Discrimination of glaucomatous optic neuropathy by digital stereoscopic analysis

Abstract

Purpose: To evaluate the diagnostic power of a novel digital stereoscopic imaging system in the diagnosis of glaucomatous optic neuropathy.

Design: Prospective cross-sectional analysis of the diagnostic accuracy of digital stereoscopic optic disc analysis in the diagnosis of glaucomatous optic neuropathy exhibiting mild to moderate field loss.

Participants: Fifty-two patients with open-angle glaucoma and 54 normal individuals were recruited. The presence of a reproducible visual field loss characteristic of glaucoma was used as the reference standard for the presence of glaucoma independent of the optic nerve head appearance. Patients were excluded if the optic disc, fundus, or visual field indicated other disease. One eye from each patient and individual was included in the study, the eye with the least field loss and a randomly designated normal eye, respectively.

Methods: Simultaneous stereoscopic optic disc photography was performed on each specified eye. Three experienced observers viewed the resultant stereoscopic image of each nerve head using a Z screen, recorded a subjective clinical diagnosis, and undertook digital stereoscopic planimetry. Separate linear regression analysis was performed, post hoc, from the planimetric results for each observer of the logarithm of neuroretinal rim (NRR) against optic disc area derived from each normal eye. Eyes with NRR areas below the 95th prediction interval of the normal cohort were then classified as glaucomatous.

Main outcome measures: Sensitivity and specificity for the detection of glaucomatous optic neuropathy.

Results: With subjective stereoscopic analysis, sensitivity for glaucoma detection among the 3 observers was 80.8%, 76.9%, and 90.4%, with respective specificities of 94.4%, 79.6%, and 79.6%. Regression analysis of the NRR in 30 degrees segments gave sensitivities between 69.2% and 80.8% and specificities between 83.3% and 90.7%. A combination of the subjective and quantitative analysis did not significantly improve discrimination.

Conclusions: The subjective analysis of digital stereoscopic images provides a useful method for the discrimination of normal and glaucomatous optic nerves. Planimetric analysis does not significantly improve the diagnostic precision of this technique.