Identifying Poor SBRT Plan Performance Using IMRT QA and Complexity Metrics
Presentations
PO-GePV-T-242 (Sunday, 7/10/2022) [Eastern Time (GMT-4)]
ePoster Forums
Purpose: Previous studies indicate that highly complex radiotherapy plans can produce more errors in dose calculation and delivery. This work aims to identify stereotactic body radiation therapy (SBRT) treatment plan characteristics associated with poor patient-specific quality assurance (QA) results for Elekta accelerators by examining complexity characteristics.
Methods: 62 SBRT plans (30 passing, 32 failing QA) were retrospectively sampled from 2017-2020. Plans were sampled for a variety of disease sites, with proportional representation among passing and failing cohorts. Each plan was delivered on an Elekta Infinity linac for IMRT QA with ArcCheck using 2%/2mm global dose/distance-to-agreement gamma criteria. Plans were analyzed for trends in complexity features, including average MLC gap, tongue and groove index, modulation complexity score, and aperture irregularity as calculated via an in-house script. T-test was used to compare complexity metrics among passing and failing QA plans. For features demonstrating difference between cohorts, receiver-operator characteristic (ROC) analysis was then performed to determine predictability of poor QA results based on complexity.
Results: Using the t-test, metrics found to have distinct distributions between plans passing and failing IMRT QA included average MLC gap, mean tongue and groove index, plan irregularity (Park et al. 2014), and edge metric (Younge, et al. 2012) (p-value <0.02). From ROC analysis, mean MLC gap and mean tongue and groove index demonstrated a marked ability to discriminate passing and failing QA
results: area under the curve (AUC) was 0.803 and 0.826, respectively.
Conclusion: Conventional patient-specific QA identified suboptimal SBRT plans, particularly with respect to MLC aperture characteristics, suggesting a potential to discriminate overmodulation. ROC analysis demonstrated that one could predict quality of plan delivery based on quantification of MLC complexity characteristics.
Keywords
Intensity Modulation, Quality Assurance, ROC Analysis
Taxonomy
TH- External Beam- Photons: Quality Assurance - IMRT/VMAT
Contact Email