ESTRO 38 Abstract book

S575 ESTRO 38

The dose linearity & repeatability of the linac was confirmed using an ion chamber, with the same measurements used to assess the dosimetric characteristics of the EPID panel. This was necessary as the as the EPID will be used as a routine dosimeter for daily QA measurements. Film was used to determine the absolute calibration of the MLCs and jaws using an in-house built copper jig. Copper is used to reduce the effect of the magnetic field on the secondary electron trajectories, such that the deposited dose more closely matches the photon fluence. The same jig was used to locate the films with respect to the absolute beam coordinate system defined by the EPID panel via a copper cross (see figure 2). Copper was also used in an in-house ring phantom used to measure the radiation isocentre radius of the MRL using a star-shot.

Conclusion A novel concept of predictive daily machine QA system was proposed that applied the historical MPC test data to develop the ARIMA forecast model and SPC-based warning level for each parameter. The system could detect the forthcoming errors where predict MPC test exceeds the SPC-based warning level for each parameter. This represents the likelihood of machine failure in the near future and requires for further investigation and/or calibration. PO-1035 Linac commissioning methodology for the Elekta Unity MR Linac I. Hanson 1 , S. Nill 2 , J. Chick 1 , U. Oelfke 2 1 The Royal Marsden NHS Foundation Trust & The Institute of Cancer Research, Radiotherapy Physics, Sutton, United Kingdom ; 2 The Royal Marsden NHS Foundation Trust & The Institute of Cancer Research, Physics, Sutton, United Kingdom Purpose or Objective The Elekta Unity MR Linac (MRL) allows for daily adaptive radiotherapy based on high quality, real-time imaging. A commissioning program has been developed to ensure the safe and accurate operation of the MRL. Measurements performed on the MRL have to overcome several challenges. Amongst these are the presence of the magnetic field and its influence and limitations on the measuring equipment that can be used, the lack of lasers and light field, the limited bore size and the lack of access to the linac head from the treatment room. We have developed novel methods and tools to address these issues. Consideration of the interdependencies of the MRL parameters was give when designing and ordering tests in order to eliminate any self-referencing loops. Material and Methods An in-house phantom was developed to provide a set of foundational measurements from which further dependent tests could be built upon (see figure 1). This phantom ensures the correct orientation of the radiation beam along the vertical and horizontal axis. The same phantom is used to measure the absolute position of the radiation beam with respect to the fixed EPID panel, and also the rotation of the EPID panel. Once this rotation had been determined the EPID panel could then be used to confirm the rotation of the collimating system.

Finally, an end-to-end test was performed using the STEEV phantom. A reference plan was generated and adapted using a daily MRL. The difference between reference predicted dose and measured dose was recorded Results The radiation beam was found to be within 0.3° in the horizontal and vertical plane. The EPID panel rotation was 0.092°. The collimator rotation averaged over all cardinal gantry angles was -0.01°. The average MLC positioning error was (- 0.12±0.23)mm. The radiation isocentre radius was 0.32mm. Repeat 1MU exposures agreed dosimetrically to within 1.4%. Finally, end-to-end testing of the adaptive workflow using the STEEV showed an excellent agreement of 0.018 % Conclusion New methodologies and tools have been developed to commission the radiation delivery system of the Elekta Unity. Measurements showed that the MRL is operating within expectations. PO-1036 Deep Learning for automatic contouring of clinical target volumes in breast cancer patients. P. Poortmans 1 , A. Henry Alexandre 1 , P. Aljabar 2 , R. Baggs 2 , M. Gooding 2 , M. Leclerc-Chalvet 2 , Y. Kirova 1 1 Institut Curie Ensemble Hospitalier, Department of Radiation Oncology, Paris cedex 05, France ; 2 Mirada Medical Ltd., Research & Development, Oxford, United Kingdom Purpose or Objective To improve consistency and decrease workload, we introduced and evaluated a system for automated contouring of clinical target volumes (CTV) in breast cancer. Material and Methods For 146 patients undergoing radiation therapy (RT) for breast cancer, CT-based images were acquired for treatment planning purposes. CTVs, including the breast, and the following lymph node regions: internal mammary chain, interpectoral and axilla levels one to four, were manually delineated on all slices following the ESTRO