Prediction of VMAT delivery accuracy using plan modulation complexity score and log-files analysis.

Area of Science:

• Medical Physics
• Radiation Oncology
• Radiotherapy Planning

Background:

• Volumetric Modulated Arc Therapy (VMAT) is a complex radiotherapy technique.
• Ensuring dosimetric accuracy in VMAT plans is critical for patient safety and treatment efficacy.
• Current quality assurance (QA) methods can be time-consuming and may not always identify all suboptimal plans.

Purpose of the Study:

• To develop a predictive model for classifying the dosimetric accuracy of VMAT plans.
• To utilize plan complexity metrics and linac log-file analysis for this classification.
• To identify overly modulated VMAT plans that may fail QA and require re-optimization.

Main Methods:

• Analysis of 612 VMAT plans (1224 arcs) using pre-treatment verification data from linac dynamic log-files.
• Comparison of predicted and measured integral fluences using gamma analysis (2%/2 mm criteria).
• Development of a logistic regression model incorporating the Modulation Complexity Score (MCS) to predict QA outcomes ('pass' or 'fail').

Main Results:

• The developed model achieved high accuracy (97.4% training, 98.0% testing).
• An optimal MCS threshold of 0.142 was identified to flag plans failing QA with 91% true positive rate.
• Confidence and action limits for gamma analysis (γ%) were established at 20.1% and 79.9%.

Conclusions:

• The predictive model effectively classifies VMAT plan dosimetric accuracy.
• The MCS threshold allows for prompt identification of overly modulated and potentially failing plans.
• Implementing this model can reduce QA failures and enhance the overall quality of VMAT treatment plans.