ESTRO 2020 Abstract book

S78 ESTRO 2020

coverage (Bartlett et al. 2013, Kügele et al. 2017). Using SGRT, the intrafraction motion of the isocenter position in between breath hold can be assessed. The use of SGRT can mitigate intrafraction motion, and hence, reduce the large dosimetric effects due to undesirable intrafraction motion (Kügele et al. 2017). SGRT have also been shown to be an efficient tool for patient positioning and motion monitoring during frameless stereotactic radiosurgery (SRS) of the brain (Cervino et al. 2010, Peng et al. 2010, Wiersma 2013, Li et al. 2015, Mancosu et al. 2016, Wen et al. 2016, Konradsson 2018). The patient’s surface serves as a surrogate for the target, including the forehead, nose, eyes, and part of the temporal bones without the need for any additional markers. Intrafraction motion tolerances in the SGRT system can be applied to ensure the accuracy of the treatment, also when couch rotations are used. Recently, SGRT has been used on deep-gantry linacs for patients receiving total-marrow irradiations (TMI). For TomoTherapy (Accuray), the TMI have shown significant dose reductions compared to conventional total body irradiations (TBI) (Haraldsson et al. 2019). The use of SGRT patient positioning and motion monitoring could further reduce the PTV margins, especially for the extremities. In conclusion, SGRT has not only increased the setup accuracy and decreased the time for patient set up, but also enabled decreased absorbed dose to OAR due to motion monitoring. This presentation will highlight the intra-fractional motion monitoring of the isocenter position during DIBH for breast cancer treatment, including the estimated dosimetric effects. SP-0149 Optical imaging for gating and tracking G. Fattori 1 1 PSI, Center for Proton Therapy, Villigen, Switzerland Abstract text Accurate modelling and detection of breathing induced organ motion is essential when treating non-static tumours with high precision radiotherapy techniques. Its indirect assessment by tracking the body surface displacement as a surrogate of the patient's anatomical changes is widespread in 4D medical imaging and similarly applied by most motion mitigation techniques. The benefits and pitfalls of using optical tracking are discussed against the alternative electromagnetic technologies, in the specific application of image-guided, gantry-based proton therapy. Despite the wide range of commercial products, it is often difficult to identify an off-the-shelf system that is flexible enough to support future trends of research and allowing for swift retrofit into existing treatment units, each with specific layout and technologies. We take the opportunity to present the solution developed at PSI that recently got into clinical operation as a couch mounted gating and motion monitoring system for 4D imaging and treatments. The technology transfer process is presented focusing on the requirements specification, technical safety measurements and quality assurance program designed to meet the clinical standards. Tracking the tumour position in real time with the treatment beam is possibly the most efficient way to treat a tumour that moves as a result of patient breathing. However, for the ultimate in refinement and precision, tumour tracking requires more than high technical specifications and low-latency performance of the monitoring device. A sophistication of the clinical workflow with tighter integration between surface tracking and 4D MR imaging together with visual feedback to the patient, are put forward as key elements for its adoption in proton therapy.

Symposium: "Big data" approaches to quality improvement

SP-0150 Integration of AI and Machine Learning in Radiotherapy QA M. Chan 1 1 memorial Sloan Kettering Cancer Center, Medical Physics, New York, Usa Abstract text The use of machine learning and other sophisticated models to aid in prediction and decision making has become widely popular across a breadth of disciplines. Within the greater diagnostic radiology, radiation oncology, and medical physics communities, promising work is being performed in tissue classification and cancer staging, decision support and outcome prediction, automated segmentation, treatment planning and quality assurance, and numerous other areas. In this presentation, machine learning approaches are explored, highlighting specific applications in machine and patient-specific quality assurance (QA). Virtual Intensity-Modulated Radiation Therapy (IMRT) QA can predict passing rates using different measurement techniques, different treatment planning systems, and different treatment delivery machines across multiple institutions. Prediction of QA passing rates can have profound implications on the current IMRT process. The presentation will cover the general concept about machine learning in dosimetry, various methods used in virtual IMRT QA, as well as their clinical applications.