Verifying a Novel Index Method of Analyzing Imaging & Radiation Coordinate Agreement with the Multimet Winston-Lutz Test
Abstract
Purpose
To verify a simplified analysis of imaging and radiation coordinate agreement across multiple geometric scenarios with thresholds that proportionally consider all independent sources of error.
Methods
A novel coordinate-agreement analysis method was developed, in which coincidence values from Sun Nuclear’s MultiMet-WL test are re-analyzed with adjusted fail criteria to account for all independent error sources. Isocenter concentricity, table walkout, table angle, collimator walkout, and gantry angle error thresholds from TG-142 are added in quadrature, where applicable, for each phantom target and delivery geometry. The relative difference between measured values and their failure threshold is averaged across all delivery geometries to yield a coordinate-agreement index. This was measured on a Varian Truebeam and Edge over 18 monthly tests. Uncertainty was characterized with 5 repeated measurements per machine. To characterize response sensitivity to each error source, measurements were taken with intentional offsets in gantry angle, table angle, and table position.
Results
The Edge monthly trend analysis demonstrated a mean index value of 0.59 (1 is perfect) and a SD of 0.07. The measurement error was 0.± 11 with 95% confidence. The three months failing TG-142 standards yielded the 1st, 4th, and 5th lowest index values at 0.46, 0.53, and 0.57 respectively. Two months that did not fail TG-142, showed significantly lower index values of 0.47.
Conclusion
This analysis method quantifies the imaging and radiation coordinate agreement, considering all relevant uncertainty components, as one de-dimensionalized value. This method is repeatable and demonstrates general agreement with standard Winston-Lutz results, but also identifies instances of poor coincidence not caught by a standard Winston Lutz analysis.