Göttingen 2025 – scientific programme
Parts | Days | Selection | Search | Updates | Downloads | Help
ST: Fachverband Strahlen- und Medizinphysik
ST 1: Computational Methods and Simulation
ST 1.3: Talk
Tuesday, April 1, 2025, 14:15–14:30, ZHG003
Fast dose predicition for treatment planning in matRad — •Ruben Trimpop1, Carsten Burgard1, Tobias Cremer2, Cornelius Grunwald1, Kevin Kröninger1, Florian Mentzel2, Marco Schlimbach1, and Jens Weingarten1 — 1TU Dortmund — 2Formerly TU Dortmund
Microbeam radiation therapy (MRT) is a preclinical method for tumor treatment, demonstrating potential for improved post-treatment outcomes. The method employs a multi-slit collimator to produce dose peaks with high dose rates, separated by valleys of lower dose rates. This method of spatially segmenting the beam, combined with applying high dose rates over a short amount of time, called FLASH-Therapy, enhances the survivability rate of healthy tissue.
A major impediment to the utilisation of this method is the large amount of time it takes for dose prediction using conventional methods. Previous research has successfully used Machine Learning (ML) networks to significantly reduce the required time for dose prediction.
In this work a ML network is integrated into matRad, an open source software for radiation treatment planning, developed for research purposes, substituting matRads internal dose prediction with a ML based dose prediction. This is one of the first steps to fully enable treatment planning for MRT.
This presentation will showcase the first results of ML integration into matRad, highlighting the advantages of ML-based dose prediction for MRT.
Keywords: Microbeam Radiation Therapy; matRad; Treatment planning; Machine Learning