Development and Evaluation of a Perpendicular Frame-by-frame Patient-specific QA Method for Large VMAT Fields Using the TrueBeam Electronic Portal Imaging System

Abstract

Background: The verification of VMAT delivery accuracy is widely performed with measurement-based QA methods and gamma index test evaluation. Having the gantry speed as an element of modulation requires that VMAT QA methods resolve the gantry angle accuracy during delivery. EPIDs have increasingly been used for VMAT QA and its minimal size limitation make it advantageous for the measurement of large fields. In this work, we implemented a gantry-resolved EPID-based QA method for patient-specific QA and evaluated its performance for large VMAT fields based on gamma index analysis and process-based tolerance and action limits.Materials and Methods: Our method created gantry-resolved pseudo-3D dose distributions from XIM files acquired in TrueBeam using dosimetry mode acquisition. The method was used for the evaluation of 35 large VMAT fields measured with two different EPID models and two energies, accounting a total of 140 fields divided into 4 different processes. Predicted portal dose distributions were calculated based on MU information contained in the image headers. An independent calibration procedure that only requires MATLAB for the full implementation of the method was developed. Gamma index analyses were performed with a two-step calculation algorithm that increases accuracy in steep dose gradient regions. Acquisition artifacts causing MU information variability and banding patterns were addressed. Gamma pass rates for pseudo-3D and composite 2Ddose distributions were used to calculate process-based tolerance and action limits following the TG-218 methodology.Results: The methods to increase gamma index calculation accuracy and reduce artifacts greatly improved the performance and reduced the variation of our results. An independent calibration procedure was successfully implemented. All calculated tolerance limits were stricter than the action limits, and no gamma pass rate from the pseudo-3D distributions with global normalization fell outside of the recommended universal action limit of 90%.Conclusion: We have demonstrated that our software is suitable for use in patient-specific QA of large VMAT fields. Our results met the recommendation of TG-218. The differences in performance among the processes illustrated they are affected by different sources of variation, indicating that improvements are possible to obtain stricter process-specific tolerance and action limits.Keywords: VMAT QA, EPID, gantry-resolved, gamma analysis, TG-218, tolerance limits