Cancer Prognosis

We are developing methods to quantify the imaging characteristics of cancer tumors from multimodality imaging data. We are investigating the predictive value of these imaging features as prognostic biomarkers. Our goal is to more comprehensively predict risk of cancer recurrence and metastasis, and incorporate this information into clinical decision making. In addition to characterizing the normal tissue, we have also conducted studies on characterizing intra-tumor heterogeneity via imaging as a biomarker for evaluating therapy response and risk for recurrence. Our work was amongst the first to show that intrinsic imaging phenotypes for breast cancer tumors correlate to gene expression profiling. More recently, we have also extended this work to lung cancer, where we showed that combining liquid biopsy data with radiomic tumor heterogeneity phenotypes can improve the prediction of response to targeted therapy and immunotherapy for non small cell lung cancer (NSCLC). We have also shown that radiomic phenotypes from CT scans can predict response to immunotherapy for NSCLC.