Detection of Glaucoma Using Operator–Dependent Versus Operator–Independant Classification in the HRT–III

Abstract

Purpose: : To compare the abilities of a new Glaucoma Probability Scoring system (GPS) and Moorfields Regression Analysis (MRA) to differentiate glaucomatous from normal eyes using HRT–III software and race–specific databases.

Methods: : In this prospective study, one eye (refractive error ≤ 5D) of consecutive normal and glaucoma patients were enrolled. All subjects underwent a full eye examination, standard achromatic perimetry (SITA–SAP, 24–2) and confocal scanning laser ophthalmoscopy (Heidelberg Retinal Tomograph, HRT–II) within one month. Glaucoma was defined based on SAP visual field (VF) loss (PSD<5% and/or GHT outside normal limits) on two consecutive VFs. Normal subjects had two normal VFs (PSD>5% and GHT within 99% normal limits) and a normal clinical examination. HRT–II exams were exported to the HRT–III software (version 3.0). HRT–III uses an enlarged race–specific database, consisting of 733 white and 215 black eyes. GPS is a new automated algorithm that takes into account optic disc topography and surrounding parapapillary retinal nerve fiber layer measurements to give a percentage for probability of damage consistent with glaucoma without contour line placement (operator independent). Race–adjusted MRA for the most abnormal sector (operator dependant contour line placement) was compared to the global race–adjusted GPS scores. MRA sectors outside the 99.9% confidence interval limits ("outside normal limits") and GPS>64% were considered abnormal.

Results: : 140 normal and 84 glaucoma patients (98 white, 126 black) were enrolled (mean age, 50 ± 14.5 years). The average VF mean deviation was –7.3 ± 6.7 db for the glaucoma group and –0.4 ± 1.1 db for the normals (p<.001). Mean GPS scores were 0.22 and 0.73 for normals and glaucoma respectively (p<.001). Sensitivity and specificity for GPS was 75.9% and 89.1% and for MRA was 71.4% and 91.4%.

Conclusions: : In this cohort, GPS software increased sensitivity at the expense of specificity compared to MRA. The development of software to detect glaucoma without a contour line is critical to improve the potential use of the HRT as a tool for glaucoma screening.

Keywords: optic disc • imaging/image analysis: clinical • clinical (human) or epidemiologic studies: systems/equipment/techniques