PhD defence R. (Rens) van Haveren

Radiotherapy is often part of the treatment for cancer patients. For each patient, an individualised radiotherapy treatment plan has to be made that ensures to deliver a sufficiently high dose to the tumour while keeping the doses delivered to the surrounding healthy tissues as low as possible to minimise radiation-induced side-effects. Erasmus-iCycle automates the treatment plan generation process and produces a consistent high-quality and Pareto-optimal treatment plan for each patient. The default planning method in Erasmus-iCycle requires solving multiple consecutive optimisation problems to generate the plan. Also, the configuration of the parameters required is done manually which is both time-consuming and labour-intensive. 

The aim of this thesis was to improve Erasmus-iCycle in two ways: (1) to reduce the computation time for automated treatment plan generation without compromising in plan quality, and (2) to automate the configuration process in order to replace the traditional manual configuration process. A new method, the lexicographic reference point method (LRPM), was developed for automated treatment planning Instead of solving multiple optimisations, the LRPM only requires to solve a single optimisation problem to automatically generate a plan which considerably reduces the computation time. In addition, an automatic configuration method was developed for the LRPM based on a set of existing high-quality treatment plans. For both prostate, and head and neck cancer patients, it was found that the LRPM can automatically generate a high-quality and Pareto-optimal treatment plan for each patient, and that the configuration of the LRPM can be automated as well. The resulting LRPM plans had quality similar or better than those generated with the default planning method in Erasmus-iCycle.The LRPM has shown great potential to automatically generate high-quality and Pareto-optimal treatment plans. In addition, the automatic configuration process of the LRPM can replace time-consuming and labour-intensive manual tweaking of parameters.

Due to corona, the PhD defences do not take place publicly in the usual way in the Senate Hall or in the Professor Andries Querido Room. The candidates will defend their dissertation either in a small group or online.