Room: Stars at Night Ballroom 2-3
Purpose: Real-time high-soft tissue contrast MRI from the MR-Linac offers the best opportunity for accurate motion tracking during radiation therapy delivery. This work aims to investigate the efficacy of real-time organ motion tracking based on 2D-to-3D registration of MRI acquired on MR-Linac.
Methods: 2D and 3D abdominal MRIs were acquired for 10 healthy volunteers using a high-field MR-Linac (Unity, Elekta AB, Stockholm, Sweden). For each volunteer, 3D respiration-gated T2 and 2D T2/T1-weighted cine in sagittal and coronal planes with planar temporal resolution of 0.6 s for 60 s were captured. Datasets were also collected on MR-compatible physical (Model 008M, CIRS) and virtual (XCAT, Duke University) 4D motion phantoms. Pancreatic head contours from the 3D T2 were populated to the cine and assumed as the ground-truth motion. We performed 2D-to-3D registration using a research software (Elekta AB) to track the pancreatic head motion. Standard deviations of the difference (SDD) between the ground-truth and tracking were analyzed.
Results: For physical (virtual) 4D motion phantoms, the system accurately tracked the motion with SDDs of 0.40 (1.17), 1.08 (1.25), and 0.74 (0.25) mm for anterior-posterior (AP), left-right (LR), and superior-inferior (SI) directions, respectively. The software tool was able to accurately track the motion of the pancreatic head for 6 of 10 volunteers having average centroid motions of 5.98, 5.78, and 18.02 mm with average SDD values of 1.23, 1.05, and 2.12 mm in the AP, LR and SI directions, respectively. The rest of the cases (4/10) exhibited larger SDDs primarily due to the extremely large and/or complex motions.
Conclusion: Feasibility to track organ motion using 2D-to-3D registration of MRIs was demonstrated. Tracking accuracy of about 2 mm was achieved for the motion of the pancreatic head for less extreme motion. Further development is ongoing to improve the tracking algorithm for large and complex motions.
Funding Support, Disclosures, and Conflict of Interest: Funding: Elekta research grants and MCWCC/FH Foundation. Martin Lachaine, David Cooper, and Fabienne Lathuiliere are employees of Elekta AB. This work is partially supported by Elekta AB.