Multi-modal magnetic resonance imaging has improved detection of prostate cancer. The ability of functional imaging modalities allows characterization of lesions diffusion and perfusion characteristics. We identify aggressive imaging regions (‘Imaging habitats’) by co-localizing multimodal characteristics and describe them using quantitative imaging metrics (radiomics). We believe these metrics along with machine learning methods allow us to identify aggressive disease patterns.
B-cell lymphoma (stage 4) is one of an aggressive disease with a median 5-year survival is about 65% and recent development of immune therapy has been effective in improving prognosis in these patients. We propose to characterize patients lesion seen on their imaging scans (PET/CT) and identify image based biomarkers to identify patients that will benefit from immune therapy and prognosticate risk of recovery.
Lung cancer is the largest cause of cancer death in the U.S. and globally. The 5-year survival for patients diagnosed with a non-small cell lung carcinoma remains dismal at 21%, largely attributed to lack of early detection. The 5-year survival for early stage IA patients is 49%. We propose to us patients CT images to prognosticate disease progression risk. Using radiomic approach to characterize nodules in 3D and these features along with advanced deep learning methods provide malignancy risk predictors that allows stratification of progression risk for the patients.