Comparison of log file-based and measurement-based QA for detecting MLC positional errors and evaluating dosimetric impacts of MLC defects

Abstract

Purpose: This study aimed to compare the sensitivity of log file-based quality assurance (LF-QA) and measurement-based quality assurance (MB-QA) for detecting multileaf collimator (MLC) positional errors and to evaluate the dosimetric impacts of MLC mechanical drive train defects. Methods: Mechanical degradation of the MLC was simulated on a TrueBeam STx system by inducing three defect types: T-nut surface wear (0.5–1.2 mm), drive screw thread wear, and motor degradation. MLC positioning accuracy was assessed using a rotational Picket Fence (PF) test, and the dosimetric impacts were evaluated on clinical breast intensity‑modulated radiation therapy (IMRT) and prostate volumetric‑modulated arc therapy (VMAT) plans. LF-QA and MB-QA were performed concurrently under identical delivery conditions. Gamma passing rates (GPRs) and dose-volume histogram (DVH) analyses were compared between baseline and defective deliveries. Results: LF-QA detected positional deviations between baseline and defective conditions (<0.14 mm; p < 0.05) but consistently underestimated the extent of the induced defects. Correspondingly, LF-QA gamma analysis (GPRs ≈ 100%) and DVH metrics (∆D < 0.2%) showed no detectable dosimetric differences. MB-QA exhibited higher sensitivity to specific MLC defects, identifying localized fluence variations for T-nut surface wear, whereas no discernible differences were observed for drive screw thread wear or motor degradation. MB-QA gamma analysis revealed localized dose differences of up to 15% in breast IMRT and 7.4% in prostate VMAT. DVH analysis further demonstrated clinically relevant dose variations in organs at risk (OARs), including the contralateral breast (ΔDmean: 5.52%) and right lung (ΔD1: 4.50%) in breast IMRT, and the penile bulb (ΔD99: 1.55%) in prostate VMAT. Conclusion: LF-QA showed limited sensitivity to sub-millimeter MLC errors and did not capture clinically meaningful dosimetric deviations under mechanically degraded conditions. MB-QA enabled superior error detection and clinically relevant dosimetric verification. These findings indicate that LF-QA alone may be insufficient for patient-specific QA and that incorporating MB-QA is essential for ensuring reliable dosimetric verification.