Skip to main content

An EPID-based virtual audit framework for multi-institutional radiotherapy audits using direct fluence comparison

Journal
Physics and imaging in radiation oncology (Q1)
Published
7 November 2026
Study design
Unclassified
Evidence level
Level 5, Expert Opinion (CEBM 5)
Authors
Maiko Kishigami, Takanori Adachi, Yoshitomo Ishihara, Hiroyuki Ito, Tomohiro Ono, Yukako Kishigami, et al.
PMID
43060554
DOI
10.1016/j.phro.2026.101124

Why clinicians should know about it

  • Picked for Medical Physics (top studies of the week, 22 November 2026): EPID-based VMAT audit framework with measured error-detection performance

Abstract

BACKGROUND AND PURPOSE: The complexity of volumetric modulated arc therapy highlights the need for robust quality assurance in multi-institutional clinical trials. Although electronic portal imaging device-based virtual audits are scalable, standardized workflows and evaluation criteria remain lacking. This study developed and evaluated a virtual audit framework comparing radiotherapy plan-derived and electronic portal imaging device-derived fluence maps. MATERIAL AND METHODS: Forty-one volumetric modulated arc therapy datasets from 15 institutions were analyzed. Plan-derived fluence maps were generated by convolving opening density matrices with a double-Gaussian electronic portal imaging device response kernel, while integrated portal images served as detector-derived fluence maps. Vendor-specific common parameters were evaluated using leave-one-machine-out validation, excluding each held-out machine's parameters before recalculating vendor-level parameters. Multileaf collimator positional, collimator-angle, and monitor-unit errors were evaluated using receiver operating characteristic analysis. RESULTS: Mean gamma passing-rate differences ranged from -0.015 to 0.018 percentage points for Varian systems and from 0.387 to 1.018 percentage points for Elekta systems. The 3%/1 mm criterion yielded areas under the receiver operating characteristic curves of 0.959, 0.938, and 0.898 for multileaf collimator positional, collimator-angle, and monitor-unit errors, respectively, and produced the highest values for multileaf collimator positional and collimator-angle errors. A 95% gamma passing rate threshold at 3%/1 mm provided high sensitivity, whereas 90% at 3%/2 mm provided high specificity. CONCLUSIONS: Leave-one-machine-out validation supported the feasibility of vendor-level parameter standardization, potentially reducing machine-specific optimization. A preliminary tiered framework combining 95% at 3%/1 mm and 90% at 3%/2 mm may identify cases requiring review in multi-institutional radiotherapy audits.

Abstract as published, via PubMed.

View on PubMedFull text at the publisherOpen in the app

For healthcare professionals. The summary is generated by AI from the published abstract, and the evidence level is assigned automatically from the study design on the Oxford CEBM hierarchy. Neither is medical advice. Read the full paper before changing practice.