Relationship between IMRT Complexity, Monte Carlo Dose Verification, and Phantom-Based QA In SBRT Plans
Abstract
Purpose
To evaluate correlations among publicly accessible IMRT plan complexity metrics and their relationships with plan uncertainty, independent dose calculation, phantom-based QA, and log-file analysis.
Methods
Ten complexity metrics based on dose, monitor units (MU), and MLC parameters (e.g., open area, perimeter, leaf travel) were computed for 5-fraction lung SBRT plans using Radformation’s ClearCheck and two open-source ESAPI scripts. All plans used 6 MV energy and were calculated with AcurosXB v16.1. Delivery was performed on a Varian TrueBeam, and measured with an SNC SRS-MapCHECK. Absolute gamma passing rates were evaluated with 3% dose / 1 mm distance / 10% threshold criteria. Independent 3D dose calculations were performed in RadMonteCarlo for initial and log-reconstructed plans using 2%/1 mm/10% criteria for the full matrix and PTV. Plan uncertainty was qualitatively assessed using ±3 mm x/y/z isocenter shifts in the Eclipse Plan Uncertainty tool.
Results
Eight metrics incorporating MLC open area were strongly correlated with each other (Pearson r = 0.56–1.0) but not with total MU or total leaf travel. The MU-duty cycle showed the strongest correlation (r = 0.72) with Monte Carlo dose differences. Log-based dose variations were minimal. Complexity metrics showed weak correlation with phantom-based QA results (max r ≈ 0.3). Plans with higher complexity exhibited more pronounced dose streaks and hotspots in nearby critical structures under simulated isocenter shifts.
Conclusion
For clinically approved SBRT IMRT plans, complexity scores showed limited correlation with consistently high gamma pass rates (>98% at 3%/1 mm). Simple indicators—such as MU-duty cycle—may provide sufficient warning of potential over-modulation and support improved plan robustness.