CD

Medical Dosimetry 38 (2013) 35-41

Medical Dosimetry

journal homepage: www.meddos.org

h

Clinical practice and evaluation of electronic portal imaging device for VMATquality assurance

Yen-Cho Huang, M.Sc.,* Chien-Yi Yeh, B.Sc.,† Jih-Hsiang Yeh, M.Sc.,* Ching-Jung Lo, M.Sc.,†

Ping-Fang Tsai, M.Sc.,† Chih-Hung Hung, Ph.D., M.D.,† Chieh-Sheng Tsai, M.D.,* and Chen-Yuan Chen, B.Sc.†

*Department of Radiation Oncology, Chang Gung Memorial Hospital, Keelung Branch, Keelung City, Taiwan; and †Department of Radiation Oncology, Chang Gung Memorial Hospital,Linkou Branch, Taoyuan County, Taiwan

A R T I C L E I N F O

Article history:

A B S T R A C T

Volumetric-modulated arc therapy (VMAT) is a novel extension of the intensity-modulated radiation therapy

Received 05 January 2012Accepted 08 May 2012

(IMRT) technique, which has brought challenges to dose verification. To perform VMAT pretreatment qualityassurance, an electronic portal imaging device (EPID) can be applied. This study’s aim was to evaluate EPIDperformance for VMAT dose verification. First, dosimetric characteristics of EPID were investigated. Then 10

selected VMAT dose plans were measured by EPID with the rotational method. The overall variation of EPIDdosimetric characteristics was within 1.4% for VMAT. The film system serving as a conventional tool forverification showed good agreement both with EPID measurements ([94.1 � 1.5]% with 3 mm/3% criteria) andtreatment planning system (TPS) calculations ([97.4 � 2.8]% with 3 mm/3% criteria). In addition, EPIDmeasurements for VMAT presented good agreement with TPS calculations ([99.1 � 0.6]% with 3 mm/3%criteria). The EPID system performed the robustness of potential error findings in TPS calculations and thedelivery system. This study demonstrated that an EPID system can be used as a reliable and efficient qualityassurance tool for VMAT dose verification.

� 2013 American Association of Medical Dosimetrists.

Keywords:IMATEPIDRapidArcQuality assuranceVMAT

ou

Introduction

Introductions of new techniques have improved radiotherapy, butat the expense of treatment complexity. Volumetric-modulated arctherapy (VMAT), a novel extension of intensity-modulated radiationtherapy (IMRT), is one these new techniques. This rotational therapydelivers prescribed dose in relatively shorter duration and has betterdose conformity, uniformity, and normal organ sparing.1–5

Many dosimetric devices for patient-specific quality assurance(QA), such as MatrixX (IBA Radiation Dynamics, Inc., Edgewood,NY), Arc Check (Sun Nuclear, Inc., Melbourne, FL), and Delta 4(ScandiDos, Inc., Uppsala, Sweden) have been developed for rota-tional therapy.5,6 Portal dosimetry (Varian Medical Systems, PaloAlto, CA) is one of those dosimetry verification devices. The currentgeneration of electronic portal imaging device (EPID) is composedof amorphous silicon and semiconductor materials. It is not only animaging tool for treatment setup verification but is also an imple-ment for dosimetry measurements.7–13 Moreover, EPID is a conve-

Reprint requests to: Yen-Cho Huang, M.Sc., Department of Radiation Oncology,hang Gung Memorial Hospital, Keelung Branch No.200, Lane 208, Jijin 1st Road, Anle

ist., Keelung City 20445, Taiwan.

E-mail: yenchohuang@hotmail.com

0958-3947/$ – see front matter Copyright � 2013 American Association of Medical Dosimetrttp://dx.doi.org/10.1016/j.meddos.2012.05.004

nient dosimetry QA tool with a relatively high resolution of 0.392mm.

GLAaS is an algorithm used to derive absolute dose maps fromportal images acquired with EPID. The algorithm was developedoriginally for pretreatment verification. It is a method to comparedosimetric measurements directly against treatment planning sys-tem (TPS) calculations. Van Esch et al. investigated EPID dosimetriccharacteristics (aS500/IAS2, Varian Medical Systems) and usedthem for IMRT dosimetry.9 Nicolini et al. also tested a new versionf the EPID GLAaS algorithm for machine quality assurance andsed it for VMAT fields.14,15 Fogliata et al. analyzed the EPID mea-

surements for VMAT quality assurance in different centers.16 How-ever, the previous publications did not use other independent do-simetric systems to verify GLAaS and TPS calculations.

The study was divided into 3 parts. The first was to evaluate dosi-metric characteristics of the EPID system (aS1000/IAS3). The tests in-cluded dose-response linearity, dose-rate dependence, and field sizesdependence. Both 6- and 10-MV photon energies were studied. In thesecond part, 10 VMAT plans were acquired by the radiographic filmsystem, EPID static method, and EPID rotational method. Finally, thefilm measurements and the EPID measurements were analyzed and

compared with the TPS calculations.

ists

lm

tww

d

6d

rc

(

d

an ope

Y.-C. Huang et al. / Medical Dosimetry 38 (2013) 35-4136

Methods and Materials

EPID acquisition description

Two C-linac iX linear accelerators (LINACs) (VarianMedical Systems, Inc.) with 120multileaf collimators (MLCs) were used in this study. VMAT, which is named RapidArc,is available on the LINACs. During VMAT treatments, MLC position, gantry speed, anddose rate are modulated to deliver desirable doses.

The EPID system was composed of an image detection unit (IDU20), an imageacquisition unit (IAS3), a robotic arm (Exact-arm), and a workstation. The detectionarea was 40 � 30 cm2, which contained 1024 � 768 pixels. Each pixel consisted of aight photodiode with a thin film transistor. The pitch between each pixel was 0.392m. The incident x-ray hit the scintillating phosphor screen (Gd2O2S:Tb) and gener-

ated optical photons. The photons were absorbed by photodiodes, and signals wereread out through an analog-to-digital converter. The signalswere converted to absolutedose maps by the dosimetric calibration with GLAaS system on the workstation. Theresponse of the detectors is a linear equation composed of primary and transmittedradiation. The total dose for each pixel is the sum of all segments. GLAaS is a method tocalibrate EPID detector into dose rather than to predict the EPID response. The detailedrationale and application were described by Nicolini et al.12,13

Compared with other dosimetric systems, hardware such as developers and phan-oms are not required for EPID systems. The system settings in this study were: thereas no additional buildupon the topof the cassette and theposition of EPID robotic armas set to 0.0/0.0/0.0 (source-to-detector distance [SDD] � 100 cm). Dark field calibra-

tion, flood field calibration, and dosimetry calibrationwere required before the first useof the system. The dark field calibration was to correct background noise signal, andflood field calibration was to equalize EPID response through the whole area.14 Indosimetry calibration, a diagonal beam profile measured at 5 cm depth of water was

Fig. 1. Three patterns for dose reproducibility test. A 10 � 10-cm2 field,

Fig. 2. The picket fence pattern for EPID dose verification: (A) TP

required for off-axis ratio correction. The calibrationswere executed for different LINACdose rates (range from 100–600 monitor units [MU]/min) and for different photonenergies (6–10 MV). In practice, only one calibration mode can be selected for onemeasurement and each calibration mode was used for a specific dose rate and energy.After calibrations, 1 calibration unit (CU) of portal dose of EPID represented 1 cGy. Inother words, raw images of EPID were converted to dose at 5 cm depth of water byGLAaS.

EPID characteristics

The dose reproducibility of EPID was evaluated by 3 different patterns, including a10 � 10-cm2 field, a 3D conformal field, and an IMRT field with sliding technique(Fig. 1). For each field, 3 measurements were compared with each other and they wereacquired on 2 LINACs with the same mode EPID system.

The dose linearity of EPID was performed with MU ranging from 10–200. Theaverage over an area of 20� 20 pixels at field centerwas recorded for differentMU. Theifferences were from the comparison of EPID to an ideal linearity.

The dose rate is varied in VMAT. Therefore, different LINAC dose rates (range 100–00 MU/min) were studied by delivering 100 MU. In addition, the response of usingifferent EPID calibration modes was evaluated.

The EPID field size dependence was investigated by setting different field sizesanging from 3 � 3 to 30 � 30 cm2. The average over an area of 20 � 20 pixels at fieldenter was normalized to that of a 10 � 10-cm2 field. The value as a function of equiv-alent square field size was compared between EPID and 0.6 cm3 farmer-type chamberExradin A12) measured at 5 cm depth in water.

One picket fence pattern proposed by Ling et al. was selected to test the EPIDosimetric system.17 The pattern was originally one of the quality assurance tests for

n treatment field, and an IMRT field. The images were achieved by EPID.

S calculation; (B) EPID measurement; (C) GAI evaluation.

pdw

G

sv

Y.-C. Huang et al. / Medical Dosimetry 38 (2013) 35-41 37

the RapidArc delivery system. As shown in Fig. 2, 7 strips were irradiated at the samedose but for different dose rates, gantry ranges, and gantry speeds. Before using thispattern to check the EPID dosimetric system, the VMAT delivery systemwas performedand verified by following the film measurement procedures of Ling et al.17

Verification of clinical VMAT plans

Ten selected VMAT plans (22 arc fields) for specific patient QA were listed in Table 1.There were 6 plans for head and neck and 4 for other sites. Optimizations and dosecalculations were done on TPS Eclipse v8.6 (Varian Medical Systems). Anisotropic ana-lytical algorithm (AAA) and 2.5-mm calculation grid size were used for calculations.Because EPID is perpendicular to the beam axis during the rotational therapy, originalVMAT plans were converted to static IMRT plans for calculation purpose. Thismethod was proposed by Iori et al.18 By keeping the sameMLC positions, MU values,and dose rates as the original plans, verification plans were defined at depth 5 cm inwater (source-to-axis distance � 100 cm).

The verification processes, which are based on the radiographic film system, arepresented in Fig. 3. As shown in process A, the static EPID measurements werecompared with the radiographic film measurements. To confirm the consistencybetween static verification plans and rotational VMAT, EPID measurements ac-quired with these 2 methods were compared with each other (process B). Thecomparisons were also done between TPS calculations and filmmeasurements (pro-cess C). Furthermore, the EPID rotational measurements were applied for the TPScalculations verification (process D). All measurements and the TPS calculationswere exported to OmniPro-I’mrt v1.6 (IBA Dosimetry, Schwarzenbruck, Germany)verification software for analysis.

Radiographic film dosimetry system

The radiographic film system used consisted of films (Kodak X-Omat V), an auto-matic film processor (Konica Minolia SRX-101A), and a film scanner (Vidar Dosimetry-ProTM Advantage). To keep accuracy of the radiographic film dosimetry system, wetook precautions as follows: careful control of processing environment (chemical solu-tions, developer immersion time, and temperature): 15-minute warm-up procedurefor film scanner; and background corrections for the scanner lights and sensors. Thescan resolution was 71 dpi with a 5 � 5 pixels average filter. Doses to the films wereconverted from scanning images on OmniPro-I’mrt software by applying calibrationcurves. Different calibration curveswere created for a specific energy (6 or 10MV)withthe same set and same batch of films. Childress et al. studied variations of opticaldensity (OD) over an 18-month period and found them to be as high as 15%.19 Torevent OD variations over time, calibration films and VMATmeasurements were irra-iated and processed sequentially. The time period between irradiation and processingas controlled as well.

Table 1Characteristics of 10 RapidArc plans

Patientno. Site Volume (mL) Energy (MV)

Arcnumbers*

1 Soft palate 662.8 6 22 Palate (left) 198.2 6 23 Buccal 900.4 6 34 Hypopharynx 1019.1 6 25 Lower gum 738.0 6 26 Hypopharynx (right) 478.1 6 27 Esophagus 550.8 6 38 Prostate 799.4 10 29 Pelvis bone 240.4 10 2

10 Rectum 975.5 10 2

* Two arcs are 2 full rotations, and 3 arcs are 2.5 full rotations.

Fig. 3. Schematic diagram of RapidArc plan verification processes. There were 4 differentRapidArc. The conversion plan was kept at the same MLC position, MU, and dose rates as the

amma agreement index analysis

In this study, gamma agreement index (GAI)was used for comparison.20 In general,the criteria were set to be a 3-mm difference in distance and 3% difference in dose.Moreover, strict criteria of 2 mm/2% were also used for comparison. All fields werenormalized to the maximum dose in each field (Dmax � 100%) and the low dose exclu-ion threshold was set to 20%. The percentage of points falling within the criteria pro-ided very important information for dose verification and comparison.

Results

EPID characteristics

The reproducibility is reported in Table 2. There was �0.5% differ-ence between 3 repeated measurements for all 3 patterns. A similarresult (�0.4%) was found in another LINAC.

The EPID dose-response of linearity variation is displayed in Fig. 4.Themaximum value is at low dose (10MU) and is 8% for 6MV and 9%for 10 MV. However, the variation between 30 and 200 MU is (0.7 �1.0)%.

The EPID dose rate dependence test is shown in Fig. 5. Each linepresents different dose rate calibrationmodes of EPID. The response isproportional to LINAC dose rate and themaximumvariation overall is2.5%. One finding is that 400 MU/min calibration has �0.8% variationfor LINAC dose rate between 100 and 600 MU/min.

The field size output factors (OFs) of EPID and farmer-type cham-ber are illustrated in Fig. 6. Variations are larger for �15 � 15-cm2

field size and up to 5.6% for 10 MV 30 � 30 cm2. Nevertheless, it wasnoted that the equivalent field sizes of VMAT control points (seg-ments) are �15 � 15 cm2 primarily, and the mean variation of thesefield sizes is (0.8 � 0.9)%.

From these results, the EPID systemshows anuncertainty of�1.4%by propagating all subsections in quadrature (Table 3).

By using the method by Ling et al., the accuracy of MLC position,movement, dose rates, and gantry speeds were verified.17 The MLCposition was within the tolerance with the film measurements. Ac-cordingly, the EPID image verification results of specific picket fencepattern are shown in Table 4. Compared with the TPS calculation, theGAI evaluation passing rates with criteria 3mm/3% are (100.0 � 0.0)%for 6MVand (99.7�0.0)% for 10MV respectively. By usingmore strict

Table 2EPID reproducibility test

Max difference* (%)

LINAC A LINAC B

ReproducibilitySingle 10 � 10-cm2 field (100 MU) 0.2 0.2Single 3D conformal field (150 MU) 0.3 0.4Single IMRT field (150 MU) 0.5 0.4

* Calculation for each pixel in 3 measurements.

verification processes (A, B, C, and D). Static verification plan was converted fromoriginal RapidArc plan.

mw

E

c

Y.-C. Huang et al. / Medical Dosimetry 38 (2013) 35-4138

criteria of 2 mm/2%, the GAI evaluation is (98.8 � 0.1)% for 6 MV and(96.3 � 0.4)% for 10 MV.

Radiographic film measurements

The filmmeasurements were compared with the static EPID mea-surements (process A, Fig. 3) and TPS calculations (process C, Fig. 3).One example of VMAT field for comparison is shown in Fig. 7. The GAIvalues for 22 arc fields of 10 patients are summarized in Table 5. Fromthe comparison between films and static EPID measurements, themean GAI passing rate was (94.1 � 1.5)% with 3 mm/3% criteria. Bycontrast, the mean GAI passing rate of the comparison between filmsand TPS calculations was (97.4 � 2.8)%.

Fig. 4. EPID dose-response linearity variation.

Fig. 5. Dose-response of EPID irradiated by 100 MU at different LINAC

EPID static and rotational measurements

The comparisons of EPID measurements between static and rota-tional VMATplans for 10 patients (process B, Fig. 3)were (97.8� 1.8)%ean GAI passing rate with strict criteria 2 mm/2% and (100 � 0.0) %ith criteria of 3 mm/3%.

PID rotational measurements and TPS calculations

An example of GAI evaluation of EPID rotational measurementompared with TPS calculation (process D, Fig. 3) is shown in Fig. 8.The red area shows the doses failing the criteria and the white arearepresents doses passing the criteria. The mean GAI passing rate wasequal to (99.1 � 0.6)% with a criteria of 3 mm/3%.

Discussion

The reproducibility test confirmed the dosimetry performanceof the EPID system, but high variation in low dose (�30 MU) wasfound. The under-response for low dose is a result of ghosting ef-fects caused by charge trapping, which was proposed by McDer-mott et al.21 They stated that the longer the irradiation time, thesmaller the relative deficit. The low dose response with differenttype of EPID have also been discovered by Van Esch et al. and Win-kler et al.9,11 It should be noted that 1 field with �30 MU may givean under-reading. However, clinical VMAT fields have �30 MU andthey are in the linear response area.

Nicolini et al.’s study has shown no differences of EPID dose-response between different LINAC dose rates with GAI 3 mm/3% cri-teria.13 The same result (�3% difference)was found in this study. EPIDdose-response is affected by both dose rate calibration mode of EPIDand dose rate of LINAC. During VMAT delivery, LINAC has a variabledose rate (�600 MU/min). EPID over-response may happen if using

dose rate with 6 different dose-rate calibration modes of EPID.

TT

TT

Y.-C. Huang et al. / Medical Dosimetry 38 (2013) 35-41 39

Fig. 6. Field size factor differences between EPID and water. The value was normalized to 10 � 10 cm2.

* The value corresponds to the mean of 20 � 20 pixels at field center.

10 MV 99.7 99.7 99.7 99.7 � 0.0 96.5 96.5 95.8 96.3 � 0.4

able 3he uncertainties of EPID dosimetric measurements

Reproducibility LinearityDose-ratedependence Field size OF

Overall simple errorpropagation

Uncertainty* (%) 0.5 0.7 0.8 0.8 1.4Condition — �30 MU For 400 MU/min EPID

calibration modeEquivalent field size

�15 � 15 cm2—

able 4he picket fence pattern for EPID dose verification

Energy

GAI value (3 mm/3%) GAI value (2 mm/2%)

1 2 3 Mean � SD 1 2 3 Mean � SD

6 MV 100.0 100.0 100.0 100.0 � 0.0 98.8 98.7 98.8 98.8 � 0.1

Fig. 7. One example of RapidArc field. (A) TPS calculation. (B) Portal dose of EPID measurement. (C) Radiographic film measurement.

past

c

TT

Y.-C. Huang et al. / Medical Dosimetry 38 (2013) 35-4140

default setting of 600 MU/min EPID calibration mode. Based on thisstudy, choosing 400 MU/min dose rate of EPID calibration mode forVMAT measurement may have less variation and it is close to theaverage of dose rates of VMAT.

The EPID verification method reported in this study uses a simpleand direct measurement, which is without computing any specifickernel. There is no extra build-upmaterial on EPIDwhile the detectorsare inside with a copper layer and a phosphor layer above. The dosi-metric characteristic of EPID is not the same as the dose to water sothe behaviors have been studied. The field size factor that includedboth collimator scatter factor and phantom scatter factor was mea-sured. It is an experimental method for pretreatment verification andthe accuracy in field center is within an accuracy of 1.4%. In flood fieldcalibration of EPID, the beam profile is assumed to be perfectly flat.This error is corrected by a diagonal beam profile in calibration pro-cess. Other complicated correction methods of EPID off-axis positionand arm backscatter for specific field size have been published by

able 5he passing rates of GAI evaluation from different system measurements

Patient no.

GAI value (%)

Process A* (3 mm/3%) Process B* (3 mm/3%)

1 92.4 100.02 96.4 100.03 96.5 100.04 93.7 100.05 93.5 100.06 94.6 100.07 94.4 100.08 92.8 100.09 92.6 100.0

10 94.3 100.0Mean � SD (%) 94.1 � 1.5 100.0 � 0.0

* The process is presented in Fig. 3.

3Fig. 8. One example of GAI dosimetric evaluation (3 mm/3%).

several groups, and they can be applied to obtain amore accurate dose(�1%).22–25

Nicolini et al. have found that the maximum displacement of Ex-act-arm (Varian Medical Systems) position was 1.5 mm as a result ofthe gravity effect.15 We performed a similar test and found the armosition shift during rotation was in different directions for differentngles. Themaximumdisplacementwas 1.2mm for Exact-arm in thistudy. For R-arm (Varian Medical Systems), the displacement was upo 9.1 mm and it should be considered in VMAT verification.26 Tomaintain positional stability of R-armduring gantry rotation, Iori et al.used a homemade holding device attached to a gantry head to deliverVMAT with EPID.18

In the quality assurance of VMAT delivery system, EPID measure-ments have shown similar results in the picket fence pattern. In Fig. 2,although the EPID image is a little different from the TPS image, EPIDexhibits higher resolution and more information. By using gammaanalysis, the difference near the penumbra area in each strip is mostlikely because of the displacement of arm position from gravity,which can also be found in the film measurements. Thus, it provedthat the EPID dosimetric system has the capability for measuring dif-ferent dose rates, gantry ranges, and gantry speeds.

Films have been used as an integral dosimetric tool but many fac-tors can affect the quality. The sensitive response of films to low x-rays contamination from segments would cause dosimetric variation.Nevertheless, the processing condition and scanner performancewere controlled cautiously to maintain film quality.

The films measurements confirmed EPID measurements andTPS calculations. The EPID measurements proved that EPID dose-responses between static and rotational VMAT were similar, whichincluded the error of EPID shift during rotation. Accordingly, it dem-onstrated that EPID rotational verification planmethod could be usedfor VMAT dose verification.

The TPS was commissioned and verified before clinical use. Be-cause dose calculationswere done in awater phantom, therewas lessuncertainty in heterogeneous correction. However,MLC transmissionfactor and tongue-and-groove effect could not bemodeled and calcu-lated precisely. This brought few errors from TPS. By contrast, EPIDmeasurement errors were from EPID uncertainties and systematicerrors. The EPID uncertainties are listed in Table 3. The systematicerrors included MLC leaf sequencing, collimator angle, beam profile,output variation, etc.

To maintain the center accuracy of EPID, the mechanical qualityassurance of EPID was performed monthly. By contrast, the stabil-ity of EPID response was investigated by Nicolini et al. and thearticle indicated the satisfactory result with a frequency variablefrom daily to monthly.13 In this study, the static 10 � 10-cm2 fieldonstancy check was also within a 0.3% difference over a period of

cess B* (2 mm/2%) Process C* (3 mm/3%) Process D* (3 mm/3%)

8.3 98.6 99.88.2 99.9 99.69.0 99.7 99.69.8 97.8 99.27.1 96.3 98.77.2 98.4 99.60.0 94.6 98.87.0 96.2 98.13.9 95.6 99.68.0 99.5 98.6.8 � 1.8 97.4 � 2.8 99.1 � 0.6

Pro

999999

10999

97

months. Although the EPID system presents good stability, the

iq

gcVb

Y.-C. Huang et al. / Medical Dosimetry 38 (2013) 35-41 41

sensitivity of hardware changes stepwise and it may deviate fromthe expectations.14 The need for the EPID recalibrations, whichncluded dark field, flood field, and dosimetry calibration, are re-uired every 3 months.Because EPID is mounted on the gantry, the gantry sag caused by

ravity cannot be detected from EPID images. The mechanical centerheck and radiation center check of the gantry should be done beforeMATpretreatment verification. The quality assurances for VMAT cane found in the study by Ling et al.17

Conclusion

The characteristics and dose response of EPID (aS1000/IAS3) werepresented in this study. The dose behaviors of EPID proved the robustand stable dose response for clinical practice. Furthermore, the filmmeasurements verified that EPID application for rotational VMAT isreliable. The goal of patient dose verification is to identify potentialerrors either in TPS calculations or in delivery systems, and EPID hasthe capability to do this. The EPID system provides a reliable andefficient tool for VMAT quality assurance.

References

1. Cozzi, L.; Dinshaw, K.A.; Shrivastava, S.K.; et al. A treatment planning study com-paring volumetric arc modulation with RapidArc and fixed field IMRT for cervixuteri radiotherapy. Radiother. Oncol. 89:180–91; 2008.

2. Fogliata, A.; Clivio, A.; Nicolini, G.; et al. Intensity modulation with photons forbenign intracranial tumours: A planning comparison of volumetric single arc, he-lical arc and fixed gantry techniques. Radiother. Oncol. 89:254–62; 2008.

3. Verbakel, W.F.; Cuijpers, J.P.; Hoffmans, D.; et al. Volumetric intensity-modulatedarc therapy vs. conventional IMRT in head-and-neck cancer: A comparative plan-ning and dosimetric study. Int. J. Radiat. Oncol. Biol. Phys. 74:252–9; 2009.

4. Nicolini, G.; Clivio, A.; Fogliata, A.; et al. Simultaneous integrated boost radiother-apy for bilateral breast a treatment planning and dosimetric comparison for volu-metric modulated arc and fixed field intensity modulated therapy. Radiol. Oncol.4:27; 2009.

5. Rao, M.; Yang, W.; Chen, F.; et al. Comparison of Elekta VMAT with helical tomo-therapy and fixed field IMRT: Plan quality, delivery efficiency and accuracy. Med.Phys. 37:1350–9; 2010.

6. Low, D.A.; Moran, J.M.; Dempsey, J.F.; et al. Dosimetry tools and techniques forIMRT.Med. Phys. 38:1313–51; 2011.

7. Bertelsen, A.; Hansen, C.R.; Johansen, J.; et al. Single arc volumetric modulated arctherapy of head and neck cancer. Radiother. Oncol. 95:142–50; 2010.

8. McCurdy, B.M.; Luchka, K.; Pistorius, S. Dosimetric investigation and portal doseimage prediction using an amorphous silicon electronic portal imaging device.Med. Phys. 28:911–24; 2001.

9. Van Esch, A.; Depuydt, T.; Huyskens, D.P. The use of an aSi-based EPID for routineabsolute dosimetric pre-treatment verification of dynamic IMRT fields. Radiother.Oncol. 71:223–34; 2004.

10. Menon, G.V.; Sloboda, R.S. Quality assurance measurements of a-Si EPID perfor-mance.Med. Dosim. 29:11–7; 2004.

11. Winkler, P.; Hefner, A.; Georg, D. Dose-response characteristics of an amorphoussilicon EPID.Med. Phys. 32:3095–105; 2005.

12. Nicolini, G.; Fogliata, A.; Vanetti, E. GLAaS: An absolute dose calibration algorithmfor an amorphous silicon portal imager. Applications to IMRT verification. Med. Phys.33:2839–51; 2006.

13. Nicolini, G.; Fogliata, A.; Vanetti, E.; et al. Testing the GLAaS algorithm for dosemeasurements on low- andhigh-energy photon beamsusing an amorphous siliconportal imager.Med. Phys. 35:464–72; 2008.

14. Nicolini, G.; Vanetti, E.; Clivio, A.; et al. Testing the portal imager GLAaS algorithmfor machine quality assurance. Radiat. Oncol. 3:14; 2008.

15. Nicolini, G.; Vanetti, E.; Clivio, A.; et al. The GLAaS algorithm for portal dosimetryand quality assurance of RapidArc, an intensity modulated rotational therapy. Ra-diat. Oncol. 3:24; 2008.

16. Fogliata, A.; Clivio, A.; Fenoglietto, P.; et al. Quality assurance of RapidArc in clinicalpractice using portal dosimetry. Br. J. Radiol. 84:534–45; 2011.

17. Ling, C.C.; Zhang, P.; Archambault, Y.; et al. Commissioning and quality assurance ofRapidArc radiotherapy delivery system. Int. J. Radiat. Oncol. Biol. Phys. 72:575–81;2008.

18. Iori, M.; Cagni, E.; Nahum, A.E.; et al. IMAT-SIM: A new method for the clinicaldosimetry of intensity-modulated arc therapy (IMAT). Med. Phys. 34:2759–73;2007.

19. Childress, N.L.; Salehpour, M.; Dong, L.; et al. Dosimetric accuracy of Kodak EDR2film for IMRT verifications.Med. Phys. 32:539–48; 2005.

20. Low, D.A.; Harms, W.B.; Mutic, S.; et al. A technique for the quantitative evaluationof dose distributions.Med. Phys. 25:656–60; 1998.

21. McDermott, L.N.; Nijsten, S.M.; Sonke, J.J.; et al. Comparison of ghosting effects forthree commercial a-Si EPIDs.Med. Phys. 33:2448–51; 2006.

22. Bailey, D.W.; Kumaraswamy, L.; Podgorsak, M.B. An effective correction algorithmfor off-axis portal dosimetry errors.Med. Phys. 36:4089–183; 2009.

23. Berry, S.L.; Polvorosa, C.S.; Wuu, C.S. A field size specific backscatter correctionalgorithm for accurate EPID dosimetry.Med. Phys. 37:2425–34; 2010.

24. Greer, P.B. Off-axis dose response characteristics of an amorphous silicon elec-tronic portal imaging device.Med. Phys. 34:3815–24; 2007.

25. Rowshanfarzad, P.; McCurdy, B.M.; Sabet, M.; et al. Measurement and modeling ofthe effect of support arm backscatter on dosimetry with a Varian EPID. Med. Phys.37:2269–79; 2010.

26. Iori, M.; Cagni, E.; Paiusco, M.; et al. Dosimetric verification of IMAT delivery with a

conventional EPID system and a commercial portal dose image prediction tool.Med. Phys. 37:377–90; 2010.