ESTRO 38 Abstract book

S482 ESTRO 38

based on validated IAEA PhSp. It offers an alternative when neither geometric details nor validated PhSp are available for a specific machine. PO-0907 Gafchromic EBT3 film for absolute dosimetry in proton therapy based on averaging of beam quality A. Resch 1 , P. Heyes 1 , H. Fuchs 1 , D. Georg 1 , P. Hugo 2 1 Medizinische Universität Wien, Department of Radiotherapy and Christian Doppler Laboratory for Medical Radiation Research for Radiation Oncology, Vienna, Austria ; 2 MedAustron Ion Therapy Centre, Medical Physics, Wiener Neustadt, Austria Purpose or Objective Gafchromic EBT3 films are a relevant dosimeter for proton therapy due to their high spatial resolution and near water equivalence. Currently their energy dependent relative effectiveness (RE) limits their usage for absolute dosimetry in particle therapy to the beam quality (e.g. the average linear energy transfer (LET)) at calibration. Previously, beam quality correction factors have been proposed based on a fluence averaging over the mixed charged particle field. In this study we investigate the RE for varying energy distributions to provide experimental evidence for a suitable averaging method. Material and Methods The RE as a function of depth was experimentally characterized for two beam types, mono-energetic (initial beam energy) and spread-out Bragg peaks (SOBPs). The initial beam energies (E 0 ) of three single-energy fields were 62, 148 and 252 MeV. Three constant-dose SOBPs (E 0 ranging from 62 to 68 MeV) were created with different energy distributions by superimposing a fraction of low LET protons (E 0 ≈ 252 MeV). Beam 1 (b1) consisted only of the SOBP. In beam 2 (b2) approximately 50% of the initial fluence and in beam 3 (b3) approximately 50% of the dose was contributed by the low LET beam. To ensure lateral charged particle equilibrium a field size of 7x7cm 2 was used. EBT3 films were calibrated in a low LET region, in the entrance plateau region of a single-energy 179.2 MeV proton beam, for several dose levels. Gate/Geant4 Monte Carlo (MC) simulations employing a validated beam model were used to calculate the absorbed dose to water as well as dose-( LET d ) and track- average LET (LET t ). Dose simulated with MC was normalized to measurements with a plane parallel ionization chamber in the SOBP. The RE of the EBT3 response was calculated as the ratio of the film dose relative to MC simulated dose. Results The RE was found to be below unity except for the 62 MeV beam at around LET d = 6 keV/μm (corresponding to points shortly before the Bragg peak). RE as a function of LET d of all beams was similar up to approximately 5 keV/μm. In Fig.1 the measured RE is shown as a function of LET d and LET t for the 3 SOBP beams. The RE of b1 to b3 was comparable but differed at high LET d . The same experimental RE (but with smaller LET binning) expressed as a function of LET t revealed the limitations of the concept of fluence averaging, as there was no one-to-one correlation of LET t and RE for beams b2 and b3. The LET t for b2 and b3 hardly increased over depth and hence could not reflect the decreasing RE over depth.

optimize the iterative process, the PhSp was cropped into smaller sections, depending on the field size and simulation configuration. The differences in the simulated results using cropped PhSp, in comparison to the corresponding results using the full PhSp, are under 3% (Figure 1), which assures no relevant information was disregarded. Small energy perturbations were applied to cropped PhSps and PDD were simulated. This process was repeated until a minimum cost value was reached and an optimal energy value was found. Next, the momenta of the particles in all three directions were perturbed, following identified constraints and statistical correlations between particle’s momentum and lateral profiles. Two distinct optimal perturbation factors were found for particles’ momentum in crossline and inline directions.

Results For validation, the optimal perturbation factors were finally applied to the entire PhSp for different field sizes. For the considered linac, an energy perturbation of 0.32 MeV results in PDD cost values under 1.5% for all fields considered, and improves cost values for lateral profiles in the inline direction (Figure 2). Perturbation factors of 0.61% and 0.05%, when applied separately to the momenta of particles in the crossline and inline direction respectively, yield cost values below 3%. Several approaches to merge both factors and investigate the method with different beam energies are under evaluation.

Conclusion A methodology for optimization of an existing PhSp was developed, enabling generation of machine-specific PhSp