Automated Log File–Based Quality Assurance of Mlc Positional Accuracy for a Magnetic Resonance–Guided Linear Accelerator
Abstract
Purpose
To develop and evaluate an automated, log-file–based MLC positional accuracy QA method for routine and longitudinal assessment of MLC performance on the ViewRay MRIdian MR-Linac system.
Methods
Four picket fence plans were generated in the ViewRay MRIdian treatment planning system at gantry angles of 0°, 90°, 180°, and 270° and delivered on the ViewRay MRIdian MR-Linac equipped with a dual-stack, dual-focused 138-leaf MLC. Each plan consisted of 11 segments with a nominal 6-mm leaf gap. An automated QA pipeline was developed in Python to parse treatment plan reports and delivery log files and perform a comprehensive analysis. Due to the unique ViewRay MLC design, upper and lower MLC stacks were analyzed independently. QA metrics included per-leaf and per-segment maximum absolute positional errors, as well as the overall maximum error relative to nominal planned positions.
Results
The maximum absolute deviation across all leaves was ≤0.3 mm, satisfying both the AAPM TG-142 action limit (±1 mm) and the more stringent AAPM MPPG 8.b. action level (±0.5 mm). Higher localized deviations were consistently observed at segments 2 and 10 and at gantry angles of 90° and 270°, consistent with expected gantry-angle dependence. Longitudinal analysis across two QA runs demonstrated stable MLC performance, with effective identification of outliers exceeding warning and action thresholds.
Conclusion
An automated, log-file–based MLC QA framework was developed for the ViewRay MRIdian MR-Linac, enabling efficient routine and longitudinal performance evaluation. The method supports integration into weekly or monthly QA workflows, with film-based methods reserved for annual independent verification.