Automated Workflow for Rapid VMAT Plan Migration across Linac Geometries without Manual Re-Planning
Abstract
Purpose
Unplanned linac downtime necessitates rapid transfer of VMAT plans to backup machines. Prior methods relying on geometric leaf mapping often fail when transferring between different MLC models or beam profiles. This study validates a workflow that augments mathematical leaf interpolation to fully adapt plans to new machine constraints without manual re-planning.
Methods
An in-house ESAPI script performed linear interpolation to map leaf positions from High-Definition (2.5 mm) to Standard (5 mm) MLC linacs. Following geometric transfer, plans underwent a final stage, fine-tuning re-optimization (MR4, Eclipse PO 15.6) using original objectives. The workflow was tested on five prostate cases and one complex Oropharynx challenge case (normalized to mandible constraints). Performance was evaluated using Global Max Dose, Conformity Index (CI), Gradient Index (GI), Homogeneity Index (HI), and Target Coverage.
Results
Geometric interpolation alone resulted in severe dosimetric failure. In the Oropharynx case, PTV and GTV coverage (V100%) collapsed from 82.6% and 83.7% (Original) to 11.9% and 14.6% (Direct Transfer) due to the inability of mapped leaves to navigate the OAR interface. However, MR4 optimization recovered coverage to 81.3% and 83.2%, matching the HD-MLC baseline, while maintaining critical OAR doses within ±7% of original values. Similarly, for the Prostate cohort, the optimizer restored conformity to baseline levels (p = 0.083), correcting the significant degradation observed in the direct transfer. While the Gradient Index increased by 4% due to coarser leaf resolution, Global Max Dose and HI remained clinically acceptable, showing deviations of 2% and 0.018, a marked improvement over the 16% and 0.14 increases in the direct transfer.
Conclusion
While geometric leaf mapping provides a starting point, it is dosimetrically insufficient for complex transfers between mismatched linacs. The addition of MR4 optimization is critical, effectively resolving discrepancies in MLC geometry and beam profiles to ensure safe, high-quality treatment continuity during machine downtime.