Indication Specific Multi-Criteria Optimization in Radiotherapy Treatment Planning using Knowledge from previously delivered Treatment Plans

A multi-criteria optimization (MCO) approach for volumetric-modulated-arc (VMAT) radiotherapy treatment plan optimization with a low number of composite optimization objectives is presented. The solution space for a number of cranial and spinal patient geometries is explored and an approximation method of the solution space using trilinear dose interpolation is investigated. A machine learning (ML) driven approach for prediction of the characteristics of the solution space with respect to the patient geometry is developed and evaluated, followed by an outlook on how to combine the approximation method and the ML approach.