ESTRO 2025 - Abstract Book

S3441

Physics - Machine learning models and clinical applications

ESTRO 2025

Conclusion: We developed a DL model that can accurately and quickly predict treatment isocenter shift vectors by using dose distributions. This method reduces the need for human intervention, enhances workflow efficiency and patient comfort by enabling consistent patient pretreatment positioning time reduction from 15min to 3min. Future work aims to further refine the model towards clinical implementation and expand it to other tumor sites.

Keywords: Dose-guided Patient Positioning, Dual-CNN model