Advanced dosimetry near metal implants using radiochromic films

Abstract

Purpose: Patients receiving radiation therapy often have metal implants in or near target volumes which increase the overall uncertainty in the radiotherapy process through interference with the applied dose calculation algorithms in the vicinity of the implant. Hence, this diploma thesis aimed to analyse edge effects caused by metal implants when irradiated with photon and proton beams using radiochromic films as well as to calibrate a new film scanner for enhanced film dosimetry measurements in order to improve future radiotherapy treatment planning in terms of accuracy near metal implants. Methods: For the calibration procedure, 4 × 4 cm2 GafChromic EBT3 films were irradiated with monitor units (MU) 0, 100, 200, 300 and 400 corresponding to 0, 1, 2, 3 and 4 Gy. Subsequently, they were scanned one by one in order to calculate a calibration curve as well as placed in equal distances along the scanner to assess the scanner inhomogeneity. Photon beam experiments took place at the Department of Radiation Oncology, Medical University of Vienna/University Hospital AKH Vienna. The dose distribution measurements with the metal phantom were conducted by irradiating 5.9 × 5.9 cm2 GafChromic EBT3 films at 15° table angle using photon energies 6 MV, 6 MV flattening-filter free (FFF), 10 MV and 10 MV FFF and evaluated using the newly calibrated scanner. Furthermore, reference films were irradiated for the respective photon energies except for 6 MV to test the thickness of the active layer of different film sheets within one lot (manufacturer identification number). The equivalent proton beam experiments took place at MedAustron. Similarly, 5.9 × 5.9 cm2 GafChromic EBT3 films were examined, but instead using a 20° table angle. Film evaluations were performed with the scanner provided at MedAustron. Dose verifications were executed in all dose distribution measurements except photon beam energy 10 MV FFF and the films were irradiated with doses of 2 Gy using three different treatment plans. The Hounsfield units (HU) plan used the original computed tomography (CT) numbers of the CT scan, the polystyrene plan assigned polystyrene density values for the phantom and the titanium plan was overwritten with the titanium density values of the metal screw as well as the polystyrene values of the phantom. Furthermore, two different phantom configurations, namely a metal screw insert and a polystyrene insert, were used. All films were scanned in portrait orientation and used the response from the red color channel. Film analysis was performed with VeriSoft (PTW Dosimetry, Freiburg, Germany) and Python scripts. For meaningful comparison, the films were properly aligned with each other as well as the treatment plans using the pinholes on each film. Furthermore, the films and treatment plans were normalized, re-calibrated and a region of interest (ROI) was determined. At line dose profiles of interest, a mean and sample standard deviation were calculated from the line and its two neighboring lines as well as all the films irradiated with a specific treatment plan. The results were then compared with the treatment plans calculated with the clinical treatment planning system (TPS) RayStation (RaySearch, Stockholm, Sweden) as well as Monte Carlo (MC) simulations performed with the independent secondary dose calculation system SciMoCa (ScientificRT GmbH, Munich, Germany). Lastly, dose differences and a confidence interval (CI) of 95 % were determined for the analysis. Results: The calibration curve and error analysis showed satisfactory results with a mean calibration error of 0.3 % and a standard deviation of 1.7 %. Examination of lateral scanner artefacts revealed discrepancies up to 13 % from the normalized pixel values when scanning films irradiated with 400 MUs in the lateral regions. Photon and proton beam experiments corresponded well with the RayStation and SciMoCa plans for polystyrene measurements, with dose differences generally contained within a 3 % margin. Photon beam experiments using HU-plans matched with MC simulations whereas comparisons with the RayStation plans exhibited dose differences up to approx. 13.6 % where the titanium screw and the tissue equivalent polystyrene material intersected. Meanwhile, Ti-plan simulations showed inaccuracies in the vicinity of the titanium screw for both plans with maximum dose differences around 13.4 % for the RayStation plans and even 25.3 % for the SciMoCa plans due to imprecise titanium screw density data. Regarding the proton beam experiments, increased variability was observed in line dose profiles for films irradiated under the HU- and Ti-plans, likely due to the fixed horizontal beamline and the lack of detailed geometrical data about the screw thread in the TPS. Overall, dose verification using an ionization chamber revealed dose differences ranging from 0.14 to 0.95 % post-output correction for photon beams, and from 0.3 to 1.0 % without such corrections for proton beams. Lastly, reference film measurements resulted in a 2.8 % standard deviation among radiochromic films within the same and across differing film sheets of the lot, as well as across different energy settings. Nonetheless, these variances were deemed negligible due to normalization applied to the line dose profiles. Conclusions: The calibration of the new film scanner was successful - with the knowledge of the row scans a correction matrix can be calculated for future scanning processes. Both the results of the proton beam experiment as well as the photon beam experiment clearly show the influence of the metal screw. For the polystyrene measurements, both the RayStation and SciMoCa plans show similarly good results. When the metal screw is present, MC simulations come much closer to the actual dose profile than the RayStation TPS. Therefore, MC can be a very useful tool in dose estimation.