Benchmarking of Low-Field MR-Linacs in a Multi-Institutional International Consortium
Glide-Hurst C, Yadav P, Gutierrez A, Gungor G, Hoffmans D, Mittauer K, Sethi A, Romaguera T, Pennell R, Palacios M, Omari E, Meeks S, Mailleux H, Green O, Fau P, Doemer A, DeWyngaert JK, Boye K, Shah A, Klueter S, Lotey R, and Bellon M. Benchmarking of Low-Field MR-Linacs in a Multi-Institutional International Consortium. J Med Phys 2019; 46(6):e200.
J Med Phys
Purpose: Recently, MR-linacs have been integrated into clinical practice, introducing new needs for QA and baseline machine characterization. This work summarizes a multi-institutional evaluation 12 low-field MR-linacs with the overarching goal of benchmarking machine performance. Methods: Acceptance and commissioning data were analyzed for 12 0.35 T ViewRay MRIdian linacs equipped with double-focused MLCs (4 mm aperture resolution). MRI-radiation isocenter accuracy was assessed. Couch transmission was measured at various beam angle incidences. Dosimetric evaluation included 6XFFF photon beam spot size, profiles, PDD curves, chamber-corrected Monte-Carlo derived relative photon OFs (0.83-25.6 cm2 field sizes), temporal output factor stability, and MLC transmission/leakage. End-to-end testing and IMRT performance were evaluated. MRI benchmarking included spatial integrity, magnetic field homogeneity (MFH) using spectral peak analysis (5-12 gantry angles), and image quality evaluation via ACR/NEMA standards. Clinical integration including QA timelines, staffing, and equipment were summarized. Results: MRI/laser/radiation isocenter coincidence was ≤0.8 mm for all MR-linacs. Couch transmission ranged from 13% to 17% (180° and 140°, respectively) requiring inclusion in treatment planning. Excellent agreement in PDD(10)x was observed (64.1 ± 0.4%) with spot sizes of 0.15 ± 0.03 mm. The largest discrepancy in corrected OFs was 0.72 ± 0.03 (0.83 cm2 field size) while all other OFs were in close agreement. Average output values within 2-18 months of initial calibration were <1% of nominal; four institutions adjusted output at ∼90 days. On average, MLC transmission and leakage were <0.3% and all IMRT plans were within 99% agreement of expected (3%/3 mm). MRI ACR and vendor-specified limits were met for all image quality metrics. Gantryangle dependence of MFH was observed (2.93 ± 1.82 ppm) with 3/12 institutions exceeding 5 ppm at a subset of angles, warranting a dynamic gantry angle-dependent shim. Conclusion: Overall, excellent agreement in multiinstitutional commissioning data was observed, providing important comparison data to others embarking on MR-linac commissioning.