ESTRO 38 Abstract book

S481 ESTRO 38

system for the CyberKnife robotic radiosurgery system is offering a 4D treatment planning and optimization feature to take into account tissue motion recorded in a 4D computed tomography (4DCT). The aim of the study is to experimentally validate the 4D Monte Carlo (MC) dose calculation before its use in the clinical routine of not visible lung lesions when tracking is not feasible. Material and Methods A thorax-moving phantom was used to simulate 3 motion sets amounting to 9 patterns. Two of them were defined with a lesion velocity of 1 mm/s and 2.5 mm/s respectively. The last set was defined to simulate 3 patients’ extracting surrogate motions from RPM files (Real-Time Position Management, Varian Medical Systems) while the superior-inferior lesion displacement was measured as the distance between the barycentre of the CTV in the full-inhale and full-exhale phases. The phantom was scanned for a static benchmark (SB) in absence of motion. An 8-respiratory phase binning was performed and images were imported in the Precision® Treatment Planning System (Accuray Inc.): 4D plans were optimized and calculated for all different motion patterns (MP Plans) with an 80% prescription to the PTV. All dose distributions were calculated with the Ray Tracing (RT) algorithm and then recalculated by a MC algorithm with a 1% uncertainty. Plans were delivered to the phantom with axial and sagittal films positioned in the lung insert. A local γ-analysis was performed with the FilmQA Pro software (Ashland) to quantitatively evaluate the agreement between the calculated and measured doses. Treatment plans were considered acceptable if the passing rate (P γ ) was greater than or equal to 90%. Results for the 3 set of motions have been compared to evaluate the 4D module performances. Results γ-analysis results for the SB plans are reported in Fig. 1. An acceptable P γ is reached in axial and sagittal planes for the RT plan when the γ-criteria are 7%/3 mm and 8.5%/3 mm, respectively. The MC calculation showed an acceptable P γ for 3%/3 mm γ-criteria in both directions. On the other hand, MP γ-analysis results are reported in Table 1 in terms of axial, sagittal and overall mean P γ ( 3%/3 mm). The overall mean passing rate for the 3 motion sets are 85.3%, 59.9% and 66.2% for RT Plans; 92.7%, 95.9% and 84.1% for MC Plans.

Conclusion The static benchmark has allowed a definition of minimum 3%/3 mm γ-criteria for following measurements because of the used experimental setup. The 4D feature allows obtaining an acceptable agreement between measured and calculated dose distributions in lung treatments when a MC calculation is performed. Future studies will follow to define a breathing-irregularity cut-off guaranteeing an acceptable passing rate for the third set. The RT calculation resulted in not acceptable accordance with measured dose distributions. PO-0906 Perturbation techniques for optimizing IAEA phase spaces for different medical linacs J.C. Martins 1 , R. Saxena 1 , S. Neppl 2 , A. Alhazmi 1 , M. Reiner 2 , C. Belka 2 , K. Parodi 1 1 Ludwig-Maximilians-Universität München, Chair of Experimental Physics - Medical Physics, Garching b. München, Germany ; 2 LMU Munich, Department of Radiation Oncology - University Hospital, München, Germany Purpose or Objective Monte Carlo techniques are very accurate for simulating dose distributions in radiotherapy, provided the model is based on detailed geometric information of the linac. Alternatively, validated phase space (PhSp) files (e.g. provided by IAEA) can be used, however these are available for a limited number of linac models. This work proposes a methodology to optimize an existing IAEA PhSp to produce fine-tuned machine-specific PhSp for an Elekta Synergy linac, by tuning the energy and momentum of the particles inside the PhSp. Material and Methods An Elekta Synergy linac coupled with an Agility collimator was modeled using Geant4. Due to unavailability of geometric information of the linac’s head, the IAEA PhSp for Elekta Precise was used as surrogate. Percentage Depth Dose (PDD) and lateral profiles, for different field sizes and 6 MeV photon beams, were measured and simulated under the same conditions. The discrepancies between measurements and simulations were quantified in term of cost values. Initially, the information inside the PhSp was perturbed randomly, aiming at finding correlations with the simulated dose profiles. Subsequent perturbations were performed following identified correlations, in the direction of decreasing cost values. To