METHODS: In this study, an in-house developed Eclipse Scripting API (ESAPI) script was used to extract five independent plan complexity metrics from N = 667 VMAT treatment fields. The correlation between metrics and portal dosimetry measurements was investigated with Pearson correlation, box plot analysis and receiver operating characteristic curves, which were used to defined the best performing metric and its threshold.
RESULTS: The incidence of fields failing the clinical PSQA criteria of 3%/2mm (NFRT) and 3%/1.5mm (SBRT) was low (N = 1). The mean MLC opening was the metric with the highest correlation with the portal dosimetry data and among the best in discriminating the requirement of PSQA. The thresholds of 16.12 mm (NFRT) and 7.96 mm (SBRT) corresponded to true positive rates higher than 90%.
CONCLUSIONS: This work presents a quantitative approach to reduce the time allocation for PSQA by identifying the most complex plans demanding a dedicated measurement. The proposed method requires PSQA for approximately 10% of the plans. The ESAPI script is distributed open-source to ease the investigation and implementation at other institutions.