Segmentation Regularized Training for Multi-Domain Deep Learning Registration applied to MR-Guided Prostate Cancer Radiotherapy

2025-07-09

AI Summary

Deformable registration of MRI images across different scanners (1.5T MR-Linac vs. 3T simulation MR) and magnetic field strengths poses a major challenge in MR-guided adaptive radiotherapy (MRgART) for prostate cancer. Method: We propose ProRSeg, a segmentation-regularized progressive registration-segmentation network, incorporating a weighted segmentation consistency loss to strengthen anatomical constraints and improve domain generalization. Contribution/Results: To our knowledge, this is the first end-to-end framework enabling simultaneous deformable registration and contour propagation across multi-center MRgART settings, supporting clinically feasible dose accumulation. Quantitative evaluation shows cross-domain Dice scores of 0.86–0.89 for bladder, rectum, and clinical target volume (CTV); dose accumulation met clinical constraints in 83.3% of patients. This work establishes a high-accuracy, verifiable technical pathway for multi-domain MRI-driven adaptive radiotherapy.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Intelligent Robots: Multi-Robot Systems

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Background: Accurate deformable image registration (DIR) is required for contour propagation and dose accumulation in MR-guided adaptive radiotherapy (MRgART). This study trained and evaluated a deep learning DIR method for domain invariant MR-MR registration. Methods: A progressively refined registration and segmentation (ProRSeg) method was trained with 262 pairs of 3T MR simulation scans from prostate cancer patients using weighted segmentation consistency loss. ProRSeg was tested on same- (58 pairs), cross- (72 1.5T MR Linac pairs), and mixed-domain (42 MRSim-MRL pairs) datasets for contour propagation accuracy of clinical target volume (CTV), bladder, and rectum. Dose accumulation was performed for 42 patients undergoing 5-fraction MRgART. Results: ProRSeg demonstrated generalization for bladder with similar Dice Similarity Coefficients across domains (0.88, 0.87, 0.86). For rectum and CTV, performance was domain-dependent with higher accuracy on cross-domain MRL dataset (DSCs 0.89) versus same-domain data. The model's strong cross-domain performance prompted us to study the feasibility of using it for dose accumulation. Dose accumulation showed 83.3% of patients met CTV coverage (D95 >= 40.0 Gy) and bladder sparing (D50 <= 20.0 Gy) constraints. All patients achieved minimum mean target dose (>40.4 Gy), but only 9.5% remained under upper limit (<42.0 Gy). Conclusions: ProRSeg showed reasonable multi-domain MR-MR registration performance for prostate cancer patients with preliminary feasibility for evaluating treatment compliance to clinical constraints.

Problem

• Develop domain-invariant MR-MR registration for prostate cancer
• Improve contour propagation accuracy across different MRI domains
• Evaluate dose accumulation feasibility in MR-guided radiotherapy

Innovation

• ProRSeg method for multi-domain MR-MR registration
• Weighted segmentation consistency loss training
• Cross-domain generalization for dose accumulation