Analyzing tumor lesions in PET/CT images using deep learning methods and physiological models

The detection of pathological lesions plays an important role in the diagnostic and therapeutic assessment of various oncological disorders. Medical imaging provides a visualized way to keep track of tumor status. On the other hand, tumor hypoxia, which characterizes a depletion of oxygen (O2) supply, is a major factor that impedes radiotherapy. Based on the above concerns the dissertation consists of two parts, which explores to automatically detect tumor using deep learning methods, and also investigates the characteristics of chronic and acute hypoxia.