SU-E-T-360: Stereotactic Radiosurgery Plan Evaluation Using High Order Central Moments




To examine the effectiveness of the coefficient of variation, skewness (third central moment), and kurtosis (fourth central moment) in quantifying and characterizing the target dose distribution in brain stereotactic radiosurgery (SRS) cases.


Twenty-one brain lesions in eighteen SRS patients treated using non-coplanar dynamic conformal arcs were randomly selected. Setup errors of these patients were extracted from clinical ExacTrac data. Retrospective plans were generated based on the ExacTrac data to simulate the effects of patient positioning errors. The coefficient of variation, skewness, and kurtosis were used to analyze the dose distribution of the planning target volume (PTV). These variables were computed from the dose-volume histogram of the PTV. For each patient, the ratios of the variables were calculated between treatment plans with and without setup errors.


The magnitude of patient setup errors ranged from 0.28 mm to 2.78 mm, with an average of 1.35 ±0.67 mm. The average ratio of coefficient of variation was 1.63 ±1.25, the ratio of the skewness was 1.40 ±0.41, and the ratio of the kurtosis was 1.26 ±0.33. These ratios were in accordance with the fact that when setup errors existed, the PTV dose distribution generally became broader, more asymmetric, and less uniform. The values of the skewness were negative, reflecting that upon setup errors part of the PTV fell into regions of lower doses. Anova analysis showed significant differences between the coefficient of variation (p=0.037), the skewness (p=0.02), and the kurtosis (p=0.01) with and without patient setup errors.


The coefficient of variation, skewness, and kurtosis were introduced to PTV dose distribution analysis. It was found that those variables were good measures of the characteristics of the PTV dose distribution. They may be useful parameters in the evaluation of treatment plan quality.