lable at ScienceDirect

Physica Medica 30 (2014) 503e508

Contents lists avai

Physica Medica

journal homepage: http: / /www.physicamedica.com

Original paper

Monte Carlo comparison of superficial dose between flattening filterfree and flattened beams

Khosrow Javedan a, Vladimir Feygelman a, Ray R. Zhang b, Eduardo G. Moros a,Candace R. Correa a, Andy Trotti a, Weiqi Li a, Geoffrey G. Zhang a,*

aRadiation Oncology Department, Moffitt Cancer Center, Tampa, FL, USAb School of Medicine and Public Health, University of Wisconsin, Madison, WI, USA

a r t i c l e i n f o

Article history:Received 12 November 2013Received in revised form27 January 2014Accepted 3 March 2014Available online 21 March 2014

PACS:87.55.km87.56.bd

Keywords:Monte CarloSkin doseFlattening filter free

* Corresponding author.E-mail addresses: geoffrey.zhang@moffitt.org,

G. Zhang).

http://dx.doi.org/10.1016/j.ejmp.2014.03.0011120-1797/� 2014 Associazione Italiana di Fisica Med

a b s t r a c t

This study investigates the superficial dose from FFF beams in comparison with the conventional flat-tened ones using a Monte Carlo (MC) method. Published phase-space files which incorporated real ge-ometry of a TrueBeam accelerator were used for the dose calculation in phantom and clinical cases. Thephoton fluence on the central axis is 3 times that of a flattened beam for a 6 MV FFF beam and 5 times fora 10 MV beam. The mean energy across the field in air at the phantom surface is 0.92e0.95 MeV for the6 MV FFF beam and 1.18e1.30 MeV for the corresponding flattened beam. At 10 MV, the values are 1.52e1.72 and 2.15e2.87 MeV for the FFF and flattened beams, respectively. The phantom dose at the depthof 1 mm in the 6 MV FFF beam is 6% � 2.5% (of the maximum dose) higher compared to the flattenedbeam for a 25 � 25 cm2

field and 14.6% � 1.9% for the 2 � 2 cm2field. For the 10 MV beam, the cor-

responding differences are 3.4% � 1.5% and 10.7% � 0.6%. The skin dose difference at selected points onthe patient’s surface between the plans using FFF and flattened beams in the head-and-neck case was6.5% � 2.3% (1SD), and for the breast case it was 6.4% � 2.3%. The Monte Carlo simulations showed thatdue to the lower mean energy in the FFF beam, the clinical superficial dose is higher without the flat-tening filter compared to the flattened beam.

� 2014 Associazione Italiana di Fisica Medica. Published by Elsevier Ltd. All rights reserved.

Introduction Since the skin dose is often a clinical concern and many clinical

As the flattening filter with its associated attenuation isremoved from the X-ray beam path, the accelerator dose rate in-creases. The other effect is substantial reduction in head scatter, asthe flattening filter is the major source of scattered photons. Flat-tening filter free (FFF) beams in radiotherapy thus have theadvantage of shorter treatment delivery time and lower out-of-field dose compared with conventional flattened beams [1,2]. Inthe absence of beam hardening, it was predicted that skin dosewould increase with FFF beams [2,3]. Indeed, the measured su-perficial dose increase of the 6 MV FFF compared to the flattenedbeamswas reported to be 6e7% of themaximumdose for field sizesbetween 2 � 2 and 10 � 10 cm2 for a TrueBeam accelerator (VarianMedical Systems, Palo Alto, CA) by Wang et al. [4] and 7% for5 � 5 cm2 for an Elekta Precise accelerator (Elekta Oncology Sys-tems, Stockholm, Sweden) by Kragl et al. [5].

gzhang78@hotmail.com (G.

ica. Published by Elsevier Ltd. All

treatment planning systems have limited accuracy in the buildupregion [6], even measurement in this region has challenges andcomplicated corrections are required for accuracy [7], the MonteCarlo (MC) method is often applied to superficial dose analysis[8,9]. Monte Carlo simulation is the most accurate technique todetermine dose deposited in tissue based on the actual radiationtransport physics [10]. Based on the fundamental interactionprinciples, the backscattered electrons and the electron contami-nation of the photon beamswhich contribute to superficial dose areappropriately calculated with Monte Carlo.

The MC method provides the tools to understand the physicsbehind the dose distribution [11,12], and many new techniques aredeveloped to enhance the speed of MC calculation [13]. AlthoughMC has been used in superficial dose investigations for FFF beams,the studies were performed before any realistic FFF linear acceler-ator became commercially available. The structures of the com-mercial FFF accelerators are not simply having the flattening filterremoved in conventional accelerators, but mostly with new de-signs. It is well known that the dose characteristics depend onlinear accelerator head design. This study investigates in detail byMC superficial dose distributions produced by FFF beams from a

rights reserved.

Figure 1. (A) PDD and (B) Profile at 1.5 cm depth for 15 � 15 cm2 6 MV FFF beamcomparison between calculation and measurement.

K. Javedan et al. / Physica Medica 30 (2014) 503e508504

TrueBeam accelerator and compares them to those produced byconventional flattened beams from the same accelerator. Thephysics behind the dose difference is also explored with the con-venience of the MC tools.

Materials and methods

Monte Carlo simulations

Published phase space files for 6 MV and 10 MV FFF and flat-tened beams generated using the actual geometry of a TrueBeamaccelerator above the collimator moving jaws [14] were used as theinput radiation source to BEAMnrc [15], an EGSnrc based MC pro-gram [16]. These programs were used to simulate the beamsdownstream of the collimator. Phase space files of various fieldsizes were generated at the phantom surface to be used as thesource for dose calculations.

The dynamicMLCmodule (DMLC) was used for the static as wellas the step and shoot beam model. The collimator Jaws were set to2 � 2, 15 �15, and 25� 25 cm2 for the static fields and 12 � 20 and10.5 � 10.5 cm2 for the clinical step and shoot fields. The MLC leafposition data for all clinical fields were exported from the Pinnaclev.9.2 treatment planning system (Philips Radiation Oncology,Fitchburg, WI) for the MC step and shoot virtual simulation. Aformatted text file containing the leaf opening data for each clinicalfield was created for the FFF and flattened beam simulationsrespectively.

In the Multileaf collimator DMLC module, the three leaf typesand their positions with respect to the reference plane positionwere specified. The leaf cross-sections were defined for the FULLleaves, the TARGET leaves and the ISOCENTER leaves. The 20 fullleaf pairs (leaf numbers 1e10 and 51e60) and 10 TARGET/ISO-CENTER leaf pairs (leaf numbers 11e50) were selected based on thereal MLC configuration. The tongue and groove and the gap widthsand the distance of driving screw hole from leaf tip were defined.

There were approximately 570 million particles incident fromthe original source in the primary TrueBeam phase space file abovethe jaws. A total of 87million particles were recycles at least 4 timesand restarted to result in approximately 240million particles in thephase space file below the jaws. The phase space files that weregenerated below the jaws were used as the source above the MLC.These phase space files were restarted many times to yield morethan 500million particles in each phase space file below theMLC atthe phantom surface. DOSXYZnrc [17] was used for dose calcula-tions in phantom and in patient. The cutoff energies for electrons(ECUT) and for photons (PCUT) were 0.521 and 0.01 MeV, respec-tively. No specific variance reduction technique was applied.BEAMDP version v1.9 [18] was used to analyze the phase space filesfor mean energy and fluence distributions. Superficial dose wascalculated in a water phantom for various field sizes, and beamincident angles, at different points across the field. The dose in thewater phantomwas calculated in a grid size of 5 � 5 � 2 mm3 withthe 2 mm along the depth.

Clinical cases

Skin dose was calculated and compared for two clinical cases,headeneck (HN) and breast. Opposed tangential fields with 6control points per field, using step and shoot technique, were usedto treat the infiltrating ductal carcinoma of the left breast to cover95% of the planning target volume (PTV, volume ¼ 915 cc) to5000 cGy as prescribed.

The squamous cell carcinoma of the right neck was treated with5 non-opposed but coplanar intensity modulated radiotherapy(step and shoot IMRT) fields. The prescription dosewas 6600 cGy to

the right alveolar ridge PTV (132 cc), and 5600 cGy to the 4 positivelymph nodes PTV (174 cc). At least 95% of each volume received theprescribed dose. The planning dose objectives for targets and or-gans at risk (OAR) were respected in both cases.

CTCREATE [19] was used to convert CT data to phantom datathat can be used in DOSXYZnrc for dose calculations. In the con-version, the material was assigned to air, lung, tissue and bonebased on the local CT number. The patients of the two selectedcases were treated using the flattened 6 MV beam. An additionalIMRT treatment planwas retrospectively generatedwith a 6MV FFFbeam in each case. The FFF and conventional plans shared the samebeam arrangements, satisfied the same objectives, and producedsimilar DVHs. The treatment plans were generated in Pinnacle v.9.2treatment planning system.

The treatment plans were then calculated using BEAMnrc andDOSXYZnrc using the new dynamic source features of BEAMnrc[20]. Once the 120 Varian MLC leaf dimensions and material wereprecisely specified in the model, the MLC leaf opening coordinateswere calculated based on the leaf dimension. In turn, the dynamicsource feature enables MLC incorporation and synchronization ofvarious MLC segments within the step and shoot IMRT field. Thedose grid size in the MC calculation for the clinical cases was3 � 3 � 3 mm3. The total dose was summed with the weightingsbased on the monitor units in each beam in the plans generated bythe treatment planning system. The superficial doses were calcu-lated as the percentages of the prescription dose covering 95% ofthe PTV. The skin dose was taken as the dose to a first voxel withdensity less than 1.0 at the airetissue interface in the phantom dataconverted from patient’s planning CT. The skin dose was compared

K. Javedan et al. / Physica Medica 30 (2014) 503e508 505

for multiple points with identical coordinates between the plans offlattened and FFF beams. There were 20 anterior and 20 posteriorpoints for the head and neck and 15 lateral, 15 medial, and 15central points evaluated for the breast case.

The depth dose and cross-beam profiles were scanned in awaterphantom using a Scanditronix (IBA) photon diode (PFD-3G) for fieldsizes �10 � 10 cm2, and the IBA CC13 ion chamber for larger fieldsizes. The point of measurement was set at 0.6 chamber radiusupstream (1.8 mm). The percentage depth dose (PDD) and profilesof the 6 MV filter free beams were compared with the MCcalculation.

Results

The phantom study

Phase space files were generated for the TrueBeam FFF andflattened 25 � 25 cm2

fields at 100 cm source to surface distance(SSD). Figure 1 shows the comparison of the PDD and profile for a15 � 15 cm2

field of the 6 MV FFF beams between the MC calcu-lation and measurement. The MC calculated PDDs and profiles inwater agreed with measurement within 2% in the high dose regionor 2 mm in the penumbra, validating the MC model.

Figure 2 shows the fluence distributions for the 6 and 10 MV25 � 25 cm2 beams. Notice the lower order of magnitude of theelectron fluence. The fluence per source particle upstream of theflattening filter is constant between FFF and flattenedmodes, so thedifferences of the fluence reflect solely the difference from the

Figure 2. Fluence distribution comparison of photons (A) and electrons (B) betweenthe FFF and flattened beams for 25 � 25 cm2

field at SSD ¼ 100 cm for a TrueBeamaccelerator. The electron fluence is more than 2 orders of magnitude lower than that ofphotons.

absence of the flattening filter. The MC calculation demonstratedthat the fluence at the center of a 6 MV FFF beam with the sameelectron beam hitting the target was 3.1 times that of a flattenedbeam, and for a 10 MV beam the difference was 5.1-fold. For thecontaminant electrons, the MC calculation showed relatively lowerfluence increase for the FFF beams. For the same 6 MV beams, thecontaminant electron fluence was 1.8 times higher than the flat-tened beam, while for the same 10 MV beam, it was 2.1 times. Thelower ratio of component of contaminant electrons to photons inthe FFF beams is another factor that makes the superficial doselower, since contaminant electrons usually contribute to the su-perficial dose. In the 6 MV 25 � 25 cm2 beam simulations, theelectron component dose at 1 mm depth contributed 12.8% � 2.1%(the presentedmean and standard deviation values are in referenceto the maximum dose if not specified) to the total dose for theflattened beam while it was 9.4% � 1.6% for the FFF beam.

Figure 3A shows the mean photon energy for these beams as afunction of off-axis distance at the phantom surface. The field wasdivided into 40 square rings (bins) and the mean energy for eachone was calculated. The mean energy of the photons was differentbetween the FFF and flattened beams. For the 10 MV beams, themean energy for the flattened beam was between 2.15 and2.87MeVwhile it was between 1.52 and 1.72MeV for the FFF beam.For the 6 MV beams it was 1.18e1.30 and 0.92e0.95 MeV for theflattened and FFF beams, respectively. Figure 3B shows the energyfluence spectra comparison for 6 MV beams between 20 � 20 and2 � 2 cm2

fields. The spectra were almost identical for the FFF

Figure 3. (A) Mean energy distribution comparison of photons between the FFF andflattened beams for 25 � 25 cm2

field at SSD ¼ 100 cm for a TrueBeam accelerator. (B)Energy fluence spectra comparison for 6 MV beams between 20 � 20 and 2 � 2 cm2

fields.

Figure 5. MC calculated percentage central axis doses in water for 10 � 10 cm2 6 MVFFF and flattened beams with various incident angles. The maximum depth dose of thecorresponding normal incident beam was used as the reference.

K. Javedan et al. / Physica Medica 30 (2014) 503e508506

beams while they were different for the flattened beams, withhigher energy in smaller beams. This trend can be seen in Fig. 3A.Similar results were also found for the 10 MV beams (data notshown).

Figure 4A compares the MC calculated shallow depth doses inwater for a 25 � 25 cm2 6 MV FFF and flattened beams. At 1 mmdepth, the FFF beam PDD was 56.7 � 1.5% while it was 50.7 � 2.0%for the flattened beam. This indicates that for the normally incidentbeams, the 6 MV FFF dose at 1 mm would be about 6% of themaximum dose, or 12% of the local dose, higher than with theflattened beam. Figure 4B shows the dose difference versus fieldsize between the FFF and flattened beams. With the 6 MV beams,the dose difference increased from 6.0% � 2.5% to 14.6% � 1.9%between the 25 � 25 cm2 and 2 � 2 cm2

fields. For the 10 MVbeams, the differences ranged from 3.4% � 1.5% to 10.7% � 0.6% forthe 25� 25 cm2 and 2� 2 cm2

fields, respectively, all in reference tothe maximum dose. The beam energy difference between the FFFand flattened beams, shown in Fig. 3, is considered to be the majorreason for the superficial dose difference between the two kinds ofbeams: softer FFF beams produce higher superficial dose.

For both FFF and flattened beams the superficial dose is essen-tially independent on the off-axis distance, when normalized to thecorresponding off-axis PDD maximum (data not shown).

The dose difference between the FFF and flattened beams atshallow depths changes with the beam incident angle. This varia-tion for a 10 � 10 cm2 6 MV FFF and flattened beams is shown inFig. 5. The relative dose at 1 mm depth was normalized to the

Figure 4. (A) MC calculated central axis superficial doses in water for 6 MV FFF andflattened beams of 25 � 25 cm2. (B) Superficial dose % difference (in reference of themaximum dose) of FFF minus flattened at 1 mm depth versus field size.

maximum depth dose of the corresponding normally incidentbeam. The dose difference between the FFF and flattened beams(FFF minus flattened) changes from 10.6% � 0.9% of Dmax at0� incidence angle to 6.1% � 1.1% at 70� angle.

Clinical cases

The two plans in both clinical cases were dosimetrically similarto each other. Figure 6 shows the histograms of skin dose differ-ences for multiple points between the plans of FFF and flattenedbeams for the HN case and the breast case. The average skin dosedifference in the HN case was 6.5% � 2.3% (1SD) with the range of(2.9e10.3%), and for the breast case it was 6.4%� 2.3% (1.5%e11.3%).A paired t-test demonstrated that the two sets of skin dose datawere significantly different with p < 0.001 for both cases.

Discussion

The superficial dose is relatively uniform across the field for theflattened beams, which is consistent with other studies [6,21].Similar to the flattened beams, the superficial dose for the FFFbeams does not change with off-axis distance.

Figure 6. Histograms of skin dose differences between the IMRT plans using FFF andflattened beams for the headeneck and breast cases. The skin dose in the FFF plans washigher than the flattened beam plans for all the points. The percentage difference wasin reference to the prescription dose.

K. Javedan et al. / Physica Medica 30 (2014) 503e508 507

The use of the maximum dose from the off-axis PDD fornormalization of the off-axis superficial dose is justified becausethe off-axis target volume needs to be covered by the same pre-scription dose. As the accelerator calibration point is on the centralaxis and the lateral dose profile is centrally peaked, more monitorunits (MU) are needed to deliver the same dose to the off-axistarget volumes when FFF beams are used in IMRT treatment [22].

The major reason for the field size dependence of the superficialdose difference between the FFF and flattened beams is the scattercomponent. This scatter originates mostly from the flattening filter,as our MC data show that contribution from the photons scatteredby the jaws andMLC to the central axis superficial dose is negligible.As shown in Fig. 3B, the energy spectra of the 20� 20 and 2� 2 cm2

6 MV FFF beams are almost identical while those of the flattenedbeams are quite different. The softening of the 20� 20 cm2

flattenedbeam by scatter makes the central axis superficial dose difference(FFF minus flattened) smaller for the larger beams.

There is less accelerator head scatter in the FFF beams comparedto the flattened beams [1,2], which tends to make the skin dose lessfor the FFF beams. Since the phase space files above the collimatorjaws were generated by Varian Medical Systems and the path in-formation of the particles were not available, the quantitativeanalysis of this effect could not be done. Based on the data in thisstudy, this effect, however, cannot completely compensate for theeffect of lower mean energy, but makes the skin dose differencesmaller.

As shown in Fig. 2, the increase of contaminant electron fluencein the FFF beams is relatively lower than the photon counterpart,which suggests that the component electron dose (i.e., contami-nation) per given dose of photons is actually considerably smaller inFFF mode than for the flattened beam. This is similar to the findingin the comparison study between the compensator- and MLC-based IMRT treatment [9]. The reason for the higher electroncontamination in the flattened beams compared to the FFF beams isthat more electrons are generated in the flattening filter due tophoton interactions, while in the compensator-based IMRT moreelectrons are generated in the compensator. However, due to theorder of magnitude difference between the photon and electronfluence, in addition to less accelerator head scatter, the effect of thelower contaminant electrons in FFF beams cannot overcome theeffect of the softer photon beam, thus the superficial dose is stillhigher.

Photon energy spectra and mean energies of FFF beams havebeen published in previous studies for Elekta Precise linear accel-erators [5,23] and for a Varian Clinac 2100C accelerator [24]. Somedifferences can be found between our study presented in this paperand the previous ones. As mentioned in Introduction, these previ-ous studies were not based on real commercially released accel-erators with FFF capability, but simulations with flattening filterremoved from conventional flattened beam accelerators while thisstudy was based on TrueBeam, a commercially released realaccelerator. Differences are also expected between Varian andElekta accelerators.

The water phantom is geometrically well defined so that thesurface position can be exactly determined and the dose at a certaindepth can be accurately calculated. This is not true for the clinicalcases. The exact surface of the patient’s body on a clinical CT datasetis more difficult to locate, thus the exact depth of the skin dose inthe clinical cases is not defined in this study. Instead, the skin dosewas taken as the dose to a first voxel with density less than 1.0 atthe airetissue interface. The average density of the voxel chosen torepresent the skinwas 0.4� 0.2 (1SD) g/cc (range 0.1e0.8 g/cc). Therelatively low reported density is due to partial volume effect in theCT images. The partial volume effect causes depth uncertaintywhen CT data are used in MC superficial dose calculation. As the

doses were read from the same voxels for the FFF and conventionalplans, this approach by comparison of the skin dose differences isjustified even without the depth being exactly known.

Thewater phantom study indicated the dose difference of 6% forlarge fields at 1 mm for the FFF and flattened 6 MV beams and it ishigher for small fields. The average skin dose difference in the MCcalculated clinical cases was 6%. The lower dose difference in theclinical cases is most likely due to the depth uncertainty because ofthe partial volume effect in MC calculations and oblique beamincidence. As shown in Fig. 5, the difference decreases with theincidence angle. Particularly in the breast case, the beam wasopposed with the exit dose contribution and tangent to the skinsurface in many locations.

Different results are expected for different linear acceleratormodels other than Varian’s TrueBeam. This in part is due to specificimplementation for various parameters such as different electronenergy between FFF and flattened beams which would result indifferent skin dose [25].

The dose to dermal skin layer has been reported in this MCinvestigation which may be different than that at the basal skinlevel of 0.07 mm.

In our clinic, HN treatments are mostly done with volumetricmodulated arc therapy (VMAT). The skin dose analysis in this studyonly applies to static beam IMRT cases and should not be extendedto the VMAT modality. The skin dose using MC for VMAT wouldrequire a separate study.

The original IAEA formatted phase space files were generated ona cylindrical surface. The new version of BEAMnrc is capable ofreading the IAEA formatted phase space file, but assuming thephase space file is on a flat surface. This small z value approxima-tion does not create large discrepancy between the calculation andmeasurement based on the PDD and profile comparisons.

Conclusions

The Monte Carlo simulations showed that due to the lowermean energy in the FFF beam, the clinical superficial dose is higherwithout the flattening filter compared to the flattened beam. Thisenergy effect overcomes the decrease in skin dose due to thereduced head scatter in the absence of the flattening filter andlower contaminant electron dose. The simulated clinical casesdemonstrated the skin dose difference between FFF and flattenedbeamswas relatively lower than that in the phantom study becauseof the oblique beam angles and exit dose in addition to the un-certainty in depth due to partial volume effect in CT images.

Conflict of interest statement

There is no conflict of interest for all authors.

References

[1] Georg D, Knöös T, McClean B. Current status and future perspective of flat-tening filter free photon beams. Med Phys 2011;38:1280e93.

[2] Titt U, Vassiliev ON, Pönisch F, Dong L, Liu H, Mohan R. A flattening filter freephoton treatment concept evaluation with Monte Carlo. Med Phys 2006;33:1595e602.

[3] Mesbahi A, Mehnati P, Keshtkar A, Farajollahi A. Dosimetric properties of aflattening filter-free 6-MV photon beam: a Monte Carlo study. Radiat Med2007;25:315e24.

[4] Wang Y, Khan MK, Ting JY, Easterling SB. Surface dose investigation of theflattening filter-free photon beams. Int J Radiat Oncol Biol Phys 2012;83:e281e5.

[5] Kragl G, af Wetterstedt S, Knäusl B, Lind M, McCavana P, Knöös T, et al.Dosimetric characteristics of 6 and 10 MV unflattened photon beams.Radiother Oncol 2009;93:141e6.

[6] Kry SF, Smith SA, Weathers R, Stovall M. Skin dose during radiotherapy: asummary and general estimation technique. J Appl Clin Med Phys 2012;13(3):20e34.

K. Javedan et al. / Physica Medica 30 (2014) 503e508508

[7] Keall PJ, Siebers JV, Libby B, Mohan R. Determining the incident electronfluence for Monte Carlo-based photon treatment planning using a standardmeasured data set. Med Phys 2003;30:574e82.

[8] Chow JCL, Jiang R, Leung MKK. Dosimetry of oblique tangential photon beamscalculated by superposition/convolution algorithms: a Monte Carlo evalua-tion. J Appl Clin Med Phys 2010;12(1):108e21.

[9] Javedan K, Zhang GG, Hoffe S, Feygelman V, Forster K. Comparing dose in thebuild-up region between compensator- and MLC-based IMRT. J Appl Clin MedPhys 2012;13(5):1e12.

[10] Kim S, LC, Zhu TC, Palta J. Photon beam skin dose analysis for different clinicalsetups. Med Phys 1998;25:860e6.

[11] Tsiamas P, Sajo E, Cifter F, Theodorou K, Kappas C, Makrigiorgos M, et al. Beamquality and dose perturbation of 6 MV flattening-filter-free linac. Phys Med2014;30:47e56.

[12] Zhang RR, Feygelman V, Harris ER, Rao N, Moros EG, Zhang GG. Is waxequivalent to tissue in electron conformal therapy planning? A Monte Carlostudy of material approximation introduced dose difference. J Appl Clin MedPhys 2013;14(1):92e101.

[13] Ezzati AO, Sohrabpour M, Mahdavi SR, Buzurovic I, Studenski MT.A comprehensive procedure for characterizing arbitrary azimuthally sym-metric photon beams. Phys Med 2014;30:191e201.

[14] Constantin M, Perl J, LoSasso T, Keall PJ. Modeling the TrueBeam linac using aCAD to Geant4 geometry implementation: dose and IAEA-compliant phasespace calculations. Med Phys 2011;38:4018e24.

[15] Rogers DWO, Faddegon BA, Ding GX, Ma C-M, We J, Mackie TR. BEAM: aMonte Carlo code to simulate radiotherapy treatment units. Med Phys1995;22:503e24.

[16] Kawrakow I, Rogers DWO. The EGSnrc code system. NRCC Report PIRS-701.Ottawa, Canada: NRCC; 2003.

[17] Kawrakow I, Walters BRB. Efficient photon beam dose calculations usingDOSXYZnrc with BEAMnrc. Med Phys 2006;33:3046e56.

[18] Ma CM, Rogers DWO. BEAMDP users manual. NRC Report PIRS 509c (rev A);2004.

[19] Ma C-M, Reckwerdt P, HolmesM, Rogers DWO, Geiser B. DOSXYZ usersmanual.NRC Report. Ottawa, Canada: National Research Council Canada; 1995.

[20] Lobo J, Popescu IA. Two new DOSXYZnrc sources for 4D Monte Carlo simu-lations of continuously variable beam configurations, with applications toRapidArc, VMAT, TomoTherapy and CyberKnife. Phys Med Biol 2010;55:4431e43.

[21] Butson MJ, Cheung T, Yu PKN, Currie M. Surface dose extrapolation mea-surements with radiographic film. Phys Med Biol 2004;49:N197e201.

[22] Zhang G, Ku L, Dilling T, Stevens C, Zhang R, Li W, et al. Volumetric modulatedarc planning for lung stereotactic body radiotherapy using conventional andunflattened photon beams: a dosimetric comparison with 3D technique.Radiat Oncol 2011;6:152.

[23] Dalaryd M, Kragl G, Ceberg C, Georg D, McClean B, af Wetterstedt S, et al.A Monte Carlo study of a flattening filter-free linear accelerator verified withmeasurements. Phys Med Biol 2010;55:7333.

[24] Vassiliev ON, Titt U, Kry SF, Pönisch F, Gillin MT, Mohan R. Monte Carlo studyof photon fields from a flattening filter-free clinical accelerator. Med Phys2006;33:820e7.

[25] Huang Y, Siochi RA, Bayouth JE. Dosimetric properties of a beam quality-matched 6 MV unflattened photon beam. J Appl Clin Med Phys 2012;13(4):71e81.