ESTRO 36 Abstract Book

S418 ESTRO 36 _______________________________________________________________________________________________

on lung-like phantoms with clinical proton and carbon beams at the Heidelberg ion-therapy center (HIT). We adopted the benchmarked model to provide a parametrization of the Bragg peak degradation on the beam and on the previously mentioned lung parameters. Throughout this work, we tested and used a Gaussian convolution of the undegraded Bragg peak (U. Titt et al, 2015) to parametrize the degradation. Furthermore, the model was used to investigate the effects on clinical spread out Bragg peak (SOBP) and on the relative biological effectiveness (RBE). Results Fluctuations in the WET were found the major degradation factor, contributing more than 75% (40%) to the cumulative distal falloff widening for a carbon (proton) Bragg peak. The simulated lung parenchyma model (Figure 1) was capable to reproduce the experimental data with a slight underestimation of the degradation parameters, yet guaranteeing the correct reproduction of all the relevant characteristics in the degraded dose distribution. The Gaussian filtration unified the description for different beam particles and provided a compact and complete characterization with specific dependencies with respect to each lung parameter. Moreover, the description was found independent from the initial beam energy resulting in deviations mainly about the SOBP distal falloff while the plateau remains unaffected. Finally, the impact on the biological dose was mainly driven by changes to the physical dose due to the limited deviations in the RBE. Conclusion We provide a comprehensive characterization of Bragg peak degradation that can readily be implemented in a TPS. Such implementation is crucial for a more complete description of lung treatments, adding to the effect of macroscopic structures (e.g. bronchi, CT resolvable) the contribution of microscopic lung parenchyma (below CT resolution). PO-0788 First assessment of Delivery Analysis tool for pre-treatment verification on the new Radixact system A. Girardi 1 , T. Gevaert 1 , C. Jaudet 1 , M. Boussaer 1 , M. Burghelea 2 , J. Dhont 1 , T. Reynders 1 , K. Tournel 1 , M. De Ridder 1 1 Universitair Ziekenhuis Brussel, Department of Radiotherapy- Universitair Ziekenhuis Brussel- Vrije Universiteit Brussel- Brussels- Belgium, Brussels, Belgium 2 Brainlab AG, BRAINLAB AG Feldkirchen Germany, Brussels, Belgium Purpose or Objective To evaluate the accuracy of the Delivery Analysis (DA) tool for patient-specific pre-treatment verification and the sensitivity to detect discrepancies in dose delivery in comparison with widespread detectors. Material and Methods The Radixact machine is equipped with the DA device for pre-treatment Quality Assurance (QA) and interfraction verification. This tool is designed to assess the consistency of the delivered treatment through the detector data and to show anatomical changes of the patient. The latter representing a powerful tool to be coupled with Adaptive Radiotherapy. The idea is to use the detector: a) to

measure the Multileaf Collimator (MLC) leaf open time, b) to compare the planned sinogram to the delivered one and c) for dose reconstruction purposes. In this study, we performed pre-treatment verification on the very first twenty heterogeneous patients treated worldwide (target volumes ranging between 98 to 4179 cc) using the DA, the Sun Nuclear MapCheck2 (MC2) and the ScandiDos Delta4 (D4).The Gamma Index was used to show the agreement between dose planning calculations and measurements. To compare the three methods, criteria were set to 2% and 3% in local dose and to 2mm and 3mm in distance, respectively, excluding doses lower than 20% of the maximum doses. The performances of the systems were analysed with a single factor ANOVA test, with a significance level of α=0.05. A possible dependence of the results from the target volume was furthermore explored with a simple linear regression analysis. Results The ANOVA test showed no statistically significance differences between the performances of the three systems, both for the 2%/2 mm and the 3%/3mm criteria (p-values equal to 0.351 and 0.660 respectively). The linear regression indicated a variation of performance as a function of target volume for the MC2 (R 2 2%/2mm =0.819 and R 2 3%/3mm =0.979) and the D4 detectors (R 2 2%/2mm =0.991 and R 2 3%/3mm =0.990), which is not highlighted for the DA system (R 2 2%/2mm =0.283 and R 2 3%/3mm =0.290). This difference could be related to the missing data due to the larger dimension of the dose map with respect to the detection area of the MC2 and D4 systems.

Conclusion This study showed that the performances of the Delivery Analysis tool for the new Radixact machine is not different from those of two other widespread detectors for pre- treatment verification. Moreover, the linear regression test showed that the performances of the system are not correlated with the target volume, as is the case for two other detectors used in the study, proving its sensitivity as a patient specific QA tool.