2296 - Feasibility and Accuracy of CBCT-MRI Image Matching for Daily Positioning in MRI-Only Pelvic Radiotherapy
Presenter(s)
X. Xin, M. Liu, D. Zhang, B. Tang, J. Ma, F. Wu, X. Wang, J. Wu, J. Li, L. Sun, J. Xie, and L. C. Orlandini; Sichuan Cancer Hospital and Research Institute, University of Electronic Science and Technology of China, Chengdu, China
Purpose/Objective(s): MRI-only workflows in pelvic radiotherapy remove the need for CT-based image registration, providing superior soft-tissue contrast for improved target delineation. However, accurate daily patient positioning remains challenging, requiring CBCT-to-MRI rather than CBCT-to-CT registration. This study assesses the clinical accuracy of CBCT-MRI registration for daily image guidance, with a focus on automatic registration reliability and the frequency of manual adjustments.
Materials/Methods: Thirty patients receiving pelvic radiotherapy for prostate (n=10), cervical (n=6), or rectal cancer (n=14) at our institution were retrospectively analyzed. Simulation MRI and CT scans were acquired on the same day to maintain anatomical consistency. Imaging was conducted using different 3 Tesla MRI scanners available, with standardized protocols to ensure comparability. To use MRI as the primary reference for daily positioning, DICOM metadata were modified for compatibility with the oncology information system and treatment workflow. All patients underwent daily CBCT-based positioning on multiple linear accelerators. In the clinical workflow, CBCT images were registered with CT, while offline MRI registrations were performed using automated bone algorithms; both registrations were followed by manual refinement when needed. Two registrations per patient were assessed at the start and mid-treatment, totaling 60 evaluations. CBCT-MRI registration accuracy was determined by comparing X, Y, and Z displacements with CBCT-CT matching. The success rate of automatic registration was assessed based on the percentage of cases requiring manual intervention. A paired t-test was used to determine statistical significance, with p < 0.05 considered significant.
Results: CBCT-MRI registration demonstrated comparable positioning accuracy to conventional CBCT-CT matching (p > 0.05), with mean displacement values summarized in Table 1. Automatic registration was successful in 92% of CBCT-CT and 88% of CBCT-MRI cases, with manual intervention required particularly for significant rectal gas or pelvic anatomical variations.
Conclusion:
A fully MRI-driven workflow for pelvic radiotherapy is feasible, with CBCT-MRI registration providing a viable alternative to CBCT-CT for daily positioning. The clinical implementation of this method requires standardization of imaging protocols, optimization of automated registration algorithms, and structured therapist training. Abstract 2296 - Table 1: Displacement comparison: CBCT-MRI vs. CBCT-CT registration, mean ± std.dev| Match Method | ?x (mm) | ?y (mm) | ?z (mm) |
| CT-CBCT | 0.50 ± 0.25 | 0.20 ± 1.04 | 1.37 ± 1.08 |
| MRI-CBCT | 0.43 ± 0.13 | 0.05 ± 1.05 | 1.63 ± 1.03 |
| p-value | 0.546 | 0.999 | 0.675 |