Section leader: dr. ir. Marius Staring
Machine learning is a powerful technique that emerged from computer science: it gives computers the ability to learn from (annotated) observations and make predictions on new data. Among others, deep learning (DL) is a paradigm in machine learning that has recently revolutionized the field of computer vision, and is starting to impact medical image analysis as well. The research section “Biomedical Machine Learning” aims to develop generic machine learning approaches for automated image analysis techniques, and to deploy these in the clinical and life-science research at LUMC.
Current projects include machine learning for disease classification and staging, for segmentation, image registration and uncertainty estimation.
Image registration techniques are historically a strong point in our group, focusing amongst others on fast (optimization) methods that render this technique usable in a time-critical intra-operative setting. An important application area is that of adaptive radiation therapy (photon as well as proton), which is an interesting setting for its demand for real-time solutions that are robust to real-life variations in patients. We explore deep learning methods for segmentation of target areas and organs-at-risk, and moreover for regression problems such as registration.
We aim for high quality open source algorithms, to promote widespread routine usage of our techniques. This is amongst others facilitated by our registration software elastix and SuperElastix, publicly available as open source via:
- ADAPTNOW: High-Precision Cancer Treatment by Online Adaptive Proton Therapy
- Registration visualization
- Medical Image Registration - Linking Algorithm and User
- Fast Image Registration for Time-critical Medical Applications
- Brain MRI Image Analysis for an Ageing Population
- Esophageal Gross Tumor Volume Segmentation using Deep Learning