Author: Parham Alaei, Timothy J Allen, Tomas Agustin Montenegro 👨🔬
Affiliation: University of Minnesota, Department of Radiation Oncology, University of Minnesota, Minneapolis 🌍
Purpose:
To assess possible relationships between treatment plan optimization parameters and gamma pass rates in quality assurance (QA) analysis of IMRT/VMAT plans.
Methods:
Plan optimization parameters were extracted from Pinnacle v18.0 and compiled alongside plan characteristics and 3%/2 mm gamma analysis pass rates. Parameters of interest were identified, and multilinear regression tests were conducted to determine correlation with gamma pass rates. Gamma analysis followed TG-218 recommendations.
Results:
Visual analysis of the scatter and histogram plots suggested that the modulation factor, field size, and minimum MLC segment area may be correlated to gamma pass rates in VMAT planning. The modulation factor and minimum segment area proved to be weak (R2=0.203), yet significant (p-value = 0.003) predictors of gamma pass rate. As the field blocking and minimum area allowed to be formed by MLCs in a segment decreased, the gamma pass rate decreased.
Analysis of IMRT plans proved more difficult, as several parameters constraining leaf motion and dose to normal tissue were predictors of gamma pass rate but were also strongly correlated to the modulation factor. However, these parameters were typically modified in response to already challenging planning cases. The modulation factor and field size were determined to be the only independent variables and were significant (p=0.0003) but weak (R2=0.328) predictors of the gamma pass rate.
Conclusion:
The plan modulation factor was the most significant predictor of gamma pass rates. In VMAT planning, reducing the minimum segment area reduced pass rates by increasing allowed MLC complexity. Analysis of several IMRT planning parameters was confounded as values were only changed in response to complex situations, even in scenarios in which the parameter change would have constrained the treatment plan further. A more controlled analysis of optimization parameters would be required to draw conclusions on these parameters.