Krzysztof J. Geras, PhD

My background is in computer science and machine learning. My main research interests are in developing novel machine learning methods (multi-modal learning, self-supervised learning, survival analysis, and evaluation of machine learning models) and applying these methods to medical image analysis.

Over the last few years my efforts were concentrated on analyzing breast imaging data. With my colleagues and students, we developed a series of AI models for mammography, breast ultrasound and breast MRI. To validate these models, I am frequently collaborating with my clinical colleagues on retrospective and prospective studies. An AI model for breast cancer screening created by my research group is currently deployed clinically and available to patients for free.

My latest efforts are directed at creating AI models that go beyond supporting humans in existing tasks by exploiting signals in data which are imperceptible to humans, even experts. I intend to use such models for cancer detection and risk prediction, as well as scientific discovery.