Australas Phys Eng Sci Med DOI 10.1007/s13246-015-0401-2
SCIENTIFIC NOTE
Use of electronic portal imaging devices for electron treatment verification T. Kairn1,2 • T. Aland1,2 • S. B. Crowe2,3 • J. V. Trapp2
Received: 7 August 2015 / Accepted: 4 November 2015 Ó Australasian College of Physical Scientists and Engineers in Medicine 2015
Abstract This study aims to help broaden the use of electronic portal imaging devices (EPIDs) for pre-treatment patient positioning verification, from photon-beam radiotherapy to photon- and electron-beam radiotherapy, by proposing and testing a method for acquiring clinicallyuseful EPID images of patient anatomy using electron beams, with a view to enabling and encouraging further research in this area. EPID images used in this study were acquired using all available beams from a linac configured to deliver electron beams with nominal energies of 6, 9, 12, 16 and 20 MeV, as well as photon beams with nominal energies of 6 and 10 MV. A widely-available heterogeneous, approximately-humanoid, thorax phantom was used, to provide an indication of the contrast and noise produced when imaging different types of tissue with comparatively realistic thicknesses. The acquired images were automatically calibrated, corrected for the effects of variations in the sensitivity of individual photodiodes, using a flood field image. For electron beam imaging, flood field EPID calibration images were acquired with and without the placement of blocks of water-equivalent plastic (with thicknesses approximately equal to the practical range of electrons in the plastic) placed upstream of the EPID, to
filter out the primary electron beam, leaving only the bremsstrahlung photon signal. While the electron beam images acquired using a standard (unfiltered) flood field calibration were observed to be noisy and difficult to interpret, the electron beam images acquired using the filtered flood field calibration showed tissues and bony anatomy with levels of contrast and noise that were similar to the contrast and noise levels seen in the clinically acceptable photon beam EPID images. The best electron beam imaging results (highest contrast, signal-to-noise and contrast-to-noise ratios) were achieved when the images were acquired using the higher energy electron beams (16 and 20 MeV) when the EPID was calibrated using an intermediate (12 MeV) electron beam energy. These results demonstrate the feasibility of acquiring clinically-useful EPID images of patient anatomy using electron beams and suggest important avenues for future investigation, thus enabling and encouraging further research in this area. There is manifest potential for the EPID imaging method proposed in this work to lead to the clinical use of electron beam imaging for geometric verification of electron treatments in the future. Keywords
Electrons Portal imaging Radiation therapy
A case study based on this work was presented at the Winter School on Scientific Publication, Australasian College of Physical Scientists and Engineers in Medicine (Queensland Branch), Brisbane, 2014. & T. Kairn [email protected] 1
Genesis Cancer Care Queensland, Brisbane, Australia
2
Science and Engineering Faculty, Queensland University of Technology, Brisbane, Australia
3
Cancer Care Services, Royal Brisbane and Women’s Hospital, Brisbane, Australia
Introduction Electronic portal imaging devices (EPIDs) are widely used for pre-treatment patient positioning verification, before and during megavoltage photon radiotherapy treatment delivery [1–3]. This study aims to help broaden their use to the geometric verification of electron radiotherapy treatment delivery.
123
Australas Phys Eng Sci Med
The amorphous-silicon (a-Si) flat panel EPIDs that augment most contemporary medical linear accelerators generate images using a metal plate and phosphor screen, immediately above (upstream of) an array of a-Si photodiodes and transistors. An early review of EPID technologies by Boyer et al explains that ‘‘X-ray interactions in the metal plate create high-energy electrons that in turn produce fluorescence in the phosphor screen’’ and provides a detailed description of the process by which the resulting distribution of optical photons is detected and read out via the a-Si layer [4]. Fundamentally, EPIDs are designed for the rapid acquisition of images produced by the interactions of megavoltage photons. Contemporary EPIDs can read out image data at a rate of several frames per second without dead-time between frames [5] and the automatic averaging of all frames acquired during each exposure [6] can lead to the production of relatively high quality images of photon treatment portals and the intervening patient anatomy [7, 8]. While still primarily used for patient setup imaging before and during photon beam radiotherapy treatments [2, 3], EPIDs are increasingly being investigated and used for alternative applications including: the generation of threedimensional (cone-beam CT) images [9–11]; evaluation of the water-equivalence of plastics and identification of metal implants [12]; routine photon beam constancy (including flatness and symmetry) checks [13, 14]; routine geometric testing of the radiation isocentre (including radiation isocentre size and coincidence with imaging isocentres) [15]; pre-treatment quality assurance of modulated radiotherapy treatment plans [16–21]; and in vivo verification of treatment delivery accuracy [22–27]. All of these applications utilise EPID images acquired using megavoltage photon beams. By contrast, references to EPID images acquired using electron beams are sparse in the literature. Use of EPIDs for electron-beam imaging has largely been limited to the investigation of the electron beams themselves; EPID images have been used to investigate the flatness and symmetry of static electron beams [28, 29] and the deliverability of modulated electron beams produced using purpose-built electron MLCs [30], for potential use in electron IMRT treatments [31]. To date, published evaluations of the EPID image quality achievable using standard, static electron radiotherapy beams have been rare [32, 33]. Jarry and Verhaegen [32] used 6–16 MeV electron beams to image a QC-3 (PipsPro) phantom (Standard Imaging, Middleton, USA) placed between slabs of water equivalent plastic, to investigate EPID image quality and demonstrate agreement with Monte Carlo modelling. Importantly, their Monte Carlo simulations of the linacphantom-EPID system showed that EPID image contrast
123
and resolution could be noticeably improved if the electron component of the imaging beam could be excluded from the calculation, leaving only the bremsstrahlung photon contribution [32]. Jarry and Verhaegen also provided examples of the use of a 9 MeV electron beam to image a head phantom, suggesting (without explicitly quantifying the contrast and noise in these images) that the appearance of bony anatomy in electron beam images might be similar to the appearance of bony anatomy in 6 MV photon beam images [32]. This study proposes a method for acquiring clinicallyuseful EPID images of patient anatomy using electron beams, which specifically evaluates image contrast in a patient-like geometry, while taking advantage of Jarry and Verhaegen’s Monte-Carlo-based observations of improved image quality in electron beams from which the electrons have been removed. This study thereby provides initial feasibility testing of the potential capabilities and limitations of EPID imaging with electron beams, with a view to enabling and encouraging further research in this area.
Method EPID images used in this study were acquired using a Varian iX linear accelerator (linac) with a Varian IDU20 aS1000 IAS3 Portal Vision system operated in service mode via the ‘AM Maintenance’ application (Varian Medical Systems, Palo Alto, USA). The linac was configured to deliver electron beams with nominal energies of 6, 9, 12, 16 and 20 MeV, as well as photon beams with nominal energies of 6 and 10 MV. A heterogeneous, approximately-humanoid phantom was used in this study, to provide an indication of the contrast and noise produced when imaging different types of tissue with comparatively realistic thicknesses. The phantom selected for this purpose was the CIRS model 002LFC thorax (Standard Imaging, Middleton, USA), which is constructed from tissue-equivalent plastic, including lung- and bone-equivalent regions. The use of this phantom was regarded as advantageous for two reasons: firstly, the use of a thorax phantom approximately models a breast boost scenario, where electrons are frequently employed [34], secondly the CIRS thorax is recommended by the IAEA for use in treatment planning system quality assurance [35, 36] and is therefore likely to be more available to physicists wishing to repeat or improve this study than many other humanoid phantoms. When acquiring all the images used in this study, the EPID panel was positioned at a source-to-detector distance (SDD) of 150 cm, with the phantom at a source-to-surface distance (SSD) of 100 cm. The phantom was set up by aligning its external markers with the transverse and
Australas Phys Eng Sci Med
sagittal lasers, and then skewed in the coronal plane by applying an arbitrary (8 ) couch rotation. This skew was applied in order to establish that the images of our longitudinally-unvarying phantom were not affected by any artefacts arising from the line-by-line readout of the stored charge in the EPID’s photodiodes [4, 37, 40, 41]. All images were acquired in integrated mode using 100 MU at a dose rate of 600 MU/min, without beam pulse synchronisation, using a 1515 cm2 square applicator with and without a block of low melting point alloy (LMPA) containing an irregularly shaped treatment aperture, to simulate the acquisition of dose during the delivery of an electron treatment fraction. All beams deliverable by the linac, including photon beams, were used. The acquired images were automatically corrected for the effects of dark current (photodiodes storing charge during beam-off), by subtracting a dark field image, and corrected for the effects of variations in the sensitivity of individual photodiodes, using a flood field image. All dark field images were acquired using the same method; accumulating data with the linac beam off. There were several important differences between the flood field images acquired for the different photon and electron beams. For imaging using the photon beams, the 6 and 10 MV integrated flood field images were acquired using the standard method; covering the EPID’s active area with an open field and delivering sufficient MU to accumulate and average 200 frames. For imaging the 6, 9, 12, 16 and 20 MeV electron beams, three different methods were used to acquire sets of flood field images. All three methods involved the delivery of sufficient MU to accumulate and average 200 frames, using a field collimated by a 2020 cm2 electron applicator, projecting to 3030 cm2 at the EPID and thus covering the central region of the EPID’s active area. Preliminary investigations utilised a set of electron flood field images acquired with the default exposure rate of 0.348 MU/frame, using open fields, as used for acquiring photon flood field data. After the resulting images were evaluated and found to show barely distinguishable anatomical features that were obscured by the bremsstrahlung photon signal (see the ‘‘Image acquisition’’ results section), a new set of flood field images were acquired, at the default exposure rate, with the beams transmitted through blocks of water-equivalent plastic upstream of the EPID. After observing that the resulting images were substantially improved compared to the preliminary electron beam images (see the ‘‘Image acquisition’’ results section), the image acquisition was optimised by varying the number of MU per frame used in both the flood field and the phantom images. The acquisition rate 5.00 MU/frame was identified as producing the best results
(see the ‘‘Image acquisition’’ results section) and was therefore used in acquiring the final, ‘optimised’, set of EPID images. The thicknesses of water equivalent plastic used to produce the ‘filtered’ flood field images are listed in Table 1. These thicknesses were chosen to approximately equal the practical range of electrons from each beam (given the limitations of the water-equivalent block thicknesses available), which are also shown in Table 1, as calculated using the electron-energy-dependent waterequivalent thickness calculation method suggested by Ding et al. [38]. This plastic effectively filtered out the primary electron beam, which would otherwise have undergone multiple scattering in the air between the linac and the EPID, and within the EPID itself, leading to excessive noise in the flood field data and thereby producing excessively noisy phantom images (as seen in the preliminary images, in the ‘‘Image acquisition’’ results section). Contributions from the electron beam were thereby removed from both the phantom images (by the thickness of the humanoid phantom itself) and the flood field images, so that the resulting images were generated using the small bremsstrahlung photon contributions to each beam (listed in Table 1). While the Portal Vision software permits EPID imaging using electron beams, the software only allows one electron beam calibration (ie. one set of dark and flood field data, for one electron beam energy) to be used. Changing imaging modalities from one electron beam energy to another therefore requires that new flood field images be acquired at the new electron beam energy (whereas the software can automatically change between ‘Lo-X’ and ‘Hi-X’ calibration datasets, without needing new flood field data acquisition, when photon beam energies are changed). Because the repeated acquisition of electron imaging calibration data throughout the treatment day would be clinically unrealistic, this study sought to identify whether any one electron beam energy could be used to produce flood field images that result in good quality images when used to correct humanoid phantom images at all other electron beam energies. The optimised images were therefore acquired by using each flood field image, for each electron beam energy, to correct a set of images acquired at all electron beam energies. In order to evaluate the quality of the large number of images produced by this study, in-house ‘Batch Image Processor’ code was written. Images from Portal Vision AM Maintenance were saved in DICOM format, with each image pixel scaled using a linear relationship with coefficients recorded in the DICOM header. The Batch Image Processor automatically read all DICOM files stored within a specified directory, reversed the pixel scaling and output a text file (in csv format) containing the
123
Australas Phys Eng Sci Med Table 1 Summary of the properties of the electron beams used in this study (nominal electron beam energies, 50% dose depths in water (R50 ), practical ranges of the electron beams in water (Rp ) and percentage doses detectable in water beyond the practical ranges (indicating the contribution of bremsstahlung photons) (Dx ), all obtained from water tank measurements), listed alongside the physical thickness (tphys ) and water equivalent thickness (twe ) of the electron filtering phantoms used in this study, as well as the ratio of electron beam practical range to phantom water-equivalent thickness (Rp =twe ) which provides an indication of electron beam filtering efficiency, in each electron beam twe (cm)
Rp =twe
Nominal energy (MeV)
R50 (cm)
Rp (cm)
Dx (%)
tphys (cm)
6
2.33
2.93
0.4
3
2.87
9
3.57
4.40
0.7
4.5
4.36
1.01
12
5.00
5.88
1.5
6
6.08
0.97
16
6.64
7.85
2.9
8
7.99
0.98
20
8.32
9.82
4.6
10
10.12
0.97
1.02
mean and standard deviations of the pixel values within userspecified regions of interest (ROIs). The resulting data was analysed using Excel (Microsoft Corporation, Redmond, USA), to quantify the contrast, contrast-to-noise ratios (CNRs) and SNRs in regions where the beam was transmitted through tissue-, lung- and bone-equivalent material, using the following relationships [7, 8, 32]: Mean in ROI1 Mean in ROI2 Contrast ¼ ð1Þ Mean in ROI1 Mean in ROI1 Mean in ROI2 : ð2Þ CNR ¼ St: Dev: in ROI1 Mean in ROI1 : ð3Þ SNR ¼ St: Dev: in ROI1 In this work, ROI1 , the reference region, was defined as an area on the image where the beam passed through tissue. ROI2 , the comparison region, was defined as an area on the image where the beam passed through either tissue and lung (referred to as ‘lung’) or where the beam passed through tissue and bone (referred to as ‘bone’). Finally, having identified a range of flood-field calibration beam energies and imaging beam energies that were capable of producing images with contrast and noise levels that were similar to photon beam images, the clinical utility of the electron beam imaging method was briefly exemplified. To simulate the likely clinical scenario of an electron treatment delivered to a breast tumour bed, several 0.6 mm thick stainless steel surgical clips (Coviden AG, New Haven, USA) were placed on top of the thorax phantom then covered with a 2.5 cm thickness of sheet bolus and imaged using an 813 cm2 field from a 16 MeV electron beam. The resulting image was qualitatively compared with an image from a 6 MV photon beam.
123
Results Image acquisition Figure 1a, b shows lateral profiles through flood field EPID acquisition calibration images. Figure 1a provides a comparison between profiles through a familiar photon flood field image (10 MV) and through electron flood field images acquired at a similar nominal energy (9 MeV) with and without the use of MU/frame optimisation and with and without the inclusion of a thick plastic electron filter (see Table 1). This figure illustrates the effect on the shape and intensity of the beam profile when a beam comprised of photons from the electron beam’s bremsstrahlung tail is transmitted to the EPID. Figure 1b shows the flood field profiles obtained using filtered beams at all five nominal electron energies used in this study. This figure shows a clear increase in imaging beam intensity (arising from the increasing bremsstrahlung fluence—see Table 1) as well as the familiar increase in the degree to which the bremsstrahlung photon beam is forward peaked, with increasing electron beam energy. This result is qualitatively similar to large-field profiles through the bremsstrahlung tail region previously measured in a water tank at various electron beam energies [39]. Figure 2a–q shows a selection of images of the thorax phantom, acquired using electron beams under different conditions. Evidently, none of these images were affected by the horizontal or vertical line artefacts that may arise when much smaller numbers of frames are used to produce each image [40, 41]. The top row of images (Fig. 2a–e) shows the results of using the five different electron energies (increasing in nominal energy from left to right), where the corresponding calibration flood field images were acquired using open fields and default (low) MU/frame. While the images acquired at 6 and 9 MeV reveal little anatomical information, the lung-tissue and bone-tissue interfaces become increasingly apparent (appearing as transitions from darkto mid-grey (lung-tissue) and light- to mid-grey (bone-tissue), angled 8 from vertical) as the electron beam energy is increased. All of the higher electron beam energy images (Fig. 2c–e) are obscured, however, by a darkening at their centres, caused by the increased bremsstrahlung photon fluence at the centre of these electron beams. Effectively, the open electron flood fields have failed to flatten the imaging beams. Figure 2f–h shows the effect of recalibrating the 6 MeV beam with the electron signal filtered using plastic blocks (see Table 1) and increasing the acquired MU/frame [(f) 1.28, (g) 2.00, (h) 5.00]. Qualitative examination indicates that these changes produce a substantial increase
Australas Phys Eng Sci Med
from left to right). Qualitative comparison of Fig. 2h–l with a–e shows how using electron filtration to acquire the flood field images that are used to calibrate the EPID imaging system results in a clearer, less noisy, higher contrast image that is less affected by the shape of the bremsstrahlung beam profile, at all electron beam energies used in this study. Figure 2m–q provides examples of the use of clinical electron treatment beams, collimated using an irregular, off-axis aperture in an LMPA block, to produce EPID images. The unblocked region in each of these images shows phantom features with a similar quality and clarity to the larger field images shown in Fig. 2h–l. The blocked regions of Fig. 2m–q show faint but observable phantom features, suggesting that the bremsstrahlung signal used to produce the EPID images is not entirely shielded out by the 1.5 cm thickness of LMPA used to form the treatment aperture or that the LMPA block is itself a source of secondary bremsstrahlung photons. Image quality
Fig. 1 a Profiles through flood field images acquired using (from lightest to darkest profile lines, highest to lowest intensity) an unfiltered 10 MV photon beam collimated to 3040 cm2 using the linac’s orthogonal jaws, an unfiltered 9 MeV electron beam collimated to 2020 cm2 using an electron applicator, the same 9 MeV electron beam filtered through 4.5 cm of water equivalent plastic imaged at maximum MU/frame, the same 9 MeV electron beam filtered through 4.5 cm of water equivalent plastic imaged at minimum MU/frame. b Profiles through flood field images acquired at minimum MU/frame using filtered electron beams with nominal energies of (from lightest to darkest profiles lines, highest to lowest intensity) 20, 16, 12, 9 and 6 MeV. For ease of comparison, all profiles are normalised to the maximum pixel value in the 10 MV flood field profile. (Note that a and b are plotted with different vertical scales.)
in image quality, compared to the preliminary 6 MeV image (Fig. 2a). This increase in image quality with increasing MU/frame suggests that the low detective quantum efficiency (DQE) of the EPID imaging system [42], combined with the reduced photon fluence in the electron beams compared with the megavoltage photon beams that are conventionally used for portal imaging (see Table 1; Fig. 1), has the potential to noticeably decrease the resulting image quality if the number of photons contributing to each image frame are insufficient to show anatomical information above quantum noise. Figure 2i–l shows effects of making these changes (filtering the flood field images using the thicknesses of plastic listed in Table 1, acquiring all images using 5.00 MU/ beam) for the remaining electron beam energies (increasing
Figures 3a–c, 4a–b and 5a–d quantify the contrast and noise apparent in the electron beam images of the thorax phantom obtained at maximum MU/frame, after calibrating the EPID using filtered flood field images, and provide comparisons with the contrast and noise apparent in standard photon beam images of the same phantom. SNR data (see Eq. 1), plotted in Fig. 3a–c, show that compared with the photon-beam images that are currently regarded as clinically acceptable, most of the electron beam images were subject to similar levels of noise, except when imaging bone-equivalent material with low-energy electron beams (Fig. 3b) or when imaging lung-equivalent material with all electron beams (Fig. 3a), where photon beam images produced substantially higher SNR results. These figures show that the SNR values obtained from the electron beam images were improved by imaging at higher energies (16 and 20 MeV) or by calibrating the EPID using flood field images acquired at intermediate energies (9 and 12 MeV). Data in Fig. 4a, b suggest that electron beam images can achieve lung-tissue and bone-tissue contrast (see Eq. 2) that has, respectively, a greater and a similar magnitude to the contrast achieved in the photon beam images. Electronbeam EPID images generally have higher lung-tissue and bone-tissue contrast when the EPID is calibrated with a higher energy electron beam than with a lower energy electron beam, although the differences due to flood field beam energy are smaller than the differences due to imaging beam energy. For example, data in Fig. 4b shows that changing the flood field beam energy from 6 to 20 MeV increases bone-
123
Australas Phys Eng Sci Med
Fig. 2 a–e Preliminary images, where the electron beam energy used for the image acquisition increases (from 6 to 20 MeV) from left to right. f–h Test images acquired using a 6 MeV electron beam, calibrated using filtered flood-field images, with increasing MU/ frame. h–l Optimised images, calibrated using filtered flood-field images and acquired using the maximum available MU/frame, where
the electron beam energy used for the image acquisition increases (from 6 to 20 MeV) from left to right. m–q Optimised images, where the electron beams are transmitted through an LMPA block incised with a treatment aperture. Inset describes the features shown in the images
tissue contrast by a factor of 1.5, for a 12 MeV imaging beam, while changing the imaging energy from 6 to 20 MeV decreases bone-tissue contrast by a factor of 4.5, for a beam calibrated with a 12 MeV flood field beam. Generally, bone-tissue contrast increases with flood field energy and decreases with imaging energy, while the opposite is the case (up to 16 MeV imaging energy) for lung-tissue contrast. CNR (Fig. 5a–d) combines the effects of contrast and noise (see Eq. 3), producing values that vary inconsistently with calibration and imaging beam energy. Despite the substantial difference between the lung SNRs in electron and photon images, lung CNR values produced by the two modalities are in close agreement when similar imaging and calibration energies are used (Fig. 5a, b). Bone CNR
values are also similar across modalities when similar imaging and calibration energies are used (Fig. 5c, d), although bone CNR declines when higher electron imaging and calibration energies are used. Results shown in Fig. 5a–d indicate that the magnitude of both lung-tissue CNR and bone-tissue CNR is increased when images acquired at any electron beam energy use flood-field calibrations produced at 12 MeV. Figure 6a, b exemplifies a potential clinical application of electron beam imaging, showing the result of imaging several surgical clips, placed on top of the thorax phantom, underneath a 2.5 cm thickness of sheet bolus, using a 16 MeV electron beam calibrated using a 12 MeV flood field image filtered by a 6 cm block of plastic. This scenario is intended to simulate the clinical use of an electron beam to
123
Australas Phys Eng Sci Med
image a breast tumour bed. Although window levels in Fig. 6a, b have been adjusted to show the surgical clips as well as the imaged anatomy as clearly as possible, the group of surgical clips is only faintly visible against the dark background of tissue- and lung-equivalent material. These small clips (6 mm thick, 6 mm long) would likely be indistinguishable if imaged in a patient with more complex bony anatomy, more heterogeneous lung tissues and a more irregular external contour than the simple thorax phantom used here. Nonetheless, comparison of the electron imaging result shown in Fig. 6a, b with the photon imaging result shown in Fig. 6c, d leads to the important observation that even in this challenging situation, the electron beam images have a similar quality to photon beam images of the same subject.
Discussion
Fig. 3 SNR in a lung, b bone and c tissue, in EPID images obtained by imaging using photon beams (black data points) and electron beams (all other data points). The nominal energy of the flood fields used to calibrate the EPID for electron beam imaging increases from 6 MeV (white data points) up to 20 MeV (dark grey data points), with increasingly dark data points used to indicate increasing flood field energy [as indicated by labels on (a)]. Lines interpolate between data points as a visual guide only
Fig. 4 a Lung-tissue and b bone-tissue contrast, in EPID images obtained by imaging using photon beams (black data points) and electron beams (data points as described in caption for Fig. 3).
The results of this study suggest that electron radiotherapy beams can be used to produce EPID images with levels of contrast and noise that are similar to the contrast and noise levels seen in clinically acceptable photon beam EPID images, if the EPID is calibrated for ‘electron’ imaging by physically filtering out the electron beam, leaving only the bremsstrahlung photon component. If only one electron beam energy can be used for EPID calibration, then an intermediate electron beam energy should be used; in this work 12 MeV was identified as preferable to higher and lower electron beam energies, when used to flood-field correct EPID images acquired with all other available images. If the filtered flood field method proposed in this work is utilised clinically, then EPID images produced during
Example error bars are shown, for clarity. Lines interpolate between data points as a visual guide only
123
Australas Phys Eng Sci Med
Fig. 5 a and b lung-tissue and c and d bone-tissue CNR, in EPID images obtained by imaging using photon beams (black data points) and electron beams (all other data points). CNR is plotted against imaging beam energy in (a) and (c), where data points as described in
caption for Fig. 3. CNR is plotted against flood-field calibration beam energy in (b) and (d), where the shading of the electron data points indicates the nominal energy of the imaging beam. Lines interpolate between data points as a visual guide only
electron treatments may become useful for post-treatment setup verification and review. In such cases the entire dose from each electron treatment fraction would be used to acquire the image; there would be no additional imaging dose to the patient. Depending on the patient management software used by the department, it may or may not be possible to schedule electron beam imaging during treatment. In departments where software permits electron beam images to be added to the treatment plan, the acquisition and use of these images would follow a similar workflow to the acquisition and use of during-treatment photon beam images. However, in departments where electron beam images cannot be added to the treatment plan, either during treatment planning, plan preparation or using the treatment console immediately prior to delivery, then the EPID image acquisition software may need to be run independently while the treatment is delivered using the linac in clinical mode. The location of the treatment aperture relative to bony anatomy or lung-tissue interfaces may be identified as easily using electron beams as using photon beams, provided that the EPID is calibrated appropriately. Localisation of the treatment aperture relative to the location of the
tumour bed, as indicated by surgical clips, has not been conclusively demonstrated in this study. Results shown in Fig. 6a–d indicate that adequate imaging of surgical clips can be challenging when using both photon and electron beams. However, only one (very small and thin) type of surgical clip was used in this work. Further testing and discussions with referring thoracic surgeons is recommended in order to evaluate the suitability of more radiologically opaque clips or markers for this purpose. The positive results of this study suggest that EPID imaging of planar homogeneous phantoms using electron beams may be useful for routine linac quality assurance, if the constancy of the (relative) symmetry and flatness of the bremsstrahlung component of each electron beam can be shown to be indicative of the constancy of the symmetry and flatness of the overall electron beam. Note that it is not possible to recommend the use of the EPID to measure absolute symmetry and flatness (such as might be measured using a water tank), due to the inherent asymmetry of EPID support arm backscatter (leading to measured asymmetry) [43] as well as the known over-response of amorphous silicon EPIDs to low energy photons (leading to measured over-flatness) [44].
123
Australas Phys Eng Sci Med Fig. 6 Images of thorax phantom, with surgical clips and bolus: a image acquired using 16 MeV electron beam (calibrated using 12 MeV filtered flood field); b magnified view of surgical clips from (a); c image acquired using 6 MV photon beam; d magnified view of surgical clips from (c). Group of surgical clips is circled in (a), (b), (c) and (d)
This proof-of-concept study suggests several important avenues for future investigation. For example, the complex relationship between lung-tissue contrast and imaging electron beam energy (decreasing contrast with increasing energy from 6 to 16 MeV before a dramatic contrast increase at 20 MeV) could be investigated further via photon beam imaging using a high-energy (up to 23 MV) linear accelerator. The physical reasons for EPID image quality improvements when calibrating with an intermediate electron beam energy (12 MeV) and imaging with a high electron beam energy (16 or 20 MeV) could be investigated via Monte Carlo simulation or by direct analysis of interaction cross-sections in lung, bone, tissue, gadolinium oxysulfide and silicon. The broader applications of this work, for electron beam quality assurance, could be investigated using repeated measurements over a long period of time or over several different linacs.
Conclusion It is possible to use electron radiotherapy beams to produce EPID images showing tissue and bony anatomy with levels of contrast and noise that are similar to the contrast and
noise levels seen in clinically acceptable photon beam EPID images. In order to achieve these results, however, it is necessary to both increase the MU/frame used to acquire the images and to physically filter the primary electron beam out of the flood field images and thereby calibrate the EPID using only the bremsstrahlung photon component of each electron beam. If these changes are not made, then the resulting electron beam images can be noisy and difficult to interpret. SNR, CNR and contrast values obtained from electron beam images acquired using an EPID calibrated using the filtered flood field method have been shown, in this study, to be comparable to SNR, CNR and contrast values obtained using standard photon beam EPID imaging, with the best results being achieved when images are acquired using the higher energy electron beams (16 and 20 MeV) when the EPID is calibrated using an intermediate (12 MeV) electron beam energy. These results demonstrate the feasibility of acquiring clinically-useful EPID images of patient anatomy using electron beams and suggest important avenues for future investigation, and therefore are expected to enable and encourage further research in this area. There is manifest potential for the EPID imaging method proposed in this
123
Australas Phys Eng Sci Med
work to lead to the clinical use of electron beam imaging for geometric verification of electron treatments in the future. Acknowledgments The authors wish to acknowledge the valued contributions of the following participants at the Australasian College of Physical Scientists and Engineers in Medicine (ACPSEM) Queensland Branch 2014 Winter School on Scientific Publication: Jacqueline Charles, Paul Charles, Allison Fox, Robin Hill, Benjamin Harris, Emma Inness, Vaughan Moutrie, Zoe¨ Moutrie, Patrick O’Connor, Bess Sutherland, Steven Sylvander, Luke Webb, Rachael Wilks and Nancy Yu. This work was supported by the Australian Research Council, the Wesley Research Institute, Premion (Genesis Cancer Care Queensland) and the Queensland University of Technology (QUT), through linkage Grant Number LP110100401.
References 1. El-Gayed AA, Bel A, Vijlbrief R, Bartelink H, Lebesque JV (1993) Time trend of patient setup deviations during pelvic irradiation using electronic portal imaging. Radiother Oncol 26(2):162–171 2. The Royal Australian and New Zealand College of Radiologists, Faculty of Radiation Oncology Position Paper: Techniques and Technologies in Radiation Oncology (2012) Horizon Scan Australia, RANZCR 3. Bridge P, Carmichael MA, Brady C, Dry A (2013) A snapshot of radiation therapy techniques and technology in Queensland: an aid to mapping undergraduate curriculum. J Med Radiat Sci 60(1):25–34 4. Boyer AL, Antonuk L, Fenster A, Van Herk M, Meertens H, Munro P, Reinstein LE, Wong J (1992) A review of electronic portal imaging devices (EPIDs). Med Phys 19(1):1–16 5. McDermott LN, Nijsten SMJJG, Sonke JJ, Partridge M, Van Herk M, Mijnheer BJ (2006) Comparison of ghosting effects for three commercial a-Si EPIDs. Med Phys 33(7):2448–2451 6. Greer PB, Popescu CC (2003) Dosimetric properties of an amorphous silicon electronic portal imaging device for verification of dynamic intensity modulated radiation therapy. Med Phys 30(7):1618–1627 7. Flampouri S, McNair HA, Donovan EM, Evans PM, Partridge M, Verhaegen F, Nutting CM (2005) Initial patient imaging with an optimised radiotherapy beam for portal imaging. Radiother Oncol 76(1):63–71 8. Kairn T, Khoei S, Markwell TS, Fielding AL, Trapp JV (2010) Contrast enhancement of EPID images via difference imaging: a feasibility study. Phys Med Biol 55(22):N533–N544 9. Pouliot J, Bani-Hashemi A, Chen J, Svatos M, Ghelmansarai F, Mitschke M, Aubin M, Xia P, Morin O, Bucci K, Roach M III, Hernandez P, Zheng Z, Hristov D, Verhey L (2005) Low-dose megavoltage cone-beam CT for radiation therapy. Int J Radiat Oncol Biol Phys 61(2):552–560 10. Hansen EK, Larson DA, Aubin M, Chen J, Descovich M, Gillis AM, Morin O, Xia P, Pouliot J (2006) Image-guided radiotherapy using megavoltage cone-beam computed tomography for treatment of paraspinous tumors in the presence of orthopedic hardware. Int J Radiat Oncol Biol Phys 66(2):323–326 11. Markwell T, Perera L, Trapp J, Fielding A (2014) Evaluation of MegaVoltage Cone Beam CT image quality with an unmodified Elekta Precise Linac and EPID: a feasibility study. Australas Phys Eng Sci Med 37(2):291–302
123
12. Moutrie V, Kairn T, Rosenfeld A, Charles PH (2015) Use of a megavoltage electronic portal imaging device to identify prosthetic materials. Australas Phys Eng Sci Med 38(1):93–100 13. Liu G, Van Doorn T, Bezak E (2002) Assessment of flatness and symmetry of megavoltage X-ray beam with an electronic portal imaging device (EPID). Australas Phys Eng Sci Med 25(2):58–66 14. Budgell GJ, Zhang R, Mackay RI (2007) Daily monitoring of linear accelerator beam parameters using an amorphous silicon EPID. Phys Med Biol 52(6):1721–1733 15. Clivio A, Vanetti E, Rose S, Nicolini G, Belosi MF, Cozzi L, Baltes C, Fogliata A (2015) Evaluation of the machine performance check application for TrueBeam Linac. Radiat Oncol 10(1):97 16. Wendling M, Louwe RJ, McDermott LN, Sonke JJ, van Herk M, Mijnheer BJ (2006) Accurate two-dimensional IMRT verification using a back-projection EPID dosimetry method. Med Phys 33(2):259–273 17. Warkentin B, Steciw S, Rathee S, Fallone BG (2003) Dosimetric IMRT verification with a flat-panel EPID. Med Phys 30(12): 3143–3155 18. Van Esch A, Depuydt T, Huyskens DP (2004) The use of an aSibased EPID for routine absolute dosimetric pre-treatment verification of dynamic IMRT fields. Radiother Oncol 71(2):223–234 19. Nicolini G, Fogliata A, Vanetti E, Clivio A, Cozzi L (2006) GLAaS: an absolute dose calibration algorithm for an amorphous silicon portal imager. Applications to IMRT verifications. Med Phys 33(8):2839–2851 20. Nicolini G, Vanetti E, Clivio A, Fogliata A, Korreman S, Bocanek J, Cozzi L (2008) The GLAaS algorithm for portal dosimetry and quality assurance of RapidArc, an intensity modulated rotational therapy. Radiat Oncol 3:24 21. Lee C, Menk F, Cadman P, Greer PB (2009) A simple approach to using an amorphous silicon EPID to verify IMRT planar dose maps. Med Phys 36(3):984–992 22. Mans A, Wendling M, McDermott LN, Sonke JJ, Tielenburg R, Vijlbrief R, Mijnheer B, Van Herk M, Stroom JC (2010) Catching errors with in vivo EPID dosimetry. Med Phys 37(6):2638–2644 23. Mijnheer B, Mans A, Olaciregui-Ruiz I, Sonke JJ, Tielenburg R, Van Herk M, Vijlbrief R, Stroom J (2010) 2D AND 3D dose verification at The Netherlands Cancer InstituteAntoni van Leeuwenhoek Hospital using EPIDs. J Phys Conf Ser 250(1): 012020 24. van Elmpt W, McDermott L, Nijsten S, Wendling M, Lambin P, Mijnheer B (2008) A literature review of electronic portal imaging for radiotherapy dosimetry. Radiother Oncol 88(3): 289–309 25. Kairn T, Cassidy D, Sandford PM, Fielding AL (2008) Radiotherapy treatment verification using radiological thickness measured with an amorphous silicon electronic portal imaging device: Monte Carlo simulation and experiment. Phys Med Biol 53(14):3903–3919 26. Kavuma A, Glegg M, Metwaly M, Currie G, Elliott A (2010) A novel method for patient exit and entrance dose prediction based on water equivalent path length measured with an amorphous silicon electronic portal imaging device. Phys Med Biol 55(2):435–452 27. Kairn T, Crowe SB, Trapp JV (2012) A simple method for EPIDbased in vivo dosimetry for radiotherapy treatments of head-andneck cancers. IFMBE Proc 39:1727–1730 28. Beck JA, Budgell GJ, Roberts DA, Evans PM (2009) Electron beam quality control using an amorphous silicon EPID. Med Phys 36(5):1859–1866
Australas Phys Eng Sci Med 29. Wang Y, Heaton R, Norrlinger B, Islam M (2013) Quality assurance of electron beams using a Varian electronic portal imaging device. Phys Med Biol 58(16):5461–5475 30. Emam I, Fan J, Jin L, El Dib A, Ma C (2009) SU-FF-T-307: a multileaf collimator quality assurance tool for modulated electron radiation therapy. Med Phys 36(6):2592 31. Gauer T, Engel K, Kiesel A, Albers D, Rades D (2010) Comparison of electron IMRT to helical photon IMRT and conventional photon irradiation for treatment of breast and chest wall tumours. Radiother Oncol 94(3):313–318 32. Jarry G, Verhaegen F (2005) Electron beam treatment verification using measured and Monte Carlo predicted portal images. Phys Med Biol 50(21):4977–4994 33. Aubin M, Faddegon B, Pouliot J (2002) Electron beam verification with an a-Si flat-panel electronic portal imaging device. Proc SPIE 4682:549–557 34. Shank B, Moughan J, Owen J, Wilson F, Hanks GE (2000) The 199394 patterns of care process survey for breast irradiation after breast-conserving surgerycomparison with the 1992 standard for breast conservation treatment. Int J Radiat Oncol Biol Phys 48(5):1291–1299 35. IAEA (2004) Technical Reports Series 430. Commissioning and quality assurance of computerized planning systems for radiation treatment of cancer. International Atomic Energy Agency, Vienna 36. IAEA (2008) TECDOC 1583. Commissioning of radiotherapy treatment planning systems: testing for typical external beam treatment techniques. International Atomic Energy Agency, Vienna
37. Louwe RJ, McDermott LN, Sonke JJ, Tielenburg R, Wendling M, Van Herk MB, Mijnheer BJ (2004) The long-term stability of amorphous silicon flat panel imaging devices for dosimetry purposes. Med Phys 31(11):2989–2995 38. Ding GX, Rogers DWO, Mackie TR (1996) Mean energy, energy-range relationships and depth-scaling factors for clinical electron beams. Med Phys 23(3):361–376 39. Faddegon B, Schreiber E, Ding X (2005) Monte Carlo simulation of large electron fields. Phys Med Biol 50(5):741–753 40. Mooslechner M, Mitterlechner B, Weichenberger H, Huber S, Sedlmayer F, Deutschmann H (2013) Analysis of a free-running synchronization artifact correction for MV-imaging with aSi: H flat panels. Med Phys 40(3):031906 41. Woodruff HC, Greer PB (2013) 3D Dose reconstruction: banding artefacts in cine mode EPID images during VMAT delivery. J Phys Conf Ser 444:012042 42. Herman MG, Balter JM, Jaffray DA, McGee KP, Munro P, Shalev S, Van Herk M, Wong JD (2001) Clinical use of electronic portal imaging: Report of AAPM Radiation Therapy Committee Task Group 58. Med Phys 28(5):712–737 43. Rowshanfarzad P, McCurdy BM, Sabet M, Lee C, OConnor DJ, Greer PB (2010) Measurement and modeling of the effect of support arm backscatter on dosimetry with a Varian EPID. Med Phys 37(5):2269–2278 44. Kirkby C, Sloboda R (2005) Consequences of the spectral response of an a-Si EPID and implications for dosimetric calibration. Med Phys 32(8):2649–2658
123
SCIENTIFIC NOTE
Use of electronic portal imaging devices for electron treatment verification T. Kairn1,2 • T. Aland1,2 • S. B. Crowe2,3 • J. V. Trapp2
Received: 7 August 2015 / Accepted: 4 November 2015 Ó Australasian College of Physical Scientists and Engineers in Medicine 2015
Abstract This study aims to help broaden the use of electronic portal imaging devices (EPIDs) for pre-treatment patient positioning verification, from photon-beam radiotherapy to photon- and electron-beam radiotherapy, by proposing and testing a method for acquiring clinicallyuseful EPID images of patient anatomy using electron beams, with a view to enabling and encouraging further research in this area. EPID images used in this study were acquired using all available beams from a linac configured to deliver electron beams with nominal energies of 6, 9, 12, 16 and 20 MeV, as well as photon beams with nominal energies of 6 and 10 MV. A widely-available heterogeneous, approximately-humanoid, thorax phantom was used, to provide an indication of the contrast and noise produced when imaging different types of tissue with comparatively realistic thicknesses. The acquired images were automatically calibrated, corrected for the effects of variations in the sensitivity of individual photodiodes, using a flood field image. For electron beam imaging, flood field EPID calibration images were acquired with and without the placement of blocks of water-equivalent plastic (with thicknesses approximately equal to the practical range of electrons in the plastic) placed upstream of the EPID, to
filter out the primary electron beam, leaving only the bremsstrahlung photon signal. While the electron beam images acquired using a standard (unfiltered) flood field calibration were observed to be noisy and difficult to interpret, the electron beam images acquired using the filtered flood field calibration showed tissues and bony anatomy with levels of contrast and noise that were similar to the contrast and noise levels seen in the clinically acceptable photon beam EPID images. The best electron beam imaging results (highest contrast, signal-to-noise and contrast-to-noise ratios) were achieved when the images were acquired using the higher energy electron beams (16 and 20 MeV) when the EPID was calibrated using an intermediate (12 MeV) electron beam energy. These results demonstrate the feasibility of acquiring clinically-useful EPID images of patient anatomy using electron beams and suggest important avenues for future investigation, thus enabling and encouraging further research in this area. There is manifest potential for the EPID imaging method proposed in this work to lead to the clinical use of electron beam imaging for geometric verification of electron treatments in the future. Keywords
Electrons Portal imaging Radiation therapy
A case study based on this work was presented at the Winter School on Scientific Publication, Australasian College of Physical Scientists and Engineers in Medicine (Queensland Branch), Brisbane, 2014. & T. Kairn [email protected] 1
Genesis Cancer Care Queensland, Brisbane, Australia
2
Science and Engineering Faculty, Queensland University of Technology, Brisbane, Australia
3
Cancer Care Services, Royal Brisbane and Women’s Hospital, Brisbane, Australia
Introduction Electronic portal imaging devices (EPIDs) are widely used for pre-treatment patient positioning verification, before and during megavoltage photon radiotherapy treatment delivery [1–3]. This study aims to help broaden their use to the geometric verification of electron radiotherapy treatment delivery.
123
Australas Phys Eng Sci Med
The amorphous-silicon (a-Si) flat panel EPIDs that augment most contemporary medical linear accelerators generate images using a metal plate and phosphor screen, immediately above (upstream of) an array of a-Si photodiodes and transistors. An early review of EPID technologies by Boyer et al explains that ‘‘X-ray interactions in the metal plate create high-energy electrons that in turn produce fluorescence in the phosphor screen’’ and provides a detailed description of the process by which the resulting distribution of optical photons is detected and read out via the a-Si layer [4]. Fundamentally, EPIDs are designed for the rapid acquisition of images produced by the interactions of megavoltage photons. Contemporary EPIDs can read out image data at a rate of several frames per second without dead-time between frames [5] and the automatic averaging of all frames acquired during each exposure [6] can lead to the production of relatively high quality images of photon treatment portals and the intervening patient anatomy [7, 8]. While still primarily used for patient setup imaging before and during photon beam radiotherapy treatments [2, 3], EPIDs are increasingly being investigated and used for alternative applications including: the generation of threedimensional (cone-beam CT) images [9–11]; evaluation of the water-equivalence of plastics and identification of metal implants [12]; routine photon beam constancy (including flatness and symmetry) checks [13, 14]; routine geometric testing of the radiation isocentre (including radiation isocentre size and coincidence with imaging isocentres) [15]; pre-treatment quality assurance of modulated radiotherapy treatment plans [16–21]; and in vivo verification of treatment delivery accuracy [22–27]. All of these applications utilise EPID images acquired using megavoltage photon beams. By contrast, references to EPID images acquired using electron beams are sparse in the literature. Use of EPIDs for electron-beam imaging has largely been limited to the investigation of the electron beams themselves; EPID images have been used to investigate the flatness and symmetry of static electron beams [28, 29] and the deliverability of modulated electron beams produced using purpose-built electron MLCs [30], for potential use in electron IMRT treatments [31]. To date, published evaluations of the EPID image quality achievable using standard, static electron radiotherapy beams have been rare [32, 33]. Jarry and Verhaegen [32] used 6–16 MeV electron beams to image a QC-3 (PipsPro) phantom (Standard Imaging, Middleton, USA) placed between slabs of water equivalent plastic, to investigate EPID image quality and demonstrate agreement with Monte Carlo modelling. Importantly, their Monte Carlo simulations of the linacphantom-EPID system showed that EPID image contrast
123
and resolution could be noticeably improved if the electron component of the imaging beam could be excluded from the calculation, leaving only the bremsstrahlung photon contribution [32]. Jarry and Verhaegen also provided examples of the use of a 9 MeV electron beam to image a head phantom, suggesting (without explicitly quantifying the contrast and noise in these images) that the appearance of bony anatomy in electron beam images might be similar to the appearance of bony anatomy in 6 MV photon beam images [32]. This study proposes a method for acquiring clinicallyuseful EPID images of patient anatomy using electron beams, which specifically evaluates image contrast in a patient-like geometry, while taking advantage of Jarry and Verhaegen’s Monte-Carlo-based observations of improved image quality in electron beams from which the electrons have been removed. This study thereby provides initial feasibility testing of the potential capabilities and limitations of EPID imaging with electron beams, with a view to enabling and encouraging further research in this area.
Method EPID images used in this study were acquired using a Varian iX linear accelerator (linac) with a Varian IDU20 aS1000 IAS3 Portal Vision system operated in service mode via the ‘AM Maintenance’ application (Varian Medical Systems, Palo Alto, USA). The linac was configured to deliver electron beams with nominal energies of 6, 9, 12, 16 and 20 MeV, as well as photon beams with nominal energies of 6 and 10 MV. A heterogeneous, approximately-humanoid phantom was used in this study, to provide an indication of the contrast and noise produced when imaging different types of tissue with comparatively realistic thicknesses. The phantom selected for this purpose was the CIRS model 002LFC thorax (Standard Imaging, Middleton, USA), which is constructed from tissue-equivalent plastic, including lung- and bone-equivalent regions. The use of this phantom was regarded as advantageous for two reasons: firstly, the use of a thorax phantom approximately models a breast boost scenario, where electrons are frequently employed [34], secondly the CIRS thorax is recommended by the IAEA for use in treatment planning system quality assurance [35, 36] and is therefore likely to be more available to physicists wishing to repeat or improve this study than many other humanoid phantoms. When acquiring all the images used in this study, the EPID panel was positioned at a source-to-detector distance (SDD) of 150 cm, with the phantom at a source-to-surface distance (SSD) of 100 cm. The phantom was set up by aligning its external markers with the transverse and
Australas Phys Eng Sci Med
sagittal lasers, and then skewed in the coronal plane by applying an arbitrary (8 ) couch rotation. This skew was applied in order to establish that the images of our longitudinally-unvarying phantom were not affected by any artefacts arising from the line-by-line readout of the stored charge in the EPID’s photodiodes [4, 37, 40, 41]. All images were acquired in integrated mode using 100 MU at a dose rate of 600 MU/min, without beam pulse synchronisation, using a 1515 cm2 square applicator with and without a block of low melting point alloy (LMPA) containing an irregularly shaped treatment aperture, to simulate the acquisition of dose during the delivery of an electron treatment fraction. All beams deliverable by the linac, including photon beams, were used. The acquired images were automatically corrected for the effects of dark current (photodiodes storing charge during beam-off), by subtracting a dark field image, and corrected for the effects of variations in the sensitivity of individual photodiodes, using a flood field image. All dark field images were acquired using the same method; accumulating data with the linac beam off. There were several important differences between the flood field images acquired for the different photon and electron beams. For imaging using the photon beams, the 6 and 10 MV integrated flood field images were acquired using the standard method; covering the EPID’s active area with an open field and delivering sufficient MU to accumulate and average 200 frames. For imaging the 6, 9, 12, 16 and 20 MeV electron beams, three different methods were used to acquire sets of flood field images. All three methods involved the delivery of sufficient MU to accumulate and average 200 frames, using a field collimated by a 2020 cm2 electron applicator, projecting to 3030 cm2 at the EPID and thus covering the central region of the EPID’s active area. Preliminary investigations utilised a set of electron flood field images acquired with the default exposure rate of 0.348 MU/frame, using open fields, as used for acquiring photon flood field data. After the resulting images were evaluated and found to show barely distinguishable anatomical features that were obscured by the bremsstrahlung photon signal (see the ‘‘Image acquisition’’ results section), a new set of flood field images were acquired, at the default exposure rate, with the beams transmitted through blocks of water-equivalent plastic upstream of the EPID. After observing that the resulting images were substantially improved compared to the preliminary electron beam images (see the ‘‘Image acquisition’’ results section), the image acquisition was optimised by varying the number of MU per frame used in both the flood field and the phantom images. The acquisition rate 5.00 MU/frame was identified as producing the best results
(see the ‘‘Image acquisition’’ results section) and was therefore used in acquiring the final, ‘optimised’, set of EPID images. The thicknesses of water equivalent plastic used to produce the ‘filtered’ flood field images are listed in Table 1. These thicknesses were chosen to approximately equal the practical range of electrons from each beam (given the limitations of the water-equivalent block thicknesses available), which are also shown in Table 1, as calculated using the electron-energy-dependent waterequivalent thickness calculation method suggested by Ding et al. [38]. This plastic effectively filtered out the primary electron beam, which would otherwise have undergone multiple scattering in the air between the linac and the EPID, and within the EPID itself, leading to excessive noise in the flood field data and thereby producing excessively noisy phantom images (as seen in the preliminary images, in the ‘‘Image acquisition’’ results section). Contributions from the electron beam were thereby removed from both the phantom images (by the thickness of the humanoid phantom itself) and the flood field images, so that the resulting images were generated using the small bremsstrahlung photon contributions to each beam (listed in Table 1). While the Portal Vision software permits EPID imaging using electron beams, the software only allows one electron beam calibration (ie. one set of dark and flood field data, for one electron beam energy) to be used. Changing imaging modalities from one electron beam energy to another therefore requires that new flood field images be acquired at the new electron beam energy (whereas the software can automatically change between ‘Lo-X’ and ‘Hi-X’ calibration datasets, without needing new flood field data acquisition, when photon beam energies are changed). Because the repeated acquisition of electron imaging calibration data throughout the treatment day would be clinically unrealistic, this study sought to identify whether any one electron beam energy could be used to produce flood field images that result in good quality images when used to correct humanoid phantom images at all other electron beam energies. The optimised images were therefore acquired by using each flood field image, for each electron beam energy, to correct a set of images acquired at all electron beam energies. In order to evaluate the quality of the large number of images produced by this study, in-house ‘Batch Image Processor’ code was written. Images from Portal Vision AM Maintenance were saved in DICOM format, with each image pixel scaled using a linear relationship with coefficients recorded in the DICOM header. The Batch Image Processor automatically read all DICOM files stored within a specified directory, reversed the pixel scaling and output a text file (in csv format) containing the
123
Australas Phys Eng Sci Med Table 1 Summary of the properties of the electron beams used in this study (nominal electron beam energies, 50% dose depths in water (R50 ), practical ranges of the electron beams in water (Rp ) and percentage doses detectable in water beyond the practical ranges (indicating the contribution of bremsstahlung photons) (Dx ), all obtained from water tank measurements), listed alongside the physical thickness (tphys ) and water equivalent thickness (twe ) of the electron filtering phantoms used in this study, as well as the ratio of electron beam practical range to phantom water-equivalent thickness (Rp =twe ) which provides an indication of electron beam filtering efficiency, in each electron beam twe (cm)
Rp =twe
Nominal energy (MeV)
R50 (cm)
Rp (cm)
Dx (%)
tphys (cm)
6
2.33
2.93
0.4
3
2.87
9
3.57
4.40
0.7
4.5
4.36
1.01
12
5.00
5.88
1.5
6
6.08
0.97
16
6.64
7.85
2.9
8
7.99
0.98
20
8.32
9.82
4.6
10
10.12
0.97
1.02
mean and standard deviations of the pixel values within userspecified regions of interest (ROIs). The resulting data was analysed using Excel (Microsoft Corporation, Redmond, USA), to quantify the contrast, contrast-to-noise ratios (CNRs) and SNRs in regions where the beam was transmitted through tissue-, lung- and bone-equivalent material, using the following relationships [7, 8, 32]: Mean in ROI1 Mean in ROI2 Contrast ¼ ð1Þ Mean in ROI1 Mean in ROI1 Mean in ROI2 : ð2Þ CNR ¼ St: Dev: in ROI1 Mean in ROI1 : ð3Þ SNR ¼ St: Dev: in ROI1 In this work, ROI1 , the reference region, was defined as an area on the image where the beam passed through tissue. ROI2 , the comparison region, was defined as an area on the image where the beam passed through either tissue and lung (referred to as ‘lung’) or where the beam passed through tissue and bone (referred to as ‘bone’). Finally, having identified a range of flood-field calibration beam energies and imaging beam energies that were capable of producing images with contrast and noise levels that were similar to photon beam images, the clinical utility of the electron beam imaging method was briefly exemplified. To simulate the likely clinical scenario of an electron treatment delivered to a breast tumour bed, several 0.6 mm thick stainless steel surgical clips (Coviden AG, New Haven, USA) were placed on top of the thorax phantom then covered with a 2.5 cm thickness of sheet bolus and imaged using an 813 cm2 field from a 16 MeV electron beam. The resulting image was qualitatively compared with an image from a 6 MV photon beam.
123
Results Image acquisition Figure 1a, b shows lateral profiles through flood field EPID acquisition calibration images. Figure 1a provides a comparison between profiles through a familiar photon flood field image (10 MV) and through electron flood field images acquired at a similar nominal energy (9 MeV) with and without the use of MU/frame optimisation and with and without the inclusion of a thick plastic electron filter (see Table 1). This figure illustrates the effect on the shape and intensity of the beam profile when a beam comprised of photons from the electron beam’s bremsstrahlung tail is transmitted to the EPID. Figure 1b shows the flood field profiles obtained using filtered beams at all five nominal electron energies used in this study. This figure shows a clear increase in imaging beam intensity (arising from the increasing bremsstrahlung fluence—see Table 1) as well as the familiar increase in the degree to which the bremsstrahlung photon beam is forward peaked, with increasing electron beam energy. This result is qualitatively similar to large-field profiles through the bremsstrahlung tail region previously measured in a water tank at various electron beam energies [39]. Figure 2a–q shows a selection of images of the thorax phantom, acquired using electron beams under different conditions. Evidently, none of these images were affected by the horizontal or vertical line artefacts that may arise when much smaller numbers of frames are used to produce each image [40, 41]. The top row of images (Fig. 2a–e) shows the results of using the five different electron energies (increasing in nominal energy from left to right), where the corresponding calibration flood field images were acquired using open fields and default (low) MU/frame. While the images acquired at 6 and 9 MeV reveal little anatomical information, the lung-tissue and bone-tissue interfaces become increasingly apparent (appearing as transitions from darkto mid-grey (lung-tissue) and light- to mid-grey (bone-tissue), angled 8 from vertical) as the electron beam energy is increased. All of the higher electron beam energy images (Fig. 2c–e) are obscured, however, by a darkening at their centres, caused by the increased bremsstrahlung photon fluence at the centre of these electron beams. Effectively, the open electron flood fields have failed to flatten the imaging beams. Figure 2f–h shows the effect of recalibrating the 6 MeV beam with the electron signal filtered using plastic blocks (see Table 1) and increasing the acquired MU/frame [(f) 1.28, (g) 2.00, (h) 5.00]. Qualitative examination indicates that these changes produce a substantial increase
Australas Phys Eng Sci Med
from left to right). Qualitative comparison of Fig. 2h–l with a–e shows how using electron filtration to acquire the flood field images that are used to calibrate the EPID imaging system results in a clearer, less noisy, higher contrast image that is less affected by the shape of the bremsstrahlung beam profile, at all electron beam energies used in this study. Figure 2m–q provides examples of the use of clinical electron treatment beams, collimated using an irregular, off-axis aperture in an LMPA block, to produce EPID images. The unblocked region in each of these images shows phantom features with a similar quality and clarity to the larger field images shown in Fig. 2h–l. The blocked regions of Fig. 2m–q show faint but observable phantom features, suggesting that the bremsstrahlung signal used to produce the EPID images is not entirely shielded out by the 1.5 cm thickness of LMPA used to form the treatment aperture or that the LMPA block is itself a source of secondary bremsstrahlung photons. Image quality
Fig. 1 a Profiles through flood field images acquired using (from lightest to darkest profile lines, highest to lowest intensity) an unfiltered 10 MV photon beam collimated to 3040 cm2 using the linac’s orthogonal jaws, an unfiltered 9 MeV electron beam collimated to 2020 cm2 using an electron applicator, the same 9 MeV electron beam filtered through 4.5 cm of water equivalent plastic imaged at maximum MU/frame, the same 9 MeV electron beam filtered through 4.5 cm of water equivalent plastic imaged at minimum MU/frame. b Profiles through flood field images acquired at minimum MU/frame using filtered electron beams with nominal energies of (from lightest to darkest profiles lines, highest to lowest intensity) 20, 16, 12, 9 and 6 MeV. For ease of comparison, all profiles are normalised to the maximum pixel value in the 10 MV flood field profile. (Note that a and b are plotted with different vertical scales.)
in image quality, compared to the preliminary 6 MeV image (Fig. 2a). This increase in image quality with increasing MU/frame suggests that the low detective quantum efficiency (DQE) of the EPID imaging system [42], combined with the reduced photon fluence in the electron beams compared with the megavoltage photon beams that are conventionally used for portal imaging (see Table 1; Fig. 1), has the potential to noticeably decrease the resulting image quality if the number of photons contributing to each image frame are insufficient to show anatomical information above quantum noise. Figure 2i–l shows effects of making these changes (filtering the flood field images using the thicknesses of plastic listed in Table 1, acquiring all images using 5.00 MU/ beam) for the remaining electron beam energies (increasing
Figures 3a–c, 4a–b and 5a–d quantify the contrast and noise apparent in the electron beam images of the thorax phantom obtained at maximum MU/frame, after calibrating the EPID using filtered flood field images, and provide comparisons with the contrast and noise apparent in standard photon beam images of the same phantom. SNR data (see Eq. 1), plotted in Fig. 3a–c, show that compared with the photon-beam images that are currently regarded as clinically acceptable, most of the electron beam images were subject to similar levels of noise, except when imaging bone-equivalent material with low-energy electron beams (Fig. 3b) or when imaging lung-equivalent material with all electron beams (Fig. 3a), where photon beam images produced substantially higher SNR results. These figures show that the SNR values obtained from the electron beam images were improved by imaging at higher energies (16 and 20 MeV) or by calibrating the EPID using flood field images acquired at intermediate energies (9 and 12 MeV). Data in Fig. 4a, b suggest that electron beam images can achieve lung-tissue and bone-tissue contrast (see Eq. 2) that has, respectively, a greater and a similar magnitude to the contrast achieved in the photon beam images. Electronbeam EPID images generally have higher lung-tissue and bone-tissue contrast when the EPID is calibrated with a higher energy electron beam than with a lower energy electron beam, although the differences due to flood field beam energy are smaller than the differences due to imaging beam energy. For example, data in Fig. 4b shows that changing the flood field beam energy from 6 to 20 MeV increases bone-
123
Australas Phys Eng Sci Med
Fig. 2 a–e Preliminary images, where the electron beam energy used for the image acquisition increases (from 6 to 20 MeV) from left to right. f–h Test images acquired using a 6 MeV electron beam, calibrated using filtered flood-field images, with increasing MU/ frame. h–l Optimised images, calibrated using filtered flood-field images and acquired using the maximum available MU/frame, where
the electron beam energy used for the image acquisition increases (from 6 to 20 MeV) from left to right. m–q Optimised images, where the electron beams are transmitted through an LMPA block incised with a treatment aperture. Inset describes the features shown in the images
tissue contrast by a factor of 1.5, for a 12 MeV imaging beam, while changing the imaging energy from 6 to 20 MeV decreases bone-tissue contrast by a factor of 4.5, for a beam calibrated with a 12 MeV flood field beam. Generally, bone-tissue contrast increases with flood field energy and decreases with imaging energy, while the opposite is the case (up to 16 MeV imaging energy) for lung-tissue contrast. CNR (Fig. 5a–d) combines the effects of contrast and noise (see Eq. 3), producing values that vary inconsistently with calibration and imaging beam energy. Despite the substantial difference between the lung SNRs in electron and photon images, lung CNR values produced by the two modalities are in close agreement when similar imaging and calibration energies are used (Fig. 5a, b). Bone CNR
values are also similar across modalities when similar imaging and calibration energies are used (Fig. 5c, d), although bone CNR declines when higher electron imaging and calibration energies are used. Results shown in Fig. 5a–d indicate that the magnitude of both lung-tissue CNR and bone-tissue CNR is increased when images acquired at any electron beam energy use flood-field calibrations produced at 12 MeV. Figure 6a, b exemplifies a potential clinical application of electron beam imaging, showing the result of imaging several surgical clips, placed on top of the thorax phantom, underneath a 2.5 cm thickness of sheet bolus, using a 16 MeV electron beam calibrated using a 12 MeV flood field image filtered by a 6 cm block of plastic. This scenario is intended to simulate the clinical use of an electron beam to
123
Australas Phys Eng Sci Med
image a breast tumour bed. Although window levels in Fig. 6a, b have been adjusted to show the surgical clips as well as the imaged anatomy as clearly as possible, the group of surgical clips is only faintly visible against the dark background of tissue- and lung-equivalent material. These small clips (6 mm thick, 6 mm long) would likely be indistinguishable if imaged in a patient with more complex bony anatomy, more heterogeneous lung tissues and a more irregular external contour than the simple thorax phantom used here. Nonetheless, comparison of the electron imaging result shown in Fig. 6a, b with the photon imaging result shown in Fig. 6c, d leads to the important observation that even in this challenging situation, the electron beam images have a similar quality to photon beam images of the same subject.
Discussion
Fig. 3 SNR in a lung, b bone and c tissue, in EPID images obtained by imaging using photon beams (black data points) and electron beams (all other data points). The nominal energy of the flood fields used to calibrate the EPID for electron beam imaging increases from 6 MeV (white data points) up to 20 MeV (dark grey data points), with increasingly dark data points used to indicate increasing flood field energy [as indicated by labels on (a)]. Lines interpolate between data points as a visual guide only
Fig. 4 a Lung-tissue and b bone-tissue contrast, in EPID images obtained by imaging using photon beams (black data points) and electron beams (data points as described in caption for Fig. 3).
The results of this study suggest that electron radiotherapy beams can be used to produce EPID images with levels of contrast and noise that are similar to the contrast and noise levels seen in clinically acceptable photon beam EPID images, if the EPID is calibrated for ‘electron’ imaging by physically filtering out the electron beam, leaving only the bremsstrahlung photon component. If only one electron beam energy can be used for EPID calibration, then an intermediate electron beam energy should be used; in this work 12 MeV was identified as preferable to higher and lower electron beam energies, when used to flood-field correct EPID images acquired with all other available images. If the filtered flood field method proposed in this work is utilised clinically, then EPID images produced during
Example error bars are shown, for clarity. Lines interpolate between data points as a visual guide only
123
Australas Phys Eng Sci Med
Fig. 5 a and b lung-tissue and c and d bone-tissue CNR, in EPID images obtained by imaging using photon beams (black data points) and electron beams (all other data points). CNR is plotted against imaging beam energy in (a) and (c), where data points as described in
caption for Fig. 3. CNR is plotted against flood-field calibration beam energy in (b) and (d), where the shading of the electron data points indicates the nominal energy of the imaging beam. Lines interpolate between data points as a visual guide only
electron treatments may become useful for post-treatment setup verification and review. In such cases the entire dose from each electron treatment fraction would be used to acquire the image; there would be no additional imaging dose to the patient. Depending on the patient management software used by the department, it may or may not be possible to schedule electron beam imaging during treatment. In departments where software permits electron beam images to be added to the treatment plan, the acquisition and use of these images would follow a similar workflow to the acquisition and use of during-treatment photon beam images. However, in departments where electron beam images cannot be added to the treatment plan, either during treatment planning, plan preparation or using the treatment console immediately prior to delivery, then the EPID image acquisition software may need to be run independently while the treatment is delivered using the linac in clinical mode. The location of the treatment aperture relative to bony anatomy or lung-tissue interfaces may be identified as easily using electron beams as using photon beams, provided that the EPID is calibrated appropriately. Localisation of the treatment aperture relative to the location of the
tumour bed, as indicated by surgical clips, has not been conclusively demonstrated in this study. Results shown in Fig. 6a–d indicate that adequate imaging of surgical clips can be challenging when using both photon and electron beams. However, only one (very small and thin) type of surgical clip was used in this work. Further testing and discussions with referring thoracic surgeons is recommended in order to evaluate the suitability of more radiologically opaque clips or markers for this purpose. The positive results of this study suggest that EPID imaging of planar homogeneous phantoms using electron beams may be useful for routine linac quality assurance, if the constancy of the (relative) symmetry and flatness of the bremsstrahlung component of each electron beam can be shown to be indicative of the constancy of the symmetry and flatness of the overall electron beam. Note that it is not possible to recommend the use of the EPID to measure absolute symmetry and flatness (such as might be measured using a water tank), due to the inherent asymmetry of EPID support arm backscatter (leading to measured asymmetry) [43] as well as the known over-response of amorphous silicon EPIDs to low energy photons (leading to measured over-flatness) [44].
123
Australas Phys Eng Sci Med Fig. 6 Images of thorax phantom, with surgical clips and bolus: a image acquired using 16 MeV electron beam (calibrated using 12 MeV filtered flood field); b magnified view of surgical clips from (a); c image acquired using 6 MV photon beam; d magnified view of surgical clips from (c). Group of surgical clips is circled in (a), (b), (c) and (d)
This proof-of-concept study suggests several important avenues for future investigation. For example, the complex relationship between lung-tissue contrast and imaging electron beam energy (decreasing contrast with increasing energy from 6 to 16 MeV before a dramatic contrast increase at 20 MeV) could be investigated further via photon beam imaging using a high-energy (up to 23 MV) linear accelerator. The physical reasons for EPID image quality improvements when calibrating with an intermediate electron beam energy (12 MeV) and imaging with a high electron beam energy (16 or 20 MeV) could be investigated via Monte Carlo simulation or by direct analysis of interaction cross-sections in lung, bone, tissue, gadolinium oxysulfide and silicon. The broader applications of this work, for electron beam quality assurance, could be investigated using repeated measurements over a long period of time or over several different linacs.
Conclusion It is possible to use electron radiotherapy beams to produce EPID images showing tissue and bony anatomy with levels of contrast and noise that are similar to the contrast and
noise levels seen in clinically acceptable photon beam EPID images. In order to achieve these results, however, it is necessary to both increase the MU/frame used to acquire the images and to physically filter the primary electron beam out of the flood field images and thereby calibrate the EPID using only the bremsstrahlung photon component of each electron beam. If these changes are not made, then the resulting electron beam images can be noisy and difficult to interpret. SNR, CNR and contrast values obtained from electron beam images acquired using an EPID calibrated using the filtered flood field method have been shown, in this study, to be comparable to SNR, CNR and contrast values obtained using standard photon beam EPID imaging, with the best results being achieved when images are acquired using the higher energy electron beams (16 and 20 MeV) when the EPID is calibrated using an intermediate (12 MeV) electron beam energy. These results demonstrate the feasibility of acquiring clinically-useful EPID images of patient anatomy using electron beams and suggest important avenues for future investigation, and therefore are expected to enable and encourage further research in this area. There is manifest potential for the EPID imaging method proposed in this
123
Australas Phys Eng Sci Med
work to lead to the clinical use of electron beam imaging for geometric verification of electron treatments in the future. Acknowledgments The authors wish to acknowledge the valued contributions of the following participants at the Australasian College of Physical Scientists and Engineers in Medicine (ACPSEM) Queensland Branch 2014 Winter School on Scientific Publication: Jacqueline Charles, Paul Charles, Allison Fox, Robin Hill, Benjamin Harris, Emma Inness, Vaughan Moutrie, Zoe¨ Moutrie, Patrick O’Connor, Bess Sutherland, Steven Sylvander, Luke Webb, Rachael Wilks and Nancy Yu. This work was supported by the Australian Research Council, the Wesley Research Institute, Premion (Genesis Cancer Care Queensland) and the Queensland University of Technology (QUT), through linkage Grant Number LP110100401.
References 1. El-Gayed AA, Bel A, Vijlbrief R, Bartelink H, Lebesque JV (1993) Time trend of patient setup deviations during pelvic irradiation using electronic portal imaging. Radiother Oncol 26(2):162–171 2. The Royal Australian and New Zealand College of Radiologists, Faculty of Radiation Oncology Position Paper: Techniques and Technologies in Radiation Oncology (2012) Horizon Scan Australia, RANZCR 3. Bridge P, Carmichael MA, Brady C, Dry A (2013) A snapshot of radiation therapy techniques and technology in Queensland: an aid to mapping undergraduate curriculum. J Med Radiat Sci 60(1):25–34 4. Boyer AL, Antonuk L, Fenster A, Van Herk M, Meertens H, Munro P, Reinstein LE, Wong J (1992) A review of electronic portal imaging devices (EPIDs). Med Phys 19(1):1–16 5. McDermott LN, Nijsten SMJJG, Sonke JJ, Partridge M, Van Herk M, Mijnheer BJ (2006) Comparison of ghosting effects for three commercial a-Si EPIDs. Med Phys 33(7):2448–2451 6. Greer PB, Popescu CC (2003) Dosimetric properties of an amorphous silicon electronic portal imaging device for verification of dynamic intensity modulated radiation therapy. Med Phys 30(7):1618–1627 7. Flampouri S, McNair HA, Donovan EM, Evans PM, Partridge M, Verhaegen F, Nutting CM (2005) Initial patient imaging with an optimised radiotherapy beam for portal imaging. Radiother Oncol 76(1):63–71 8. Kairn T, Khoei S, Markwell TS, Fielding AL, Trapp JV (2010) Contrast enhancement of EPID images via difference imaging: a feasibility study. Phys Med Biol 55(22):N533–N544 9. Pouliot J, Bani-Hashemi A, Chen J, Svatos M, Ghelmansarai F, Mitschke M, Aubin M, Xia P, Morin O, Bucci K, Roach M III, Hernandez P, Zheng Z, Hristov D, Verhey L (2005) Low-dose megavoltage cone-beam CT for radiation therapy. Int J Radiat Oncol Biol Phys 61(2):552–560 10. Hansen EK, Larson DA, Aubin M, Chen J, Descovich M, Gillis AM, Morin O, Xia P, Pouliot J (2006) Image-guided radiotherapy using megavoltage cone-beam computed tomography for treatment of paraspinous tumors in the presence of orthopedic hardware. Int J Radiat Oncol Biol Phys 66(2):323–326 11. Markwell T, Perera L, Trapp J, Fielding A (2014) Evaluation of MegaVoltage Cone Beam CT image quality with an unmodified Elekta Precise Linac and EPID: a feasibility study. Australas Phys Eng Sci Med 37(2):291–302
123
12. Moutrie V, Kairn T, Rosenfeld A, Charles PH (2015) Use of a megavoltage electronic portal imaging device to identify prosthetic materials. Australas Phys Eng Sci Med 38(1):93–100 13. Liu G, Van Doorn T, Bezak E (2002) Assessment of flatness and symmetry of megavoltage X-ray beam with an electronic portal imaging device (EPID). Australas Phys Eng Sci Med 25(2):58–66 14. Budgell GJ, Zhang R, Mackay RI (2007) Daily monitoring of linear accelerator beam parameters using an amorphous silicon EPID. Phys Med Biol 52(6):1721–1733 15. Clivio A, Vanetti E, Rose S, Nicolini G, Belosi MF, Cozzi L, Baltes C, Fogliata A (2015) Evaluation of the machine performance check application for TrueBeam Linac. Radiat Oncol 10(1):97 16. Wendling M, Louwe RJ, McDermott LN, Sonke JJ, van Herk M, Mijnheer BJ (2006) Accurate two-dimensional IMRT verification using a back-projection EPID dosimetry method. Med Phys 33(2):259–273 17. Warkentin B, Steciw S, Rathee S, Fallone BG (2003) Dosimetric IMRT verification with a flat-panel EPID. Med Phys 30(12): 3143–3155 18. Van Esch A, Depuydt T, Huyskens DP (2004) The use of an aSibased EPID for routine absolute dosimetric pre-treatment verification of dynamic IMRT fields. Radiother Oncol 71(2):223–234 19. Nicolini G, Fogliata A, Vanetti E, Clivio A, Cozzi L (2006) GLAaS: an absolute dose calibration algorithm for an amorphous silicon portal imager. Applications to IMRT verifications. Med Phys 33(8):2839–2851 20. Nicolini G, Vanetti E, Clivio A, Fogliata A, Korreman S, Bocanek J, Cozzi L (2008) The GLAaS algorithm for portal dosimetry and quality assurance of RapidArc, an intensity modulated rotational therapy. Radiat Oncol 3:24 21. Lee C, Menk F, Cadman P, Greer PB (2009) A simple approach to using an amorphous silicon EPID to verify IMRT planar dose maps. Med Phys 36(3):984–992 22. Mans A, Wendling M, McDermott LN, Sonke JJ, Tielenburg R, Vijlbrief R, Mijnheer B, Van Herk M, Stroom JC (2010) Catching errors with in vivo EPID dosimetry. Med Phys 37(6):2638–2644 23. Mijnheer B, Mans A, Olaciregui-Ruiz I, Sonke JJ, Tielenburg R, Van Herk M, Vijlbrief R, Stroom J (2010) 2D AND 3D dose verification at The Netherlands Cancer InstituteAntoni van Leeuwenhoek Hospital using EPIDs. J Phys Conf Ser 250(1): 012020 24. van Elmpt W, McDermott L, Nijsten S, Wendling M, Lambin P, Mijnheer B (2008) A literature review of electronic portal imaging for radiotherapy dosimetry. Radiother Oncol 88(3): 289–309 25. Kairn T, Cassidy D, Sandford PM, Fielding AL (2008) Radiotherapy treatment verification using radiological thickness measured with an amorphous silicon electronic portal imaging device: Monte Carlo simulation and experiment. Phys Med Biol 53(14):3903–3919 26. Kavuma A, Glegg M, Metwaly M, Currie G, Elliott A (2010) A novel method for patient exit and entrance dose prediction based on water equivalent path length measured with an amorphous silicon electronic portal imaging device. Phys Med Biol 55(2):435–452 27. Kairn T, Crowe SB, Trapp JV (2012) A simple method for EPIDbased in vivo dosimetry for radiotherapy treatments of head-andneck cancers. IFMBE Proc 39:1727–1730 28. Beck JA, Budgell GJ, Roberts DA, Evans PM (2009) Electron beam quality control using an amorphous silicon EPID. Med Phys 36(5):1859–1866
Australas Phys Eng Sci Med 29. Wang Y, Heaton R, Norrlinger B, Islam M (2013) Quality assurance of electron beams using a Varian electronic portal imaging device. Phys Med Biol 58(16):5461–5475 30. Emam I, Fan J, Jin L, El Dib A, Ma C (2009) SU-FF-T-307: a multileaf collimator quality assurance tool for modulated electron radiation therapy. Med Phys 36(6):2592 31. Gauer T, Engel K, Kiesel A, Albers D, Rades D (2010) Comparison of electron IMRT to helical photon IMRT and conventional photon irradiation for treatment of breast and chest wall tumours. Radiother Oncol 94(3):313–318 32. Jarry G, Verhaegen F (2005) Electron beam treatment verification using measured and Monte Carlo predicted portal images. Phys Med Biol 50(21):4977–4994 33. Aubin M, Faddegon B, Pouliot J (2002) Electron beam verification with an a-Si flat-panel electronic portal imaging device. Proc SPIE 4682:549–557 34. Shank B, Moughan J, Owen J, Wilson F, Hanks GE (2000) The 199394 patterns of care process survey for breast irradiation after breast-conserving surgerycomparison with the 1992 standard for breast conservation treatment. Int J Radiat Oncol Biol Phys 48(5):1291–1299 35. IAEA (2004) Technical Reports Series 430. Commissioning and quality assurance of computerized planning systems for radiation treatment of cancer. International Atomic Energy Agency, Vienna 36. IAEA (2008) TECDOC 1583. Commissioning of radiotherapy treatment planning systems: testing for typical external beam treatment techniques. International Atomic Energy Agency, Vienna
37. Louwe RJ, McDermott LN, Sonke JJ, Tielenburg R, Wendling M, Van Herk MB, Mijnheer BJ (2004) The long-term stability of amorphous silicon flat panel imaging devices for dosimetry purposes. Med Phys 31(11):2989–2995 38. Ding GX, Rogers DWO, Mackie TR (1996) Mean energy, energy-range relationships and depth-scaling factors for clinical electron beams. Med Phys 23(3):361–376 39. Faddegon B, Schreiber E, Ding X (2005) Monte Carlo simulation of large electron fields. Phys Med Biol 50(5):741–753 40. Mooslechner M, Mitterlechner B, Weichenberger H, Huber S, Sedlmayer F, Deutschmann H (2013) Analysis of a free-running synchronization artifact correction for MV-imaging with aSi: H flat panels. Med Phys 40(3):031906 41. Woodruff HC, Greer PB (2013) 3D Dose reconstruction: banding artefacts in cine mode EPID images during VMAT delivery. J Phys Conf Ser 444:012042 42. Herman MG, Balter JM, Jaffray DA, McGee KP, Munro P, Shalev S, Van Herk M, Wong JD (2001) Clinical use of electronic portal imaging: Report of AAPM Radiation Therapy Committee Task Group 58. Med Phys 28(5):712–737 43. Rowshanfarzad P, McCurdy BM, Sabet M, Lee C, OConnor DJ, Greer PB (2010) Measurement and modeling of the effect of support arm backscatter on dosimetry with a Varian EPID. Med Phys 37(5):2269–2278 44. Kirkby C, Sloboda R (2005) Consequences of the spectral response of an a-Si EPID and implications for dosimetric calibration. Med Phys 32(8):2649–2658
123
