Evaluating and optimizing test strategies in automated perimetry

This article summarizes some previous and ongoing research on the optimization of test strategies in automated perimetry for both the detection and the evaluation of glaucomatous visual field damage. A variety of methods, ranging from a large screening program to computer simulation, were employed to investigate the performance of different test strategies. We detail methods to optimize the number and distribution of test locations for visual field screening and show that effective screening programs can be designed with as few as 20 test locations. Computer-simulation experiments in threshold perimetry were done using empirical models for both the normal and glaucomatous visual field. We evaluated the test performance of the 30-2 program of the Humphrey Field Analyzer in estimating the thresholds in 50 normal and 50 glaucomatous visual fields. We also studied the interaction between various thresholding strategies and the number of test locations in the visual field to decide how clinical visual field information can be obtained most accurately and efficiently.