崔 一

My research focuses on quantitative analysis of medical images so as to identify imaging biomarkers for cancer prognostication, and furthermore, to build computational models for patient stratification and personalized medicine. Advances in imaging technology have made possible identification of a variety of phenotypes beyond the molecular level—from anatomical to functional to metabolic—captured by medical images. Therefore, mining the information embedded in these medical imaging data will lend new insights into the disease process and even new therapeutic strategies. At this time, however, existing analyses of medical images are mainly qualitative, depending heavily on labor-intense visual assessment. Development of quantitative, automated and robust approaches is thus in pressing need given the vast availability of imaging data. I have therefore proposed innovative prognostic models for glioblastoma (GBM) and pancreatic cancer using MR or PET images. I have also identified novel PET imaging biomarkers for early prediction of response to radiochemotherapy in head and neck cancer patients. My near-term aim is to investigate these models and biomarkers for their biological significance and to correlate these findings with known oncogenes. If validated prospectively, they may be used as non-invasive surrogates for genomic signatures and for individualized therapy in the future.