Monte Carlo study of LDR seed dosimetry with an applicationin a clinical brachytherapy breast implant

C. Furstossa�

Medical Physics Unit, McGill University, Montreal General Hospital, 1650 Cedar Avenue, Montreal,Quebec H3G 1A4, Canada

B. ReniersMedical Physics Unit, McGill University, Montreal General Hospital, 1650 Cedar Avenue, Montreal,Quebec, H3G 1A4, Canada and Department of Radiation Oncology (MAASTRO), GROW, University HospitalMaastricht, Maastricht 6202 AZ, The Netherlands

M. J. Bertrand and E. PoonMedical Physics Unit, McGill University, Montreal General Hospital, 1650 Cedar Avenue, Montreal,Quebec H3G 1A4, Canada

J.-F. CarrierDépartement de Radio-Oncologie et Centre de Recherche du CHUM, Hôpital Notre-Dame,1560 rue Sherbrooke Est, Montréal, Québec H2L 4M1, Canada

B. M. Keller and J. P. PignolDepartment of Radiation Oncology, Sunnybrook Health Sciences Centre, University of Toronto, Toronto,Ontario M4N 3M5, Canada

L. BeaulieuDépartement de Radio-Oncologie et Centre de Recherche en Cancérologie, CHUQ Pavillon L’Hôtel-Dieude Québec, 11 Côte du Palais, Québec G1R 2J6, Canada and Département de Physique, de GéniePhysique et d’Optique, Université Laval, Québec G1K 7P4, Canada

F. Verhaegenb�

Medical Physics Unit, McGill University, Montreal General Hospital, 1650 Cedar Avenue, Montreal,Quebec H3G 1A4, Canada and Department of Radiation Oncology (MAASTRO), GROW, University HospitalMaastricht, Maastricht 6202 AZ, The Netherlands

�Received 15 October 2008; revised 12 March 2009; accepted for publication 20 March 2009;published 23 April 2009�

A Monte Carlo �MC� study was carried out to evaluate the effects of the interseed attenuation andthe tissue composition for two models of 125I low dose rate �LDR� brachytherapy seeds �Medi-Physics 6711, IBt InterSource� in a permanent breast implant. The effect of the tissue compositionwas investigated because the breast localization presents heterogeneities such as glandular andadipose tissue surrounded by air, lungs, and ribs. The absolute MC dose calculations were bench-marked by comparison to the absolute dose obtained from experimental results. Before modeling aclinical case of an implant in heterogeneous breast, the effects of the tissue composition and theinterseed attenuation were studied in homogeneous phantoms. To investigate the tissue compositioneffect, the dose along the transverse axis of the two seed models were calculated and compared indifferent materials. For each seed model, three seeds sharing the same transverse axis were simu-lated to evaluate the interseed effect in water as a function of the distance from the seed. A clinicalstudy of a permanent breast 125I implant for a single patient was carried out using four dosecalculation techniques: �1� A TG-43 based calculation, �2� a full MC simulation with realistictissues and seed models, �3� a MC simulation in water and modeled seeds, and �4� a MC simulationwithout modeling the seed geometry but with realistic tissues. In the latter, a phase space filecorresponding to the particles emitted from the external surface of the seed is used at each seedlocation. The results were compared by calculating the relevant clinical metrics V85, V100, and V200

for this kind of treatment in the target. D90 and D50 were also determined to evaluate the differencesin dose and compare the results to the studies published for permanent prostate seed implants inliterature. The experimental results are in agreement with the MC absolute doses �within 5% forEBT Gafchromic film and within 7% for TLD-100�. Important differences between the dose alongthe transverse axis of the seed in water and in adipose tissue are obtained �10% at 3.5 cm�. The

comparisons between the full MC and the TG-43 calculations show that there are no significant

1848 1848Med. Phys. 36 „5…, May 2009 0094-2405/2009/36„5…/1848/11/$25.00 © 2009 Am. Assoc. Phys. Med.

1849 Furstoss et al.: LDR seed dosimetry: Monte Carlo study 1849

differences for V85 and V100. For V200, 8.4% difference is found coming mainly from the tissuecomposition effect. Larger differences �about 10.5% for the model 6711 seed and about 13% for theInterSource125� are determined for D90 and D50. These differences depend on the composition of thebreast tissue modeled in the simulation. A variation in percentage by mass of the mammary glandand adipose tissue can cause important differences in the clinical dose metrics V200, D90, and D50.Even if the authors can conclude that clinically, the differences in V85, V100, and V200 are acceptablein comparison to the large variation in dose in the treated volume, this work demonstrates that thedevelopment of a MC treatment planning system for LDR brachytherapy will improve the dosedetermination in the treated region and consequently the dose-outcome relationship, especially forthe skin toxicity. © 2009 American Association of Physicists in Medicine.�DOI: 10.1118/1.3116777�

Key words: low dose rate brachytherapy, Monte Carlo, breast implant, I-125, MCNPX, film125

dosimetry, TLD dosimetry, 6711, InterSource

I. INTRODUCTION

Low dose rate �LDR� permanent breast seed implants havebeen used as a form of partial breast irradiation for earlystage breast cancer patients.1,2 The main advantage of thistechnique is the single treatment session versus 5–6 weeks oftreatment for external beam radiotherapy. The planning andpostprocedure evaluation are carried out using the CT breastimages calculating the isodose distribution according to thesource positions with the AAPM Radiation Therapy TaskGroup 43 formalism �TG-43�.3,4 The TG-43 dose calculationdoes not account for the attenuation between the differentseeds since the dose is calculated individually for each seedand then summed to consider all seeds. Moreover, the patienttissue heterogeneities are ignored assuming water in thewhole calculation area. The combined effects of these ap-proximations have been previously studied for permanentprostate LDR seed implants.5–8 LDR brachytherapy forbreast is different from prostate in that the tissue compositionis much more heterogeneous with glandular and adipose tis-sues, lungs, ribs, and the surrounding air such that uncertain-ties using the TG-43 formalism may result in significantover- or underdosage.

The objective of this paper is to evaluate theoretically andexperimentally the combined effect of interseed attenuationand tissue heterogeneity for permanent breast seed implantsin a realistic case. The dosimetry of permanent breast seedimplants was carried out with Monte Carlo �MC� simula-tions. Although for radioprotection issues, real patient treat-ments are performed with Pd-103 seeds,2 for this study ofMC dose calculation and experiments, implants were mod-eled with I-125 seeds due to isotope availability in our cen-ter. This allowed experimental validation of the dosimetricmodels. Two different models of iodine seed presenting verydifferent construction geometry were considered for the im-plant. The paper is organized as follows. First, the LDRsources, phantoms, dosimetry techniques, and the MC simu-lations are described. Then, experimental benchmarks of thesimulations in homogeneous phantoms are presented. Theeffects of tissue composition and interseed attenuation are

studied to determine the geometry parameters which impact

Medical Physics, Vol. 36, No. 5, May 2009

the two effects. Finally, the results of a study for a permanentbreast seed implant in a single patient are presented for thetwo different LDR seed models.

II. MATERIALS AND METHODS

II.A. Brachytherapy LDR sources

Two models of iodine seed were studied: The model 6711seed9,10 �Medi-Physics, Arlington Heights, IL� and theInterSource125 seed11,12 �IBt, Seneffe, Belgium�. The model6711 seed consists of a silver right circular cylindrical rodencased inside a titanium cylindrical capsule. Two titaniumhalf spheres are located at each boundary of the titaniumcylinder. The iodine is distributed around the silver rod in ahalide layer of AgBr and AgI �Fig. 1�. The InterSource125

model is designed as two concentric titanium cylinderswelded at the extremities. An imaging marker consisting of aplatinum ring covers the center of the internal cylinder. A 125Iradioactive coating is applied around the internal cylinderand the platinum marker �Fig. 1�. All seeds come with aspecified air kerma strength value measured by the manufac-turer at a given time. These two models were chosen becausethey had very different design characteristics with the pres-ence of a silver rod for the 6711 model and the presence ofonly air for the InterSource125 at the center of the seed.13

II.B. Phantoms and detectors

Two experiments were performed, combining two differ-ent phantoms and two types of dosimeters in order to vali-date the absolute MC dose calculations.

Titanium shell Silver rodIodine uniformly

distributed

0.8 mm

L=2.8 mm

0.4 mm

4.5 mm Platinum/Iridium

radio-opaque

markerIodine-125

Laser

welded

seal

Titanium outer tube Titanium inner tube

m

4.5 mm

L=3.7 mm

0.8 mm0.5

mm

FIG. 1. Schematic drawing of the model 6711 seed �Ref. 10� �left� and the125

IBt InterSource seed �Ref. 12� �right�.

1850 Furstoss et al.: LDR seed dosimetry: Monte Carlo study 1850

II.B.1. Experiment 1

A model 6711 seed was implanted inside a plastic 4French catheter, which was centered at the surface of a cir-cular piece of Gafchromic EBT film �ISP Corp., Wayne, NJ�mounted on a plastic jig to stretch the film. Then, the jig waspositioned in a water tank in order to mimic photon attenu-ation and scatter similar to the human body �Fig. 2�. After 1day of irradiation, the film was read with a flat-bed scanner.14

According to the calibration curve of the film �dose as afunction of the optical density� carried out with an 192Irsource and the air kerma strength of the source on the day ofthe experiment, the isodoses were determined. EBT calibra-tion was carried out with 125I and 192Ir sources: Within themeasurement uncertainty, the same calibration curve was ob-tained. Due to the higher number of measurement points ob-tained, the calibration with the 192Ir source was used. Theexperiment was repeated to improve the statistical uncer-tainty of the results. The overall accuracy of dose measure-ment was estimated to be �6% �uncertainty in conversionfrom optical density to dose in the calibration procedure,uncertainty in seed air kerma strength, and uncertainty in netoptical density from experimental film scanning�.15

II.B.2. Experiment 2

A phantom made of acrylic slices was used to representthe breast. Each slice had a thickness of 0.5 cm and wasdesigned in order to model a conical shape �Fig. 4�a��. Fiveinterseeds were snugly fit in shallow slots machined in oneslice. In another slice, at 0.5 cm from the sources, 30 TLD-100 ribbons �Harshaw, Gamble Technology, Mississauga,Ontario, Canada� were inserted into slots. After irradiation,the TLDs were read with an automatic reader Harshaw QS5500. The heating treatments for experiment and calibrationwere similar. For calibration, the TLDs were irradiated usinga Theratron T-1000 60Co unit �Theratronics InternationalLtd., Kanata, Ontario, Canada�. To take the TLD energy de-pendence into account, a correction factor of 1.40 was ap-plied between the calibration carried out with a 60Co beamand the measurements with 125I radiation. This correction

FIG. 2. Isodose measurement with the model 6711 seed and EBT Gafchro-mic® film in a water tank.

factor was determined averaging the factors obtained in vari-

Medical Physics, Vol. 36, No. 5, May 2009

ous studies.16–18 The statistical uncertainty for the TLDs wascalculated averaging the standard deviation of the calibrationfactors over four calibrations. An uncertainty in the positionof the seed in the slice was also incorporated. The total un-certainty for absolute dose determination with TLDs wasevaluated as �7%.

II.C. Monte Carlo calculation

II.C.1. Code description

The Monte Carlo code MCNPX �Ref. 19� version 2.5.0 wasused in this study. One of the main advantages of this versionis the calculation time speedup20 to obtain results �“tallies”�for a large number of repeated structures �“lattices”�, such asvoxels. Compton scattering, coherent scattering, photoelec-tric effect, and pair production were modeled in each simu-lation. Due to the low energy secondary electrons producedby 125I, the thick-target bremsstrahlung model was used.21

The distributed cross-section data were from ENDF/B-VI,22

where the photon interaction data were based on the EPDL97tables �specification with the card MPLIB=04�.23,24

II.C.2. Seed modeling, absolute dose, and validation

The modeled seed geometries in MCNPX were as close aspossible to reality �Fig. 1�. The welding at the two extremi-ties of the silver cylinder for the 6711 seed model and thetitanium cylinders for the InterSource125 model was approxi-mated by modeling cones at each cylinder end. The densityand the atomic composition used for the titanium, the silver,the air, and the platinum were taken from de Laeter25 and theICRU report 44.26 The iodine spectrum used in simulationswas taken from the TG-43 update.4 The iodine was distrib-uted as a cylindrical source around the silver rod in a halidelayer �2 �m thick mixture of AgBr and AgI with a density of6 g cm−3�10 for the model 6711 seed and for theInterSource125 seed, around the radio-opaque marker �as alayer with 9 �m thickness and 0.5 mm length� and the innertitanium tube �as two layers with 15 �m thickness and 0.8mm length, one at each end of the seed�.

The dose was scored in the medium of the area of interestwith the photon track length tally F6 since the transport ofthe very short secondary electron tracks do not need to bemodeled. The unit of the dose obtained by this tally wasMeV g−1 particle−1. The calculated dose MCD was con-verted into dose D in Gy using the equation below,

D = MCD ·SK,actual

MCSK· �

0

t

e−�tdt . �1�

In this equation, MCSK is the air kerma strength calculatedby the MC code �MeV cm2 g−1 particle−1� and SK,actual is theair kerma strength of the source provided by the manufac-turer corrected for the source decay between the calibrationday and the day of the experiment �cGy cm2 h−1�, then cor-rected by the integral for the source decay during the time tof the experiment. To calculate the air kerma strength MCSK

according to the definition in the TG-43 update, the seed was

simulated in vacuum and only the photons of energy greater

1851 Furstoss et al.: LDR seed dosimetry: Monte Carlo study 1851

than 5 keV were considered. The dose was determined in asmall air voxel �0.2�1�1 mm3� at a distance of 5 cm fromthe source. Due to the quasi-isotropy and the small size ofthe source, the volume averaging artifact is inferior to 1% at5 cm from the source according to the “good practice forMonte Carlo calculations” of the TG-43 update report.

The geometry of the model 6711 seed was validated bycomparing the radial dose function and the anisotropy func-tion published by Dolan et al.10 with MC calculations inwater. For the InterSource125 model, the source modelingwas directly taken from Reniers et al.12 The validation of thesource geometry is presented in Reniers et al.12 For the ab-solute dose calculation, validation was carried out comparingMC calculation results with dose measurements. First of all,the seed was modeled in a catheter placed at the surface of anEBT Gafchromic® film located in a 30�30�30 cm3 watertank. All layers �clear polyester, active layers, and surfacelayer� of the EBT film were modeled. The atomic composi-tion of the film was taken from Devic et al.14 The dose wasscored in the two active layers in 0.6�0.6�0.017 mm3

voxels. Then, the conic breast phantom considered to be inPMMA with a density of 1.19 g cm−3 was modeled with fiveInterSource125 seeds in the same slice. The dose was scoredin 1 mm3 cubes with TLD material �lithium fluoride with adensity of 2.635 g cm−3� at the experimental locations totake into account the TLD attenuation. The absolute dose Dobtained by MC calculation was compared with the mea-sured dose Dmeas calculating the difference �D according tothe equation below:

�D�%� =Dmeas − D

D� 100. �2�

For each simulation with Gafchromic® film and TLDs, thenumber of incident particles was about 8�108 in order toobtain a statistical uncertainty below 5% in the area of inter-est.

II.C.3. Tissue composition effect and interseedeffect calculations

The dose normalized by the air kerma strength of theInterSource125 and the model 6711 seeds were calculated ac-cording to Eq. �1� along the transverse axis of the seed inwater, acrylic, and in different biological tissues: Mammarygland tissue, adipose tissue,26 and breast tissue �50%–50%by mass adipose-glandular tissue�27 to evaluate the tissuecomposition effect. Following the study of Taylor et al.,28 thevoxel size was 0.1�0.1�0.1 mm3 for distances inferior to1 cm, 0.5�0.5�0.5 mm3 for distances between 1 and 5 cm,and 1�1�1 mm3 for distances superior to 5 cm. Then, astudy was carried out to evaluate the interseed effect for bothiodine seed models. According to Meigooni et al.,29 the in-terseed effect was evaluated by taking the ratio of the dose toa point deposited by several seeds over the sum of dosesdeposited individually by each seed at the same point. Forthis study, three seeds were modeled in water sharing thesame transverse plane with 1 cm separation. The dose D1+2+3

was scored at different distances from the seed, then com-

Medical Physics, Vol. 36, No. 5, May 2009

pared to the sum of the doses D1, D2, and D3 scored at thesame point when only one seed was present. Equation �3�was used to evaluate the interseed effect,

Interseed effect =D1+2+3

D1 + D2 + D3. �3�

A statistical uncertainty below 2% was obtained.

II.C.4. Postimplant study

Dose calculations were performed using CT images of apermanent breast implant carried out at the SunnybrookHealth Sciences Centre in Toronto. The treatment procedurewas described by Pignol et al.1 Sixty-four 103Pd seeds wereimplanted with an initial air kerma strength of 2.5 U. A pre-scription dose of 90 Gy was given to cover the planningtarget volume �PTV� defined as “the lumpectomy cavity plusa margin of 1 cm modified to 5 mm deep to the skin surfaceand also along the fascia pectoralis.” The CT images and the64 seed positions used to create the input file with the MCcode were extracted from the CT exam performed 2 monthsafter the implant. With in-house software BrachyGUI,30 theDICOM CT images were converted into material maps ac-cording to the CT calibration curve �one material for eachvoxel�. Six materials �air, lung, adipose tissue, breast tissue,bone cartilage, and cortical bone� were determined. Since itis not possible to define a continuous density for the materi-als with the MCNPX code, several densities were assigned tothe materials �2 for air, 5 for lung, 3 for adipose tissue, 1 forbreast tissue, 1 for bone cartilage, and 2 for cortical bone�.This material map was used to create the input file for theMC code. About 300 000 voxels of 2�2�2 mm3 were de-fined. The 64 seeds were modeled at the positions deter-mined from the CT images and considered to be parallel tothe axis of needle insertion �Z axis�. For each seed model,three input files were created.

�1� A full MC �FMC� input file: This corresponds to theinput file with the 6 materials and the 64 seeds.

�2� A MC input file without seeds �MCWS�: This corre-sponds to the input file with the six materials and noseed models included. A phase space file, replacing thefull seed model, was sampled at all seed positions toreproduce the radiation emitted by the external surfaceof the seed. This approach ignores the interseed attenu-ation.

�3� A MC input file with only water �MCW�: This corre-sponds to 64 seeds in homogeneous water.

The comparison between the MCWS and the FMC gavethe effect of the interseed attenuation, and the comparisonbetween the MCW and the FMC gave the effect of the tissuecomposition. These comparisons were based on dosimetrymetrics V200, V100, and V85 calculated in the PTV since forthis kind of treatment, only dose variation in volume is clini-cally pertinent.31 D90 and D50 were calculated too in order toevaluate physically the differences and compare the results

with the literature where only dosimetric studies of prostate

1852 Furstoss et al.: LDR seed dosimetry: Monte Carlo study 1852

permanent seed implant were published. For each simulation,it was necessary to take the voxels partially filled by a seedinto account. In fact, MCNPX assumes that all voxels haveequal volume and, consequently, underestimates the dosewhere the seed partially occupies a voxel since the dose iscalculated by the ratio of the deposited energy over the prod-uct of the density by the volume. An independent calculationwas performed to determine stochastically the effective vol-ume of each voxel,32 i.e., the volume not occupied by seeds.The doses calculated by MCNPX were thus corrected by ageometry factor for all voxels partially filled by seeds.

Each MC simulation was run at least three times withdifferent initial random number seeds. The standard devia-tion of the various calculated dosimetry parameters was de-termined for both seed models. The relative statistical uncer-tainties, defined as 1 standard deviation, are given in Table I.

II.C.5. Influence of the breast tissue composition

The breast tissue was composed of mammary gland andadipose tissue whose percentage differs from individual toindividual and also depends on the age of the patient.33 The50%–50% adipose-glandular tissue composition representedan average breast tissue.27 Moreover, it was very difficult todifferentiate breast tissue from adipose tissue on the CT im-ages because the two densities were similar �0.95 g cm−3 forthe adipose tissue and 1.02 g cm−3 for the breast tissue� andthe boundary between the two tissues could not be distin-guished visually. To study in more detail the effect of thebreast tissue composition on the dose metric evaluation, MCcalculations were carried out. All tissues were replaced bybreast tissue whose percentage of adipose tissue and mam-mary tissue was varied. Five simulations with the 64 model6711 seeds were carried out taking as breast tissue composi-tions 0%–100%, 75%–25%, 50%–50%, 25%–75%, and100%–0% glandular-adipose tissue, respectively. The dosemetrics calculated in the PTV for each breast compositionwere compared to the dose metrics calculated with the MCWinput file.

II.D. TG-43 calculation technique

Another calculation technique for the permanent breastseed implant consisted of using the AAPM TG-43 updatedformalism.4 In all cases, the general 2D formalism was ap-plied approximating the source as a line source. The radialdose and anisotropy function data used for the model 6711were from Dolan et al.10 and, for the InterSource125 seed,from Reniers et al.12 The comparison between the FMC andthe TG43-U1 gave the combined impact of the interseed ef-

TABLE I. Relative statistical uncertainties for various

Model of seedV85

�%�V100

�%�

6711 �0.1 �0.1InterSource125 �0.1 �0.1

fect and the tissue composition.

Medical Physics, Vol. 36, No. 5, May 2009

II.E. Dosimetry comparison

Dose metrics extracted from DVHs�V85,V100,V200,D90,D50,� were used to compare the varioussets of calculations. The relative difference �DX betweentwo calculation techniques A and B for DX values was givenby

�DX�%� =DX

B − DXA

DXB · 100. �4�

For VX values, the relative difference �VX was

�VX�%� =VX

B − VXA

VXB · 100. �5�

III. RESULTS AND DISCUSSION

III.A. Source geometry and MC absolutedose validation

The comparison between the values published by Dolan etal.10 and the MC calculation are shown in Tables II and IIIfor the radial dose function and the anisotropy function re-spectively. All the results are in good agreement, with a�2.6% maximum difference. The isodose distributions cal-culated by MCNPX and measured by EBT Gafchromic film ina water phantom with a single model 6711 seed are presented

etric parameters.

V200

�%�D90

�%�D50

�%�

�0.1 0.05–0.30 �0.1�0.1 0.10–0.40 �0.1

TABLE II. The radial dose function, g�r�, for the model 6711 calculated inwater and compared to the results of Dolan et al. �Ref. 10� by the relativedifference ���.

r�cm�

g�r��

�%�Dolan et al. This study

0.1 1.109 1.121 1.10.2 1.108 1.127 1.80.3 1.103 1.118 1.40.4 1.089 1.104 1.40.5 1.075 1.088 1.20.75 1.041 1.050 0.91 1.000 1.000 0.01.5 0.905 0.920 1.72 0.805 0.817 1.53 0.626 0.633 1.14 0.475 0.479 0.95 0.355 0.357 0.66 0.263 0.263 0.27 0.194 0.194 �0.18 0.143 0.144 0.59 0.106 0.105 �1.1

dosim

1853 Furstoss et al.: LDR seed dosimetry: Monte Carlo study 1853

in Fig. 3�a�. The relative difference between the calculationand the measurement is in general within �5%, as shown inFig. 3�b� according to the transverse axis of the seed �verticaldirection, y axis�. In Fig. 3�c�, important and asymmetricdifferences �up to �15%� are noted on the longitudinal axisof the seed �horizontal direction, x axis�. According to thestudy carried out by Dolan et al.,10 these differences aremainly due to a possible 0.4 mm offset of the vertical posi-tion of the silver rod in the Ti shell and the end weld thick-ness variations ��0.15 mm� when the seed is produced bythe manufacturer. Figure 4�b� represents the relative differ-ence between the calculation and the TLD measurement in abreast acrylic phantom. The relative differences are in gen-eral within �7% except for some points far from the sourceswhere the relative difference exceeds 10%.

III.B. Tissue composition effect

The dose normalized by air kerma strength calculated fordifferent tissues, acrylic, and water along the transverse axisof the seed are shown as a function of the distance for theInterSource125 seed �Fig. 5�a�� and the model 6711 seed �Fig.5�b��. The largest dose differences are observed for water andadipose tissue, which is a main breast constituent. Differ-ences between the TG43-U1 and the FMC are therefore ex-pected for the permanent breast seed implant due to the tis-sue composition. Moreover, differences between the dose inadipose tissue and breast tissue are also prominent; about10% at 3.5 cm and up to 40% at 7 cm. The approximation ofan average tissue of 50% adipose and 50% glandular for thebreast can lead to discrepancies in comparison to the realtissue composition of the patient. This point will be dis-cussed in more detail in Sec. III D.

The dose ratio along the transverse axis of theInterSource125 seed and the 6711 seed is represented in Fig.5�c�. The differences between the two models increase as afunction of the distance from the seed. In fact, the 6711 seedis designed with a silver marker inside the seed. The photonscoming from the 125I interact with the silver, creating twocharacteristic x rays at 22.0 and 22.2 keV. Therefore, the

TABLE III. The anisotropy function, F�r ,��, for the model 6711 seed calculadifference ���.

Angle, ��deg�

F�0.5 cm,�� F�1 cm,��

Dolanet al.

Thisstudy

�

�%�Dolanet al.

Thisstudy

�

�%�Dolet a

0 0.199 0.197 �0.9 0.280 0.277 �1.0 0.375 0.310 0.306 �1.3 0.399 0.398 �0.2 0.48

10 0.363 0.360 �0.9 0.451 0.454 0.6 0.5315 0.484 0.496 2.4 0.554 0.562 1.4 0.6220 0.605 0.591 �2.4 0.649 0.651 0.4 0.7030 0.781 0.792 1.4 0.791 0.798 0.9 0.8150 0.936 0.949 1.4 0.936 0.953 1.8 0.9470 0.995 0.987 �0.8 0.998 1.002 0.4 0.9990 1.000 1.000 1.000 1.000 1.00

photons of the 6711 model, on average, are less penetrating

Medical Physics, Vol. 36, No. 5, May 2009

than the photons of the InterSource125 model which does notemit these two characteristic x rays since the InterSource125

seed does not include any silver.

III.C. Interseed attenuation effect

Figure 6 shows the interseed effect according to Eq. �3�for three seeds sharing the same transverse plane as a func-tion of the distance for the two seed models. The averageinterseed effects are 0.86 and 0.90 �statistical uncertainty:2%� for the model 6711 and the InterSource125, respectively.The seed attenuation is more important for the model 6711than the InterSource125 seed due to the silver rod in 6711 thatabsorbs more radiation. No correlation between the distanceand the interseed effect is found. In fact, it depends on theseed type,13 the seed configuration,24 and the number ofseeds per unit volume of implanted tissue.7

III.D. Study of a permanent breast iodineseed implant

Tables IV and V present the results of the dose metriccomparison for a breast brachytherapy treatment with the6711 seeds and the InterSource125 seeds, respectively. For thetwo seed models, the V85 and the V100 relative differences�Eq. �5�� are very close to 0% whatever the studied effect is.The prescription dose covers the whole PTV, independent ofthe tissue material and the seed modeling. For example, theFMC simulation, where the dose in the PTV is the lowest

TABLE IV. Comparison of dosimetric parameters for the PTV for a breastbrachytherapy treatment with 6711 seeds.

Interseedattenuation

effect

Tissuecomposition

effectComparison

TG43-U1 and FMC

�V85 �%� 0.03 0.02 0.05�V100 �%� 0.09 0.05 0.1�V200 �%� 2.5 6.7 8.9�D90 �%� 2.4 7.7 10.3�D50 �%� 2.2 8.4 10.8

water and compared to the results of Dolan et al. �Ref. 10� by the relative

2 cm,�� F�3 cm,�� F�5 cm,��

Thisstudy

�

�%�Dolanet al.

Thisstudy

�

�%�Dolanet al.

Thisstudy

�

�%�

0.381 0.6 0.444 0.445 0.3 0.523 0.524 0.20.469 �2.2 0.531 0.519 �2.3 0.593 0.596 0.50.524 �2.2 0.583 0.569 �2.5 0.641 0.642 0.10.613 �1.5 0.662 0.645 �2.5 0.710 0.697 �1.90.682 �2.6 0.728 0.715 �1.8 0.764 0.766 0.30.812 0.1 0.830 0.819 �1.3 0.844 0.855 1.30.959 1.9 0.952 0.961 0.9 0.956 0.932 �2.50.973 �2.4 1.001 0.987 �1.4 0.999 0.982 �1.71.000 1.000 1.000 1.000 1.000

ted in

F�

anl.

906202170

1854 Furstoss et al.: LDR seed dosimetry: Monte Carlo study 1854

since it is scored in a material different than water �Fig. 5�and the interseed attenuation is taken into account, gives99.90% for the V100 parameter. The relative differences forV200 are higher by about 2.0% for the interseed attenuationeffect and by about 6.7% for the tissue composition effect.For the latter, the positive sign of the difference is due to thefact that the dose scored in water is higher than the dosescored in adipose and breast tissue since the mass energy-absorption coefficients in water are larger. These differencesare almost the same for the two seed models. Looking atonly the clinical context, the differences due to the TG-43U1dose calculation approximations have consequences only forV200 �about 8.4%� and are mainly due to the tissue approxi-mation. In permanent breast seed implant, the evaluation of

-20.0

-15.0

-10.0

-5.0

0.0

5.0

10.0

15.0

20.0

-2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2

y (cm)

Relativedifference(%)

(b)

Relativedifference(%)

y (cm)

FIG. 3. �a� Comparison between absolute isodoses �cGy� determined with anstrength of the seed was 8.8 U and the irradiation time 24 h. ��b� and �c��Gafchromic film and MC simulation for a 6711 model seed as a function of raround the origin due to the presence of the seed.

the dose to skin is critical in predicting possible acute skin

Medical Physics, Vol. 36, No. 5, May 2009

toxicity. In addition to the interseed attenuation and the tissuecomposition effects, the TG43-U1 introduces an approxima-tion by considering the calculation area as a spherical watervolume, whereas the MC code takes the real geometry of thepatient into account. For the points close to the outer breastsurface, due to missing backscatter volume, the MC dosesare lower than the TG43-U1 doses. These differences mayhave no clinically impact. Pignol et al.32 showed that thiskind of treatment is well tolerated, and after more than 2years, there is no recurrence for the treated patients. Moreaccurate MC evaluation of the TG-43U1 approximations onthe skin toxicity will need to be carried out.

The differences for D90 and D50, which are more clini-cally significant for the permanent prostate seed implant, are

-20.0

-15.0

-10.0

-5.0

0.0

5.0

10.0

15.0

20.0

-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0-2.0

x (cm)

Relativedifference(%)

Relativedifference(%)

x (cm) (c)

Gafchromic film and MC simulation for a 6711 model seed. The air kermative differences of absolute dose between the measurements with an EBTtively the y axis �y=0 cm� and the x axis �x=0 cm�. No points are present

.0

EBTRelaespec

about 2% for the interseed attenuation effect and for the tis-

1855 Furstoss et al.: LDR seed dosimetry: Monte Carlo study 1855

sue composition effect, about 8.1% for the 6711 model, and10.6% for the InterSource125 model. The fact that the tissuecomposition impact is higher for the InterSource125 seedsthan the 6711 seeds is due to the difference in the spectrumbetween the two seeds and also the nonmonotonous variationin the mass energy-absorption coefficients with energy below100 keV.

Even if the mass energy attenuation coefficients for waterand tissue are significantly different at 30 keV, the dose met-ric differences between the TG43-U1 and MC calculations inthe studied case do not exceed 15%. DeMarco et al.34 madethe same observation for their prostate treatment where thedosimetric differences comparing homogeneous water versusheterogeneous CT simulation geometries are about 6%. Incomparison to the literature, Carrier et al.7 found for a26 cm3 prostate treatment with the same seed density as theone considered in this study �3.7 seeds /cm3� an interseedattenuation of 4.4% for D90 and 3.6% for D50. It is slightlyhigher than the interseed attenuation found with this breasttreatment. However, they noted that the interseed attenuationdecreased with the volume of the prostate, which confirmsour results since the PTV breast volume is 17 cm3. For theeffect of the tissue composition, Carrier et al.7 obtainedabout 4.4% for D90 and D50. This is lower than the differ-

(a)

20 60 80

10

20

30

40

50

60

70

90

80

40 100 120 140Position (mm)

-4.2-9.82.2

2.2

12.26.9-0.8

0.30.7-1.4-0.8

10.010.5

-5.6-8.1 -10.4

4.2

-5.9-4.4

-5.1 -2.1

-0.8 -1.2

5.0 10.7

0.5

-3.6

20 60 80

1.5

-1.7

Position(mm)

Position (mm)

4.2 2.2

2.2-3.6

0.3

-1.7-9.8 -4.2

-0.8 12.20.5

5.0 10.7 10.5 10.0

-5.6-5.9

10.41.5 -8.1

-4.4

0.7-1.4

-0.8

6.9-0.8 -1.2

-5.1

(b)

FIG. 4. �a� Acrylic breast phantom. The size is 15 cm length, 13 cm width,and 8 cm for the maximal height. �b� Percentage differences between abso-lute dose determined with TLD measurement and MC simulations forInterSource125 seeds. The seeds �crosses� are located in a slice 5 mm awayfrom the slice holding the TLDs.

ences found for the breast treatment, but it can be explained

Medical Physics, Vol. 36, No. 5, May 2009

by the fact that the differences between the mass energyattenuation coefficients of water and prostate are smaller thanthe differences between those of water and adipose or breast.Chibani et al.6 obtained about 3%–4% for the interseed at-tenuation for a 36.5 cm3 prostate treatment with2.3 seeds /cm3. Due to the smaller volume of the breast tu-mor, the interseed attenuation is lower for our studied treat-ment. They too noted that the interseed attenuation wassmaller for 125I seeds than for 103Pd seeds due to the higherphoton energy for the iodine. For the same reason, the tissuecomposition effect can be expected to be higher with 103Pdseeds. The differences between the FMC and the TG-43 dosemetrics should be higher for 103Pd seeds.

These differences can decrease or increase according tothe composition of the modeled breast tissue. In Sec. II C 5,we mentioned that the composition of the breast tissue candiffer from individual to individual and depends on the ageof the patient. The 50%–50% adipose-glandular tissue com-position represents an average breast tissue.27 The study ofthe influence of the breast tissue composition shows that,whereas for V85 and V100 the differences are very close to0%, for V200, D90, and D50 the relative differences betweenthe dose metrics calculated in water and the different breastcompositions vary from about �3% when the breast compo-sition is only glandular tissue to 13% when the breast com-position is only adipose tissue �Fig. 7�. These results confirmthe fact that the breast composition strongly influences thedose metrics. Even if clinically the impact of the breast com-position influences only V200, a technique to determine moreprecisely the location and the composition of the varioustissues on the CT images should be investigated in order toimprove the determination of the true dose received by thepatients. This study could be possibly based on dual-energyCT imaging.35,36

IV. CONCLUSION

MCNPX MC calculations for two LDR seed models forpermanent brachytherapy breast implants were validatedagainst measurements in a water phantom and an acrylicbreast phantom. The interseed attenuation study reveals thatthis effect is higher for the 6711 seed than for theInterSource125 seed due to the presence of the silver markerin the 6711 seed, which causes more attenuation. The dosecalculated along the transverse axis of the seeds in different

TABLE V. Comparison of dosimetric parameters for the PTV for a breastbrachytherapy treatment with InterSource125 seeds.

Interseedattenuation

effect

Tissuecomposition

effectComparison

TG43-U1 and FMC

�V85 �%� 0.05 0.03 0.05�V100 �%� 0.05 0.05 0.05�V200 �%� 1.5 6.8 7.9�D90 �%� 2.3 10.8 13.7�D90 �%� 2.0 10.3 12.4

tissues, acrylic, and water shows important differences be-

1856 Furstoss et al.: LDR seed dosimetry: Monte Carlo study 1856

(a) (b)

(c)

6711 seed

0.01

0.1

1

10

100

0 1 2 3 4 5 6 7

r (cm)

Dose/S K(.10-3h.m-2)

watermammary glandadiposebreastPMMA

InterSource125 seed

0.01

0.1

1

10

100

0 1 2 3 4 5 6 7

r (cm)

Dose/S K(.10-3h.m-2)

watermammary glandadiposebreastPMMA

0.8

0.9

1

1.1

1.2

1.3

1.4

1.5

0 1 2 3 4 5 6 7r (cm)

Ratio

(Dose/S K)_InterSource1

25/(Dose/SK)_6711 water

mammary glandadiposebreastPMMA

FIG. 5. Dose normalized by air kerma strength versus the transverse distance from a seed for different media for: �a� The InterSource125 seed and �b� the 6711125

seed �right panel�. �c� shows the ratio of the dose of InterSource seed over 6711 seed as a function of the transverse distance for different media.

0.7

0.8

0.9

1

1.1

1.2

3 3.5 4 4.5 5 5.5 6 6.5 7

Distance from center (cm)

Inte

rse

ed

eff

ec

t

InterSource125

6711

125

FIG. 6. Comparison of the interseed effect as a function of distance for seed125

models InterSource and 6711.

Medical Physics, Vol. 36, No. 5, May 2009

FIG. 7. Dose metric differences between the FMC and the MCW as a func-tion of the different compositions of breast taken into account in the FMC.V85 is not represented because the differences are equal to 0% whatever the

composition is.

1857 Furstoss et al.: LDR seed dosimetry: Monte Carlo study 1857

tween water, breast, and adipose tissue. The study of a per-manent breast iodine seed implant shows that the differences,for the dose metrics clinically significant, are minimal for V85

and V100 �about 0.05%� and about 8% for V200 between theFMC and the TG43-U1 calculations. All the differences havea positive sign, meaning that the TG-43U1 formalism over-estimates the dose. These discrepancies are dependent on thecomposition and the location of the breast tissue on the CTimages. To determine precisely the dose metrics, a techniqueshould be investigated in order to improve the conversion ofthe CT images into material maps for the breast tissue. Evenif these differences are acceptable from a clinical perspec-tive, especially since much larger variations are seen in thetreated volume between the prescribed dose and the highdose sleeve around the seeds, the development of a brachy-therapy treatment planning system taking the interseed at-tenuation and the tissue heterogeneity into account for LDRbrachytherapy will yield the real dose received by the treatedpatients allowing improvement in the dose-outcome relation-ship. Moreover, since the evaluation of the skin dose is criti-cal in predicting possible acute skin toxicity, more accurateevaluation using MC simulation is warranted.

ACKNOWLEDGMENTS

This work is supported by Grant No. 017133 of the Na-tional Cancer Institute of Canada �NCIC�. One of the authors�F.V.� was supported by the Fonds de Recherche en Santé duQuébec �FRSQ�. E.P. is a CIHR Strategic Training Fellow inthe Excellence in Radiation Research for the 21st CenturyProgram. R van Gils is thanked for construction of the phan-toms. This project was also made possible through the gen-erous support from the Canadian Breast Cancer Foundation-Ontario Chapter.

a�Present address: Departement de Radio-Oncologie, HopitalMaisonneuve-Rosemont, 5415 Boulevard de l’Assomption, Montreal,Quebec H1T 2M4, Canada.

b�Author to whom correspondence should be addressed. Electronic mail:frank.verhaegen@maastro.nl

1J. P. Pignol, B. Keller, E. Rakovitch, and R. Sankreacha, “First report ofa permanent breast seed implant �PBSI� as adjuvant radiation for earlystage breast cancer,” Int. J. Radiat. Oncol., Biol., Phys. 64, 176–181�2006�.

2B. Keller, R. Sankreacha, E. Rakovitch, P. O’Brien, and J. P. Pignol, “Apermanent breast seed implant as partial breast radiation therapy forearly-stage patients: A comparison of palladium-103 and iodine-125 iso-topes based on radiation safety considerations,” Int. J. Radiat. Oncol.,Biol., Phys. 62, 358–365 �2005�.

3R. Nath, L. L. Anderson, G. Luxton, K. A. Weaver, J. F. Williamson, andA. S. Meigooni, “Dosimetry of interstitial brachytherapy sources: Recom-mendations of the AAPM Radiation Therapy Committee Task Group No.43,” Med. Phys. 22, 209–234 �1995�.

4M. J. Rivard, B. M. Coursey, L. A. DeWerd, W. F. Hanson, M. SaifulHuq, G. S. Ibbott, M. G. Mitch, R. Nath, and J. F. Williamson, “Update ofAAPM Task Group No. 43 Report: A revised AAPM protocol for brachy-therapy dose calculations,” Med. Phys. 31, 633–674 �2004�.

5G. S. Burns and D. E. Raeside, “The accuracy of single-seed dose super-position for I-125 implants,” Med. Phys. 16, 627–631 �1989�.

6O. Chibani, J. F. Williamson, and D. Todor, “Dosimetric effects of seedanisotropy and interseed attenuation for 103Pd and 125I prostate implants,”Med. Phys. 32, 2557–2566 �2005�.

7J. F. Carrier, L. Beaulieu, F. Therriault-Proulx, and R. Roy, “Impact ofinterseed attenuation and tissue composition for permanent prostate im-

plants,” Med. Phys. 33, 595–604 �2006�.

Medical Physics, Vol. 36, No. 5, May 2009

8J. F. Carrier, M. D’Amours, F. Verhaegen, B. Reniers, A. G. Martin, E.Vigneault, and L. Beaulieu, “Postimplant dosimetry using a Monte Carlodose calculation engine: A new clinical standard,” Int. J. Radiat. Oncol.,Biol., Phys. 68, 1190–1198 �2007�.

9C. C. Ling, E. D. Yorke, I. J. Spiro, D. Kubiatowicz, and D. Bennett,“Physical dosimetry of 125I seeds of a new design for interstitial implant,”Int. J. Radiat. Oncol., Biol., Phys. 9, 1747–1752 �1983�.

10J. Dolan, Z. Li, and J. F. Williamson, “Monte Carlo and experimentaldosimetry of an 125I brachytherapy seed,” Med. Phys. 33, 4675–4685�2006�.

11A. S. Meigooni, M. M. Yoe-Sein, A. Y. Al-Otoom, and K. T. Sowards,“Determination of dosimetric characteristics of InterSource125 Iodinebrachytherapy source,” Appl. Radiat. Isot. 56, 589–599 �2002�.

12B. Reniers, S. Vynckier, and P. Scalliet, “Dosimetric study of a newInterSource125 iodine seed,” Med. Phys. 28, 2285–2288 �2001�.

13H. Afsharpour, M. D’Amours, B. Coté, J. F. Carrier, F. Verhaegen, and L.Beaulieu, “A Monte Carlo study on the effect of seed design on theinterseed attenuation in permanent prostate implants,” Med. Phys. 35,3671–3681 �2008�.

14S. Devic, J. Seuntjens, E. Sham, and E. B. Podgorsak, “Precise radiochro-mic film dosimetry using a flat-bed document scanner,” Med. Phys. 32,2245–2253 �2005�.

15S.-T. Chiu-Tsao, D. Medich, and J. Munro III, “The use of newGAFCHROMIC® EBT film 125I seed dosimetry in Solid Water® phan-tom,” Med. Phys. 35, 3787–3799 �2008�.

16A. S. Meigooni, J. A. Meli, and R. Nath, “A comparison of solid phan-toms with water for dosimetry of 125I brachytherapy sources,” Med. Phys.15, 695–701 �1988�.

17K. A. Weaver, “Response of LiF powder to 125I photons,” Med. Phys. 11,850–854 �1984�.

18K. A. Weaver, “Anisotropy functions of 125I and 192Ir seed sources,” Med.Phys. 25, 2271–2278 �1998�.

19MCNPX TM User’s Manual LA-CP-05-0369, edited by D. B. Pelowitz�Los Alamos National Laboratory, NM, 2005�.

20J. S. Hendricks et al., MCNPX, Version 2.5.e �Los Alamos Laboratory,NM, 2004�.

21J. F. Briesmeister, “MCNP - A general Monte Carlo N-particle transportcode, version 4C,” in Los Alamos National Laboratory, Report No. LA-13709-M, 2000 , pp. 2–56.

22D. E. Cullen and R. E. MacFarlane, Lawrence Livermore National Labo-ratory Report No. UCRL-JC-148180, 2002 �unpublished�.

23D. E. Cullen, J. H. Hubbell, and L. Kissel, “ENDLIB-97: LLNL librariesof atomic data, electron data and photon data in evaluated nuclear datalibrary �ENDL� type format,” Radiation Safety Information Computa-tional Center, Data Library DLC-179, Oak Ridge National Laboratory,1997.

24T. D. Bohm, P. M. DeLuca, and L. A. DeWerd, “Brachytherapy dosimetryof 125I and 103Pd sources using an updated cross section library for theMCNP Monte Carlo transport code,” Med. Phys. 30, 701–711 �2003�.

25J. R. de Laeter, “Atomic weights of the elements 1987,” Pure Appl.Chem. 60, 841–854 �1988�.

26“Tissue substitutes in radiation dosimetry and measurement,” ICRU Re-port No. 44, 1989.

27G. R. Hammerstein, D. W. Miller, D. R. White, M. E. Masterson, H. Q.Woodard, and J. S. Laughlin, “Absorbed radiation dose in mammogra-phy,” Radiology 130, 485–491 �1979�.

28R. E. P. Taylor, G. Yegin, and D. W. O. Rogers, “Benchmarking Brachy-Dose: Voxel based EGSnrc Monte Carlo calculations of TG-43 dosimetryparameters,” Med. Phys. 34, 445–457 �2007�.

29A. S. Meigooni, J. A. Meli, and R. Nath, “Interseed effects on dose for125I brachytherapy implants,” Med. Phys. 19, 385–390 �1992�.

30E. Poon, Y. Le, J. F. Williamson, and F. Verhaegen, “BrachyGUI: Anadjunct to an accelerated Monte Carlo photon transport code for patient-specific brachytherapy dose calculations and analysis,” J. Phys.: Conf.Ser. 102, 0120018 �2008�.

31J.P. Pignol, E. Rakovitch, B. Keller, R. Sankreacha, and C. Chartier,“Tolerance and acceptance of partial breast irradiation using permanentbreast 103Pd seed implant �PBSI�,” Int. J. Radiat. Oncol., Biol., Phys.73�5�, 1482–1488 �2009�.

32J. F. Briesmeister, “MCNP—A general Monte Carlo N-particle transportcode, version 4C,” in Los Alamos National Laboratory, Report No. LA-

13709-M, 2000, pp. 2–183.

1858 Furstoss et al.: LDR seed dosimetry: Monte Carlo study 1858

33S. Eklund, A. Thilander, W. Leitz, and S. Mattsson, “The impact of ana-tomic variations on absorbed radiation doses in mammography,” Radiat.Prot. Dosim. 49, 167–170 �1993�.

34J. J. DeMarco, J. B. Smathers, C. M. Burnison, Q. K. Ncube, and T. D.Solberg, “CT-based dosimetry calculations for 125I prostate implants,” Int.

J. Radiat. Oncol., Biol., Phys. 45, 1347–1353 �1999�.

Medical Physics, Vol. 36, No. 5, May 2009

35M. Bazalova, J. F. Carrier, L. Beaulieu, and F. Verhaegen, “Tissue seg-mentation in Monte Carlo treatment planning: A simulation study usingdual-energy CT images,” Radiother. Oncol. 86, 93–98 �2008�.

36M. Bazalova, J. F. Carrier, L. Beaulieu, and F. Verhaegen, “Dual-energyCT-based material extraction for tissue segmentation in Monte Carlo dose

calculations,” Phys. Med. Biol. 53, 2439–2456 �2008�.