Grant: CIHR Fellowship

Mouth cancer (OC) has a worldwide impact with a poor chance of survival. Cancer spreading to the neck (LN+) is a major contributor to poor survival and morbidity, which can be prevented by surgically removing lymph nodes (END) at the earliest stage. However, not all OCs require END as only 25% will develop LN+. Therefore, to avoid over-treating and under-treating the patients, an objective biological marker is needed, yet currently lacking, to accurately identify patients who need END. We have shown that LN+ OC can be detected by characterizing the microscale physical characteristics of cancer cells and abnormality in certain microRNA and inflammatory gene expression. These new biomarkers can detect aggressive early-stage OC with > 90% accuracy compared to currently pathological methods. This project is to build deep learning models for multi-modal data integration to verify and validate the reported signatures case-controlled biopsy and surgical samples. This project's success will help create new risk-assessment methods to improve the decision in END and OC outcomes by avoiding unnecessary health and societal impact caused by unnecessary END or salvage neck treatment.