Despite progress in cancer treatment, many patients still do not respond well to prescribed therapy and may experience severe side effects without benefit. To avoid this, we need methods that can predict the best therapy for a given patient. In this thesis, we study how immune system fights cancer to improve such predictions. When cancer develops, it reshapes the surrounding tissue to support its growth, while immune cells come to the tumor site to eliminate it. This complex tumor-immune microenvironment contains information crucial for predicting response to treatment. To study this, we developed an analytical framework that uses a deep learning model to detect immune cells in microscopic images of tumor. We also designed a novel method summarizing how cells are arranged in space, reflecting the “strategy” immune system chose to fight tumor. Applying this approach to salivary gland cancer identified immune cell patterns linked to longer remission after treatment.
Evgenia Martynova (1990) obtained her bachelor’s degree in Applied Mathematics and Informatics in Voronezh, Russia, and then worked as a software engineer. In 2019, she decided to return to studies and obtained her master's degree in Computing Science (Data Science track), cum laude, at Radboud University in 2021. In the same year, she began her PhD in the Computational Immunology Group at Radboudumc. Currently, she continues her academic career as a postdoctoral researcher in the Computational Pathology Group of Radboudumc.