Informatics Colloquium 7.12. RationAI: Rational and conservative artificial intelligence (not only) in digital pathology
Informatics Colloquium 7.12. 2021, 14:00 lecture hall D2
doc. RNDr. Tomáš Brázdil, Ph.D., FI MU
Abstract: I will present the new research group RationAI concerned with explainable artificial intelligence (AI) techniques in biomedicine. I will concentrate on a single case study, analyzing whole slide images (WSI) from digital pathology. Our primary concern is to find out how good cutting-edge methods are and to make them useful for pathologists. Our aims will be illustrated using our deep learning-based system for cancer detection in WSI of prostate cancer. Subsequently, I will explain how explainability can be achieved using (slightly modified) standard methods for explainable AI and how pathologists interpret our explanations. Finally, I will present further plans of the RaionAI group in pathology and other areas of AI in biomedicine.