Geometric variability of organs at risk in head and neck radiotherapy

Many head and neck cancer patients live with side effects due to the radiation therapy. Xerostomia (having a dry mouth) is a serious and common side effect, which can persist long after the treatment. Predictive models can be used to calculate the probability of the occurrence of a certain complication, the ' normal tissue complication probability (NTCP) '. NTCP models for xerostomia often use the mean dose to the major salivary gland (parotid gland). The NTCP can be used to select patients for an advanced treatment. However, there is variation in the delineation of organs at risk (such as the parotid gland). Also, the shape and size of the parotid gland change during the course of radiotherapy. These variations in geometry lead to variations in the mean dose, and therefore to inconsistent values of NTCP.
In this dissertation, the size and the effect of variations in delineation and changing anatomy are studied, and solutions to reduce these variations or limit its consequences are proposed. The most important results are internationally approved delineation guidelines, which have been implemented world-wide. This will reduce interobserver variability and allow the generic application of NTCP models.
In addition, a method is proposed to select patients with a large dose variation in the parotid gland for adaptive radiotherapy. In this way, the risk of xerostomia will be reduced and quality of life improved.