Development of imaging-based response predictors for personalized radiotherapy in head and neck cancer

Tumor response to chemoradiotherapy is heterogeneous in patients with head and neck cancer. At the same time, head and neck radiotherapy can lead to significant toxicity in treated patients. Personalization of treatment could improve response to treatment while minimizing side effects. The largest bottleneck to employ personalization approaches are the lack of methods for response and toxicity prediction. In this thesis we therefore provide improved approaches for response prediction. In part one, we present improved MRI techniques to measure response before and early during treatment. In part two we present dose response models for osteoradionecrosis of the mandible incorporating key spatial information and the equivalent uniform dose as a generalizable dose variable across different fractionation schemes. The presented MRI techniques and dose response models can now be validated in larger groups of patients, after which they could contribute to personalized treatment planning and decision making processes.