This thesis investigated broadly speaking the association between biology-based endpoints and artificial intelligence derived imaging biomarkers. The most important group of biomarkers (also called ‘radiomics signature’) in this thesis derived from CT and FDG-PET was able to accurately classify both lung and head and neck cancer patients as hypoxic (low tumour oxygen) or non-hypoxic in external datasets not seen before by the AI models. Other important findings in this thesis were that peritumoral regions (3 and 5mm around head and neck tumours) on CT did not have predictive value for overall survival, recurrence or metastasis. This thesis for instance also demonstrated that we can potentially enhance frozen section histology results by the addition of radiomics CT imaging biomarkers in solitary pulmonary nodules, having the potential to aid the surgeon in decision making about the most adequate type of surgery to remove the solitary nodule.
|Award date||11 Jan 2022|
|Place of Publication||Maastricht|
|Publication status||Published - 2022|
- tumour biology