ESTRO 2024 - Abstract Book

S4288

Physics - Intra-fraction motion management and real-time adaptive radiotherapy

ESTRO 2024

Conclusion:

A novel patient-specific training strategy was developed for real-time target tracking with a Transformer model. We demonstrated its superiority over unsupervised and supervised strategies, with training times that align with existing online adaptive pre-treatment workflows.

Keywords: Deep Learning, Tumor Tracking, MRIgRT

References:

[1] Keall, Paul J., et al. "ICRU REPORT 97: MRI-Guided Radiation Therapy Using MRI-Linear Accelerators." Journal of the ICRU 22.1 (2022): 1-100.

[2] Chen, Junyu, et al. "Transmorph: Transformer for unsupervised medical image registration." Medical image analysis 82 (2022): 102615.

Funding: DFG GRK2274

1792

Digital Poster