Room: Exhibit Hall | Forum 4
Purpose: Determine prognostic indicators for IMRT QA failure based on plan complexity metrics.
Methods: ESAPI scripts extracted plan information to calculate various IMRT complexity metrics previously published in recent literature. Information such as leaf movement/location, control point MU, gantry position/speed, etc. were utilized to calculate complexity metrics. 10 different metrics were calculated, each identifying different aspects of a plan to possibly indicate IMRT QA failure.130 non-SRS VMAT plans (36 of which failed IMRT QA) and 38 SRS VMAT plans (20 of which failed IMRT QA) were analyzed with scripts written to calculate complexity metrics and store data within SQL databases. All plans were remeasured on one linac to ensure identical measurement conditions using a cylindrical diode array. Gamma pass rates for 2%/2mm were calculated with a pass rate of 90% considered passing IMRT QA. ROC curves were generated by varying the thresholds associated with each metric to determine a planâ€™s pass/fail status.
Results: Two different sets of the 10 metrics studied were found to give the highest predictability of specific planâ€™s IMRT QA result. For non-SRS VMAT plans: Plan MU Factor (PMUF), ratio of dose per fraction to total MU delivered, and Plan Area Jaw Ratio (PAJR), a ratio of the average aperture open by the leaves to that of the jaw size, weighted by MU, produced a Sensitivity of 77.4% and Specificity of 88.9%. For SRS VMAT plans: Plan Average Modulation (PAM), a measure of the MUs delivered as a function of aperture area throughout each arc, and PAJR, as previously described, produced a Sensitivity of 83.3% and a Specificity of 95.0%.
Conclusion: No one complexity metric studied here had the ability to predict IMRT QA failures. However, depending on plan type, a combination of metrics may indicate a failure prior to IMRT QA measurement.