Speaker
Description
Guillaume Landry is W2 professor at the Department of Radiation Oncology of the LMU University Hospital in Munich. He leads the LMU Adaptive Radiation Therapy Lab focusing on AI-based automation in ART. The lab has recently sought broader engagement with the medical AI community and radiotherapy by organizing the TrackRAD2025 and DoseRAD2026 data challenges. Of relevance to this talk, DoseRAD2026 aims at identifying the best algorithms for real-time dose calculation in ART at MRI-linacs, potentially bypassing the need for synthetic CT images.
Summary
Monte Carlo (MC) simulation is the accuracy gold standard for radiotherapy dose calculation, but its computational cost is often a bottleneck. Even GPU implementations may take tens of seconds per plan, constraining online adaptation and precluding real-time dose-guided radiotherapy.
This talk examines whether deep learning can predict dose directly from images at MC-level accuracy while running fast enough for real-time use. I frame the problem across four clinically distinct scenarios: photon and proton therapy, each on CT and MRI. The MRI cases are especially compelling, since direct prediction could bypass synthetic CT generation and enable intra-fraction dose reconstruction at MRI-linacs.
To benchmark these approaches, I introduce DoseRAD2026, a MICCAI 2026 grand challenge built on the paired CT-MRI SynthRAD2025 dataset. It provides the first public beam-level dose database with Geant4 ground truth, ranking submissions jointly on dosimetric accuracy and runtime under a strict one-second-per-beam ceiling. I close by asking what such a benchmark reveals about the AI-versus-MC trade-off, and what still stands between these methods and the clinic.