Personalized treatment is one of the major challenges in modern medicine. To enable individually tailored treatments, medical imaging has become part of the standard diagnostic work-up, allowing a non-invasive anatomical and functional representation of organs. In the last decades, anatomy-based computed tomography (CT) and functional-based 18F-fluorodeoxyglucose positron emission tomography (18F-FDG PET) have been two corner stones in medical imaging. However, clinical evaluation of imaging data remains subject to intra-observer and inter-observer variation. Moreover, most imaging data contain subtle information reflecting underlying pathophysiological properties, which cannot be detected visually. Quantifying medical imaging improves reliability, precision, and speed of the assessment and may contribute to overcome such limitations. The rapidly emerging field of radiomics generally quantifies large amounts of medical imaging data and applies a large number of quantitative image features to characterize these underlying pathophysiological properties. In this thesis, these quantitative radiomic features were validated and applied to objectively evaluate and adjust the clinical decision-making in the treatment of patients with esophageal cancer and vascular graft infections. This thesis resolved the initial lack of standardization for feature extraction and image processing. Moreover, we exposed the mechanisms of confounding factors which drive the reliability of 18F-FDG PET radiomic features. We constructed several prediction models based on 18F-FDG PET/CT radiomic features along with clinical markers and biological tumor markers to adjust the clinical decision-making in patients with esophageal cancer. Some of these prediction models were able to provide up to a 30% absolute improvement in discriminatory accuracy beyond the current standard prediction methods. Furthermore, 18F-FDG PET radiomic features were applied to improve the non-invasive diagnosis of aorta-iliac graft infections after vascular graft reconstructions. We concluded that quantitative PET may be a potential tool in improving PET interpretation of patients suspected of aorta-iliac graft infections.