ERC Starting Grant for Geert Litjens

The European Research Council awarded Geert Litjens with a grant for his project AIS-CaP: Interpretable Artificial Intelligence across Scales for Next-Generation Cancer Prognostics.

Microscopic assessment of cancer tissue by pathologists is one of the cornerstones of diagnostic and treatment decision making for patients. However, the current assessment protocols, referred to as cancer grading systems, are limited to features that can feasibly be extracted by humans, leaving vast amounts of information captured in cancer tissue unused. Machine learning has already shown that it is capable of improving the diagnostic performance and efficiency of pathologists in applying the current grading systems. However, applications of ML have been mostly limited to replicating what humans can do, instead of trying to go beyond that, mainly due to methodological limitations and knowledge gaps around AI transparency and explainability. Geert Litjens is set to address these knowledge gaps. He will use innovative learning algorithms as well as latest advances in self-supervision, vision transformers, and natural language processing to make deep neural networks more effective in medical applications and endow them with transparency and explainability. His project will validate the methodology in the largest dataset of oncological whole slide images in the world. Finally, by publicly releasing all developed tools and data, the proposed project aims to have a scientific multiplier effect in the fields of oncology, computational pathology and machine learning, including possible applications in the fields of remote sensing and autonomous vehicles.