Medical Dosimetry journal homepage: www.meddos.org
Dosimetric verification of stereotactic radiosurgery/stereotactic radiotherapy dose distributions using Gafchromic EBT3 Davide Cusumano, M.Sc.,* Maria L. Fumagalli, M.Sc.,† Marcello Marchetti, M.D.,‡ Laura Fariselli, M.D.,‡ and Elena De Martin, M.Sc.† *
School of Medical Physics, University of Milan, Milan, Italy; †Health Department, Fondazione IRCCS Istituto Neurologico Carlo Besta, Milan, Italy; and Department of Neurosurgery, Radiotherapy Unit, Fondazione IRCCS Istituto Neurologico Carlo Besta, Milan, Italy
‡
A R T I C L E I N F O
A B S T R A C T
Article history: Received 19 June 2014 Received in revised form 17 December 2014 Accepted 7 January 2015
Aim of this study is to examine the feasibility of using the new Gafchromic EBT3 film in a high-dose stereotactic radiosurgery and radiotherapy quality assurance procedure. Owing to the reduced dimensions of the involved lesions, the feasibility of scanning plan verification films on the scanner plate area with the best uniformity rather than using a correction mask was evaluated. For this purpose, signal values dispersion and reproducibility of film scans were investigated. Uniformity was then quantified in the selected area and was found to be within 1.5% for doses up to 8 Gy. A high-dose threshold level for analyses using this procedure was established evaluating the sensitivity of the irradiated films. Sensitivity was found to be of the order of centiGray for doses up to 6.2 Gy and decreasing for higher doses. The obtained results were used to implement a procedure comparing dose distributions delivered with a CyberKnife system to planned ones. The procedure was validated through single beam irradiation on a Gafchromic film. The agreement between dose distributions was then evaluated for 13 patients (brain lesions, 5 Gy/die prescription isodose 80%) using gamma analysis. Results obtained using Gamma test criteria of 5%/1 mm show a pass rate of 94.3%. Gamma frequency parameters calculation for EBT3 films showed to strongly depend on subtraction of unexposed film pixel values from irradiated ones. In the framework of the described dosimetric procedure, EBT3 films proved to be effective in the verification of high doses delivered to lesions with complex shapes and adjacent to organs at risk. & 2015 American Association of Medical Dosimetrists.
Keywords: Dosimetric verification Gafchromic EBT3 Stereotactic radiosurgery/radiotherapy High dose
Introduction In the framework of the clinical management of neoplastic lesions, stereotactic radiosurgery (SRS) and stereotactic radiotherapy (SRT), despite being relatively new, play a well-established role nowadays. Their ability to deliver extremely steep dose gradients allows the treatment of such lesions to high therapeutic doses, while effectively ensuring organs at risk (OARs) sparing. The safety and effectiveness of the clinical outcome for such modalities heavily rely on an extremely accurate treatment administration. For this reason, SRS/SRT systems are usually provided with imaging devices that allow a precise localization of the target volumes. Millimeter and even submillimeter spatial accuracy of the administered dose can thus be obtained during the whole delivery process. In this perspective, consistency between calculated and delivered dose distributions is of crucial importance,
and actually irradiated dose distribution verification becomes a relevant step to assess the quality of the radiotherapy treatment. Owing to the complexity of SRS/SRT delivery techniques, the choice of a proper detector for individual patient plan verification is critical. Certain characteristics, such as high spatial resolution, weak or absent energy, and dose-rate dependence are necessary for measurements in radiation treatments with high dose gradients. Considering also their near-tissue equivalence, EBT-type Gafchromic films (ISP, Wayne, NJ) are suitable detectors for the dosimetry of radiosurgical treatments.1,2 General characteristics, cautions, and techniques about using Gafchromic films for dosimetric purposes have been widely addressed elsewhere.3-5 In this study, we intended to evaluate and optimize the use of the new EBT3 films to implement a dosimetric protocol specific for radiosurgery patient plan verification.
Methods Reprint requests to: Davide Cusumano, School of Medical Physics, University of Milan, Via Celoria 16, Milano 20133, Italy. E-mails: [email protected], [email protected] http://dx.doi.org/10.1016/j.meddos.2015.01.001 0958-3947/Copyright Ó 2015 American Association of Medical Dosimetrists
A quality assurance (QA) dosimetric protocol was developed using Gafchromic EBT3 (batch number: AO4041203) in conjunction with the flatbed scanner Epson Expression 10000 XL (Seiko Epson Corp., Nagano, Japan).
D. Cusumano et al. / Medical Dosimetry ] (2015) ]]]–]]]
2
This protocol was then optimized to evaluate dose distributions effectively delivered with a CyberKnife system, version 9.6 (Accuray, Sunnyvale, CA) in comparison with planned ones.
borders. The relevance of using net exposure values in the comparison of dose distributions was evaluated for a sample of patients using gamma analysis tool.10 Sensitivity and high-dose threshold evaluation
Determination of image acquisition parameters A scan result can be influenced by many different factors. Some features of the Gafchromic scanning system have been addressed elsewhere and will be adopted in our study (scanner warm up, transmission mode, 48 bit-RGB mode, tagged image file format, and portrait scanning film orientation), while others require a specific customized analysis.6,7 In particular, scanning resolution in published works is usually set at 72 dpi consistently with manufacturing recommendations.1,5 We considered this resolution to be suboptimal in the case of CyberKnife system, which is equipped with target tracking tools allowing for correction and subsequent verification also of interfraction submillimeter movements. For this reason, in this work, we evaluated the feasibility of using a higher resolution by investigating the effects on image quality in conjunction with scanning area dimension influence. Five acquisitions of the same unirradiated film were performed both including the whole scanner plate (41.91 30.99 cm2) and limited to the film area (25.10 20.48 cm2) for 3 different image resolutions (72, 150, and 300 dpi). Differences on the image reading were evaluated in percentage standard deviation (SD) (2σ %) with respect to the mean signal.
To correctly evaluate the outcome of dose distribution verifications, sensitivity variation with dose of the dosimetry system was analyzed. Measurements were performed irradiating 4 4 cm2 films with doses ranging from 1 to 8 Gy (0.3, 0.5, 0.8, 1.0, 1.2, 2.0, 2.2, 3.0, 3.2, 4.0, 4.2, 5.0, 5.2, 6.0, 6.2, 7.0, 7.2 8.0 and 8.2 Gy). Films were irradiated within a solid water slab phantom, exposed perpendicularly to the beam axis, at 5 cm depth from the phantom surface (isocentric setup) and with 15 cm solid water thickness placed to produce backscattered radiation. An absolute dose measurement was contextually performed for verification, positioning an ion chamber Farmer FC65-P (Scanditronix Medical AB, Uppsala, Sweden) at 7 cm depth from the phantom surface. The chamber reading at 7 cm was scaled to the value of film depth, at 5 cm. (Preliminary measurements were performed at 7 and 5 cm depth to obtain the applied conversion factor.) Four films were irradiated for each dose value and every film was digitized, applying the same scan settings used for uniformity evaluation. Pixel values interval between the dose levels, which differ by 20 cGy, were considered and the higher distinguishable dose levels were examined to establish a dose threshold for film QA verification employing our procedure. Patient selection for validation of the optimized dosimetric procedure was consequently performed according to this dose threshold.
Scanning area selection and evaluation Dose characterization and measurement Another problem is the effect of scanning nonuniformity on the radiochromic measurements, which varies depending on the film optical density. Some studies have proposed a dose-dependent 2-dimensional correction mask.6,8 This is advisable in case of extended dose distributions, such as those associated with head and neck or prostate tumors with irradiation of the lymph node chain. Owing to its specific application to radiosurgery treatments, in this work, we proposed a different approach. In particular, we verified the feasibility of determining the maximum uniformity area on our scanner plate and scanning the Gafchromic film in that area. This method can be taken into consideration because of the specific nature of radiosurgery treatments, being that the involved lesion and corresponding verification film dimensions are generally not bigger than a few linear centimeters (maximum linear dimension 5.5 cm for this study). The first step was to evaluate scanner response in the direction of the charge-coupled device array movement.8 Five unexposed Gafchromic films, belonging to the same batch were scanned to minimize the effect of uniformity variability among different films.7 Each film was positioned in the center of the scanner plate, with the long side parallel to the scanning direction (portrait scanning orientation). It was also acquired 5 times, in order to allow removal of the scanner noise by averaging different images. Film analysis was always performed using ImageJ software, version 1.43u (NIH, Bethesda, MD) and selecting the red channel. Stripes of 2 cm that corresponded to the border of the Gafchromic sheet were excluded from this analysis, being usually affected by a higher dishomogeneity of the active layer or by cutting damage. To estimate uniformity in the landscape, direction was estimated for the remaining film area, divided into 14 rectangular 21.5 1.0 cm2 regions of interest (ROIs), symmetrically identified with respect to the center (ROI 0) of the Gafchromic sheet. Each ROI was then collapsed to produce a linear profile, obtained by averaging the corresponding values of the 5 films (each film scanned 5 times).9 An index, named Δ, was introduced to give a quantitative evaluation of uniformity of the ROIs. The Δ index is defined as Δ¼
2σ % opX 4
ð1Þ
where σ is the SD in pixel values and oPx4 is the mean pixel value of the profile analyzed. The same procedure was adopted also for the portrait direction analysis and the collected data were used to determine a 10 10 cm2 scanner region presenting the best uniformity. All further film considerations were carried out limitedly to this region, and uniformity was evaluated anew in this area both for unirradiated and irradiated 1010-cm2 films (6 MV Linac, 4 and 8 Gy). Unlike other studies involving the Epson Expression 10000XL scanner, the dose escalation up to 8 Gy is necessary because of the high doses found in patient-specific SRS/SRT QA plans. All films irradiated in this study were digitized 24 hours after exposure to ensure that the polymerization process in Gafchromic EBT3 active layer was completely and similarly carried out.4,8 Subtraction of the unirradiated film values Evidence aroused during our film analyzes that patient QA results using Gafchromic EBT3 films depend on the subtraction of the corresponding unirradiated pixel values from the irradiated films. The extent of pixel value variability for unirradiated films was evaluated for 13 pieces belonging to different sheets later used for patient dose distribution verification. Film dimensions were 8 8 cm2 but only the central 7 7 cm2 part was considered to exclude cutting damage of the
In this study, dosimetric verification of dose distributions of radiosurgical clinical interest (up to 8 Gy) was performed. The dose-response curve was determined using the same experimental setup described in the previous paragraph. After subtraction of the unirradiated film values, images of films exposed to the same dose were averaged using ImageJ and the mean pixel value was calculated for the resulting one over a 2 2 cm2 centered ROI area. Experimental data were fitted by a fourth-degree polynomial curve.6 Calibration curve was used in a QA protocol to verify patient-specific dose distributions, delivered with a CyberKnife system. Clinically administered dose distributions were reported on an EasyCube cubic phantom (Sun Nuclear, Melbourne, FL), maintaining both treatment beams ballistic and monitor units. A Gafchromic EBT3 film was inserted in the phantom between the 2 central slabs, oriented in the axial direction. Its position with respect to the clinical volume of interest was registered by automatically aligning the centers of the 2 structures (CyberKnife Multiplan Treatment Planning System option, version 4.6). Maximum dimensions of the film were chosen to include the tumoral lesion and always smaller than 10 10 cm2. Phantom alignment on the treatment couch was guaranteed by matching the positions of 8 fiducials (located in 2 different slabs— cranial and caudal—with respect to the film position) on digitally reconstructed radiograph (DRR) and live radiographic images. Films were always scanned before irradiation for unirradiated values recording. Irradiated films were then scanned applying the measures described in the previous paragraphs.
Results Determination of image acquisition parameters Results regarding image acquisition resolution and signal dispersion are reported in Table 1. Percentage SD (2σ %) with respect to the mean signal confirms to remain low also at 150 dpi, and inclusion of the whole scanner area in the image acquisition process gives the best result. Scanning area selection and evaluation Unirradiated film ROIs (21.5 1 cm2), collapsed to linear profiles, are shown in Fig. 1. As previously reported, linear profiles were used to determine the highest uniformity area on the scanner plate. Maximum Table 1 Signal dispersion values for 3 image resolutions in the case of acquisitions including whole scanner plate or limited to film area Resolution (dpi)
2σ % (Film area)
2σ % (Whole plate)
72 150 300
0.80 0.68 0.86
0.80 0.64 0.80
D. Cusumano et al. / Medical Dosimetry ] (2015) ]]]–]]]
3
Fig.1. Linear profile in the landscape direction and corresponding to the central axis of the unirradiated film (ROI 0).
Table 2 Maximum and mean Δ values for the 10 10 cm2 region for blank and irradiated (4 and 8 Gy) films Film type
Blank 4 Gy 8 Gy
Landscape
Portrait
Mean Δ value (%)
Max Δ value (%)
Mean Δ value (%)
Max Δ value (%)
0.38 0.62 1.02
0.40 1.01 1.24
0.70 1.24 1.21
0.74 1.47 1.50
Fig. 2. Pixel value dispersion as a function of unirradiated piece number.
normalized SDs values Δ across the film were calculated to be 1.88 and 0.71 for portrait and landscape directions, respectively. Linear profile regions with the lower Δ values for both landscape and portrait directions were identified to produce the 10 10 cm2 region presenting the best uniformity and maximum and mean Δ values for the this region are listed in Table 2 for unirradiated and irradiated (4 and 8 Gy) films. As expected, Δ values are higher in the portrait direction and increase for the irradiated film. Maximum Δ value was found to be within 1.5% in the portrait direction for the film irradiated at 8 Gy. Subtraction of the unirradiated film values Subtraction of the unirradiated film values topic has been addressed in this study for EBT3 Gafchromic films in pixel value dispersion analyses. Figure 2 shows pixel value dispersion as a function of unirradiated piece number for the 13 considered films. Maximum inhomogeneity value for 1 film was 0.8% (1 SD, piece no 7) whereas considering mean pixel values of film pieces coming from different sheets dispersion raised to 1.4% (1 SD). Sensitivity and high-dose threshold evaluation In our study, the EBT3 films sensibility dose dependence was analyzed for a fine sampling of dose values by studying the pixel values interval between 2 consecutive levels differing by 20 cGy (range: 0.3 to 8.2 Gy). In Fig. 3, mean pixel values oPx4 corresponding to each film dose are shown with the associated 2σ SD error bars. The hypothesis of linear behavior was assumed for pixel values
increment in the interval portion included between the upper and lower extremities of the error bars. According to this consideration, we found our dosimetry system to be able to discriminate dose values of radiosurgical clinical interest, with sensitivity up to 6.2 Gy varying from 75Px/cGy to 14Px/cGy. At greater than the 6.2Gy threshold dose level, error bars are not so clearly separated. Dose characterization and measurement Following the previous results, a dosimetry protocol was established for dose distribution verification. Calibration data (dose-response curve is shown in Fig. 4) were introduced into the Wellhofer-Scanditronix OmniPro ImRT 1.7 analysis software. Planned and Gafchromic dose distributions were also imported into the software and renormalized to the prescription dose (100% ¼ 5 Gy). The protocol was first validated irradiating a single beam (collimator aperture ¼ 6 cm, prescription dose ¼ 5 Gy) on a Gafchromic film with the experimental setup described in the dose characterization and measurement paragraph. Planned and effectively irradiated dose distributions were compared using gamma analysis tool10 (ΔDM ¼ 3% ΔdM ¼ 1 mm as acceptance criteria), as shown in Fig. 5. The pass rate for single beam gamma analysis confirmed to be very good (99.18%). The previously described dosimetry procedure was then used to verify the agreement between expected and measured dose distributions for 13 intracranial lesions. Average dimensions of the target in the axial plane were 3.2 cm (range: 1.5 to 5.5 cm) and
D. Cusumano et al. / Medical Dosimetry ] (2015) ]]]–]]]
Dose (Gy)
4
8.20 8.00 7.20 7.00 6.20 6.00 5.20 5.00 4.20 4.00 3.20 3.00 2.20 2.00 1.20 1.00 0.80 0.50 0.30 0
3000
6000
9000
12000
15000
18000
21000
24000
27000
30000
33000
Fig.3. On the x-axis, mean pixel values oPx4 are shown with the associated 2σ standard deviation error bars. On the y-axis, corresponding film dose values are reported.
40000 35000
Pixel Value
30000 25000 20000 15000 10000 5000 0 0
1
2
3
4
5
6
7
8
9
Dose (Gy) Fig.4. CyberKnife dose calibration curve.
(5 Gy/die, I95% range: 76% to 86%). Patient dose distributions relative to one treatment fraction were delivered on 8 8 cm2 Gafchromic films, thus ensuring not to exceed the threshold dose level of 6.2 Gy established with the sensitivity analysis. Comparison was carried out using the gamma analysis tool on a ROI composed by the entire film extent with the exclusion of a 1-cmwide frame region. In this way, analysis errors deriving from cutting damage of the film or writings on the border were avoided. Acceptance criteria selected for this study were ΔDM ¼ 5% ΔdM¼1 mm. The agreement between expected and irradiated dose distributions was evaluated according to the following γ frequency parameters, determined for the 13 patients11: (1) Mean γ value (2) Percentage of points with γ o 1, γ o 1.5, and γ 4 2
Fig. 5. Gamma value resulting from comparison of planned and delivered dose distribution for single beam irradiation. (Color version of figure is available online.)
2.5 cm (range: 1.1 to 4.3 cm) for anterior-posterior and laterolateral directions, respectively. For patients, prescription dose was 25 Gy to the isodose enclosing 95% (I95%) of the treatment volume
The obtained results are reported in Table 3 as averages of the γ parameters over all the lesions (group 1). Averaged gamma frequency parameters calculated without unirradiated film values subtraction (group 2) are also reported in Table 3, showing a pass-rate reduction of approximately 20% when compared with group 1 for which the subtraction process was performed.
D. Cusumano et al. / Medical Dosimetry ] (2015) ]]]–]]] Table 3 Averaged gamma frequency parameters with (group 1) and without (group 2) blank film subtraction
Group 1 Group 2
0 o γ o 1 (%)
1 o γ o 1.5 (%)
1.5 o γ o 2
94.30 74.21
3.89 12.27
1.82 13.52
Discussion Considering the results regarding image acquisition parameters in contradistinction to other studies, we decided to implement a scanning film procedure at a resolution of 150 dpi. We also decided to always include the whole plate area, giving this a better image reading in signal dispersion than scanning the film area alone. This choice allows to keep signal dispersion range at a minimum and to investigate submillimeter resolutions that are comparable to SRS/SRT spatial accuracy requirements. Values for the Δ index reported in Table 2 are consistent with the 1.71% and 1% SDs (for unirradiated films and films irradiated at 3 Gy, respectively) found by Alnawaf et al.9 when considering normalized optical density variation deriving from of a 1-cm linear profile centered on the plate of an Epson 10000XL scanner. Additionally, Huet et al. scanned a piece of a transparent blue sheet of 3 3 cm2 at 15 different positions. The pixel value dispersion, extracted from the 5 3 positions, was 0.7%. According to these considerations, our results allow us to perform dose distribution verification also at high doses without any correction factors on the identified area, which was marked on the scanner plate for proper film positioning and evaluation. The necessity of subtracting the unirradiated film values has been considered for EBT2 films. The behavior of the film uniformity was found by Huet et al.4 to be better than 0.3% for an unirradiated Gafchromic films and around 1.2% in dose for an irradiated sheet. These results were reported to be in good agreement with those obtained by Richley et al., whereas Hartmann et al.13 found a larger inhomogeneity (3.7% in pixel value).4,12 In this study, film uniformity was evaluated for EBT3 films. The results reported in Fig. 2 led to the decision of implementing subtraction of the unirradiated film values by scanning each Gafchromic film before irradiation and calculating the mean value of the pixels of the resulting image. The mean value was then subtracted from each pixel of the same film after irradiation to obtain net exposure values correctly estimating the administered dose. As the aim of this study was the evaluation of high dose distributions, we also deemed it necessary to perform sensitivity analyses to identify the higher dose distinguishable in the framework of our dosimetry protocol using EBT3 films. Results in Fig. 3 show dose levels to be clearly separated up to 6.2 Gy. We therefore considered this value to be the upper limit for a reliable escalation of doses, which could be checked in a dose distribution comparison process. According to these considerations, film QA verification for this study was performed for patients with a prescription fraction dose of 5 Gy. To our knowledge, still only a few articles in literature address the use of EBT3 films in routine patient dose distribution verifications, usually considering standard dose levels ( 2 Gy). Casanova Borca et al.14 reported results of 10 intensity-modulated radiation therapy cases as mean value and SD of the percentage of points satisfying the constraint gamma o1. Measurements of EBT3 films showed averaged fractions of passed gamma values greater than 99% using 4%, 3 mm (dose difference tolerance 8 and 10 cGy for head and neck and pelvis plans) and 3%, 3 mm gamma evaluation criteria. Anyway, when the tightest criteria were used (2%, 2 mm), an average gamma pass-rate reduction of approximately 15% was
5
observed. Micke and Lewis15 compared an irradiated film of an intensity-modulated radiation therapy plan (max dose ¼ 220 cGy) and an irradiated volumetric modulated arc therapy beam (max dose ¼ 125 cGy) to the calculated plans to evaluate a new dosimetry protocol. Gamma function was calculated with test criteria of 2% dose agreement within 2 mm. In our study, dose tolerance level ΔDM was chosen to be 5% in consideration of dose inhomogeneity requirements within the target and geometrical complexity of delivery for the CyberKnife system (more than a hundred non-coplanar beams for each treatment plan), and the statistical fluctuations of the signal during film scanning procedure. Also, it is the tolerance level clinically accepted for patient point dose ionization chamber (cc13/TNC) quality checks in the practice of our institution. The distance tolerance level ΔdM was chosen to be 1 mm in consideration both of the steep dose gradients characterizing peripheral target dose for brain lesions, usually in strict proximity with the OARs and of the CyberKnife system's ability to correct for millimeter intrafraction target movements. One thing to be considered is that clinical quality controls involve the management of discrete dose distributions, and their spatial resolution has to be higher than the tolerance ΔdM for the γ index to be properly calculated. This is ensured in our clinical practice by dose calculation at the high (voxel) resolution for patient and phantom plans and by film scanning procedure at 150 dpi. Comparison between measured and planned dose distributions for the 13 cases considered in this study showed a very good agreement. Averaged gamma frequency parameters calculated with (group 1) and without unirradiated film values subtraction (group 2) are reported in Table 3, showing a good pass rate for group 1. The gamma analysis tool was used to confirm also for EBT3 films the necessity for the unirradiated pixel values correction, being the pass rate for group 2 reduced by approximately 20% with respect to group 1.
Conclusions Despite being relatively new, nowadays SRS/SRT systems play a well-established role in the treatment of complex shape lesions often adjacent to OARs. They deliver extremely steep dose gradients also within the target and the use of the appropriate dosimeter is fundamental in order to obtain an effectively representative evaluation of the agreement between expected and measured dose distributions. This study showed Gafchromic EBT3 films to be effective for such verifications, producing reliable results when escalation from standard ( 2 Gy) to higher SRS/SRT doses was performed. Reduced film dimensions necessary for dose verification of radiosurgical targets allow to consider nonuniform response of the scanning procedure and selected area on the Epson Expression plate as parts of the statistical signal fluctuation. Anyway, in this study, gamma frequency parameters calculation showed subtraction of unexposed film pixel values from irradiated ones to be of crucial importance also for EBT3 films. Noting that the procedure used did not always give the best possible representation for the unirradiated pixel values, we plan to evaluate a more accurate estimate to be used in the subtraction step as a future improvement. Additionally, owing to sensitivity considerations, our QA procedure is limited to a threshold for up to the maximum dose that can be verified. Using the information from the 3 red, green, and blue channels in the scanning step might help to overcome this limit, thus allowing to perform verification of higher dose values. This could be useful to expand the type of treatments that could undergo dosimetric QAs.10
6
D. Cusumano et al. / Medical Dosimetry ] (2015) ]]]–]]]
References 1. Fiandra, C.; Ricardi, U.; Ragona, R.; et al. Clinical use of EBT model Gafchromic film in radiotherapy. Med. Phys. 33:4314–9; 2006. 2. Wilcox, E.; Daskalov, G.M. Evaluation of Gafchromics EBT film for CyberKnifes dosimetry. Med. Phys. 34:1967–74; 2007. 3. Devic, S. Radiochromic film dosimetry: past, present and future. Phys. Med. 27:122–34; 2011. 4. Huet, C.; Dagois, S.; Derreumaux, S.; et al. Characterization and optimization of EBT2 radiochromic films dosimetry system for precise measurements of output factors in small fields used in radiotherapy. Radiat. Meas. 47:40–9; 2012. 5. Martisikova, M.; Ackermann, B.; Jakel, O. Analysis of uncertainties in Gafchromic EBT film dosimetry of photon beams. Phys. Med. Biol. 53:7013–27; 2008. 6. Saur, S.; Frengen, J. Gafchromic EBT film dosimetry with a flatbed CCD scanner: a novel background correction method and full dose uncertainty analysis. Med. Phys. 35:3094–101; 2008. 7. Devic, S.; Seuntjens, J.; Sham, E.; et al. Precise radiochromic film dosimetry using a flat-bed document scanner. Med. Phys. 32:2245–53; 2005. 8. Ferreira, B.C.; Lopes, M.C.; Capela, M. Evaluation of an Epson flatbed scanner to read Gafchromic EBT films for radiation dosimetry. Phys. Med. Biol. 54:1073–85; 2009.
9. Alnawaf, H.; Yu, P.K.; Butson, P.K. Comparison of Epson scanner quality for radiochromic film evaluation. J Appl. Clin. Med. Phys. 5:314–22; 2012. 10. Low, D.A. A technique for the quantitative evaluation of dose distributions. Med. Phys. 25:656–61; 1998. 11. De Martin, E.; Fiorino, C.; Broggi, S.; et al. Agreement criteria between expected and measured field fluences in IMRT of head and neck cancer: the importance and use of the gamma histograms statistical analysis. Radiother. Oncol. 85:399–406; 2007. 12. Reinhardt, S.; Hillbrand, M.; Wilkens, J.J.; et al. Comparison of Gafchromic EBT2 and EBT3 films for clinical photon and proton beams. Med. Phys. 39 (8):5257–62; 2012. 13. Hartmann, B.; Martisikova, M.; Jäkel, O. Technical note: homogeneity of Gafchromic EBT2 film. Med. Phys. 37:1753–6; 2010. 14. Casanova Borca, V.; Pasquino, M.; Russo, G.; et al. Dosimetric characterization and use of GAFCHROMIC EBT3 film for IMRT dose verification. J. Appl. Clin. Med. Phys. 14(2):158–71; 2013. 15. Micke, A.; Lewis, D. Multichannel film dosimetry with nonuniformity correction. Med. Phys. 35:3078–85; 2011.
