Comparing ophthalmoscopy, slide viewing, and semiautomated systems in optic disc morphometry.

Purpose: To compare disc measurements obtained by indirect ophthalmoscopy, the Heidelberg Retina Tomograph (HRT), stereoscopic slide viewing (SSV) of color transparencies, and the Topcon ImageNet System (ImageNet).

Design: Population-based cross-sectional study.

Methods: From the Rotterdam Study, 324 subjects (567 eyes) were nonselectively included. All underwent a full ophthalmologic examination in mydriasis. Vertical cup/disc ratios (VCDRs) were compared between all four methods and disc area (mm(2)), neural rim area (mm(2)), cup area (mm(2)), and cup volume (mm(3)) between HRT and ImageNet.

Results: Mean VCDR for ophthalmoscopy was 0.25 (standard error [SE], 0.007), for HRT 0.42 (SE, 0.008), for SSV 0.39 (SE, 0.010), and for ImageNet 0.50 (SE, 0.006). The correlation for VCDR between ophthalmoscopy, the two devices, and SSV was 0.42, respectively 0.57; between ImageNet and HRT 0.75. The 97.5th percentiles of the VCDR for ophthalmoscopy, HRT, SSV, and ImageNet were 0.80, 0.73, 0.80, and 0.73, respectively; the 99.5th percentiles thus were 0.90, 0.79, 0.86, and 0.79. The mean disc area, rim area, cup area, and cup volume were 2.08, 1.63, 0.45 mm(2) and 0.09 mm(3) for HRT, and 2.39, 1.77, 0.61 mm(2) and 0.16 mm(3) for ImageNet, respectively. The corresponding correlations for these four parameters were 0.67, 0.42, 0.81, and 0.82.

Conclusions: Different techniques lead to considerable differences in disc morphometric values. ImageNet produced higher mean values compared with HRT and ophthalmoscopy. Ophthalmoscopy showed the lowest correlations and SSV the highest ones with the two semiautomated devices. Between ImageNet and HRT the correlation for all parameters was high except for the neural rim area.