Abstract
Abstract: :
Purpose: To compare the diagnostic utility of RNFL measures obtained with the GDx scanning laser polarimeter with variable corneal compensation (GDx VCC). Methods: Three hundred ninety normal eyes and 253 glaucomatous eyes were imaged using the GDx VCC. Data was collected from 5 eye clinics in the US (SUNY College of Optometry, New Jersey College of Medicine and Dentistry, New York Eye and Ear Infirmary, Eye Care of San Diego, and La Mesa Vision Care Optometry). All eyes underwent complete ophthalmic exams to verify their health status. Glaucoma was defined as the presence of optic disc abnormality, and the patient breakdown by field severity was: 86 patients with MD > -2, 68 patients with MD between -2 and –5, 56 patients with MD between –5 and –10, and 43 patients with MD less than –10. Receiver operating characteristic (ROC) curves were generated for RNFL parameters (NFI, TSNIT average, TSNIT standard deviation, TSNIT superior average, and TSNIT inferior average). Correlations were also calculated with the glaucoma patients comparing the output parameters with mean defect (MD) from standard automated perimetry (SAP). Results: The area under the ROC curve was .90 +/- .01 for the Nerve Fiber Indicator (NFI) parameter, which was derived using a support vector-machine (SVM) algorithm. The area under the ROC curve for the other parameters was: .80 +/- .02 for the TSNIT average; .84 +/- .016 for the TSNIT standard deviation; .85 +/- .016 for the TSNIT superior average; .79 +/- .02 for the TSNIT inferior average. Significant correlations were found between MD and all parameters (NFI, r = -.52; TSNIT avg, r = .38; TSNIT standard deviation, r = .53; TSNIT superior average, r = .48; TSNIT inferior average, r = .41), where p < .001 in all comparisons. Conclusions: The GDx VCC can effectively discriminate between healthy eyes and glaucomatous eyes. The significant association between the parameters and visual field MD values suggest that the RNFL defects measured in the patients strongly reflect disease severity.
Keywords: imaging/image analysis: clinical • imaging/image analysis: clinical • nerve fiber layer