ADAPTIVE RADIOTHERAPY BY GENERALIZED DEEP LEARNING BASED ON PHYSICAL PRINCIPLES

Monitoring the effects of photon and proton radiotherapy by sophisticated MRI techniques and predicting patient-specific outcomes via deep learning methods will cause a breakthrough in treatment selection (proton versus photon treatment) and prediction-based threatment adaptation. Generalized machine learning methods will be developed that can cope with highly diverse (functional) imaging data. This will be achieved by exploiting databases from collaborating partners (Holland PTC, AMC, Erasmus MC) and introduction of governing physical principles, in particular MRI fingerprinting techniques for normalization of the data.