Correlation in retinal nerve fiber layer thickness between two OCT units

Abstract

Purpose: It is expected that spectral-domain optical coherence tomography (SD-OCT) will replace time-domain (TD)-OCT in the near future. In this study, we applied a robust set of statistical analyses to evaluate the correlation between retinal nerve fiber layer (RNFL) thickness measurements obtained by SD-OCT and those obtained by TD-OCT in a Chinese population with different stages of glaucoma.

Methods: A total of 40 patients with primary open-angle glaucoma, 31 with primary angle-closure glaucoma (PACG), 31 with suspected glaucoma (GS), 25 with ocular hypertension (OH), and 52 normal subjects were enrolled in this prospective, cross-sectional study. Eyes of all participants were imaged by a single trained operator using the Stratus TD-OCT (fast RNFL scan mode) system and the Cirrus SD-OCT (optic disc cube mode) system during the same visit. The correlations between RNFL thickness measurements obtained from the two OCT instruments were assessed using Pearson correlation analysis and generalized estimating equation (GEE), mixed effect, linear regression models, after adjusting for the effect of individual variation.

Results: There was good correlation in RNFL thickness parameters between the two OCT devices, particularly in average RNFL thickness (Pearson Correlation, 0.8798 vs. GEE mixed model, 0.9470). The proposed GEE mixed model method showed better correlation than the Pearson correlation analysis in each RNFL thickness parameter between the two OCT measurements.

Conclusions: Although the RNFL thickness measurements obtained by the TD-OCT system correlated well with those obtained by the SD-OCT system, clinicians should be cautious when interpreting RNFL thickness data for any subject undergoing longitudinal follow-up with different OCTs.