ESTRO 2025 - Abstract Book

S3472

Physics - Optimisation, algorithms and applications for ion beam treatment planning

ESTRO 2025

Conclusion: Latent space distance and ensemble model variance demonstrated potential for assessing the reliability of CNN based LET predictions on independent datasets. The findings support the use of these techniques for estimating model performance also on datasets for which ground-truth LET data are unavailable.

Keywords: linear energy transfer, uncertainty estimation

References: [1] Starke, S., Kieslich, A., Palkowitsch, M., Hennings, F., G C Troost, E., Krause, M., Bensberg, J., Hahn, C., Heinzelmann, F., Bäumer, C., Lühr, A., Timmermann, B., & Löck, S. (2024). A deep-learning-based surrogate model for Monte-Carlo simulations of the linear energy transfer in primary brain tumor patients treated with proton-beam radiotherapy. Physics in medicine and biology, 69(16), 10.1088/1361-6560/ad64b7. https://doi.org/10.1088/1361 6560/ad64b7

1831

Proffered Paper Automated patient-specific beam angle optimisation in IMPT – a novel fast algorithm with validation Wens Kong 1 , Merle Huiskes 2 , Steven J.M. Habraken 2,3 , Eleftheria Astreinidou 2 , Coen R.N. Rasch 2 , Ben J.M. Heijmen 1 , Sebastiaan Breedveld 1 1 Department of Radiotherapy, Erasmus MC Cancer institute, Erasmus University Medical Center, Rotterdam, Netherlands. 2 Department of Radiation Oncology, Leiden University Medical Center, Leiden, Netherlands. 3 Medical Physics and Informatics, HollandPTC, Delft, Netherlands Purpose/Objective: Patient-specific beam angle optimisation (BAO) in IMPT is a complex combinatorial problem. Manual iterative planning to establish patient-tailored angles is an often prohibitively lengthy process for large patient groups with no guarantee of optimality. This study aimed to develop a fully-automated, dosimetry-driven approach for optimisation of patient-specific coplanar beam configurations, designated pBAO (proton-BAO), and validate it for head and neck patients.