Key words: rectal cancer, image analysis, treatment outcome, prediction models

"Multiparametric image modelling predicting treatment response in rectal cancer"

Patients with advanced rectal cancer are typically treated with chemo- and radiation therapy (chemoradiation) followed by a major surgery where the entire rectum is removed. Some patients respond so well to chemoradiation that here is hardly any tumor present after treatment. This is why major surgery is increasingly omitted in patients that respond well to chemoradiation and instead these patients are treated “organ-saving” – i.e. with only a minor intervention or even without surgery.

In this thesis, the possibility to predict treatment outcome to chemoradiation beforehand based on the available clinical data and imaging data has been studied using prediction models. If it is known upfront whether a patient will respond well to chemoradiation, this may offer possibilities to further personalize treatment to a specific patient. The predictive values of various imaging techniques were compared. In addition, the influence of data variations on reproducibility was investigated. The results of these studies offer important insights that are valuable for the development of future prediction models.

Click here for the full dissertation.