A key goal encompassing the various projects is to develop new and robust AI image analysis methodologies to extract objective and quantitative measurements of tumor changes to aid and complement radiologist interpretation of images. To this end, we have been developing solutions to objectively quantify volumetric and imaging tumor texture heterogeneity measures to early detect developing cancer resistance to treatments as well as to early diagnose tumor recurrence following treatments. We have developed novel solutions to reliably segment and track tumor changes from multiple imaging modalities, including computed tomography (CT), cone-beam CT, and MRI images to enable reproducible and high-throughput automated radiomics analysis, as well as methods to early detect tumor regrowth from surgical endoscopy images.