3rd ESTRO Forum 2015 S755 deliver the higher surface and superficial doses than treatment techniques based linear accelerator. It showed adequate dose(over 75% of prescribed dose) at 1mm depth in skin region. Purpose/Objective: The aim of this study was two folds: (i) to implement a software that automatically assess multileaf collimator (MLC) performance in dynamic mode (ii) to analyse MLC performance over a long period of continuous operation. Materials and Methods: This study was carried out

more »

... with a Trilogy linac performing Rapidarc® treatments equipped with a Millenium 120 MLC (Varian) . Qualimagiq software solution was used (v5.4.1, Qualiformed, La Roche-Sur-Yon, France). Tests were as follows : (i) four sliding window at gantry 0°, 90°, 270° and 180° (CQ1); (ii) four picket fence at gantry 0°, 90°, 270° and 180° (CQ2) ; (iii) two picket fence with gantry rotation with and without intentional errors (CQ3 and CQ4) ; (iv) dose rate versus gantry speed (CQ5) ; (v) dose rate versus leafs velocities (CQ6). Images were acquired at 6MV with an AS1000 portal imager (EPID) and transferred to the software for data processing and analysis. All these tests were performed once a week for 9 months. Images were corrected from EPID mechanical slack. Acceptable tolerance levels and their uncertainties were taken from publications, manufacturer recommendations, repeatability and short-term reproducibility study. Results: Weekly images acquisition and associated analysis take about 25 min and 5 min respectively. CQ1, CQ2, CQ4, CQ5 and CQ6 agreed with recommendations. For CQ3, leafs positions were over 0.2mm limit. Further investigations showed that both leaf bank shifted during gantry rotation (0,74mm±0,04mm, clockwise and counter clockwise). Conclusions: A simple tool to understand and to master for assessing periodically MLC performance in dynamic mode was implemented. Apart from the CQ3 test, they were all within tolerances. Thanks to this tool, we have understood why CQ3 was out of tolerance. A sag in MLC carriage due to the gravity effect happens during gantry rotation. This phenomenon would not have been detected with log files analysis and will be further investigated on other accelerators (2100C/D and True Beam, Varian).