The vacancy is part of the ART-PIVOT (Adaptive RadioTherapy – Personalized Intrafractional Volumetric Oncological Targeting) project, which is funded by NWO and co-funded by industrial partners through the open technology programme. The aim of ART-PIVOT to enable fast and continuous volumetric modulated arc therapy (VMAT) for MR-linacs, which is expected to dramatically increase the treatment delivery speed.

As PhD student you will first develop a set of high-performance VMAT therapy plan optimization algorithms, ensuring a low computational latency for plan generation as well as an effective deliverability on the treatment machine. In addition to the high numerical efficiency of the optimizer that you will develop, the algorithms should also leverage the hardware acceleration provided by modern GPU’s and FPGA’s, with support from our industrial partners Elekta and AimValley. You will then build an in-silico demonstrator, showcasing the capabilities of the implemented therapy planning framework for online/real-time adaptation and comparing the dosimetric performance with conventional multi-beam MR-linac deliveries.

Background

A new generation of radiotherapy treatment machines called MR-linac allows for the delivery of radiation cancer treatments with unprecedented accuracy and precision. This is mainly because such machines can deliver radiation while simultaneously providing MR images, facilitating the clear identification of the tumour and surrounding healthy tissue. In turn, this allows for precise therapy planning and adaptation, in accordance with the location of the target area at the time of treatment, potentially under the effect of anatomical and physiological motion. One of the drawbacks, however, is that the creation and delivery of a therapeutic plan on the MR-linac currently requires a considerably longer time compared to their counterpart on conventional CT-guided linacs. The reason behind this is the fact that the treatment is delivered as a sequence of radiation beams with extended pauses between each partial segment. In addition, plan generation itself implies solving a complex optimization problem to simultaneously maximize the radiation delivered to the target areas as well as healthy tissue sparing. The aim of the project is to develop a time-effective solution for therapy plan creation and delivery for MR-linacs.