BMFTR KI-ROBUST

The joint BMFTR project KI-ROBUST addresses the growing demand for high-quality radiotherapy treatment in the face of increasing cancer incidence and a shortage of clinical specialists. Its central goal is to make treatment planning more efficient, reliable, and trustworthy by combining uncertainty-aware artificial intelligence (AI) with advanced methods of robust optimization. To this aim researchers from the LMU University Hospital Department of Radiation Oncology, the Fraunhofer ITWM Kaiserlautern and the LMU Chair for Mathematical Foundations of Artificial Intelligence will team up.

At the LMU ART Lab, the project primarily focuses on developing uncertainty-aware deep learning–based auto-segmentation models for both organs-at-risk and tumor volumes, adressing epistemic as well as aleatoric uncertainties. The output of these models will serve as starting point for developing inverse robust treatment planning strategies at ITWM and will be further enhanced by explainable AI approaches
developed at the LMU Chair for Mathematical Foundations of Artificial Intelligence.

By integrating these innovations, the project seeks to enable faster, more resilient, and clinically transparent radiotherapy planning.