ESTRO 38 Abstract book

S931 ESTRO 38

Results Dose calculation and dose measurements remain within tolerances of NCS15 confirming the accuracy of the Monte Carlo Algorithm for different SSD's and all investigated inserts. Large fields remain in output very close to 1/(SSD)² and steeper measured dose fall off for small fields is predicted within tolerance with the Raystation Monte Carlo algorithm. The Distance To Agreement for the R50 values with increasing incidence angle remains 2 mm or smaller, again within NCS15 tolerances. The best fit for the line profiles behind the air, lung and bone cavities was obtained within 3 mm for the δ2 region (high dose gradient) and within 4% in dose for the δ4 region (photon tail), both within the tolerance of NCS15. Conclusion According to NCS15 for non-standard treatment setups a dose agreement of 3 % in the d1 region of 4 % in de d4 region and for the d2 region a distance to agreement of 3 mm has to be obtained between measurements and calculations. This is investigated using different air filled and liquid field ionization chambers and diodes in several setups. This confirms the accuracy of the Monte Carlo code of Raystation for prediction of dose distribution within acceptable timeframes (<1000000 histories) for standard clinical treatment planning. [1] Quality Assurance of 3-D treatment planning systems for external photon and electrons beams, NCS15, Bruinvis I.A.D. et all 2005 EP-1728 1-year experience with automated transit in vivo dosimetry in a busy multicenter department E. Bossuyt 1 , R. Weytjens 1 , S. De Vos 1 , R. Gysemans 1 , D. Verellen 1 1 Iridium Kankernetwerk Antwerpen, Physics department radiotherapy, Wilrijk, Belgium Purpose or Objective Being a busy department with 5600 new patients/year and several satellite centers, efficiency, standardization and automation are key for a QC program. A web-based system was installed in our center early 2017, for pre-treatment and in-vivo QA based on phantomless EPID and/or log files (Sun Nuclear Corporation). DICOM data is pushed to the server. Images and log files are actively retrieved. Calculation and analysis occur automatically in the background. A clinical validation of the system’s performance on detection of errors and reducing workload will be reported. Special attention was given to identify and mitigate false positive (FP) results. Material and Methods In this study results are reported from an analysis for all patients treated between October 2017 and August 2018. During treatment every field or arc was controlled by means of logfiles for all patients receiving photon treatment. In addition, the first 3 days and then weekly, transit EPID images were generated. These were compared using relative 2D analysis. Recently absolute verification was introduced comparing the images to calculated data, further enhancing detectable errors. Appropriate actions were undertaken based on a decision tree derived from an initiating training period. Results No relevant patient errors were detected with analysis of the logfiles, 15% of integrated images failed. One third were FP, due to incomplete dose accumulation (beam stop, 6%) or dispositioning of the imager (27%). Creating a report in Aria helps detecting these FP’s. The remainder concerned patient related issues. Most common causes were patient positioning and anatomy changes such as weight loss (41%) and deviations in bladder or rectal filling (6%). Rare observations were shrinkage of tumor, hematoma, pneumonia, air cavity in the tumor and accidental translation of the table between fields. Actions to mitigate the most dominant causes of errors were

Conclusion Bolus considerably improves the dose distribution in thin chest wall target. It is advisable also for other chest wall thicknesses to achieve better target coverage and decrease the overdose volume. Dose estimation in presence of bolus is more robust and less sensitive to the calculation algorithm. If the improved dose distribution with bolus translates into fewer chest wall recurrences needs to be verified in clinical trials. EP-1727 Validation of the Raystation Monte Carlo Code using dedicated ionization chambers G. Pittomvils 1 , E. Bogaert 1 , P. Thysebaert 2 , C. De Wagter 1 , Y. Lievens 1 1 Ghent University Hospital, Dienst Radiotherapie, Gent, Belgium ; 2 Odisee Hogeschool-Universiteit, Campus Terranova, Brussels, Belgium Purpose or Objective Electron Monte Carlo dose calculation algorithms are capable of predicting dose distributions within tolerances defined by international and national organizations for standard and non-standard setups, such as extended SSD treatments, oblique incidences and heterogeneous tissues. At Ghent University Hospital the Raystation planning system was introduced in 2016 and the electron Monte Carlo was also commissioned, allowing the dose prediction of electron dose deliveries for all treatment modalities. Material and Methods Raystation Monte Carlo Code is verified using plane parallel chambers; Roos Chamber and a linear array using liquid filled plane parallel chambers (PTW, Freiburg, Germany) and an PTW Electron Diode type 6012. The clinical calculation grid size of 2mm³ was selected using 250000 histories resulting in a statistical dose accuracy of 1.5%. The NCS15 protocol (Nederlandse Commissie Stralingsdosimetrie)[1] is used to validate the results. The dose output versus SSD (95.5-115cm) is evaluated for all applicators at Zref using the Roos chamber. The output of 19 different inserts is verified using de Electron Diode at SSD=100 cm. The dose reduction at the standard R50 value for a 14x14cm² applicator was verified for beam incidence angles ranging from 0° to 30°. The calculated dose is verified with a linear array for lung and air inserts of different dimensions with and without 1 cm build up material behind the 3D plus shaped cavity. The same linear array is used to verify the dose behind a cylindrical insert with three different bone density inserts.