Feasibility of a remote, automated daily delivery verification of volumetric‐modulated arc therapy treatments using a commercial record and verify system

Volumetric-modulated arc therapy (VMAT) is an effective but complex technique for delivering radiation therapy. VMAT relies on precise combinations of dose rate, gantry speed, and multileaf collimator (MLC) shapes to deliver intensity-modulated patterns. Such complexity warrants the development of correspondingly robust performance verification systems. In this work, we report on a remote, automated software system for daily delivery verification of VMAT treatments. The performance verification software system consists of three main components: (1) a query module for retrieving daily MLC, gantry, and jaw positions reported by the linear accelerator control system to the record and verify system; (2) an analysis module which reads the daily delivery report generated from the database query module, compares the reported treatment positions against the planned positions, and compiles delivery position error reports; and (3) a graphical reporting module which displays reports initiated by a user anywhere within the institutional network or which can be configured to alert authorized users when predefined tolerance values are exceeded. The utility of the system was investigated through analysis of patient data collected at our clinic. Nearly 2500 VMAT fractions have been analyzed with the delivery verification system at our institution. The average percentage of reported MLC leaf positions within 3 mm, gantry positions within 2°, and jaw positions within 3 mm of their planned positions was 92.9% ± 5.5%, 95.9%± 2.9%, and 99.7% ± 0.6%, respectively. The level of agreement between planned and reported MLC positions decreased for treatment plans requiring larger MLC leaf movements between control points. Differences in the reported MLC position error between the delivery verification system and data extracted manually from the control system were noted; however, the differences are likely systematic and, therefore, may be characterized if appropriately accounted for. Further investigation is needed to confirm the utility and accuracy of the system.