Effect of Human Perception On Winston Lutz QA Process
A Moghadam*, R Hamilton, C Watchman, University of Arizona, Tucson, AZMO-B-Salon EF-10 Monday 10:00:00 AM - 12:00:00 PM Room: Salon EF
Purpose: This study demonstrates the effect of human bias in performing the QA tests and specifically quantifies the amount of uncertainty due to observer bias for the case of Winston Lutz (WL) test for Stereotactic Radio Surgery (SRS)
Methods:Twenty four WL tests for previously treated patients were used in this analysis. Each test was analyzed using a Matlab program which required the user to place a series of circles around the BB and cone regions. The positional differences were then determined by the program. A second program was developed to analyze the displacement of the circle centers automatically. The automated program uses Center of Mass calculation along with an optimization algorithm to find the outer irradiation circle and BB shadow circle. The optimization process involves finding the circle which maximizes the summation of pixel values enclosed by the circle for the outer irradiation circle. The BB circle is found by a minimum summation of pixel values. The difference in the measured deviation between the two methods of measurement was then evaluated.
Results:The average displacement measured by the manual process was found to always be smaller than the amount measured by the automated program. Our analysis also showed that 17% of the cases examined would have failed the WL criterion had the automated method been used.
Conclusion:Observer bias in WL analysis is a significant factor in the QA process. This analysis indicated that nearly 1/6 of observer determined cases would fail as compared to the objective method. Also, it appears that the observer will innately conclude the deviation to be smaller than the automated method describe.