ICRU 38 Dose and Volume Specification for Reporting Intracavitary
interstitial brachytherapy sources: Recommendations of the AAPM Ra- Therapy in Gynecology, International Commission on Radiation Units
diation Therapy Committee Task Group No. 43’,’’ Med. Phys. 27, 265 and Measurements ~ICRU 38, Bethesda, MD, 1985!.
24
~2000!. ICRU 60 Fundamental Quantities and Units for Ionizing Radiation, In-
5
R. W. Kline, ‘‘Comment on ‘Dosimetry of interstitial brachytherapy ternational Commission on Radiation Units and Measurements ~ICRU
sources: Recommendations of the AAPM Radiation Therapy Committee 60, Bethesda, MD, 1998!.
25
Task Group No. 43’,’’ Med. Phys. 23, 1579 ~1996!. J. F. Williamson, ‘‘Low energy photon source dosimetry,’’ in Proceedings
6
A. Y. C. Fung, ‘‘Comment on ‘Dosimetry of interstitial brachytherapy of 2000 World Congress on Medical Physics and Biomedical Engineer-
sources: Recommendations of the AAPM Radiation Therapy Committee ing, edited by G. D. Fullerton ~IEEE Engineering Medicine and Biology
Task Group No. 43’,’’ Med. Phys. 25, 2477 ~1998!. Society, Chicago, IL, 2000!, MO A313-01.
7 26
J. F. Williamson, B. M. Coursey, L. A. DeWerd, W. F. Hanson, R. Nath, M. J. Rivard, ‘‘Refinements to the geometry factor used in the AAPM
M. J. Rivard, and G. Ibbott, ‘‘Recommendations of the American Asso- Task Group Report No. 43 necessary for brachytherapy dosimetry calcu-
ciation of Physicists in Medicine on 103Pd interstitial source calibration lations,’’ Med. Phys. 26, 2445–2450 ~1999!.
27
and dosimetry: Implications for dose specification and prescription M. J. Rivard, ‘‘Neutron dosimetry for a general 252Cf brachytherapy
~AAPM Report No. 69!,’’ Med. Phys. 27, 634 – 642 ~2000!. source,’’ Med. Phys. 27, 2803–2815 ~2000!.
8 28
J. F. Williamson, B. M. Coursey, L. A. DeWerd, W. F. Hanson, R. Nath, E. Kouwenhoven, R. van der Laars, and D. R. Schaart, ‘‘Variation in
M. J. Rivard, and G. Ibbott, ‘‘On the use of apparent activity (A app) for interpretation of the AAPM TG-43 geometry factor leads to unclearness
treatment planning of 125I and 103Pd interstitial brachytherapy sources: in brachytherapy dosimetry,’’ Med. Phys. 28, 1965–1966 ~2001!.
29
Recommendations of the American Association of Physicists in Medicine J. A. Meli, ‘‘Let’s abandon geometry factors other than that of a point
Radiation Therapy Subcommittee on Low-Energy Brachytherapy Source source in brachytherapy dosimetry,’’ Med. Phys. 29, 1917–1918 ~2002!.
30
Dosimetry,’’ Med. Phys. 26, 2529–2530 ~1999!. M. J. Rivard, B. M. Coursey, L. A. DeWerd, W. F. Hanson, M. S. Huq, G.
9 Ibbott, R. Nath, and J. F. Williamson, ‘‘Comment on ‘Let’s abandon
J. F. Williamson, B. M. Coursey, L. A. DeWerd, W. F. Hanson, and R.
Nath, ‘‘Dosimetric prerequisites for routine clinical use of new low en- geometry factors other than that of a point source in brachytherapy do-
ergy photon interstitial brachytherapy sources,’’ Med. Phys. 25, 2269– simetry’,’’ Med. Phys. 29, 1919–1920 ~2002!.
31
2270 ~1998!. M. J. Rivard, ‘‘Comments on ‘Dosimetry of interstitial brachytherapy
10 sources: Recommendations of the AAPM Radiation Therapy Committee
Radiological Physics Center, The M.D. Anderson Cancer Center, Hous-
ton, TX http://rpc.mdanderson.org/rpc/htm/Home – htm/Low-energy.htm Task Group No. 43’,’’ Med. Phys. 26, 2514 ~1999!.
32
last accessed December 1, 2003. D. C. Moss, ‘‘Improved analytical fit to the TG-43 radial dose function,
11 g(r),’’ Med. Phys. 27, 659– 661 ~2000!.
B. S. Hilaris, D. Nori, and L. L. Anderson, Atlas of Brachytherapy ~Mac-
33
Millan Publishing Co., New York, 1988!. P. Lindsay, J. Battista, and J. Van Dyk, ‘‘The effect of seed anisotropy on
12
W. Lee, B. D. T. Daly, T. A. DiPetrillo, D. M. Morelli, A. C. Neuschatz, brachytherapy dose distributions using 125I and 103Pd,’’ Med. Phys. 28,
J. Morr, and M. J. Rivard, ‘‘Limited resection for non-small cell lung 336 –345 ~2001!.
34
cancer: Observed local control with implantation of I-125 brachytherapy J. F. Corbett, J. J. Jezioranski, J. Crook, T. Tran, and I. W. T. Yeung, ‘‘The
seeds,’’ Ann. Thorac. Surg. 75, 237–243 ~2003!. effect of seed orientation deviations on the quality of 125I prostate im-
13
Y. Yu, L. L. Anderson, Z. Li, D. E. Mellenberg, R. Nath, M. Schell, F. M. plants,’’ Phys. Med. Biol. 46, 2785–2800 ~2001!.
35
Waterman, A. Wu, and J. C. Blasko, ‘‘Permanent prostate seed implant R. Nath, M. J. Rivard, B. M. Coursey, L. A. DeWerd, W. F. Hanson, M.
brachytherapy: Report of the American Association of Physicists in S. Huq, G. Ibbott, and J. F. Williamson, ‘‘Status of the American Asso-
Medicine Task Group Report No. 64,’’ Med. Phys. 26, 2054 –2076 ciation of Physicist in Medicine Radiation Therapy Committee Subcom-
~1999!. mittee’s Low-Energy Interstitial Brachytherapy Source Dosimetry: Pro-
14
C. J. Mettlin, G. P. Murphy, D. S. Rosenthal, and H. R. Menck, ‘‘The cedure for the development of consensus single source dose
national cancer data base report on prostate carcinoma after the peak in distributions,’’ Med. Phys. 29, 1349 ~2002!.
36
incidence rates in the U.S.,’’ Cancer 83, 1679–1684 ~1998!. J. F. Williamson, B. M. Coursey, L. A. DeWerd, W. H. Hanson, R. Nath,
15
H. Ragde, A-A. A. Elgamal, P. B. Snow, J. Brandt, A. A. Bartolucci, B. S. and G. Ibbott, ‘‘Guidance to users of Nycomed Amersham and North
Nadir, and L. J. Korb, ‘‘Ten-year disease free survival after transperineal American Scientific, Inc., I-125 interstitial sources: Dosimetry and cali-
sonography-guided iodine-125 brachytherapy with or without 45-gray ex- bration changes: Recommendations of the American Association of
ternal beam irradiation in the treatment of patients with clinically local- Physicists in Medicine Radiation Therapy Committee Ad Hoc Subcom-
ized, low to high Gleason grade prostate carcinoma,’’ Cancer 83, 989– mittee on Low-Energy Seed Dosimetry,’’ Med. Phys. 26, 570–573
1001 ~1998!. ~1999!.

37 60
J. F. Williamson, ‘‘Comparison of measured and calculated dose rates in S.-T. Chiu-Tsao and L. L. Anderson, ‘‘Thermoluminescent dosimetry for
103
water near I-125 and Ir-192 seeds,’’ Med. Phys. 18, 776 –786 ~1991!. Pd seeds ~model 200! in solid water phantom,’’ Med. Phys. 18, 449–
38
D. M. Gearheart, A. Drogin, K. Sowards, A. S. Meigooni, and G. S. 452 ~1991!.
61
Ibbott, ‘‘Dosimetric characteristics of a new 125I brachytherapy source,’’ A. S. Meigooni, V. Mishra, H. Panth, and J. F. Williamson, ‘‘Instrumen-
Med. Phys. 27, 2278 –2285 ~2000!. tation and dosimeter-size artifacts in quantitative thermoluminescent do-
39
R. Nath and N. Yue, ‘‘Dose distribution along the transverse axis of a new simetry of low-dose fields,’’ Med. Phys. 22, 555–561 ~1995!.
125 62
I source for interstitial brachytherapy,’’ Med. Phys. 27, 2536 –2540 L. da Rosa and H. Nette, ‘‘Thermoluminescent dosimeters for exposure
~2000!. assessment in gamma or x radiation fields with unknown spectral distri-
40
J. I. Monroe and J. F. Williamson, ‘‘Monte Carlo-aided dosimetry of the bution,’’ Appl. Radiat. Isot. 39, 191–197 ~1988!.
63
Theragenics TheraSeed® Model 200 103Pd interstitial brachytherapy R. K. Das, Z. Li, H. Perera, and J. F. Williamson, ‘‘Accuracy of Monte
seed,’’ Med. Phys. 29, 609– 621 ~2002!. Carlo photon transport simulation in characterizing brachytherapy dosim-
41 eter energy response artifacts,’’ Phys. Med. Biol. 41, 995–1006 ~1996!.
B. N. Taylor and C. E. Kuyatt, ‘‘Guidelines for evaluating and expressing
64
the uncertainty of NIST measurement results,’’ NIST Technical Note J. F. Williamson, J. F. Dempsey, A. S. Kirov, J. I. Monroe, W. R. Binns,
1297 ~U.S. Government Printing Office, Washington, DC, 1994!. and H. Hedtjärn, ‘‘Plastic scintillator response to low-energy photons,’’
42
H. Hedtjarn, G. A. Carlsson, and J. F. Williamson, ‘‘Monte Carlo-aided Phys. Med. Biol. 44, 857– 872 ~1999!.
65
dosimetry of the Symmetra model 125.S06 125I, interstitial brachytherapy M. Schell, C. C. Ling, Z. C. Gromadzki, and K. R. Working, ‘‘Dose
seed,’’ Med. Phys. 27, 1076 –1085 ~2000!. distributions of model 6702 I-125 seeds in water,’’ Int. J. Radiat. Oncol.,
43
J. H. Hubbell, ‘‘Review of photon interaction cross section data in the Biol., Phys. 13, 795–799 ~1987!.
66
medical and biological context,’’ Phys. Med. Biol. 44, R1–R2 ~1999!. S-T. Chiu-Tsao, L. L. Anderson, K. O’Brien, and R. Sanna, ‘‘Dose rate
44
J. F. Williamson, ‘‘Dosimetric characteristics of the DraxImage Model determination for 125I seeds,’’ Med. Phys. 17, 815– 825 ~1990!.
67
LS-1 I-125 interstitial brachytherapy source design: A Monte Carlo in- A. Piermattei, L. Azario, G. Monaco, A. Soriani, and G. Arcovito, ‘‘
vestigation,’’ Med. Phys. 29, 509–521 ~2002!. p-type silicon detector for brachytherapy dosimetry,’’ Med. Phys. 22,
45
M. J. Rivard, ‘‘Monte Carlo calculations of AAPM Task Group Report 835– 839 ~1995!.
68
No. 43 dosimetry parameters for the MED3631-A/M 125I source,’’ Med. J. F. Williamson, H. Perera, Z. Li, and W. R. Lutz, ‘‘Comparison of
Phys. 28, 629– 637 ~2001!. calculated and measured heterogeneity correction factors for 125I, 137Cs,
46
J. F. Williamson, ‘‘Monte Carlo modeling of the transverse-axis dose and 192Ir brachytherapy sources near localized heterogeneities,’’ Med.
distribution of the Model 200 103Pd interstitial brachytherapy source,’’ Phys. 20, 209–222 ~1993!.
69
Med. Phys. 27, 643– 654 ~2000!. M. Ahmad, D. P. Fontenla, S-T. Chiu-Tsao, C. S. Chui, J. E. Reiff, L. L.
47
M. S. Levenson, D. L. Banks, K. R. Eberhardt, L. M. Gill, W. F. Guthrie, Anderson, D. Y. C. Huang, and M. C. Schell, ‘‘Diode dosimetry of mod-
H. K. Liu, M. G. Vangel, J. H. Yen, and N. F. Zhang, ‘‘An approach to els 6711 and 6712 125I seeds in a water phantom,’’ Med. Phys. 19, 391–
combining results from multiple methods motivated by the ISO GUM,’’ 399 ~1992!.
J. Res. Natl. Inst. Stand. Technol. 105, 571–579 ~2000!. 70
C. C. Ling, M. C. Schell, and E. D. Yorke, ‘‘Two-dimensional dose dis-
48
Z. Li, J. J. Fan, and J. R. Palta, ‘‘Experimental measurements of dosim- tribution of 125I seeds,’’ Med. Phys. 12, 652– 655 ~1985!.
etric parameters on the transverse axis of a new 125I source,’’ Med. Phys. 71
T. D. Bohm, D. W. Pearson, and R. K. Das, ‘‘Measurements and Monte
27, 1275–1280 ~2000!. Carlo calculations to determine the absolute detector response of radio-
49
Z. Li, J. F. Williamson, and H. Perera, ‘‘Monte Carlo calculation of chromic film for brachytherapy dosimetry,’’ Med. Phys. 28, 142–146
kerma-to-a-point in the vicinity of media interfaces,’’ Phys. Med. Biol. ~2001!.
38, 1825–1840 ~1993!. 72
A. Niroomand-Rad, C. R. Blackwell, B. M. Coursey, K. P. Gall, J. M.
50
T. Kron, L. DeWerd, P. Mobit, J. Muniz, A. Pradhan, M. Toivonen, and Galvin, W. L. McLaughlin, A. S. Meigooni, R. Nath, J. E. Rogers, and C.
M. Waligorski, ‘‘A checklist for reporting of thermoluminescence dosim- G. Soares, ‘‘Radiochromic film dosimetry: Recommendations of AAPM
etry ~TLD! measurements,’’ Phys. Med. Biol. 44, L15–L17 ~1999!. Radiation Therapy Committee Task Group 55,’’ Med. Phys. 25, 2093–
51
J. F. Williamson, ‘‘Monte Carlo evaluation of specific dose constants in 2115 ~1998!.
water for 125I seeds,’’ Med. Phys. 15, 686 – 694 ~1988!. 73
D. Flühs, M. Heintz, F. Indenkämpen, C. Wieczorek, H. Kolanoski, and
52
National Nuclear Data Center. Nuclear data from NuDat, a web-based U. Quast, ‘‘Direct reading measurement of absorbed dose with plastic
database maintained by the National Nuclear Data Center, Brookhaven scintillators—The general concept and applications to ophthalmic plaque
National Laboratory, Upton, NY, USA; last database update reported as dosimetry,’’ Med. Phys. 23, 427– 434 ~1996!.
74
August 12, 2000; http://www.nndc.bnl.gov/nndc/nudat/ last accessed Feb- A. S. Kirov, C. Hurlbut, J. F. Dempsey, S. B. Shrinivas, J. W. Epstein, W.
ruary 17, 2004. R. Binns, P. F. Dowkontt, and J. F. Williamson, ‘‘Towards two-
53
Lund/LBNL Nuclear Data Search ~1999!. Version 2.0, February 1999, by dimensional brachytherapy dosimetry using plastic scintillator: New
S. Y. F. Chu, L. P. Ekström, and R. B. Firestone, providing the WWW highly efficient water equivalent plastic scintillator materials,’’ Med.
Table of Radioactive Isotopes, maintained by the Lawrence Berkeley Phys. 26, 1515–1523 ~1999!.
75
National Laboratory, Berkeley, CA, USA, and the Department of Physics, S-T. Chiu-Tsao, A. de la Zerda, J. Lin, and J. H. Kim, ‘‘High-sensitivity
Lund University, Sweden; http://nucleardata.nuclear.lu.se/nucleardata/toi/ GafChromic film dosimetry for 125I seed,’’ Med. Phys. 21, 651– 657
last accessed February 17, 2004. ~1994!.
54 76
S. M. Seltzer, ‘‘Calculation of photon mass energy-transfer and mass J. F. Dempsey, D. A. Low, A. S. Kirov, and J. F. Williamson, ‘‘Quantita-
energy-absorption coefficients,’’ Radiat. Res. 136, 147–170 ~1993!. tive optical densitometry with scanning-laser film digitizers,’’ Med. Phys.
55
S. M. Seltzer and J. H. Hubbell, ‘‘Tables and graphs of mass attenuation 26, 1721–1731 ~1999!.
77
coefficients and mass energy-absorption coefficients for photon energies G. H. Chan and W. V. Prestwich, ‘‘Dosimetric properties of the new 125I
1 keV to 20 MeV for elements Z51 to 92 and some dosimetric materi- BrachySeed™ model LS-1 source,’’ Med. Phys. 29, 190–200 ~2002!.
als,’’ Japanese Society of Radiological Technology, ISSN 1340-7716 78
M. J. Maryanski, G. S. Ibbott, P. Eastman, R. J. Schulz, and J. C. Gore,
~1995!. ‘‘Radiation therapy dosimetry using magnetic resonance imaging of poly-
56
M. J. Rivard, D. S. Waid, and J. G. Wierzbicki, ‘‘Mass attenuation coef- mer gels,’’ Med. Phys. 23, 699–705 ~1996!.
ficients of Clear-Pb® for photons from 125I, 103Pd, 99mTc, 192Ir, 137Cs, and 79
M. F. Chan, A. Y. C. Fung, Y-C. Hu, C-S. Chui, H. Amols, M. Zaider, and
60
Co,’’ Health Phys. 77, 571–578 ~1999!. D. Abramson, ‘‘The measurement of three dimensional dose distribution
57
ICRU 37 Stopping Powers for Electrons and Positrons, International of a ruthenium-106 ophthalmological applicator using magnetic reso-
Commission on Radiation Units and Measurements ~ICRU 37, Bethesda, nance imaging of BANG polymer gels,’’ J. Appl. Clin. Med. Phys. 2,
MD, 1989!. 85– 89 ~2001!.
58 80
Private communication with Steve Seltzer of NIST, September 13, 2001. B. F. Hasson, ‘‘Chemical dosimetry in the near-zone of brachytherapy
59
N. S. Patel, S-T. Chiu-Tsao, J. F. Williamson, P. Fan, T. Duckworth, D. sources,’’ Med. Phys. 25, 2076 ~1998!.
81
Shasha, and L. B. Harrison, ‘‘Thermoluminescent dosimetry of the Sym- A. S. Meigooni, Z. Bharucha, M. Yoe-Sein, and K. Sowards, ‘‘Dosimetric
metra™ 125I model I25.S06 interstitial brachytherapy seed,’’ Med. Phys. characteristics of the Best® double-wall 103Pd brachytherapy source,’’
28, 1761–1769 ~2001!. Med. Phys. 28, 2568 –2575 ~2001!.

82
A. S. Meigooni, Z. Li, Vi. Mishra, and J. F. Williamson, ‘‘A comparative brachytherapy dosimetry data recommended by the AAPM Radiation
study of dosimetric properties of Plastic Water and Solid Water in brachy- Therapy Committee Task Group 43,’’ Med. Phys. 26, 153–160 ~1999!.
106
therapy applications,’’ Med. Phys. 21, 1983–1987 ~1994!. R. Nath, A. S. Meigooni, and J. A. Meli, ‘‘Dosimetry on transverse axes
83
J. F. Williamson and A. S. Meigooni, ‘‘Quantitative dosimetry methods in of 125I and 192Ir interstitial brachytherapy sources,’’ Med. Phys. 17, 1032–
brachytherapy,’’ in Brachytherapy Physics, edited by J. F. Williamson, B. 1040 ~1990!.
107
R. Thomadsen, and R. Nath ~Medical Physics Publishing, Madison, WI, K. A. Weaver, V. Smith, D. Huang, C. Barnett, M. C. Schell, and C. Ling,
1995!, pp. 87–133. ‘‘Dose parameters of 125I and 192Ir seed sources,’’ Med. Phys. 16, 636 –
84
A. S. Meigooni, J. A. Meli, and R. Nath, ‘‘A comparison of solid phan- 643 ~1989!.
toms with water for dosimetry of 125I model 6702 brachytherapy 108
J. F. Williamson, ‘‘On the dosimetric influences of air kerma for 103Pd
sources,’’ Med. Phys. 15, 695–701 ~1988!. and 125I brachytherapy sources,’’ in Proceedings of 2000 World Congress
85
K. A. Weaver, ‘‘Response of LiF powder to 125I photons,’’ Med. Phys. 11, on Medical Physics and Biomedical Engineering, edited by G. D. Fuller-
850– 854 ~1984!. ton ~IEEE Engineering Medicine and Biology Society, Chicago, IL,
86
C. Reft, ‘‘The energy response of LiF TLD-100 to low and high energy 2000!, MO E309-02.
photons and to high energy electrons for varying dosimeter thickness,’’ 109
Z. Chen and R. Nath, ‘‘Dose rate constant and energy spectrum of inter-
Phys. Med. Biol. 33, 96 ~1988!. stitial brachytherapy sources,’’ Med. Phys. 28, 86 –96 ~2001!.
87
A. S. Meigooni, D. M. Gearheart, and K. Sowards, ‘‘Experimental deter- 110
E. Mainegra, R. Capote, and E. López, ‘‘Radial dose functions for 103Pd,
mination of dosimetric characteristics of Best double-walled I-125 125
I, 169Yb, and 192Ir brachytherapy sources: an EGS4 Monte Carlo
brachytherapy source’’ Data sheet provided by Best Industries, October 2, study,’’ Phys. Med. Biol. 45, 703–717 ~2000!.
2000. 111
R. Capote, E. Mainegra, and E. Lopez, ‘‘Anisotropy functions for low
88
S. D. Davis, C. K. Ross, P. N. Mobit, L. Van der Zwan, W. J. Chase, and energy interstitial brachytherapy sources: an EGS4 Monte Carlo study,’’
K. R. Shortt, ‘‘The response of LiF thermoluminescent dosemeters to Phys. Med. Biol. 46, 135–150 ~2001!.
photon beams in the energy range from 30 kV X rays to 60Co gamma 112
J. H. Hubbell, J. S. Coursey, J. Hwang, and D. S. Zucker. Bibliography of
rays,’’ Radiat. Prot. Dosim. 106, 33– 43 ~2003!. Photon Total Cross Section ~Attenuation Coefficient! Measurements ~on-
89
I. Kawrakow and D. W. O. Rogers, The EGSnrc code system: Monte line version 2.2!: http://physics.nist.gov/photoncs last accessed February
Carlo simulations of electron and photon transport. NRCC Report PITS- 17, 2004, National Institute of Standards and Technology, Gaithersburg,
701, National Research Council of Canada, Ottawa, 2001. MD, 1998.
90
J. F. Briesmeister, MCNP-A General Monte Carlo N-Particle Transport 113
S. N. Rustgi, ‘‘Photon spectral characteristics of a new double-walled
Code System, Version 4C, LA-13709-M, 2000. iodine-125 source,’’ Med. Phys. 19, 927–931 ~1992!.
91
J. F. Williamson, ‘‘Monte Carlo evaluation of kerma at a point for photon 114
R. Nath and A. Mellilo, ‘‘Dosimetric characteristics of a double wall 125I
transport problems,’’ Med. Phys. 14, 567–576 ~1987!. source for interstitial brachytherapy,’’ Med. Phys. 20, 1475–1483 ~1993!.
92
G. M. Daskalov, R. S. Baker, D. W. O. Rogers, and J. F. Williamson, 115
A. S. Meigooni, D. M. Gearheart, and K. Sowards, ‘‘Experimental deter-
‘‘Multigroup discrete ordinates modeling of 125I 6702 seed dose distribu- mination of dosimetric characteristics of Best 125I brachytherapy source,’’
tions using a broad energy-group cross section representation,’’ Med. Med. Phys. 27, 2168 –2173 ~2000!.
Phys. 29, 113–124 ~2002!. 116
R. Nath and N. Yue, ‘‘Dosimetric characterization of an encapsulated
93
J. F. Williamson, R. Baker, and Z. Li, ‘‘A convolution algorithm for
interstitial brachytherapy source of I-125 on a tungsten substrate,’’ J.
brachytherapy dose computations in heterogeneous geometries,’’ Med.
Brachytherapy 1, 102–109 ~2002!.
Phys. 18, 1256 –1265 ~1991!. 117
94 K. Sowards and A. S. Meigooni, ‘‘Monte Carlo evaluation of the dosim-
Z. Chuanyu and F. Inanc, ‘‘Integral-transport-based deterministic brachy-
etric characteristics of the Best model 2301 I-125 brachytherapy source,’’
therapy dose calculations,’’ Phys. Med. Biol. 48, 73–93 ~2003!.
95 Appl. Radiat. Isot. 57, 327–333 ~2002!.
K. Weaver, ‘‘Anisotropy functions for 125I and 103Pd sources,’’ Med. 118
M. J. Rivard, ‘‘Comment on ‘Experimental measurements of dosimetric
Phys. 25, 2271–2278 ~1998!.
96 parameters on the transverse axis of a new 125I source’,’’ Med. Phys. 28,
J. J. DeMarco, R. E. Wallace, and K. Boedeker, ‘‘An analysis of MCNP
704 ~2001!.
cross-sections and tally methods for low-energy photon emitters,’’ Phys. 119
R. E. Wallace and J. J. Fan, ‘‘Report on the dosimetry of a new design
Med. Biol. 47, 1321–1332 ~2002!. 125
97 Iodine brachytherapy source,’’ Med. Phys. 26, 1925–1931 ~1999!.
E. Storm and H. Israel, ‘‘Photon cross-sections from 1 keV to 100 MeV 120
R. E. Wallace and J. J. Fan, ‘‘Evaluation of a new brachytherapy iodine-
for elements Z51 to Z5100,’’ Nucl. Data, Sect. A 7, 566 –575 ~1970!.
98 125 source by AAPM TG43 formalism,’’ Med. Phys. 25, 2190–2196
W. Alberti, S. Divoux, B. Pothmann, P. Tabor, K-P. Hermann, and D.
~1998!.
Harder, ‘‘Autoradiography for Iodine-125 seeds,’’ Int. J. Radiat. Oncol., 121
J. G. Wierzbicki, M. J. Rivard, D. S. Waid, and V. E. Arterbery, ‘‘Calcu-
Biol., Phys. 25, 881– 884 ~1993!.
99
A. S. Meigooni, J. F. Williamson, and R. Nath, ‘‘Single source dosimetry lated dosimetric parameters of the IoGold 125I source model 3631-A,’’
for interstitial brachytherapy,’’ in Brachytherapy Physics, edited by J. F. Med. Phys. 25, 2197–2199 ~1998!.
122
Williamson, B. R. Thomadsen, and R. Nath ~American Institute of Phys- J. G. Wierzbicki, D. S. Waid, M. J. Rivard, and V. E. Arterbery, ‘‘Radio-
ics, New York, 1995!, pp. 209–233. graphic characterization and energy spectrum of the IoGold 125I source
100
G. J. Kutcher, L. Coia, M. Gillin, W. F. Hanson, S. Liebel, R. J. Morton, model 3631-A/S,’’ Med. Phys. 26, 392–294 ~1999!.
123
J. R. Palta, J. A. Purdy, L. E. Reinstein, G. K. Svensson, M. Weller, and J. F. Briesmeister MCNP-A General Monte Carlo N-Particle Transport
L. Wingfield, ‘‘Comprehensive QA for radiation oncology: Report of the Code System, Version 4B, LA-12625-M, 1997.
124
AAPM Radiation Therapy Committee Task Group 40,’’ Med. Phys. 21, M. J. Rivard, J. G. Wierzbicki, F. Van den Heuvel, R. C. Martin, and R.
581– 618 ~1994!. R. McMahon, ‘‘Clinical brachytherapy with neutron emitting 252Cf
101
B. Fraass, K. Doppke, M. Hunt, G. Kutcher, G. Starkschall, R. Stern, and sources and adherence to AAPM TG-43 dosimetry protocol,’’ Med. Phys.
J. Van Dyke, ‘‘American Association of Physicists in Medicine Radiation 26, 87–96 ~1999!.
125
Therapy Committee Task Group 53: Quality assurance for clinical radio- M. J. Rivard, ‘‘Dosimetry for 252Cf neutron emitting brachytherapy
therapy treatment planning,’’ Med. Phys. 25, 1773–1829 ~1998!. sources: Protocol, measurements, and calculations,’’ Med. Phys. 26,
102
R. Nath, L. L. Anderson, J. A. Meli, A. J. Olch, J. A. Stitt, and J. F. 1503–1514 ~1999!.
126
Williamson, ‘‘Code of practice for brachytherapy physics: Report of the M. J. Rivard, ‘‘Measurements and calculations of thermal neutron fluence
AAPM Radiation Therapy Committee Task Group No. 56,’’ Med. Phys. rate and neutron energy spectra resulting from moderation of 252Cf fast
24, 1557–1598 ~1997!. neutrons: Applications for neutron capture therapy,’’ Med. Phys. 27,
103 1761–1769 ~2000!.
R. Nath, A. S. Meigooni, P. Muench, and A. Melillo, ‘‘Anisotropy func-
127
tions for Pd-103, I-125, and Ir-192 interstitial brachytherapy sources,’’ D. S. Waid, M. J. Rivard, and J. G. Wierzbicki, ‘‘The 0.897 factor is not
Med. Phys. 20, 1465–1473 ~1993!. used to convert S K,85std to S K,99std when performing Monte Carlo calcula-
104
R. S. Sloboda and G. V. Menon, ‘‘Experimental determination of the tions to calculate the dose rate constant,’’ Med. Phys. ~accepted!.
128
anisotropy function and anisotropy factor for model 6711 I-125 seeds,’’ G. S. Ibbott and R. Nath, ‘‘Dose-rate constant for Imagyn 125I brachy-
Med. Phys. 27, 1789–1799 ~2000!. therapy source,’’ Med. Phys. 28, 705 ~2001!.
105 129
E. E. Furhang and L. L. Anderson, ‘‘Functional fitting of interstitial G. S. Ibbott, A. S. Meigooni, and D. M. Gearheart, ‘‘Monte Carlo deter-

mination of dose rate constant,’’ Med. Phys. 29, 1637–1638 ~2002!. AAPM Radiation Therapy Committee on 125I sealed source dosimetry,’’
130
J. F. Williamson, B. M. Coursey, L. A. DeWerd, W. F. Hanson, G. Ibbott, Int. J. Radiat. Oncol., Biol., Phys. 40, 697–702 ~1998!.
R. Nath, and M. J. Rivard, ‘‘Important notice for radiation therapy physi- 146
H. Kubo, ‘‘Exposure contribution from Ti K x rays produced in the tita-
cists using 125I or 103Pd brachytherapy sources,’’ Radiation Therapy nium capsule of the clinical 125I seed,’’ Med. Phys. 12, 215–220 ~1985!.
Committee on Low-Energy Interstitial Brachytherapy Dosimetry Sub- 147
R. Loevinger, ‘‘Wide-angle free-air chamber for calibration of low-
committee, Recommendations of the American Association of Physi- energy brachytherapy sources,’’ Med. Phys. 20, 907 ~1993!.
cists in Medicine, http://rpc.mdanderson.org/rpc/htm/Home – htm/Low- 148
International Organization for Standardization Guide to the Expression of
energy%20documents/ImpNotice – EndUsers – v12.pdf last accessed Feb- Uncertainty in Measurement, ISO, Geneva, Switzerland ~1993!. Jointly
ruary 17, 2004. published by the International Bureau of Weights and Measures ~BIPM!,
131
R. Nath, N. Yue, K. Shahnazi, and P. J. Bongiorni, ‘‘Measurement of International Electrotechnical Commission ~IEC!, International Federa-
dose-rate constant for 103Pd seeds with air kerma strength calibration tion of Clinical Chemistry ~IFCC!, ISO, International Union of Pure and
based upon a primary national standard,’’ Med. Phys. 27, 655– 658 Applied Chemistry ~IUPAC!, International Union of Pure and Applied
~2000!. Physics ~IUPAP!, and the International Organization for Legal Metrology
132
A. S. Meigooni, S. Sabnis, and R. Nath, ‘‘Dosimetry of 103Pd brachy- ~IOLM!.
therapy sources for permanent implant,’’ Endocurietherapy/Hyperthermia 149
A. S. Kirov and J. F. Williamson, ‘‘Monte Carlo-aided dosimetry of the
Oncology 6, 107–117 ~1990!.
133 Source Tech Medical Model STM1251 I-125 interstitial brachytherapy
N. Yue and R. Nath, ‘‘Experimental determination of the anisotropy func-
source,’’ Med. Phys. 28, 764 –772 ~2001!; Med. Phys. 29, 262–263
tion for the Model 200 103Pd ‘‘light seed’’ and derivation of the anisot-
~2002!.
ropy constant based upon the linear quadratic model,’’ Med. Phys. 29, 150
S. M. Seltzer, P. J. Lamperti, R. Loevinger, M. G. Mitch, J. T. Weaver,
1120–1129 ~2002!.
134 and B. M. Coursey, ‘‘New national air-kerma-strength standards for 125 I
R. E. Wallace and J. J. Fan, ‘‘Dosimetric characterization of a new design
103 and 103 Pd brachytherapy seeds,’’ J. Res. Natl. Inst. Stand. Technol. 108,
palladium brachytherapy source,’’ Med. Phys. 26, 2465–2470 ~1999!.
135 337–358 ~2003!, online version http://nvl.nist.gov/pub/nistpubs/jres/108/
G. Luxton, ‘‘Comparison of radiation dosimetry in water and solid phan-
tom materials for I-125 and Pd-103 brachytherapy sources: EGS4 Monte 5/j85sell.pdf last accessed February 17, 2004.
151
Carlo study,’’ Med. Phys. 21, 631– 641 ~1994!. L. A. DeWerd, M. S. Huq, I. J. Das, G. S. Ibbott, W. F. Hanson, T. W.
136
R. E. Wallace, ‘‘On the conversion of dosimetric data among phantom Slowey, J. F. Williamson, and B. M. Coursey, ‘‘Procedures for establish-
materials for 125I and 103Pd brachytherapy sources using build-up factor ing and maintaining consistent air-kerma strength standards for low-
methods with spectral conditions,’’ Med. Phys. 25, A172 ~1998!. energy, photon-emitting brachytherapy sources: Recommendations of the
137
Z. Li, J. R. Palta, and J. J. Fan, ‘‘Monte Carlo calculations and experi- Calibration Laboratory Accreditation Subcommittee of the American As-
mental measurements of dosimetry parameters of a new 103Pd source,’’ sociation of Physicists in Medicine,’’ Med. Phys. 31, 675– 681 ~2004!.
152
Med. Phys. 27, 1108 –1112 ~2000!. J. F. Williamson and Z. Li, ‘‘Monte Carlo aided dosimetry of the mi-
138
M. J. Rivard, ‘‘A discretized approach to determining TG-43 brachy- croSelectron pulsed and high dose rate 192Ir sources,’’ Med. Phys. 22,
therapy dosimetry parameters: Case study using Monte Carlo calculations 809– 819 ~1995!.
153
for the MED3633 103Pd source,’’ Appl. Radiat. Isot. 55, 775–782 ~2001!. M. J. Rivard, ‘‘Comprehensive Monte Carlo calculations of AAPM Task
139
G. Luxton and G. Jozsef, ‘‘Radial dose distribution, dose to water and Group Report No. 43 dosimetry parameters for the Model 3500 I-Plant
125
dose rate constant for monoenergetic photon point sources from 10 keV I brachytherapy source,’’ Appl. Radiat. Isot. 57, 381–389 ~2002!.
154
to 2 MeV: EGS4 Monte Carlo calculation,’’ Med. Phys. 26, 2531–2538 G. M. Daskalov, E. Löffler, and J. F. Williamson, ‘‘Monte Carlo-aided
~1999!. dosimetry of a new high dose-rate brachytherapy source,’’ Med. Phys. 25,
140
Å. K. Carlsson and A. Ahnjesö, ‘‘The collapsed cone superposition algo- 2200–2208 ~1998!.
155
rithm applied to scatter dose calculations in brachytherapy,’’ Med. Phys. J. Earle, R. W. Kline, and D. M. Robertson, ‘‘Selection of iodine 125 for
27, 2320–2332 ~2000!. the Collaborative Ocular Melanoma study,’’ Arch. Opthalmol. 105, 763–
141
T. P. Loftus, ‘‘Standardization of cesium-137 gamma-ray sources in terms 764 ~1987!.
156
of exposure units ~roentgens!,’’ J. Res. Natl. Bur. Stand., Sect. A 74A, M. A. Astrahan, G. Luxton, G. Jozsef, P. E. Liggett, and A. Petrovich,
1– 6 ~1970!. ‘‘Optimization of 125I opthalmic plaque brachytherapy,’’ Med. Phys. 17,
142 1053–1057 ~1990!.
T. P. Loftus, ‘‘Exposure standardization of iodine-125 seeds used for
157
brachytherapy,’’ J. Res. Natl. Bur. Stand. 89, 295–303 ~1984!. S. Nag, J. M. Quivey, J. D. Earle, D. Followill, J. Fontanesi, and P. T.
143 Finger, ‘‘The American Brachytherapy Society recommendations for
V. H. Ritz, ‘‘Standard free-air chamber for the measurement of low en-
ergy x rays ~20-100 kilovolts-constant-potential!,’’ J. Res. Natl. Bur. brachytherapy of uveal melanomas,’’ Int. J. Radiat. Oncol., Biol., Phys.
Stand., Sect. C 64C, 49–53 ~1960!. 56, 544 –555 ~2003!.
144 158
J. T. Weaver, T. P. Loftus, and R. Loevinger, ‘‘Calibration of gamma-ray- A. L. Krintz, W. F. Hanson, G. S. Ibbott, and D. S. Followill, ‘‘A reanaly-
emitting brachytherapy sources,’’ NBS Special Publication 250-19 1988. sis of the Collaborative Ocular Melanoma study medium tumor trial eye
145
H. D. Kubo, B. M. Coursey, W. F. Hanson, R. W. Kline, S. M. Seltzer, R. plaque dosimetry,’’ Int. J. Radiat. Oncol., Biol., Phys. 56, 889– 898
E. Shuping, and J. F. Williamson, ‘‘Report of the ad hoc committee of the ~2003!.

Task Group No. 43-U1

Uploaded by

Copyright:

Available Formats

Task Group No. 43-U1

Uploaded by

Document Information

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

Task Group No. 43-U1

Uploaded by

Copyright:

Available Formats

Update of AAPM Task Group No.

43 Report: A revised AAPM protocol

TABLE OF CONTENTS Chamber. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 666

Medical Physics, Vol. 31, No. 3, March 2004

Medical Physics, Vol. 31, No. 3, March 2004

Medical Physics, Vol. 31, No. 3, March 2004

where r denotes the distance ~in centimeters! from the center

Medical Physics, Vol. 31, No. 3, March 2004

Medical Physics, Vol. 31, No. 3, March 2004

Medical Physics, Vol. 31, No. 3, March 2004

Medical Physics, Vol. 31, No. 3, March 2004

Medical Physics, Vol. 31, No. 3, March 2004

Medical Physics, Vol. 31, No. 3, March 2004

NIST date used by ADCL CONL % difference in L

Medical Physics, Vol. 31, No. 3, March 2004

Line source approximation Point source approximation

Amersham Amersham Best NASI Bebig Imagyn

systematic ~type B! uncertainties of Monte Carlo-based dose y5 f ~ x 1 ,...,x N ! ,

Line source approximation Point source approximation

Theragenics 200 NASI MED3633

0.10 0.911 0.494

Medical Physics, Vol. 31, No. 3, March 2004

TABLE IV. F(r, u ) for Amersham model 6702.

0 0.385 0.420 0.493 0.533 0.569 0.589

f an(r) 0.986 0.960 0.952 0.951 0.954 0.954

TABLE V. F(r, u ) for Amersham model 6711.

0 0.333 0.370 0.442 0.488 0.520 0.550

f an(r) 0.973 0.944 0.941 0.942 0.943 0.944

TABLE VI. F(r, u ) for Best model 2301.

0 0.367 0.454 0.922 0.902 0.894 0.893 0.858

f an(r) 0.945 0.987 0.968 0.971 0.969 0.991 0.969

Medical Physics, Vol. 31, No. 3, March 2004

TABLE VII. F(r, u ) for NASI model MED3631-A/M.

0 1.038 0.690 0.702 0.667 0.718 0.771

f an(r) 1.288 1.008 0.952 0.945 0.948 0.948

0 0.302 0.429 0.512 0.579 0.610 0.631 0.649 0.684

f an(r) 1.122 0.968 0.939 0.939 0.938 0.940 0.941 0.949

0 0.241 0.337 0.362 0.424 0.454

f an(r) 0.867 0.886 0.894 0.897 0.879

Medical Physics, Vol. 31, No. 3, March 2004

0 0.619 0.694 0.601 0.541 0.526 0.504 0.497 0.513 0.547

TABLE XI. F(r, u ) for NASI model MED3633.

0 1.024 0.667 0.566 0.589 0.609 0.733

f an(r) 1.257 0.962 0.903 0.895 0.898 0.917

Medical Physics, Vol. 31, No. 3, March 2004

Repetitive measurements 4.5%

Quadrature sum 4.5% 7.3%

Total combined uncertainty in L 8.7%

Monte Carlo uncertainties

Statistics 0.3% 1.0%

Quadrature sum 2.5% 5.0%

Medical Physics, Vol. 31, No. 3, March 2004

Medical Physics, Vol. 31, No. 3, March 2004

Medical Physics, Vol. 31, No. 3, March 2004

27.202 0.406 20.074 0.224

Medical Physics, Vol. 31, No. 3, March 2004

0 0.0000 0.0124 75.5268 23.1781 1.2827

Medical Physics, Vol. 31, No. 3, March 2004

Medical Physics, Vol. 31, No. 3, March 2004

Medical Physics, Vol. 31, No. 3, March 2004