| Popis: |
Purpose: While a commonly-used dosimeter in radiotherapy, the use of radiochromic films in particle therapy is limited due to their strong dependence on beam quality. In the case of protons, previous research has reported an underresponse of EBT3 films of up to 20% around the Bragg-peak, which has largely been associated with the linear energy transfer (LET). The main focus of this work lies on the investigation of beam quality to quantify and correct for the under-response of EBT3 films. Since exact measurements are necessary, additionally film analysis is reviewed in terms of the analytical relation between optical density (OD) and dose, the time-dependence of calibrations, and different readout methods of the scanned film image, with the aims of mitigating systematic errors and reducing statistical uncertainties. Methods: Measurements of single energy beams of 62.4, 148.2 and 252.7MeV and spread-out bragg peaks (SOBPs) were performed at the treatment and research facility MedAustron (Wiener Neustadt, Austria) in a water phantom, using EBT3 radiochromic films and an ionization chamber (Roos chamber) as validation. These experiments were accompanied by Monte-Carlo simulations using GATE v8.0, in order to infer LET and spectral information. Corrections were derived by fitting the acquired relative effectiveness (RE) of the films to the dose averaged LET, and by optimizing a weighting function over the LET spectra at points of different RE. Results: Performing the film readout using the single red channel displayed a measurement uncertainty ranging from 13% to 2.6% for doses of 0.25 to 2.5Gy, compared to higher values using the dual or triple channel readout method. Between 0.5 and 2.5Gy all examined calibration functions exhibited residuals below 2%. Calibrations with a separation of approximately half a year showed deviations in the order of 10% around 0.5Gy between each other. EBT3 films displayed an under-response of up to 30% at the Bragg peak and the modulated peak of the SOBPs at the lowest beam energies used. A functional relationship between relative effectiveness and fluence averaged LET could not be confirmed, while it was possible to describe the quenching as a function of the dose averaged LET of protons, within the confidence interval. Applying a parametrized arcustangens function of dose averaged LET to the nominal dose, agreed within 6% with the smoothed uncorrected film dose. Optimization of weights over the LET spectrum at different RE yielded a nonlinear dependence of the RE over the LET spectrum. Conclusion: For the dose range examined, single channel read out of the red channel was found to yield the most accurate results. While differences in choice of calibration function were negligible, performing the film calibration within measurement session was shown to have a large effect on accuracy. The film quenching was found to be a nonlinear function of LET, and dose averaged LET of protons, while fluence averaging was found to be conceptually limited. An online function of dose averaged LET, a parametrized arcustangens, yielded the most accurate and robust correction of the film dose under-response. The spectral correction was more susceptible to measurement and simulation inaccuracies, compared to the averaged LET approaches. Due to the high measurement uncertainties associated with steep dose gradients, the presented corrections are only valid up to approximately R80 (depth at which the dose has decreased to 80% of the maximum) and the dose fall-off, for the single energy beams and SOBPs, respectively. For more exact corrections, film dosimetry needs to be improved in terms of accuracy and precision. |
