ESTRO 2025 - Abstract Book

S3431

Physics - Machine learning models and clinical applications

ESTRO 2025

3714

Digital Poster A Digital Anatomy-Based Vertebra-Guided Affine Registration Network (VerGAReg) peilin wang 1 , Weiwei Liu 2,3 , Weihu Wang 2,3 , Hao Wu 2,3 , Joseph Lai 1 , Yibao Zhang 3,2 , Jing Cai 1 , Tian Li 1 1 Department of Health Technology and Informatics, the Hong Kong Polytechnic University, hong kong, Hong Kong. 2 Department of Radiation Oncology, Peking University Cancer Hospital & Institute, Beijing, China. 3 Key Laboratory of Carcinogenesis and Translational Research (Ministry of Education/Beijing), Beijing Cancer Hospital & Institute, Beijing, China Purpose/Objective: To develop an automatic affine registration method tailored for abdominal MRI-guided radiation therapy [1], addressing the limitations of manual registration and the underperformance of existing deep-learning algorithms [2] caused by deformable respiratory motion.

Material/Methods:

We propose VerGAReg, a vertebra-guided affine registration network with two main modules: a deep learning based key point extraction module and a parameter fitting module. The key point extraction uses dual-channel