Statistical quality control for volumetric modulated arc therapy (VMAT) delivery using machine log data

TL;DR

This study develops a statistical quality control method using machine log data to monitor and improve the accuracy of VMAT delivery, demonstrating its effectiveness in real clinical cases.

Contribution

It introduces a novel application of statistical process control to machine log data for real-time monitoring of VMAT delivery errors.

Findings

Sigma_GA depends on gantry speed

SPC effectively monitors delivery errors

SBRT VMAT shows smaller geometric errors

Abstract

The aim of this study is to set up statistical quality control for monitoring of volumetric modulated arc therapy (VMAT) delivery error using machine log data. Eclipse and Clinac iX linac with the RapidArc system (Varian Medical Systems, Palo Alto, USA) is used for delivery of the VMAT plan. During the delivery of the RapidArc fields, the machine determines the delivered motor units (MUs) and gantry angle position accuracy and the standard deviations of MU (sigma_MU; dosimetric error) and gantry angle (sigma_GA; geometric error) are displayed on the console monitor after completion of the RapidArc delivery. In the present study, first, the log data was analyzed to confirm its validity and usability; then, statistical process control (SPC) was applied to monitor the sigma_MU and sigma_GA in a timely manner for all RapidArc fields: a total of 195 arc fields for 99 patients. The sigma_MU and sigma_GA were determined twice for all fields, that is, first during the patient-specific plan QA and then again during the first treatment. The sigma_MU and sigma_GA time series were quite stable irrespective of the treatment site; however, the sigma_GA was strongly dependent on the gantry rotation speed. The sigma_GA of the RapidArc delivery for stereotactic body radiation therapy (SBRT) was smaller than that for the typical VMAT. Therefore, SPC was applied for SBRT cases and general cases respectively. Moreover, the accuracy of the potential meter of gantry rotation is important, since the sigma_GA can be dramatically changed due to its condition. By applying SPC to the sigma_MU and sigma_GA, we could monitor the delivery error efficiently. However, the upper and lower limits of SPC need to be determined carefully with full knowledge of the machine and log data.