Automation and individualization of radiotherapy treatment planning in head and neck cancer patients

Each patient treated with photon or proton radiotherapy requires a personalized treatment plan. The creation of a treatment plan is however an iterative and time consuming process. On the other hand, the daily geometrical patient variation and weight changes during the fractionated treatment course also requires plan adaptations. The limited availability of proton therapy requires an adequate selection procedure, such that only patients with a potential benefit are referred. The referral of so-called ‘model-based’ indications (such as head and neck cancer patients) relies on normal tissue complication probabilities derived from the photon and proton dose distributions. The increasing demand for treatment plans and the discussion of healthcare savings requires a next step in the direction of fully automated treatment planning. In this dissertation we introduce and evaluate new algorithms that contribute to treatment planning automation for photon and proton radiotherapy as well as individualization of the treatment plan. In the latter, normal tissue complication probability models have been implemented directly in the treatment plan optimization process, leading to further reduction of the estimated complication rate. In a next step, the work as presented in this dissertation will be implemented in the clinic. Direct benefit for patients is expected with decreased treatment preparation times and more efficient referral for proton therapy, whereas high treatment quality is guaranteed