Quantitative patient‐specific quality assurance prediction using MLC mean leaf gap and PTV volume

TL;DR

This paper introduces a method to predict the quality of radiation therapy plans using a simple metric, helping to identify complex cases that need extra checks.

Contribution

A new method using MLC mean leaf gap to predict QA outcomes during treatment planning for SBRT-VMAT.

Findings

MLG shows a significant positive correlation with gamma pass rates (R² = 0.39, p < 0.001).

For C-arm LINAC plans, MLG < 2.02 cm predicts QA failure with 87% sensitivity and 74% specificity.

For ring gantry plans, MLG < 2.13 cm predicts QA failure with 100% sensitivity and 60% specificity.

Abstract

Rigorous patient‐specific quality assurance (PSQA) is essential to radiation therapy safety. As logfile‐based PSQA gains adoption, quantitative methods to select certain high‐complexity treatment plans for additional measurement‐based PSQA can help to ensure a comprehensive QA program. We propose a simple metric to predict PSQA measurement results for stereotactic body radiation therapy (SBRT)–volumetric modulated arc therapy (VMAT) plans based on historical QA results, integrated into the treatment planning process for the early detection of potential issues. This method can be used to screen treatment plans for measurement‐based QA, saving time while maintaining safety standards. We identified 46 SBRT–VMAT plans for C‐arm LINACs and 18 plans for a ring gantry LINAC that underwent PSQA with gamma analysis thresholds of 3%, 2 mm in our clinic. We developed a script to compute the MLC mean leaf gap (MLG) width as a proxy for treatment plan deliverability, which can be run during the planning process. We analyzed the correlation of MLG with gamma pass rates and performed a receiver operating characteristic (ROC) analysis of its performance as a binary predictor of QA measurement results. We found a significant positive correlation between gamma pass rate and overall MLG (R
2 = 0.39, p < 0.001) for SBRT–VMAT plans. A criterion of MLG < 2.02 cm predicts PSQA failure with a sensitivity of 0.87 and a specificity of 0.74 for C‐arm LINAC plans. For ring gantry plans, MLG < 2.13 cm showed a sensitivity of 1.00 and specificity of 0.60. A simple descriptive metric based on historical QA results can be used to screen treatment plans that might need further analysis with logfile‐ or measurement‐based PSQA.

Figures