ESTRO 37 Abstract book

S999

ESTRO 37

2 German Cancer Research Center DKFZ, Clinical Cooperation Unit Radiation Oncology, Heidelberg, Germany 3 German Cancer Research Center DKFZ, Translational Radiation Oncology, Heidelberg, Germany 4 University of Heidelberg, Department of Physics, Heidelberg, Germany 5 University of Pavia, Department of Physics, Pavia, Italy 6 Centro Nazionale di Adroterapia Oncologica, Pavia, Italy 7 Heidelberg Ion Therapy Center HIT, Heidelberg, Germany Purpose or Objective Accuracy in dose delivery is of high concern in particle therapy. To mitigate the uncertainties in treatment delivery, positioning errors, differences in beam shape, as well as uncertainties in biological modeling should be considered in routine treatment planning. In order to verify the robustness of treatment plans within a clinically viable time-frame, a fast dose calculation platform which can consider these uncertainties is needed. Material and Methods An analytical dose calculation program, FROG (Fast Recalculation and Optimization on GPU), has been developed in-house affording re-calculation of treatment plans within few minutes. FROG uses general purpose computation on GPU and facility specific databases to precisely calculate the dose and relative biological effectiveness (RBE)-weighted dose distribution of already delivered and planned treatments with proton, 4 He, 12 C and 16 O ion beams. Thus, several robustness scenarios for complex clinical cases can be taken into account and analyzed within a clinically relevant time frame. Results As a proof of concept, a robustness analysis of physical and biological uncertainties for select patient cases has been performed. For example, two head cases that have been irradiated with protons at the Heidelberg Ion Therapy Center which have organs at risk (OAR) in close proximity to the clinical target volume (CTV) have been investigated. A ±2mm positional shift in three orthogonal directions (and combinations of those shifts) of the patients was simulated and the differences in dose of the 27 cases has been analyzed by means of dose volume histogram and corresponding dose volume metrics D 98 , D 50 and D 02 . For the two patients, CTV and a selected OAR close to the CTV have been analyzed. Analysis (mean with one standard deviation as an error) showed that the chiasm in patient 1 has a D 98 of 3.70±1.25 Gy (RBE), a D 50 of 32.51±4.48 Gy (RBE) and a D 02 of 49.12±1.28 Gy (RBE), the CTV has a D 98 of 50.55±0.67 Gy (RBE), a D 50 of 53.49±0.27 Gy (RBE) and a D 02 of 58.88±1.86 Gy (RBE). Analysis for patient 2 showed that the brainstem has a D 98 of 3.07±0.83 Gy (RBE), a D 50 of 28.62±2.60 Gy (RBE) and D 02 of 51.90±0.99 Gy (RBE), the CTV has a D 98 of 47.67±0.84 Gy (RBE), a D 50 of 52.04±0.30 Gy (RBE) and a D 02 of 55.07±0.73 Gy (RBE). Conclusion For treatment plans with OARs close to the CTV and where differences in beam delivery or biological modeling could change the outcome of the treatment, robustness analysis is a vital tool to predict the possible outcomes of the treatment and possible unwanted dose to OARs. By using GPUs for dose calculation and Monte Carlo generated databases, FROG allows for fast and accurate robustness analysis. EP-1851 Evaluation of lateral density heterogeneity handling in a novel GPU-based pencil beam algorithm T. Tessonnier 1 , S. Mein 2,3 , B. Kopp 2,4 , K. Choi 5 , T. Haberer 6 , J. Debus 2,4,6 , A. Abdollahi 2,4,6 , A. Mairani 5,6 1 Centre François Baclesse - Caen, Radiation Oncology, Caen, France 2 German Cancer Research Center DKFZ, Imaging and

Material and Methods MC simulations of a 6 MV photon beams were performed by using EGSnrc codes. BEAMnrc code was used to model an Elekta Synergy linear accelerator: the input file consisted of 8 component modules (CM), from the source to the lower diaphragms. The DOSXYZnrc code was used to simulate percentage depth doses (PDD) and profiles in a cubic water phantom. In particular, in the first stage, PDD and profiles of symmetric fields were simulated for several field sizes: 5cm x 5cm, 10cm x 10cm, 20cm x 20cm and 40cm x 40cm. In the second stage, asymmetric fields were also simulated: hemifields from 0 to 2.5cm in X direction, for the 5cm x 5cm, and from 0 to 5cm for the 10cm x 10cm field etc. The largest fields were considered of interest because they are used clinically in the case of 3DCRT planning for breast with supraclavicular involvement. Results Profiles of symmetric fields were analysed and compared with the ones obtained from Pinnacle 3 TPS. The uncertainties of MC simulations were less than 1% in the central field area and less than 2% in the penumbra region, the profiles of the two sets coincide within that uncertainty (graph 1(MC- markers, TPS- line)). In the second stage of this work, asymmetric profiles for the same set of field sizes were simulated. As shown in graph 2 , profiles of simulated hemifields are in agreement with ones obtained from TPS within the mentioned range of uncertainty for MC simulation.

Conclusion When new planning technique is being used it is of importance to physicist to analyse how that technique is appropriate to the physics model provided by the TPS, since the model is a compromise between measured data for different beam setups. Beam data for the simulated linac, were compared to the ones from the TPS. The shape of the profile of hemifields reflects the observed worsening at the border of the field in the longitudinal direction (Y), with a loss of dose coverage in treatment planning when a one isocenter technique is used compared to two isocenters technique. EP-1850 Fast robustness analysis in particle therapy with FROG B. Kopp 1,2 , S. Mein 1,3,4 , K. Choi 5,6 , T. Haberer 7 , J. Debus 1,2,7 , M. Alber 1,7 , A. Mairani 6,7 1 Heidelberg University Hospital, Department of Radiation Oncology, Heidelberg, Germany