Development of an accurate EPID-based output measurementand dosimetric verification tool for electron beam therapy

Aiping Ding, Lei Xing, and Bin Hana)

Department of Radiation Oncology, Stanford University School of Medicine, Stanford, Calilfornia 94305

(Received 3 September 2014; revised 12 May 2015; accepted for publication 1 June 2015;published 17 June 2015)

Purpose: To develop an efficient and robust tool for output measurement and absolute dose verifi-cation of electron beam therapy by using a high spatial-resolution and high frame-rate amorphoussilicon flat panel electronic portal imaging device (EPID).Methods: The dosimetric characteristics of the EPID, including saturation, linearity, and ghostingeffect, were first investigated on a Varian Clinac 21EX accelerator. The response kernels of theindividual pixels of the EPID to all available electron energies (6, 9, 12, 16, and 20 MeV) werecalculated by using Monte Carlo (MC) simulations, which formed the basis to deconvolve an EPIDraw images to the incident electron fluence map. The two-dimensional (2D) dose distribution atreference depths in water was obtained by using the constructed fluence map with a MC simulatedpencil beam kernel with consideration of the geometric and structural information of the EPID.Output factor measurements were carried out with the EPID at a nominal source–surface distanceof 100 cm for 2×2, 3×3, 6×6, 10×10, and 15×15 cm2 fields for all available electron energies, andthe results were compared with that measured in a solid water phantom using film and a Farmer-typeion chamber. The dose distributions at a reference depth specific to each energy and the flatness andsymmetry of the 10×10 cm2 electron beam were also measured using EPID, and the results werecompared with ion chamber array and water scan measurements. Finally, three patient cases withvarious field sizes and irregular cutout shapes were also investigated.Results: EPID-measured dose changed linearly with the monitor units and showed little ghostingeffect for dose rate up to 600 MU/min. The flatness and symmetry measured with the EPID werefound to be consistent with ion chamber array and water scan measurements. The EPID-measuredoutput factors for standard square fields of 2×2, 3×3, 6×6, 10×10, 15×15 cm2 agreed withfilm and ion chamber measurements. The average discrepancy between EPID and ion chamber/filmmeasurements was 0.81%±0.60% (SD) and 1.34%±0.75%, respectively. For the three clinical cases,the difference in output between the EPID- and ion chamber array measured values was foundto be 1.13%±0.11%, 0.54%±0.10%, and 0.74%±0.11%, respectively. Furthermore, the γ-indexanalysis showed an excellent agreement between the EPID- and ion chamber array measured dosedistributions: 100% of the pixels passed the criteria of 3%/3 mm. When the γ-index was set to be2%/2 mm, the pass rate was found to be 99.0%±0.07%, 98.2%±0.14%, and 100% for the threecases.Conclusions: The EPID dosimetry system developed in this work provides an accurate and reli-able tool for routine output measurement and dosimetric verification of electron beam therapy.Coupled with its portability and ease of use, the proposed system promises to replace the currentfilm-based approach for fast and reliable assessment of small and irregular electron field dosimetry.C 2015 American Association of Physicists in Medicine. [http://dx.doi.org/10.1118/1.4922400]

Key words: electron, EPID, Monte Carlo, dosimetry, quality assurance

1. INTRODUCTION

Electron beam therapy (EBT) is a valuable radiotherapymodality for treatment of superficial tumors. Electron beamsare characterized by a reasonably uniform dose from thesurface to a certain depth (depending on energy) and thenby a rapid falloff in dose.1 In practice, electron dosimetryis intractable, especially when dealing with a small and/orirregularly shaped field. While the most accurate approach isto acquire the depth doses and profile in a water phantom,2,3

film dosimetry done with plastic water slabs at a referencedepth has been a widely adopted technique for output

measurement and electron dosimetry. There is a continuousneed for a fast and robust electron dosimetry solution toovercome various issues associated with the film dosimetry,such as the large dosimetric uncertainties and laborious filmprocessing procedure.

The development of amorphous silicon (a-Si) flat panelelectronic portal imaging device (EPID) has opened upnew possibilities for quality assurance (QA) and dosimetricverification of EBT. EPID has many advantages, includinggood sensitivity, high resolution, large active detectionarea, ease-of-use, real-time image acquisition, and fast dataprocessing. Over the past two decades, several research groups

4190 Med. Phys. 42 (7), July 2015 0094-2405/2015/42(7)/4190/9/$30.00 © 2015 Am. Assoc. Phys. Med. 4190

4191 Ding, Xing, and Han: An accurate EPID-based electron beam output measurement and dosimetric verification tool 4191

have reported the use of EPID for dosimetric verification ofintensity modulated radiotherapy (IMRT)4–7 and volumetricmodulated arc therapy (VMAT).8–12 Reports on using EPIDfor EBT dosimetry are, however, relatively sparse. Jarryand Verhaegen13 reported the use of bremsstrahlung part ofan electron beam for portal image quality analysis. Becket al.14 investigated the possibility of using an EPID fordaily constancy check of electron beam. Recently, Chatelainet al.15 studied the potential of using an EPID for multileafcollimator (MLC) modulated electron radiotherapy (MERT)QA. Similar to that previously proposed by Chang et al.16 forIMRT studies,17,18 the acquired EPID images were convertedinto absolute doses with an empirical method based on themeasurement of field-size-dependent scatter factors obtainedusing an ion chamber at the depth of 1.5 cm. Although theinitial results are encouraging for MERT QA, the approachis only valid for square and rectangular fields and cannotcorrect for the discrepancy existing in the measurement.15

The accuracy of this measurement-based calibration anddosimetric procedure is limited for general electrondosimetry application, and large discrepancies have beenobserved.15 As has been discussed in a number of previouspublications in photo dosimetry,4,19 a kernel-based calibrationof the EPID system is essential for accurate radiationdosimetry.

This work is devoted to establishing a robust methodof using a high spatial-resolution a-Si EPID for outputmeasurement and dosimetric verification of EBT. Byincorporation of Monte Carlo (MC)-derived pixel-by-pixelresponse of the EPID, the underlying physical propertiesand dose deposition process are considered. Coupled witha number of unique features of EPID dosimetry, such asease-to-use, high spatial and temporal resolution and digitalreadouts, the system provides an urgently needed dosimetrytool for EBT. To the best of our knowledge, this represents the

first attempt to develop a convolution method with inclusion ofscattering kernels derived from MC simulations. In Sec. 2.A,we describe the EPID data acquisition process and systemcalibration procedure. The measurement methods of a fewimportant dosimetric characteristics of the system, includingsaturation, linearity, and ghosting effect, are presented inSec. 2.B. Issues related to the implementation of the proposedmethod and electron dosimetry are discussed in Secs. 2.C–2.E.We conclude in Sec. 4 with a summary of the study and futureperspectives of EPID-based electron dosimetry.

2. MATERIALS AND METHODS2.A. Overall process of data acquisition

A Clinac 21EX accelerator (Varian Medical Systems, PaloAlto, CA) with five available electron energies of 6, 9, 12,16, and 20 MeV was used for the study. Instead of usingthe onboard EPID, a portable independent PerkinElmer, a-Siflat panel detector (PerkinElmer, Inc., Sunnyvale, CA) wasadapted because of its high spatial and temporal resolution.The size of the detector was 20.48× 20.48 cm2 with theminimum pixel size of 0.2 mm. The maximum frame rateof the portable EPID was 50 frames per second (fps). Thedetector was connected to a host computer through a GigabitEthernet network cable. The images were acquired in a “cine-mode” so that each individual frame was recorded over theentire beam-on time. An average frame was also automaticallycalculated and saved separately after each measurement.

As shown in Fig. 1, electron beams were delivered to theEPID in a fixed source–surface distance (SSD) of 100 cm.No buildup material was placed on the top of the EPID. Allmeasurements were carried out with gantry at 0◦ position.

Before data acquisition, background and flood imageswere required for offset and gain corrections. The offset

F. 1. Experimental setup of a portable EPID detector.

Medical Physics, Vol. 42, No. 7, July 2015

4192 Ding, Xing, and Han: An accurate EPID-based electron beam output measurement and dosimetric verification tool 4192

correction was performed with the electron beam off to takeinto account the dark current of each pixel. For reliable offsetcorrection, an averaged image (EPIDDF) of 100 frames ofdark field (DF) images was obtained, and the EPIDDF wasthen subtracted from the EPID data. To homogenize thedifference in sensitivity of the pixels, an additional correctionof the EPID image, flood-field (FF) correction, was carriedout before data analysis. In this correction, the EPID wasirradiated with incident electrons covering the entire imagingplane (20.48×20.48 cm2). A total of 300 frames of suchimages were acquired and averaged to create the FF correctionimage (EPIDFF). The DF- and FF-corrected image was givenby

EPIDraw|corrected=EPIDraw−EPIDDF

EPIDFF−EPIDDF. (1)

2.B. Dosimetric characteristics of the EPID

2.B.1. Saturation behavior and linearity of the EPID

The EPID used in this study had a high resolution (0.2 mm)and high readout speed (50 fps), ideal for high dose ratebeams. When acquiring images, the detector was operated ina “free running” mode (cine-mode) which sent out framescontinuously according to the selected frame collecting time.To investigate the saturation behavior of the detector, weperformed a series of measurements with a dose rate of600 MU/min.

The linearity of EPID response with respect to the monitorunits (MUs) was assessed for all electron energies. A conesize of 15×15 cm2 was used with a range of MUs from10 to 500 MU and a dose rate of 600 MU/min at a SSDof 100 cm. Each measurement was performed three timesand the images were measured in a cine-mode over eachbeam-on time. The EPID response (REPID,MU) was evaluatedby averaging the detector signal over the central 1×1 cm2

square area (50×50 pixels). To reduce the effect of ghosting,the time interval between two subsequent measurements wasset to 2 min.

2.B.2. Ghosting effect

Ghosting effect (gain ghosting and image lag) of an a-SiEPID is an image artifact caused by the trapped chargesin the photodiodes and a signal delay.20,21 This effect wasexamined by exposing the EPID to a 15×15 cm2 field withtwo subsequent sets of measurements of known time intervals.It is known that the ghosting effect depends on the number ofMUs and becomes more pronounced for short time intervals.22

Thus, various MUs with a time interval of 1 min between twosubsequent measurements were investigated to examine theghosting effect in this study. The pixel values in the centralsquare field of 2×2 cm2 (100×100 pixels) were averaged,and the value was used to evaluate the variations of the EPIDsignals as a result of earlier exposure. To study the decay ofthe signal, measurements were continued even after the beamwas off.

2.C. Conversion of EPID images to incidentelectron fluence

The scintillator layer used in the EPID to convert theincident radiation beam to optical signal is made of phosphorterbium-doped gadolinium oxysulphide (Gd2O2S:Tb). Todetermine the incident electron beam fluence, it is necessary tocalibrate the EPID to establish a relationship between EPIDpixel values and radiation dose.23 We extended previouslyestablished photon EPID dosimetry system for electrondosimetry.12 With detailed structure and composition ofthe EPID, 4 application for tomographic emission() Monte Carlo software24 was used to model thepixel response of the EPID. The source model of electronenergy spectrum used in the MC simulations was basedon the energy integration method and was evaluated usingtreatment planning systems (TPS) commissioning beam data,as illustrated in Fig. 2. To accurately simulate the physicsresponse of the EPID to electron beams, the optical photonstracking function was activated in the MC simulationprocess. The physical process of electron beam in the EPIDdosimetry system was accurately simulated from the builduplayer. Then the energy deposition in the GOS scintillatorplate was recorded and the generation of optical photonsinitiated. Finally, the optical simulation module in the simulated the transport of the optical photon in fibers andtallied the absorption of the optical photon in the amorphoussilicon active thin-film transistor (TFT)/diode array. Theoptical properties such as the surface type and refractiveindex were defined and stored in a table for the simulation.Optical photons were detected by using a dielectric-metalboundary and a digitizer was set up to record and analyzethe optical absorption in the . With this detail, EPIDmodeled and simulated using , for each energy used inthis study, a deconvolution kernel Kde(x,y) was generated toaccount for the underlying bremsstrahlung, dose depositionin the Gd2O2S:Tb screen, and the optical photon scatteringprocess. To maximum deblur the EPID-measured images, anaccelerated and damped iterative deconvolution method wasdeveloped based on the Richardson–Lucy (RL) algorithm.25

The RL algorithm maximizes the likelihood to estimate the

F. 2. Electron energy spectrums used in the Monte Carlo simulations.

Medical Physics, Vol. 42, No. 7, July 2015

4193 Ding, Xing, and Han: An accurate EPID-based electron beam output measurement and dosimetric verification tool 4193

original signal when deconvolved with the Monte Carlogenerated Kde(x,y).26 The incident electron fluence Ψe(x,y)on the EPID was therefore reconstructed from the correctedEPID raw image and the Kde(x,y) through the followingequation:

Ψe(x,y)= (EPIDraw���corrected

(x,y)) ⊗−1 (Kde(x,y)). (2)

2.D. Conversion of the reconstructed incident fluenceto dose in a solid water phantom

Water, plastic/solid water slabs, or phantoms are commonlyused for many dosimetry measurements and routine checksbecause of the ease of set up and the reproducibility ofdetector placement. As the EPID is not water-equivalent,the incident electron fluence reconstructed from the EPIDwas converted to dose in a solid water phantom by using apencilbeam dose kernel Kpb(x,y) generated with the Xcode (version 2.6).27 The number of source electrons wasselected in such a way that an acceptable level of statisticaluncertainty (<1% at 3 cm laterally off the pencil beam axisand <3% at 10 cm laterally off the axis for each simulationwere achieved). The incident electron fluence Ψe(x,y) fora given field was convolved with the Kpb(x,y) to generatea two-dimensional (2D) relative dose distribution in a solidwater equivalent phantom at a reference depth (which wasprechosen depending on the electron energy) using

Dsw(x,y)=Ψe(x,y)⊗Kpb(x,y). (3)

All deconvolution and convolution algorithms weredeveloped in-house and implemented using software(The Mathworks, Inc., Natick, MA). The electron beams fromthe Varian Clinac 21EX accelerator used in this study werecalibrated to deliver 1 cGy/MU at the depth of dose maximum(dmax) for a 15×15 cm2 reference field with a nominal SSDof 100 cm. To determine the absolute dose, 100 MU (100cGy) was delivered to EPID with a SSD of 100 cm for the15×15 cm2 reference field. The EPID-measured image datawere recorded as DEPID, and a calibration factor FABS for

F. 3. The pixel values in a central square field of 2×2 cm2 as a function oftime. The cine-mode (33 ms/frame) was used to record all the signal frames.

F. 4. The in-plane MC generated dose-glare kernel Kde(x, y) used in thedeconvolution of EPID-measured raw images to the incident primary fluence.

each energy was then obtained by calculating the ratio of 100cGy and DEPID(FABS= 100 cGy/DEPID). The calculated FABSwas used to convert the EPID-measured relative dose to theabsolute dose in a solid water phantom.

2.E. Output factor and dosimetricverification procedure

2.E.1. Output factor measurements

For an electron beam with a given energy, the output factoris defined as the ratio of a dose of a given field size (orapplicator size) to the dose of a reference field (15×15 cm2)at the depths near dmax with a nominal SSD of 100 cm.Because of the rapid dose falloff of the electron depth-dosecurves, measurements of the output factor of an electron beamdiffer from that of a photon beam. For small fields, there isa general lack of lateral scatter equilibrium and the dmaxvaries extensively with the shape of the field.2,3 This makes itdifficult to determine the output factor of an electron beam.Moreover, measurements at depths greater than the dmax couldhave unacceptable uncertainties.2,3

In this study, the output factors were all measured withthe EPID system at the depth near dmax. The results were

F. 5. The in-plane pencil beam kernel Kpb(x, y) used in the reconstructionof water-equivalent dose distribution from the primary electron fluence.

Medical Physics, Vol. 42, No. 7, July 2015

4194 Ding, Xing, and Han: An accurate EPID-based electron beam output measurement and dosimetric verification tool 4194

F. 6. The entire process of dose reconstruction from the raw data of EPID to the dose results in a solid water phantom using Monte Carlo simulated kernels.

compared with those obtained using films and a Farmer-type ion chamber. Standard cones in three sizes of 15×15,10×10, and 6×6 cm2 were used in the measurement. Forsquare fields of 15× 15, 10× 10, and 6× 6 cm2, standardcutouts and cones with the same sizes were used. For 3×3and 2×2 cm2 fields, custom-made square cerrobend cutoutswere adopted with a 6×6 cm2 cone. Square fields of 2×2to 15×15 cm2 were measured and compared against filmdosimetry measurements. For field sizes larger than 6×6 cm2,ion chamber measurements were also performed, providingadditional validation of the system. In these film and ionchamber measurements, the reference depth in a solid waterphantom was 1.4 cm for 6 MeV, 2.0 cm for 9 MeV, 3.0 cmfor 16 MeV, and 2.5 cm for 12 and 20 MeV, respectively.Electron fields of three clinical cases with irregular cutoutshapes were also measured using EPID and compared withthe measurements done with an ion chamber array (PTW729).

2.E.2. Flatness and symmetry measurements

Uniformity of beam intensity across the electron-beam fieldrepresents an important characteristic of electron beams.2,28 Inroutine machine QA, the uniformity is examined by measuringelectron beam flatness and symmetry. The former is definedas (100× (Dmax−Dmin/Dmax+Dmin)) within a specific area,where Dmax and Dmin are the maximum and minimum beamintensities measured along a major axis.2,28 Electron beamsymmetry is defined as

�100×2×

�Dleft−Dright/Dleft+Dright

��,

where Dleft and Dright are the respective intensities measured atpoints equidistant from the central axis along a major axis ofthe beam.2,28 The value of symmetry is chosen at the distancefor which the difference between Dleft and Dright is greatest.The area is defined as 80% of the actual field size.

The EPID dosimetry system was applied to check theflatness and symmetry of the electron beams. For this

F. 7. Measured output factors of 2×2, 3×3, 6×6, 10×10, and 15×15 cm2 fields for all electron energies (6, 9, 12, 16, and 20 MeV) with EPID, film, and anion chamber. The average discrepancy between EPID and ion chamber/film measurements was 0.81%±0.60% (ion chamber) and 1.34%±0.75% (film).

Medical Physics, Vol. 42, No. 7, July 2015

4195 Ding, Xing, and Han: An accurate EPID-based electron beam output measurement and dosimetric verification tool 4195

T I. Measurements of flatness and symmetry using PTW729, water-scan, and EPID for 6, 12, and 20 MeV electrons, respectively, of 10×10 cm2 with aapplicator of 10×10 cm2.

Flatness (%) Symmetry (%)

Energy (MeV) Applicator (cm2) Axis PTW729 Water-scan EPID PTW729 Water-scan EPID

610×10

Radial 4.59 (0.01) 4.52 (0.03) 4.20 (0.01) 0.10 (0.00) 0.64 (0.00) 0.01 (0.00)Transverse 4.75 (0.01) 4.41 (0.02) 3.80 (0.01) 0.06 (0.00) 0.38 (0.00) 0.02 (0.00)

12 10×10Radial 4.34 (0.01) 3.77 (0.02) 4.04 (0.01) 0.17 (0.00) 1.40 (0.01) 0.01 (0.00)Transverse 4.17 (0.01) 3.70 (0.03) 3.97 (0.01) 0.16 (0.00) 0.07 (0.00) 0.08 (0.00)

20 10×10Radial 1.80 (0.00) 1.88 (0.01) 1.12 (0.00) 0.38 (0.00) 0.18 (0.00) 0.27 (0.01)Transverse 1.54 (0.00) 1.69 (0.01) 1.51 (0.01) 0.08 (0.00) 1.79 (0.01) 0.22 (0.00)

application, the EPID-measured 2D dose distribution wasanalyzed for the flatness and symmetry along the radial axis(in-plane) and the transverse axis (cross-plane), and the resultswere compared with the measurements done with water scanand PTW729 ion chamber array (PTW, Freiburg, Germany).

3. RESULTS AND DISCUSSION3.A. Dosimetric characteristics of the EPID system

3.A.1. Saturation and linearity

All acquired images used unsigned 16 bit integer to storethe pixel data in our system. It was found that saturation likelyoccurred when pixel values exceeding the maximum value of65 535. An appropriate image capture setting is essential toavoid signal saturation from happening, especially for highdose rate beam and long acquisition time. In this study, ashorter integration time of 33 millisecond (ms)/frame wasselected for the dose rate of 600 MU/min (i.e., a frame was

taken and read out every 33 ms). With this setting, REPID,MUwas found to be proportional to the amount of MUs with therelative standard deviation of the REPID,MU per MU less than0.2%. The excellent linearity forms the basis for quantitativedosimetry of the system.

3.A.2. Ghosting effect

When exposing the EPID to a 15×15 cm2 field with a doserate of 600 MU/min at 100 cm SSD, the two subsequent setsof signals of the central 2×2 cm2 square field were assessedto determine the influence of the first exposure on the secondone. The time interval between two subsequent measurementswas 1 min and the maximum increase in detector sensitivitybetween two subsequent measurements was found to be lessthan 0.17% for all available energies (6, 9, 12, 16, and20 MeV). The fluctuation of pixel value from frame-to-frameduring the beam-on was caused mainly by the time of delayin the selected integration time (Fig. 3). The inset of Fig. 3

F. 8. EPID-, PTW729-, and water-scan measured absolute dose profiles of 10×10 cm2 field for 6, 12, and 20 MeV electron beams, respectively. The radialand transverse profiles were shown in the first and second row, respectively. The profiles agreed well with small discrepancies (less than 5%) observed in theshoulder and trail regions of the profiles for 20 MeV.

Medical Physics, Vol. 42, No. 7, July 2015

4196 Ding, Xing, and Han: An accurate EPID-based electron beam output measurement and dosimetric verification tool 4196

F. 9. (a) Three patient cases with different cutout sizes. (b) PTW-measured 2D absolute dose maps. (c) EPID reconstructed 2D absolute dose maps. Thehorizontal and vertical black solid lines in the dose maps indicated the chosen planes for profile comparisons. (d) Selected radial relative dose profiles from thereconstructed EPID dose images (red solid lines) and from the PTW729 measurements (dark green diamonds). (e) Selected transverse relative dose profiles fromthe reconstructed EPID dose images (red solid lines) and from the PTW729 measurements (dark green diamonds).

showed that the pixel values in a central square field of2×2 cm2 dropped rapidly after beam off and the residualsignal was found to be below 0.5% within 1 s.

3.B. Monte Carlo simulation of pixel-by-pixelresponse of the EPID

A deconvolving dose kernel Kde(x,y), accounting for theunderlying bremsstrahlung, dose deposition in EPID screenand the optical photon scattering process, was generated fromtheMC simulations. Figure 4 shows the change of Kde(x,y)

as a function of distance from the central axis for a 6 MeV elec-tron beam. The incident electron fluence on the EPID wastherefore reconstructed from the corrected EPID raw imagesusing Eq. (2).

To obtain the dose in a solid water phantom from the EPIDimaging data, a series of pencilbeam dose kernels Kpb(x,y)were generated from the X MC simulations. Figure 5shows an example of a pencil beam kernel Kpb(x,y) profilescored in a solid water phantom. Figure 6 shows the entireprocess of dose reconstruction from the raw data of EPID tothe dose results in a solid water phantom.

Medical Physics, Vol. 42, No. 7, July 2015

4197 Ding, Xing, and Han: An accurate EPID-based electron beam output measurement and dosimetric verification tool 4197

T II. Percent difference between EPID- and PTW729- measured output factors for the three patient cases.Relative dose output difference, SD, and γ indices with two criteria of 3%/3 mm and 2%/2 mm were alsopresented.

Percent difference between EPID- andPTW729- measured output factors (%)

γ index analysis

Patient-case No. 3% (local)/3 mm (%) 2% (local)/2 mm (%)

1 1.13 (0.11) 100 99.0 (0.07)2 0.54 (0.10) 100 98.2 (0.14)3 0.74 (0.11) 100 100 (0.00)

3.C. Output factor and dosimetric measurements

3.C.1. Output factors

The EPID-measured output factors for various field sizesat a SSD of 100 cm are plotted in Fig. 7 along with thatobtained using film and ion chamber. Overall, it is found thatthe average discrepancy between EPID and ion chamber/filmoutput factor measurements was 0.81%±0.60% (ion chamber)and 1.34%±0.75% (film), respectively. The maximum differ-ence was found to be 2.60% with the film for the measurementof 12 MeV with a 2×2 cm2 field. The discrepancies weremostly attributed to the loss of side-scatter electron equi-librium with the decrease in the field size and the increase inelectron energy.29

To further validate the dosimetric verification methoddeveloped in this study, an additional set of output mea-surements was performed at two extended SSDs of 105 and110 cm for all five available energies (6, 9, 12, 16, and 20 MeV)with a cone size of 15×15 cm2. The results were comparedwith the measurements done with a Farmer-type ionizationchamber. The average difference between the results obtainedby the EPID and ion chamber was found to 1.11%±0.39%and 0.71%±0.23% for 6, 9, 12, 16, and 20 MeV electronbeams.

3.C.2. Flatness and symmetry measurements

The results of EPID-, water-scan, and PTW729-meausreddose flatness and symmetry in radial and transverse directionsare listed in Table I. The flatness resulted from EPID wascompatible with that from water-scan and PTW729 measure-ments. Only a small average difference of 0.40%±0.23% (withwater-scan) and 0.43%± 0.34% (with PTW729) was found.The bilateral system data of the EPID measurements werealso consistent with that obtained using PTW729 and film.As compared to the water phantom scan data, the maximumdiscrepancy was less than 1.57% for the worst scenario of 20MeV electron in the transverse axis direction.

Figure 8 shows that the EPID-measured radial and trans-verse dose profiles of a 10×10 cm2 field for three electronenergies (6, 12, and 20 MeV). The profile data obtained usinga PTW729 detector and water tank scan are also plotted inFig. 8 for comparisons. Overall, the profiles obtained usingdifferent ways agreed well including the penumbra regions.Small discrepancies (less than 5%) were also observed in theshoulder and trail regions of the profiles for 20 MeV beam,presumably because of the undesirable over response of EPID

to low energy particles at off-axis positions, as observed inother studies.30–32

3.D. Clinical cases with irregular cutout shapes

In practice, a cutout is used to shape the electron beamto conform the shape of the tumor target. The three patientcases [Fig. 9(a)] with different cutouts and cones of 15× 15and 10×10 cm2 were delivered using a 6 MeV electron beamand a dose rate of 600 MU/min at a SSD of 100 cm. Usingthe proposed method in this study, 2D absolute dose mapswere reconstructed [Fig. 9(c)] from the EPID raw images, andthe results were compared with those [Fig. 9(b)] measuredusing a PTW729 ion chamber array. The EPID-based methodexhibited a better resolution than the PTW729. In Figs. 9(d)and 9(e), selected radial and transverse relative dose profilesfrom the reconstructed EPID dose images and from the PTWmeasurements were plotted. The differences in relative doseoutput for the three patient cases were summarized in Table II.The relative dose output difference for the three patient caseswas 1.13% ± 0.11%, 0.54% ± 0.10%, and 0.74% ± 0.11%,respectively. To further evaluate the potential of the EPID,γ-index criteria of 3% (local)/3 mm and 2% (local)/2 mmwere calculated. The results, as presented in Table II, showedthat 100% of the pixel passed the criteria of 3% (local)/3 mm.For the 2% (local)/2 mm γ criterion, 99.0%±0.07%, 98.2%±0.14%, and 100% of the points passed.

A further test on a more problematic cutout of a 3×9 cm2

size with a 15×15 cm2 cone for 6 MeV was measured usingthe EPID. The results were compared with the measurementsdone with a PTW 31014 PinPoint ionization chamber (PTW,Freiburg, Germany). It was found that the relative dose outputdifference was small (∼1.76%). As shown in Fig. 7, we hadalready validated even small cutouts (e.g., 2×2 and 3×3 cm2)with a 6×6 cm2 cone and results agreed well with the filmmeasurements, indicating the success of our method in dealingwith different cutouts.

4. CONCLUSIONS

In this study, dosimetric properties of a portable EPIDwere investigated for the purpose of adapting EPID for outputmeasurement and dosimetric verification of EBT. The EPIDexhibited a good linearity and negligible ghosting effect. Theflatness and symmetry measured with the EPID were foundto be consistent with PTW729 and water phantom scan data.The EPID-measured output factors for standard square fields

Medical Physics, Vol. 42, No. 7, July 2015

4198 Ding, Xing, and Han: An accurate EPID-based electron beam output measurement and dosimetric verification tool 4198

of 2×2, 3×3, 6×6, 10×10, and 15×15 cm2 were found toagree with film and ion chamber data. Different patient caseswith various field sizes and irregular cutout shapes were alsoexamined. Small differences (<1.13%± 0.11%) between theEPID- and PTW729- measured output values were observed.The success of EPID-based system was also supported by theγ index analysis. Consequently, it can be concluded that thehigh frame-rate, high spatial resolution, and portability of thenewly developed EPID dosimetric system provide an effi-cient and reliable solution for routine, pretreatment dosimetricverification of EBT. It addresses an important unmet clinicalneed for a fast and reliable small and irregular field outputmeasurement and dosimetric verification.

ACKNOWLEDGMENTS

The authors wish to thank Dr. Minghui Lu from Perkin-Elmer, Inc. (Sunnyvale, CA) for technical assistance regardingPerkinElmer a-Si flat panel detector and for useful discussions.This work was supported in part by the National Institutes ofHealth (NIH) (Nos. 1R01 EB016777 and 1R01 CA176553).

a)Author to whom correspondence should be addressed. Electronic mail:hanbin@stanford.edu

1E. C. Halperin, L. W. Brady, C. A. Perez, and D. E. Wazer, Perez and Brady’sPrinciples and Practice of Radiation Oncology (Wolters Kluwer| LippincottWilliams & Wilkins, Philadelphia, PA, 2013).

2F. M. Khan, K. P. Doppke, K. R. Hogstrom, G. J. Kutcher, R. Nath, S. C.Prasad, J. A. Purdy, M. Rozenfeld, and B. L. Werner, “Clinical electron-beam dosimetry: Report of AAPM Radiation Therapy Committee TaskGroup No. 25,” Med. Phys. 18, 73–109 (1991).

3B. J. Gerbi, J. A. Antolak, F. C. Deibel, D. S. Followill, M. G. Herman,P. D. Higgins, M. S. Huq, D. N. Mihailidis, E. D. Yorke, K. R. Hogstrom,and F. M. Khan, “Recommendations for clinical electron beam dosimetry:Supplement to the recommendations of Task Group 25,” Med. Phys. 36,3239–3279 (2009).

4B. Warkentin, S. Steciw, S. Rathee, and B. G. Fallone, “Dosimetric IMRTverification with a flat-panel EPID,” Med. Phys. 30, 3143–3155 (2003).

5A. Van Esch, T. Depuydt, and D. P. Huyskens, “The use of an aSi-basedEPID for routine absolute dosimetric pre-treatment verification of dynamicIMRT fields,” Radiother. Oncol. 71, 223–234 (2004).

6M. van Zijtveld, M. L. Dirkx, H. C. de Boer, and B. J. Heijmen, “3D dosereconstruction for clinical evaluation of IMRT pretreatment verification withan EPID,” Radiother. Oncol. 82, 201–207 (2007).

7T. Fuangrod, H. C. Woodruff, P. Rowshanfarzad, D. J. O’Connor, R. H.Middleton, and P. B. Greer, “An independent system for real-time dynamicmultileaf collimation trajectory verification using EPID,” Phys. Med. Biol.59, 61–81 (2014).

8G. Asuni, T. A. van Beek, S. Venkataraman, I. A. Popescu, and B. M.McCurdy, “A Monte Carlo tool for evaluating VMAT and DIMRT treatmentdeliveries including planar detectors,” Phys. Med. Biol. 58, 3535–3550(2013).

9M. Bakhtiari, L. Kumaraswamy, D. W. Bailey, S. de Boer, H. K. Malhotra,and M. B. Podgorsak, “Using an EPID for patient-specific VMAT qualityassurance,” Med. Phys. 38, 1366–1373 (2011).

10A. Mans, P. Remeijer, I. Olaciregui-Ruiz, M. Wendling, J. J. Sonke, B.Mijnheer, M. van Herk, and J. C. Stroom, “3D dosimetric verification ofvolumetric-modulated arc therapy by portal dosimetry,” Radiother. Oncol.94, 181–187 (2010).

11H. C. Woodruff, T. Fuangrod, P. Rowshanfarzad, B. M. McCurdy, and P. B.Greer, “Gantry-angle resolved VMAT pretreatment verification using EPIDimage prediction,” Med. Phys. 40, 081715 (9pp.) (2013).

12B. Han, G. Luxton, S. Yu, M. Lu, L. Wang, E. Mok, and L. Xing, “SU-C-105-06: Development of a high resolution EPID solution for small fielddosimetry,” Med. Phys. 40, 84 (2013).

13G. Jarry and F. Verhaegen, “Electron beam treatment verification usingmeasured and Monte Carlo predicted portal images,” Phys. Med. Biol. 50,4977–4994 (2005).

14J. A. Beck, G. J. Budgell, D. A. Roberts, and P. M. Evans, “Electronbeam quality control using an amorphous silicon EPID,” Med. Phys. 36,1859–1866 (2009).

15C. Chatelain, D. Vetterli, D. Henzen, P. Favre, D. Morf, S. Scheib, M. K.Fix, and P. Manser, “Dosimetric properties of an amorphous silicon EPIDfor verification of modulated electron radiotherapy,” Med. Phys. 40, 061710(14pp.) (2013).

16J. Chang, G. S. Mageras, C. C. Ling, and W. Lutz, “An iterative EPIDcalibration procedure for dosimetric verification that considers the EPIDscattering factor,” Med. Phys. 28, 2247–2257 (2001).

17C. Talamonti, M. Casati, and M. Bucciolini, “Pretreatment verification ofIMRT absolute dose distributions using a commercial a-Si EPID,” Med.Phys. 33, 4367–4378 (2006).

18J. Chen, C. F. Chuang, O. Morin, M. Aubin, and J. Pouliot, “Calibration of anamorphous-silicon flat panel portal imager for exit-beam dosimetry,” Med.Phys. 33, 584–594 (2006).

19P. Winkler, A. Hefner, and D. Georg, “Implementation and validation ofportal dosimetry with an amorphous silicon EPID in the energy range from6 to 25 MV,” Phys. Med. Biol. 52, N355–N365 (2007).

20L. N. McDermott, R. J. Louwe, J. J. Sonke, M. B. van Herk, and B. J.Mijnheer, “Dose–response and ghosting effects of an amorphous siliconelectronic portal imaging device,” Med. Phys. 31, 285–295 (2004).

21L. N. McDermott, S. M. Nijsten, J. J. Sonke, M. Partridge, M. van Herk, andB. J. Mijnheer, “Comparison of ghosting effects for three commercial a-SiEPIDs,” Med. Phys. 33, 2448–2451 (2006).

22P. Winkler, A. Hefner, and D. Georg, “Dose–response characteristics of anamorphous silicon EPID,” Med. Phys. 32, 3095–3105 (2005).

23W. van Elmpt, L. McDermott, S. Nijsten, M. Wendling, P. Lambin, and B.Mijnheer, “A literature review of electronic portal imaging for radiotherapydosimetry,” Radiother. Oncol. 88, 289–309 (2008).

24S. Jan, G. Santin, D. Strul, S. Staelens, K. Assie, D. Autret, S. Avner, R.Barbier, M. Bardies, P. M. Bloomfield, D. Brasse, V. Breton, P. Bruyndon-ckx, I. Buvat, A. F. Chatziioannou, Y. Choi, Y. H. Chung, C. Comtat, D.Donnarieix, L. Ferrer, S. J. Glick, C. J. Groiselle, D. Guez, P. F. Honore,S. Kerhoas-Cavata, A. S. Kirov, V. Kohli, M. Koole, M. Krieguer, D. J. vander Laan, F. Lamare, G. Largeron, C. Lartizien, D. Lazaro, M. C. Maas, L.Maigne, F. Mayet, F. Melot, C. Merheb, E. Pennacchio, J. Perez, U. Pietrzyk,F. R. Rannou, M. Rey, D. R. Schaart, C. R. Schmidtlein, L. Simon, T. Y.Song, J. M. Vieira, D. Visvikis, R. Van de Walle, E. Wieers, and C. Morel,“GATE: A simulation toolkit for PET and SPECT,” Phys. Med. Biol. 49,4543–4561 (2004).

25D. S. Biggs and M. Andrews, “Acceleration of iterative image restorationalgorithms,” Appl. Opt. 36, 1766–1775 (1997).

26L. A. Shepp and Y. Vardi, “Maximum likelihood reconstruction for emissiontomography,” IEEE Trans. Med. Imaging 1, 113–122 (1982).

27D. B. Pelowitz, X User’s Manual, Los Alamos National Laboratory,Los Alamos, NM, 2005.

28R. Nath, P. J. Biggs, F. J. Bova, C. C. Ling, J. A. Purdy, J. van de Geijn,and M. S. Weinhous, “AAPM code of practice for radiotherapy accelerators:Report of AAPM Radiation Therapy Task Group No. 45,” Med. Phys. 21,1093–1121 (1994).

29G. G. Zhang, D. W. Rogers, J. E. Cygler, and T. R. Mackie, “Monte Carloinvestigation of electron beam output factors versus size of square cutout,”Med. Phys. 26, 743–750 (1999).

30P. Vial, P. B. Greer, P. Hunt, L. Oliver, and C. Baldock, “The impact of MLCtransmitted radiation on EPID dosimetry for dynamic MLC beams,” Med.Phys. 35, 1267–1277 (2008).

31M. Wendling, R. J. W. Louwe, L. N. McDermott, J. J. Sonke, M. van Herk,and B. J. Mijnheer, “Accurate two-dimensional IMRT verification using aback-projection EPID dosimetry method,” Med. Phys. 33, 259–273 (2006).

32P. B. Greer, “Correction of pixel sensitivity variation and off-axis responsefor amorphous silicon EPID dosimetry,” Med. Phys. 32, 3558–3568 (2005).

Medical Physics, Vol. 42, No. 7, July 2015