This project aims at pushing the boundaries of interventional image processing. We do so by 1) developing tools that enable machine learning in scenarios where little data is available, and 2) by devising predictive models that benefit intra-operative decision making.