ESTRO 38 Abstract book

S967 ESTRO 38

disadvantage, but can be addressed by leveraging the capabilities of CPUs. The advantages of dynamic VR tuning would be offset by thread divergence issues on GPU. The essential metric of code performance is its scaling capability with the number of CPU cores to be ready for the next level of parallelization in hardware. Real-time MC is rapidly becoming reality both with muscle and brain. EP-1787 Commissioning of the RayStation treatment planning system in a multi-vendor context A. Savini 1 , F. Rosica 2 , V. D'Errico 1 , T. Licciardello 1 , E. Menghi 1 , F. Bartolucci 2 , F. Christian 2 , G. Orlandi 2 , A. Sarnelli 1 1 Istituto Scientifico Romagnolo per lo Studio e la Cura dei Tumori IRST IRCCS, Medical Physics Unit, Meldola, Italy ; 2 Ausl 4, Medical Physics Unit, Teramo, Italy Purpose or Objective To present our procedure and results regarding the commissioning of the RayStation treatment planning system (TPS) in a multi-vendor context with several LINACs, beam qualities and multi-leaf collimators (MLCs). Material and Methods RayStation clinical version 7.0 was considered in this work. The available dose calculation algorithm was the collapsed cone convolution (version 3.3). LINACs, beam qualities, MLCs and radiotherapy techniques considered for implementation are reported in Tab.1. The clinical commissioning included several steps such as the acquisition of basic dosimetric measurements (i.e. percentage depth doses, profiles and output factors), the dosimetry of ad-hoc abutted fields for modeling the MLC, the beam modeling for reproducing experimental measurements and, finally, the validation of the model for each radiotherapy technique. Validation for basic dosimetry was conducted using the γ-confidence limit (CL) between measured and calculated dosimetric curves as addressed in the ESTRO booklet n.7. Validation for intensity modulated techniques was conducted following the AAPM-TG119 protocol. Finally, for each beam quality considered for implementation of intensity-modulated techniques, pre-treatment verifications of ten clinical cases were performed as end-to-end test.

Conclusion The RayStation TPS was modeled for different LINACs, beam qualities, MLCs and irradiation techniques. Following the presented procedure for modeling, we found a final accuracy that was comparable across the considered devices and clinically acceptable for each combination. EP-1788 Dose distribution for electron beam using Monte Carlo simulation with GATE J. Leste 1 , M. Chauvin 1 , T. Younes 1 , L. Vieillevigne 1 , M. Bardies 1 , X. Franceries 1 , R. Ferrand 1 , N. Pierrat 2 , L. Bartolucci 2 , L. Simon 1 1 CRCT, UMR 1037 Inserm, Toulouse, France ; 2 Institut Curie, Dpt. de Radiothérapie, Paris, France Purpose or Objective Limits of Treatment Planning Systems (TPS) for dose calculation of photon beams have been widely studied. There are fewer studies using Monte Carlo (MC) for electron beams. Although the use of advanced variance reduction technique (Macro Monte Carlo, MMC) to meet time clinical requirements leads to limited results in complex cases (heterogeneity, irregular surfaces). This study aims to assess a commercial algorithm, Fast electron Monte Carlo (eMC, Varian, Palo Alto, CA), based on MMC, and an in-house GATE model of a radiotherapy linac (TrueBeam, Varian) for electrons beams in complex cases. Algorithms are also evaluated for clinical cases (head & neck, breast). Material and Methods Validation of GATE model is performed with homogenous water phantom. Percentage depth dose curves (PDD) and lateral dose profiles (LDP) are compared with experimental measurement (6, 9, 12 and 18 MeV). For complex cases including heterogeneities (lung, bone) and irregular surface (step), results of dose calculation for eMC and GATE are compared using Gafchromic EBT3 (Ashland ISP, Wayne, NJ) films. LDPs are acquired at two different depths and normalized in a homogenous water region. Global Gamma Index Pass Rate (GIPR) 1D is computed to compare the data sets. For clinical case an absolute dose calibration of our MC model is achieved. Ten clinical cases are tested for several energies and for different inserts

Results γ-CL evaluation is reported in Fig.1a. The RayStation beam modeling was able to satisfy the required criteria (γ-CL < 1) for 99.4% of the analyzed dosimetric curves. Out of tolerances results were found for large fields (i.e. 30x30 and 40x40 cm2) at large depth (i.e. 20 cm) with little clinical relevance. For TG119 verifications, γ -index pass- rates calculated with TG119 criteria (3%, 3mm, global normalization) are reported in Fig.1b (only VMAT results are shown). For all cases the pre-treatment verifications gave clinical acceptable results (i.e. pass-rate > 95%). For pre-treatment verifications of VMAT clinical cases, γ - index pass-rates with stricter criteria (3%, 3mm local) were all above 90%.