Deep learning applications in lung cancer imaging

Lung cancer is the leading cause of cancer-related mortalities worldwide and is the second most commonly diagnosed cancer in both men and women. This PhD research explores the development of computational image analysis and machine learning methods to extract meaningful information from routine medical imaging data of lung cancer patients. Within a diagnostic context, such information may be used to predict clinically relevant information such as expected survival, as well as stratify patients into low- and high-risk groups to guide treatment decisions. Within a therapeutic context, the methods developed herein may be used to detect and delineate tumors in medical images, thereby aiding in the delivery of radiation therapy. Experimentation focused on real world applicability, generalizability, and robustness using large-scale multi-institution data and by publicly disseminating the resulting models and computational source code. This research was conducted in cooperation with clinicians and researchers from the Harvard Medical School, Dana-Farber Cancer Institute, Brigham and Women's Hospital, and Maastricht University.