Reference photon dosimetry data and reference phase space data for the 6 MV photonbeam from Varian Clinac 2100 series linear acceleratorsSang Hyun Cho, Oleg N. Vassiliev, Seungsoo Lee, H. Helen Liu, Geoffrey S. Ibbott, and Radhe Mohan

Citation: Medical Physics 32, 137 (2005); doi: 10.1118/1.1829172 View online: http://dx.doi.org/10.1118/1.1829172 View Table of Contents: http://scitation.aip.org/content/aapm/journal/medphys/32/1?ver=pdfcov Published by the American Association of Physicists in Medicine Articles you may be interested in Peripheral dose measurements for 6 and 18 MV photon beams on a linear accelerator with multileaf collimator Med. Phys. 35, 4396 (2008); 10.1118/1.2977533 Photon spectral characteristics of dissimilar 6 MV linear accelerators Med. Phys. 35, 1698 (2008); 10.1118/1.2900001 On the discrepancies between Monte Carlo dose calculations and measurements for the 18 MV Varian photonbeam Med. Phys. 34, 1206 (2007); 10.1118/1.2712414 A Monte Carlo model for calculating out-of-field dose from a Varian 6 MV beam Med. Phys. 33, 4405 (2006); 10.1118/1.2360013 Dose enhancement close to platinum implants for the 4, 6, and 10 MV stereotactic radiosurgery Med. Phys. 31, 2787 (2004); 10.1118/1.1797531

Reference photon dosimetry data and reference phase space datafor the 6 MV photon beam from Varian Clinac 2100 series linearaccelerators

Sang Hyun Cho,a) Oleg N. Vassiliev, Seungsoo Lee, H. Helen Liu,Geoffrey S. Ibbott, and Radhe MohanDepartment of Radiation Physics, The University of Texas M. D. Anderson Cancer Center,1515 Holcombe Boulevard, Unit 547, Houston, Texas 77030

(Received 5 January 2004; revised 15 October 2004; accepted for publication 15 October 2004;published 20 December 2004)

The current study presents thereference photon dosimetry data(RPDD) andreference phase spacedata (RPSD) for the 6 MV photon beam from Varian 2100 series linear accelerators. The RPDDprovide the basic photon dosimetry data, typically collected during the initial commissioning of anew linear accelerator, including output factors, depth dose data, and beam profile data in air and inwater. The RPSD provide the full phase space information, such as position, direction, and energyfor each particle generated inside the head of any particular linear accelerator in question. Thedosimetric characteristics of the 6 MV photon beam from the majority of the aforementionedaccelerators, which are unaltered from the manufacturer’s original specifications, can be fullydescribed with these two data sets within a clinically acceptable uncertaintys,±2%d. The currentstudy also presents a detailed procedure to establish the RPDD and RPSD using measured data andMonte Carlo calculations. The RPDD were constructed by compiling our own measured data andthe average data based on the analysis of more than 50 sets of measured data from the RadiologicalPhysics Center(RPC) and 10 sets of clinical dosimetry data obtained from 10 different institutionsparticipating in the RPC’s quality assurance monitoring program. All the measured data from theRPC and the RPC-monitored institutions were found to be within a statistically tight range(i.e.,1s<1% or less) for each dosimetric quantity. The manufacturer’s standard data, except for in-airoff-axis factors that are available only from the current study, were compared with the RPDD,showing that the manufacturer’s standard data could also be used as the RPDD for the photon beamstudied in this study. The RPSD were obtained from Monte Carlo calculations using theBEAMnrc/

DOSXYZnrc code system with 6.2 MeV(a spread of 3% full width at half maximum) and 1.0 mmfull width at half maximum as the values of the energy and radial spread of a Gaussian electronpencil beam incident on the target, respectively. The RPSD were capable of generating Monte Carlodata that agreed with the RPDD within the acceptance criteria adopted in the current study(e.g., 1%or 1 mm for depth dose). A complete set of the RPDD and RPSD from the current study is availablefrom the RPC website(http://rpc.mdanderson.org) or via mass storage media such as DVD orCD-ROM upon request. ©2005 American Association of Physicists in Medicine.[DOI: 10.1118/1.1829172]

Key words: reference photon dosimetry data, reference phase space data, 6 MV photon beam,Monte Carlo calculations

I. INTRODUCTIONr ras os th

-ac-ma

t arac-tsamfor

n asimel

blishrs ofd do-r. Dueimetion’s

thesim-ulti-waydata

d the

The modern computer-controlled linear accelerators fodiation therapy are becoming more reproducible, in termtheir structures and dosimetric characteristics, as long amake and model of the machines are identical.1,2 Consequently, dosimetry data commonly applicable for linearcelerators of the same make, model, and nominal energyexist. The availability of such data will significantly benefiwide range of radiation oncology and medical physics ptices. As discussed elsewhere,2 some of the possible benefiand applications include the simplification of photon becommissioning, the creation of generic beam modelsmodel-based and/or Monte Carlo-based dose calculatiogorithms, and the generation of the standard photon dotry data for the quality assurance(QA) of nationwide clinica

trials.

137 Med. Phys. 32 (1), January 2005 0094-2405/2005/32 (1

-fe

y

l--

Although no serious attempt has been made to estacommonly applicable dosimetry data, some manufacturelinear accelerators have established their own standarsimetry data for their recent models,3 which may be used focross-checking of each user’s measured dosimetry datato lack of investigation, however, it is uncertain at this thow closely these standard data match each institumeasured dosimetry data. Clearly, any investigation oncommon applicability of the manufacturers’ standard doetry data is not a trivial task, because it requires a minstitutional effort and needs to be coordinated in such athat the consistency in measurement techniques andanalysis could be maintained.

Over the last few years, some investigators adopte

Monte Carlo method to characterize photon beams from

137)/137/12/$22.50 © 2005 Am. Assoc. Phys. Med.

, anctria-

cienwe

s anteratelera

tar-stabodethada

ntereendat

ap-rica

h nbotthar, anetryvelcepoton

lesthastesar

ficu, re

. Ththe

todicaas

entMVlin-

ffecdy t

f aton

gCarl

phase

VD

datacel-can

intynt toppli-the

y al-etryake,acht thatition,e the

etryratordata,Pho-edgee ofilardata.ted

ht beable.arloetry

colli-

sed

th

ee ofene in

ize

138 Cho et al. : Reference photon dosimetry data for 6 MV photon beam 138

medical linear accelerators of the same make, modelnominal energy.3–10 However, it seems difficult to extrageneric beam models from their work, due to some vations in the results among the studies and, often, insuffiamounts of data presented. Some of these variationspartly due to the measured data(e.g., depth dose data) usedduring the Monte Carlo modeling of linear accelerators. Aresult, it is unclear at this time whether or not the MoCarlo beam models studied so far can be used to genecommonly applicable dosimetry data set for linear accetors of the same make, model, and nominal energy.

In the meantime, Cho and Ibbott2 showed in their recenpreliminary study that the similarity in the dosimetric chacteristics between machines could be quantified by elishing the standard dosimetry data for each make, mand nominal energy. This preliminary study suggestedthe dosimetry data needed to establish such standardmight be obtained from the Radiological Physics Ce(RPC) at M. D. Anderson Cancer Center, which has baccumulating a consistently measured set of dosimetrythrough its on-site dosimetry audit program. A similarproach was also taken in the past, especially by the AmeAssociation of Physicists in Medicine(AAPM) Task Group46 to establish some standard dosimetry data, althougtask group report has ever been published. Cho and Ibstudy further demonstrated that the overall dosimetric cacteristics of the machines with the same make, modelnominal energy could be described by a single dosimdata set within a clinically acceptable tolerance les,±2%d. The same study also pointed out that the conof common dosimetry data could be valid, at least for phbeams whose dosimetric characteristics are generallysensitive to minute changes in the machine structurethose for electron beams. In addition, the study suggethat the use of the Monte Carlo method would be necesto complement measurements, because it might be difto establish such commonly applicable dosimetry dataferred to as thereference photon dosimetry data(RPDD),entirely based on the RPC-measured dosimetry datadefinition of the RPDD is presented in Sec. II, along withdefinition of thereference phase space data(RPSD).

A major project is currently being planned at the RPCestablish the RPDD and RPSD for the most common melinear accelerators. The current study was conductedpilot study for this project. The main goal of the currstudy was to establish the RPDD and RPSD for the 6photon beam from the most popular dual photon energyear accelerator model, the Varian Clinac 2100 series(includ-ing all variations of this model such as C, C/D, and 21EX). Asecondary goal was to demonstrate the feasibility and etiveness of the approaches adopted in the current stuachieve the main goal. The basic photon dosimetry data(i.e.,RPDD), typically collected during the commissioning onew accelerator, were determined for the 6 MV phobeam. Also, the phase space files(i.e., RPSD) correspondinto these dosimetry data were generated by the Monte

calculations. The entire study was conducted following the

Medical Physics, Vol. 32, No. 1, January 2005

d

tre

a-

-l,tta

a

n

o’s-d

t

sndylt-

e

la

-o

o

approaches described in Sec. II. The detailed data andspace files are available from the RPC website(http://rpc.mdanderson.org) or via mass storage media such as Dor CD-ROM upon request.

II. METHODS AND MATERIALS

A. Definitions

1. Definitions of the RPDD and the RPSD

The RPDD are defined as a set of photon dosimetryby which the dosimetric characteristics of any linear acerator with the same make, model, and nominal energybe described within a clinically acceptable uncertas,±2%d. These data are for reference only and not meareplace the clinical data. Also, the RPDD may not be acable for those machines significantly modified frommanufacturers’ original specifications. Therefore, we maternatively define the RPDD as a set of photon dosimdata for a generic linear accelerator of a particular mmodel, and nominal energy. Accordingly, the RPSD for egeneric linear accelerator can be defined as a data seprovides the full phase space information, such as posdirection, and energy for each particle generated insidhead of a linear accelerator.

The RPDD may include typical basic photon dosimdata closely related with the structures of a linear accelehead. Therefore, the RPDD would include depth doseoutput factor, and beam profile data in water and in air.ton dosimetry data related with accessories such as wfactors are to be excluded from the RPDD for the saksimplicity, even if one might be able to observe a simstandard behavior as seen for other basic dosimetrySimilarly, no data related with a physical wedge mouninside the linear accelerator head(e.g., universal wedge) willbe considered for the RPDD. However, these data migpresented as some ancillary data to the RPDD, if availThe RPSD can be used as input data for the Monte Csimulation to generate the RPDD and any other dosimdata related with accessory devices such as multileafmators(MLC), dynamic wedges, etc.

2. Definitions of variables and dosimetricquantities

The definitions of variables and dosimetric quantities uin this study are provided in the following.

• Fractional depth dose(FDD) and percentage depdose(PDD):FDD5(rdg at depth for FS along CAX/rdg atdmax

for FS along CAX) PDD5FDD3100 where rdg: thmeter reading of an ion chamber; FS: the field siza given beam; CAX: the central axis of a givbeam;dmax: the depth at which the absorbed dosmedium is maximum for a given beam.

• Off-axis factor(OAF):OAF=rdg at off-axis distance/rdg at CAXNote OAF is usually defined for the largest field s

2

available(e.g., 40340 cm for photon beams), at a

thered-to-te

tom0s

tonuredPDDsisipatch

urre

stafDD,odetican ainstruu-100alysdoAF

sticawerea

es oure

t theined

potitu-

axisorredif

PCnd

read%uslyPC’sPC-

ostlye. Aa, al-C’s

corre-

edthesedevia-sid-

te setac

in theypicalnsivend

eamscan-

M.etricSecs.-

areddarddata

to beater

stan-

erecer-vali-

thePDDare

ere

la-100

the

139 Cho et al. : Reference photon dosimetry data for 6 MV photon beam 139

depth in a phantom or at the isocenter in air. Allin-air OAFs reported in this study were measuwith farmer-type ionization chambers, at a sourceaxis distance(SAD) of 100 cm, with an appropriaplastic buildup cap for the 6 MV photon beam.2

• Output factor(OF):OF5[(rdg atdref for FS)/(FDD at dref for FS)]/[(rdgat dref for RFS)/(FDD atdref for RFS)] wheredref: thereference depth for measurements in a phanRFS: the reference field size(RFS is normally a 1310 cm2 field for photon beams.); other variableare as defined earlier.

B. Creation of the RPDD by compilationof measured data

A set of the RPDD was created for the 6 MV phobeam from Varian Clinac 2100 series by compiling measdata from various sources. Basically, each entry in the Rwas derived from the average data based on the analymeasured data from the RPC and the institutions particing in the RPC’s QA monitoring program. When no suaverage data were available, measured data from the cstudy were entered into the RPDD.

1. The RPC-measured data

Typical photon dosimetry data measured by the RPCphysicists during an on-site dosimetry audit are OF, Fin-air OAF, and wedge factors. The RPC uses several mof popular Farmer-type cylindrical chambers and an idenmodel of electrometers for measurements in water and iDetails about the RPC’s measurement techniques and imentation can be found elsewhere.2 The RPC has accumlated a sufficient number of data on the Varian Clinac 2series linear accelerators and, therefore, a statistical anon the data could reveal some similarity among photonsimetry data. The RPC-measured FDD, OF, and in-air Odata were extracted from the RPC database for a statianalysis. At least 50 sets of the RPC-measured dataused to obtain the averages and standard deviations fordosimetric characteristic at selected depth and field sizoff-axis distances reported in this study. The RPC-measdata for each dosimetric quantity were sorted so thamaximum and minimum values could also be determduring this procedure.

2. Clinical data from the institutions participatingin the RPC’s monitoring program

The RPC-measured data were obtained from schecking of clinical data from the RPC-monitored insttions and, therefore, the final results(i.e., averages) wereavailable only for selected depths, field sizes, and off-distances. As a complementary way to obtain the data csponding to a wider range of field sizes and depths, 10ferent sets of the data were extracted from the Rmonitored institutions’ clinical data for the field sizes a

depths unavailable from the RPC’s measurements. The inst

Medical Physics, Vol. 32, No. 1, January 2005

;

of-

nt

f

lslr.-

is-

lechrd

-

---

tutions’ clinical data showed an acceptable statistical sp(i.e., 1s<1% or less and max/min ratio of less than 4).Each institution’s clinical data selected here were previocompared at selected depths and field sizes during the Rspot-checking, showing agreement between the Rmeasured and the RPC-monitored institution’s data, mbetter than ±1% and no worse than ±2% in any cassimilar agreement was assumed for the extracted datthough these data were not directly verified by the RPmeasurements. In general, the basic dosimetry datasponding to small(e.g., less than 636 cm2) and large fieldsizes(e.g., larger than 30330 cm2) needed to be determinby this approach. Similar to the RPC-measured data,data were used to determine the averages, standardtions, and min/max for each dosimetric characteristic conered in this step.

3. Dosimetry data from in-house measurementsand manufacturer

More dosimetry data were needed to create a compleof the RPDD for the 6 MV photon beam from Varian Clin2100 series because the average data determinedabove-noted steps were not as comprehensive as tclinical data. Therefore, two more data sets, comprehedepth dose(DD) data including the dose buildup region ain-water beam profiles, were extracted from our own bcommissioning data. These data were measured in aning water tank with a Wellhofer IC-10 chamber(6 mm in-ner diameter) on a Clinac 2100 C/D linear accelerator atD. Anderson Cancer Center that exhibits the same dosimcharacteristics as shown in the average data obtained inII B 1 and II B 2 (i.e., Tables I–III) within a typical measurement uncertaintys,1%d.

The manufacturer’s standard data were also compwith our own measured data. The manufacturer’s standata provide a comprehensive set of the basic dosimetry(i.e., DD data, OF, and in-water beam profiles) for a widerange of field sizes and depths. The data are reportedmeasured for the Clinac 21EX model in a scanning wtank using a IC-15 ionization chamber(6 mm innerdiameter).3 The agreement between the manufacturer’sdard data and our own measured data was within 1%(or1 mm where applicable). Therefore, these two data sets wconsidered to be identical within the measurement untainty and no separate comparison was made during thedation of the Monte Carlo beam model. Accordingly,manufacturer’s standard data may substitute for the Rdetermined in this study, except for the in-air OAFs thatavailable only from the current study.

C. Creation of the RPSD by Monte Carlo calculations

The Monte Carlo calculations to obtain the RPSD wconducted by the procedure shown in Fig. 1. TheBEAMnrc/

DOSXYZnrc code11–13 was used for the Monte Carlo calcutions along with a model for the head of Varian Clinac 2series derived from detailed specifications provided by

i-manufacturer. The following structures were modeled: target,

perctednte

e daeredpubctr, noudyutpu

re

sultsters

-thetargtheTh

tthe

-air

-

pac

sizings.ter-conho-

thewa-le it

laradialat 5,inty

ncecen-

each

d in-s at

s-ngif aed at

ntilund.hing

hese

redcal

file

of

le toadia-

nu-

ether

ltedthe

etryoreir ofthefieldn-espe-

SD.

140 Cho et al. : Reference photon dosimetry data for 6 MV photon beam 140

primary collimator, flattening filter, monitor chamber, up(Y) and lower(X) jaws. The phase space data were colleat a plane located right below the lower jaws. The MoCarlo output factors calculated using these phase spacdid not properly take into account the effect of backscattphotons from the jaws to the monitor chamber. Sincelished literature14,15 provide correction factors for this effefor the linear accelerator model studied here, howevefurther Monte Carlo simulation was performed in this stto derive separate correction factors. The calculated ofactors from the current study were corrected using thesults from Liuet al.14

The matching between the RPDD and Monte Carlo rewas accomplished by varying only two transport parameof the BEAMnrc/DOSXYZnrc code: the energysEed and the radial spreadsRed of a Gaussian electron beam incident ontarget. Other parameters, such as the densities of theand flattening filter, were not allowed to vary, assumingmanufacturer’s material specifications were accurate.manufacturer’s nominal value forEe is 6.0 MeV with aGaussian spread of 3% full width at half maximum(FWHM)and the nominalRe value is 1 mm FWHM.3,8 The adjustmenof Ee and Re values was guided by an observation thatdepth dose is strongly dependent on the choice ofEe valuewhereas so-called “horn” of in-water beam profile and inOAF are dependent on bothEe and Re values.3,6,8 For in-stance, the size of the horn increases as theRe value decreases. The same effect can also be seen as theEe valuedecreases.

During the data matching process, particle phase sdata were generated using theBEAMnrc code for a 434 cm2 field, a reference field(i.e., 10310 cm2), and thelargest field available(i.e., 40340 cm2). In addition, duringthe generation of phase space data for the largest fieldin-air OAFs at 100 cm SAD were also calculated followthe approach described by Sheikh-Bagheri and Roger8 Indetail, in-air OAFs were approximated as the ratios of wakerma-weighted photon fluence between the central andcentric radial scoring bins. The water-kerma-weighted p

FIG. 1. Flow chart for the Monte Carlo simulation to generate the RP

ton fluence was calculated by weighting each photon

Medical Physics, Vol. 32, No. 1, January 2005

ta

-

t-

et

e

e

e,

-

reaching the scoring plane located at 100 cm SAD byproduct of its energy, the mass-absorption coefficient forter at that energy, and one over the cosine of the angmakes with thez axis.8 The central scoring bin was a circuregion with a radius of 2.5 cm and the subsequent rscoring bins were 2.5-cm-wide annular regions located10, and 15 cm from the center. The statistical uncertas1s d for the calculated in-air OAF at each off-axis distawas less than 0.5%. The uncertainty associated with thetral bin was added in quadrature to that associated withsubsequent bin during the normalization of the data.

After phase space data were obtained, DD data anwater beam profiles for the above-mentioned field size100 cm source-to-surface distance(SSD) were calculated uing the DOSXYZnrc code. All the calculated data includiin-air OAFs were compared with the RPDD to checkmatch was found. In-water beam profiles were comparleast atdmax and several other clinically useful depths(e.g.,3, 5, 10 cm, etc.). The whole procedure was repeated ua match between Monte Carlo data and RPDD was foThe following agreement levels were used as the matccriteria.

(a) Depth dose

• Dose fall-off region (between dmax and 30 cmdepth): within 1% of measured local dose for tmajority s,95%d of the data points and no worthan 1.5% in any case

• Dose buildup region: within 1 mm of measudepth dose curve or within 3% of measured lodose

(b) In-water beam profiles:

• Plateau region: within 2% of measured profile• Penumbra region: within 2 mm of measured pro

(c) Output factor and in-air off-axis factor: within 1%measured value

The above criteria are better than or at least comparabthose used for validating beam models of model-based rtion treatment planning systems.16,17

TheBEAMnrc calculations were first started with the mafacturer’s nominalEe andRe values, followed byDOSXYZnrc

calculations. Then, theBEAMnrc/DOSXYZnrc calculations werrepeated for a number of times using the values from ogroups’ work such as Sheikh-Bagheri and Rogers8 and Kealet al.3 The results from these initial calculations indicathat an optimum pair of values could be found aroundvalues reported by Keallet al. (i.e., 6.2 MeV and 1.3 mmFWHM), considering the matching criteria and dosimdata used in this study. A further attempt to find a moptimum pair of values was made and resulted in a pavalues, 6.2 MeV and 1.0 mm FWHM, that producedmatch within the criteria used in this study across thesizes ranging from 434 to 40340 cm2. These values geerally produced a better agreement for beam profiles,

2

cially for a 40340 cm field, than the values reported by

om

pro--ons 0.7lung

off.ejecet

es fothe

red.dose o

teFo

has

manpt-

in ththe

eterandderaace

0sta-noForhanmbrme

fopro

orebe

eamr.PC

es athantion

eringarge

welldatathan

may

and

m ad theThethenple.

a canaarlo

roms per-d haven

741

ris was

RPC-than

s wastatandard

141 Cho et al. : Reference photon dosimetry data for 6 MV photon beam 141

Keall et al. and provide some similarity to the values frmanufacturer and published literature(see Table IV). All theRPSD in this study were generated using these values.

The Monte Carlo calculations were done on a dualcessor machine equipped with two 2.4 GHz Xeon® processors running Redhat® Linux 8.0. The electron and photcutoff energies, AE/ECUT and AP/PCUT, were chosen aand 0.01 MeV, respectively. The selective bremsstrahsplitting parameters wereNmin=20, Nmax=200,Rf =50 (for a40340 cm2 field), and Russian Roulette was switchedThe photon forcing option was not used and the range rtion was turned on with ESAVE=0.7 MeV for the targregion and ESAVEIGLOBAL=1.0 MeV for the rest of thlinear accelerator head. The definitions and descriptionthe BEAMnrc/DOSXYZnrc parameters are omitted here forsake of brevity and can be found elsewhere.11–13

The size of scoring voxels during theDOSXYZnrc simula-tion varied between 1.031.030.2 cm3 and 1.031.031.0 cm3, depending on the spatial resolution requiUsually, a smaller voxel size was chosen for the depthbuildup region and the beam penumbra region. The sizthe water phantom for theDOSXYZnrc simulation was 50350350 cm3. The total number of histories for the MonCarlo calculations varied depending on each situation.example, the number of histories for generating a pspace file and DD data for a 10310 cm2 field were on theorder of 108 and 109, respectively. During theDOSXYZnrc

simulation, the phase space sources were recycledtimes(up to eight times) for the most cases to obtain acceable statistical uncertainty. According to Walterset al.,18 therecycling of phase space sources is accurately reflecteduncertainty estimation of photon beam simulation. Whenphase space source was restarted more than once, theDOSX-

YZnrc simulation was repeated after adjusting a paramNRCYCL, by taking into account the number of missingrejected particles during the previous run, to avoid an unestimation of the uncertainty.13,18The typical CPU time on2.4 GHz Xeon® processor for the generation of phase spfiles was about 2 ms per history, whereas it took about 7msper history on average for the simulation in water. Thetistical uncertaintys1sd was mostly less than 0.5% andworse than 0.7% for the CAX depth dose simulation.in-water beam profile simulation, it was generally less t1% except for a few data points outside the beam penuregion. These values were better than typical measureuncertainties and approximately half of the values usedthe acceptance criteria adopted during the data matchingcess(e.g., 1% or 2%).

III. RESULTS

A. In-air off-axis factor

Although the beam profiles measured in water are mclinically relevant, in-air beam profiles are also known touseful for deriving the characteristics of electron pencil bincident on the target in the head of a linear accelerato8 InTable I, the reference data from the analysis of the R

measured in-air OAF are presented. As shown in this table

Medical Physics, Vol. 32, No. 1, January 2005

-

r

ef

re

y

e

,

-

antr-

-

the difference between the maximum and minimum valueach off-axis point was up to about 5%, slightly more tthat for FDD and OF data. The largest standard devias1sd associated with the sample was about 1%. Consida very small standard deviation associated with a fairly lamount of data points(i.e., 216 at each off-axis point), anoverall statistical tightness in the sample is deemeddemonstrated. Table I also shows that the Monte Carloare in agreement with the RPC-measured data better1%. The results presented in Table I are in-air OAF andnot be identical to in-water OAF(or off axis ratio), primarilydue to the difference in scatter condition between waterair.

B. Output factor

The reference photon output factors determined frostatistical analysis of measured data from the RPC anRPC-monitored institutions are presented in Table II.maximum and minimum values vary within 4% for allfield sizes. The standard deviations1sd was always less tha1%, showing an overall statistical tightness in the samThe statistical characteristics in the RPC-measured datalso be seen in a scatter plot(Fig. 2). Figure 3 presentscomparison between the data from Table II and Monte C

TABLE I. The RPC-measured in-air OAF for the 6 MV photon beam fVarian Clinac 2100 series accelerators. The statistical analysis waformed with 216 RPC-measured data sets. The RPC data presentebeen chosen as the RPDD in the current study. The standard deviatio(s.d.)is 1 s associated with the sample. The data from Monte Carlo(MC) calcu-lations are also presented. The statistical uncertaintys1sd of each MC-calculated OAF is less than 0.5%.

Off-axisdistance(cm)

OAF(RPC)

s.d.(RPC)

Max(RPC)

Min(RPC)

OAF(MC) MC/RPC

5 1.030 0.0070 1.049 1.001 1.027 0.9910 1.042 0.0095 1.061 1.015 1.036 0.9915 1.055 0.0114 1.078 1.029 1.056 1.00

TABLE II. The 6 MV photon beam output factor atdmax, 100 cm SSD, foVarian Clinac 2100 series linear accelerators. The statistical analysperformed with more than 150 RPC-measured data sets and 10monitored institutions’ clinical data sets. Outliers that deviated more±4s were discarded during the analysis. The number of such outliertwo and four for 20320 cm2 and 30330 cm2 fields, respectively. The dapresented have been chosen as the RPDD in the current study. The sdeviation(s.d.) is 1s associated with the sample.

Field sizescm2dOutputfactor s.d. Max Min

434a 0.923 0.0074 0.937 0.913636 0.957 0.0046 0.971 0.945

10310 1.000 ¯ ¯ ¯

15315 1.032 0.0042 1.046 1.02020320 1.053 0.0051 1.066 1.03930330 1.080 0.0063 1.100 1.06140340a 1.099 0.0101 1.108 1.075

a

,Based on clinical data from the institutions monitored by the RPC.

ontee re

andard

thetoredn isdoses. Inn forose

t fact

cC-corrected”

142 Cho et al. : Reference photon dosimetry data for 6 MV photon beam 142

calculations, showing agreement within 1% when the MCarlo calculated output factors were corrected using thsults from Liuet al.14

C. Depth dose

The reference photon PDD data at selected depthsfield sizes are presented in Table III along with the stan

FIG. 2. A scatter plot of the RPC-measured 6 MV photon outpu

FIG. 3. Comparison between measured and Monte Carlo(MC) calculated o2100 series linear accelerators. “RPDD/RPC data” means that the pre

denotes MC data corrected by applying the results from Ref. 14. The erro

Medical Physics, Vol. 32, No. 1, January 2005

-

d

deviations associated with the sample. Similar to OF,data were obtained from the RPC and the RPC-moniinstitutions. As seen in Table III, the standard deviatioless than about 1%, suggesting that the 6 MV depth(DD) characteristic changes very little between machineFig. 4, a scatter plot of the RPC-measured data is showa 636 cm2 field at selected depths. In Fig. 5, the depth d

or at 100 cm SSD,dmax for Varian Clinac 2100 series linear accelerators.

t factors for the 6 MV photon beam atdmax, 100 cm SSD, for Varian Clinated RPC data have been chosen as the RPDD in the current study. “M

utpusen

r bars represent 1s associated with the data.

thents

mes

fol-

ancr

Drgyafor

le, arentthain th

dosen thisresult,for

tonsure-s seen

datara

s for.

0.1350.3120.4700.498

t 50 sets

143 Cho et al. : Reference photon dosimetry data for 6 MV photon beam 143

data from Monte Carlo calculations are compared withdata from Table III and ionization chamber measuremeThe data are normalized at 10 cm depth for a 10310 cm2

field. Also, the measured DD data are shifted, by 0.6 tithe radius of the ionization cavity(i.e., 0.6rcav<2 mm for anIC-10 chamber), to the effective point of measurements,lowing the recommendation of the TG-51 report.19 Figure 6presents a detailed comparison between Monte Carlomeasured data, showing agreement within the matchingteria for DD data. Similar to a well-known problem in Dmatching over the dose buildup region for higher ene(e.g., 18 MV) and large field sizes,3,6,20 the agreement for40340 cm2 field was relatively poor as compared to thatsmaller field sizes(e.g., 434 cm2, 10310 cm2, etc.). How-ever, agreement within 1 mm appears to be achievableast for the 6 MV photon beam, according to the curresults. Note that no comparison at the depths less0.5 cm was made because the ionization chamber used

TABLE III. Reference photon dosimetry data: PDD at selected depths

Physical/effective

depth(cm)

434cm2 s.d.

636cm2 s.d.

10310cm2

5.0/4.8 84.39 0.213 85.78 0.699 86.9010.0/9.8 62.55 0.227 64.63 0.547 67.2215.0/14.8 46.22 0.297 48.24 0.515 51.1820.0/19.8 34.31 0.296 36.05 0.463 38.80

aThe data for 434 cm2, 30330 cm2, and 40340 cm2 are based on the 10of the RPC-measured data. The standard deviation(s.d.) is 1s associatechambers that the RPC uses, is applied to the physical depth to obta

FIG. 4. A scatter plot of the RPC-measured 6 MV photon depth dose d

Medical Physics, Vol. 32, No. 1, January 2005

.

di-

t

ne

current study might not be suitable for accurate depthmeasurements over such a region. Further discussion oissue is beyond the scope of the current study and, as athe RPDD from the current study do not include DD datathe depth less than 0.5 cm.

D. Beam profiles in water

In Figs. 7 and 8, in-water profiles for the 6 MV phobeam obtained from Monte Carlo calculations and meaments are compared at various depths and field sizes. Ain these figures, Monte Carlo data agreed with measuredwithin 2% in a flat region or within 2 mm in the penumbregion.

E. Optimum values for Ee and Re

The current study determined the two key parameterthe BEAMnrc simulation,Ee and Re, as shown in Table IV

e 6 MV photon beam from Varian Clinac 2100 series.a

20320cm2 s.d.

30330cm2 s.d.

40340cm2 s.d.

473 87.96 0.553 88.33 0.236 88.64.411 70.15 0.428 71.55 0.232 72.28.419 55.06 0.419 56.99 0.463 57.94.409 42.82 0.377 45.24 0.232 46.25

rent institutions’ clinical data. The rest of the data are based on at leasth the sample. About 2 mm shift, approximately 0.6rcav for typical farmere effective depth.

2

for th

s.d.

0.000

diffed wiin th

ata for a 636 cm field from Varian Clinac 2100 series linear accelerators.

thiseiouses

ationasod-the

ionizationto

ata” points.

144 Cho et al. : Reference photon dosimetry data for 6 MV photon beam 144

The results from published literature are also shown intable for comparison. The value ofEe determined from thcurrent study is the same as that found by two prevstudies,3,6 while the value ofRe for a Gaussian beam vari

FIG. 5. The 6 MV photon depth dose curves for various field sizes atchamber-measured DD are shifted to the effective point of measuremethe size of the symbols. The size of the symbol for the “RPDD/RPC d

FIG. 6. Difference between calculated and measured local doses. The

Medical Physics, Vol. 32, No. 1, January 2005

between 1 and 2 mm among the listed studies. The variin Ee and Re could be due to a number of factors suchdifference in measured data used for the Monte Carlo meling, possible difference in the computational model of

cm SSD. “MC” denotes the data from Monte Carlo calculations. Theby 0.6cav s,2 mmd. The statistical uncertaintys1sd of MC data are comparableis exaggerated for a better visibility, compared to that for other data

100ntsr

percent differences are given asfsMonte Carlo-measuredd / smeasureddg3100.

longn the

145 Cho et al. : Reference photon dosimetry data for 6 MV photon beam 145

FIG. 7. The 6 MV photon beam profiles in water for a 434 cm2 field at 3.0 cm depth and 10310 cm2 field at 1.5 cm depth. The profiles are presented athe direction of X-jaw for a 434 cm2 field and Y-jaw for a 10310 cm2 field. They axis is approximately scaled based on the CAX dose ratio betwee

two cases.

ction of

FIG. 8. The 6 MV photon beam profile in-water for a 40 cm340 cm field at 1.5, 5.0, and 10.0 cm depths. The profiles are presented along the dire X-jaw. The y axis is approximately scaled based on the CAX dose ratio between the cases.

Medical Physics, Vol. 32, No. 1, January 2005

eria

n

lease

rgestf0o the

ble todern

ot

146 Cho et al. : Reference photon dosimetry data for 6 MV photon beam 146

linear accelerator head, difference in data matching critetc.

F. Phase space files

In this study, the RPSD(i.e., phase space files) have beedetermined for the following field sizes: 434 cm2, 636 cm2, 10310 cm2, 15315 cm2, 20320 cm2, 30330 cm2, and 40340 cm2. The typical size of each firanges from about 1 to 5 Gbytes as the field size incre

TABLE IV. Comparison between various studies inof the electron pencil beam incident on the targe(apencil beam is 6 MeV for all of the listed studies

Author Accelerator model

Ding (Ref. 7) Clinac 21EX

Fix et al. (Ref. 5) Clinac 2100 C/DHartmann-Siantar

et al. (Ref. 6)Clinac high energy

(generic)Liu et al. (Ref. 4) Clinac 2100C

Sheikh-Bagheri andRogers(Ref. 8)

Clinac high energy(generic)

Keall et al. (Ref. 3) Clinac 21EX

This work Clinac 2100(generic)

FIG. 9. The distribution of the RPC-measured output factors for a 20320 cmincluded in this plot. After rejecting two outlying data points, the total nmean is 157. The correlation coefficient for a Gaussian fit is 0.95.

Medical Physics, Vol. 32, No. 1, January 2005

,

s

from the smallest(i.e., 434 cm2) to the largest(i.e., 40340 cm2). The phase space files for the smallest and lafield sizes contain full phase space information(i.e., type oparticle, energy, and direction) for about 30 million and 17million particles, respectively, that can be used as input tDOSXYZnrc code for in-phantom simulation.

IV. DISCUSSION

The approaches adopted in this study can be applicathe development of the RPDD and RPSD for all the mo

s of the values for the energysEed and radial spreadsReded from Ref. 3). The nominal energy of the electron

Ee (MeV)

FWHMspread inEe (%) Re

6.02 17 1.2 mm FWHMGaussian

6.05 0 Pencil beam6.2 0 2 mm diameter

cylindrical6.5 0 4 mm diameter

cylindrical5.7 3 2.0 mm FWHM

Gaussian6.2 3 1.3 mm FWHM

Gaussian6.2 3 1.0 mm FWHM

Gaussian

ld shown in Fig. 2. Outliers deviating more than ±4s from the mean are ner of data pointssNd is 165 and the number of data points within ±2s from the

termtdapt

.

2 fieumb

etrictherconrtainy arers’l the

daum-dat

PDD

rdet thethe

nifi-ons,

pleliberma-guidf memis-

, itthenifi-ine.ines, i

epthions

to e

the

r tounddataeffi-suethe

MV. Thred

of thinstPC

allyn--airere

anu-D forpro-

ly.that

pted

althan-

. Theysi-thorstheoyer-ll at

rati-entsrate-

de-

mail:

by the

limi-utputed.

theplan-

tonclini-

ueg-cula-

. Al-use,etric

for

dosearlo

tagearam-

n ofMed.

tualMed.

d T.

147 Cho et al. : Reference photon dosimetry data for 6 MV photon beam 147

computer-controlled linear accelerators, whose dosimcharacteristics can be duplicated from one unit to anoAlternatively, the manufacturers’ standard data may besidered as a potential candidate for the RPDD of a cemachine and nominal energy in question, as long as theavailable. However, the applicability of the manufacturstandard data as the RPDD may not be claimed, untidata have been thoroughly verified against the measuredobtained in a consistent manner for a sufficiently large nber of machine units. For instance, the Varian standardfor the 6 MV photon beam may be considered as the Rfrom now on, owing to the current study.

A number of important issues are discussed here in oto avoid possible confusion and misinterpretation abouRPDD. First of all, it should be emphasized again thatRPDD might not be applicable for those machines sigcantly altered from the manufacture’s original specificatiespecially in the head of the linear accelerator. For examthe RPDD may not be applicable to those machines deately modified for certain clinical procedures or thosechines that underwent a major repair such as a wavechange. In such cases, as usual, a comprehensive set osurements should be performed during the initial comsioning (or re-commissioning) process to determine(orverify) the basic photon dosimetry data. At this timeseems difficult to make any firm recommendation forcriteria that can be used to identify those machines sigcantly different from the manufacturer’s generic machNevertheless, based on the way the RPDD are determone may consider a certain machine out of specificationits basic dosimetry data at selected field sizes and ddeviate from the RPDD, more than two standard deviatassociated with the RPDD(e.g., 2s<2%). This could be areasonable criterion because the measured data usedtablish the RPDD are distributed within ±2s of the meanwith 95% probability assuming a Gaussian distribution inmeasured data. In fact, as demonstrated for a 20320 cm2

field output factor(Fig. 9), the RPC-measured data appeabe normally distributed, showing 95% of data to be fowithin ±2s from the mean. Figure 9 also shows that thecan be fit by a Gaussian function with a correlation cocient of 0.95. A further, more rigorous study about this ismay be necessary, along with further refinement ofRPDD.

V. CONCLUSIONS

This study presents the RPDD and RPSD for the 6photon beam from Varian 2100 series linear acceleratorsRPDD were constructed by compiling our own measudata and the average data obtained from the analysismeasured data from the RPC and the RPC-monitoredtutions. All the measured data from the RPC and the Rmonitored institutions were found to be within a statistictight range(i.e., 1s<1% or less) for each dosimetric quatity. The manufacturer’s standard data, except for inOAFs that are available only from the current study, w

compared with the average data and our own measured da

Medical Physics, Vol. 32, No. 1, January 2005

.-

e

ta

a

r

,-

ea-

d,fs

s-

e

ei--

during the generation of the RPDD, showing that the mfacturer’s standard data could also be used as the RPDthe photon beam studied in this study. The RPSD wereduced by the Monte Carlo calculations using theBEAMnrc/

DOSXYZnrc code system with theEe andRe values, 6.2 MeV(a spread of 3% FWHM) and 1.0 mm FWHM, respectiveThe RPSD were capable of generating Monte Carlo dataagreed with the RPDD within the acceptance criteria adoin the current study(e.g., 1% or 1 mm for depth dose).

ACKNOWLEDGMENTS

This investigation was supported in part by Public HeService Grant No. CA 10953 awarded by the National Ccer Institute, Department of Health and Human Servicesauthors acknowledge the contribution of the RPC staff phcists to the measured data used in this study. The auappreciate the following people for various help overcourse of current investigation: Dr. Lei Dong, RSteadham, Paul Holguin, and Dr. Al Smith at M. D. Andson Cancer Center; Dr. Jeff Siebers and Dr. Paul KeaMedical College of Virginia. The authors express their gtude to anonymous referees for their constructive commduring the peer-review process. Finally, the authors are gful to Varian Medical Systems for providing us with thetailed machine specifications and standard data.

a)Author to whom correspondence should be addressed; electronicscho@mdanderson.org

1R. J. Watts, “Comparative measurements on a series of acceleratorssame vendor,” Med. Phys.26, 2581–2585(1999).

2S. H. Cho and G. S. Ibbott, “Reference photon dosimetry data: A prenary study of in-air off-axis factor, percentage depth dose, and ofactor of the Siemens Primus linear accelerator,” J. Appl. Clin. MPhys. 4, 300–306(2003).

3P. J. Keall, J. V. Siebers, B. Libby, and R. Mohan, “Determiningincident electron fluence for Monte Carlo-based photon treatmentning using a standard measured data set,” Med. Phys.30, 574–582(2003).

4H. H. Liu, T. R. Mackie, and E. C. McCullough, “A dual source phobeam model used in convolution/superposition dose calculations forcal megavoltage x-ray beams,” Med. Phys.24, 1960–1974(1997).

5M. K. Fix, M. Stampanoni, P. Manser, E. J. Born, R. Mini, and P. Rsegger, “A multiple source model for 6 MV photon beam dose caltions using Monte Carlo,” Phys. Med. Biol.46, 1407–1427(2001).

6C. L. Hartmann Siantar, R. S. Walling, T. P. Daly, B. Faddegon, Nbright, P. Bergstrom, A. F. Bielajew, C. Chuang, D. Garrett, R. K. HoD. Knapp, D. J. Wieczorek, and L. J. Verhey, “Description and dosimverification of the PEREGRIN Monte Carlo dose calculation systemphoton beams incident on a water phantom,” Med. Phys.28, 1322–1337(2001).

7G. X. Ding, “Energy spectra, angular spread, fluence profiles anddistributions of 6 and 18 MV photon beams: Results of Monte Csimulations for a Varian 2100EX accelerator,” Phys. Med. Biol.47,1025–1046(2002).

8D. Sheikh-Bagheri and D. W. O. Rogers, “Sensitivity of megavolphoton beam Monte Carlo simulations to electron beam and other peters,” Med. Phys.29, 379–390(2002).

9D. Sheikh-Bagheri and D. W. O. Rogers, “Monte Carlo calculationine megavoltage photon beam spectra using the BEAM code,”Phys. 29, 391–402(2002).

10M. Fippel, F. Haryanto, O. Dohm, F. Nusslin, and S. Kriesen, “A virphoton energy fluence model for Monte Carlo dose calculation,”Phys. 30, 301–311(2003).

11D. W. O. Rogers, B. A. Faddegon, G. X. Ding, C.-M. Ma, J. We, an

taR. Mackie, “BEAM: A Monte Carlo code to simulate radiotherapy treat-

kh-arch

Na-

ut-from

am

osention/

, andtionadio-

ory

, R.er-Phys.

andams,”

148 Cho et al. : Reference photon dosimetry data for 6 MV photon beam 148

ment units,” Med. Phys.22, 503–524(1995).12D. W. O. Rogers, C.-M. Ma, G. X. Ding, B. R. Walters, D. Shei

Bagheri, and G. G. Zhang, “BEAMnrc users manual,” National ReseCouncil Report PIRS-0509(A) rev. G, Ottawa, Canada, 2002.

13B. R. Walters and D. W. O. Rogers, “DOSXYZnrc users manual,”tional Research Council Report PIRS-794, Ottawa, Canada, 2002.

14H. H. Liu, T. R. Mackie, and E. C. McCullough, “Modeling photon oput caused by backscattered radiation into the monitor chambercollimator jaws using a Monte Carlo technique,” Med. Phys.27, 737–744(2000).

15G. X. Ding, “Using Monte Carlo simulation to commission photon beoutput factors—A feasibility study,” Phys. Med. Biol.48, 3865–3874(2003).

16G. Starkschall, R. E. Steadham, R. A. Popple, S. Ahmad, and I. I. R“Beam-commissioing methodology for a three-dimensional convolusuperposition photon dose algorithm,” J. Appl. Clin. Med. Phys.1, 8–27

Medical Physics, Vol. 32, No. 1, January 2005

,

(2000).17B. Fraass, K. Doppke, M. Hunt, G. Kutcher, G. Starkshall, R. Stern

J. Van Dyk, “American Association of Physicists in Medicine RadiaTherapy Committee Task Group 53: Quality assurance for clinical rtherapy treatment planning,” Med. Phys.25, 1773–1829(1998).

18B. R. Walters, I. Kawrakow, and D. W. O. Rogers, “History by histstatistical estimators in the BEAM code system,” Med. Phys.29, 2745–2752 (2002).

19P. R. Almond, P. J. Biggs, B. M. Coursey, W. F. Hanson, M. S. HuqNath, and D. W. O. Rogers, “AAPM’s TG-51 protocol for clinical refence dosimetry of high-energy photon and electron beams,” Med.26, 1847–1870(1999).

20G. X. Ding, “Dose discrepancies between Monte Carlo calculationsmeasurements in the buildup region for a high energy photon beMed. Phys.29, 2459–2463(2002).