Discrimination between normal and glaucomatous eyes using Stratus optical coherence tomography in Taiwan Chinese subjects

Abstract

Background: We differentiated between normal and glaucomatous eyes in the Taiwan Chinese population based solely on the quantitative assessment of summary data reports from Stratus optical coherence tomography (OCT) by comparing their area under the receiver operating characteristic (ROC) curve.

Methods: One randomly selected eye from each of the 62 patients with early glaucomatous damage (mean deviation -2.8 +/- 1.8 dB) and from each of the 62 age- and sex-matched normal individuals were included in the study. Measurements of glaucoma variables (retinal nerve fiber layer thickness and optic nerve head analysis results) were obtained by Stratus OCT. Twenty-one OCT parameters were included in a linear discriminant analysis (LDA) using forward selection and backward elimination to determine the best combination of parameters for discriminating between glaucomatous and healthy eyes based on ROC curve area.

Results: The average RNFL thickness was the best individual parameter for differentiating between normal eyes and glaucomatous eyes (ROC curve area 0.793). The maximum area under the ROC curve of six input parameters (average RNFL thickness; 10, 11, and 12 o'clock segment thicknesses; cup area; and vertical integrated rim area) generated by the forward selection method was 0.881. Whereas the maximum area under the ROC curve of 15 input parameters (average RNFL thickness; 1, 3, 4, 6, 8-10, 12 o'clock segment thicknesses; vertical integrated rim area; horizontal integrated rim area; disc area; cup to disc area ratio; cup to disc horizontal ratio; and cup to disc vertical ratio) generated by backward elimination method was 0.929.

Conclusions: The performance of individual parameters obtained from Stratus OCT is fairly reliable for differentiating the early glaucomatous eyes from normal eyes. However, the discriminant power increases when LDA with forward selection and backward elimination methods is applied.