Purchase this article with an account.
J.M. Shewry, P.G. Schlottmann, F.C. Ikeji, E.T. White, T.A. Ho, D.F. Garway–Heath; Comparison of the Retinal Nerve Fibre Layer Thickness in Normal Subjects and Ocular Hypertensive and Glaucoma Patients as measured by GDxVCC and OCT3. . Invest. Ophthalmol. Vis. Sci. 2004;45(13):3339.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Purpose:To evaluate agreement between measurements of the retinal nerve fibre layer thickness (RNFLT) in normal subjects and ocular hypertensive and glaucoma patients using Scanning Laser Polarimetry and Optical Coherence Tomography. Methods:54 eyes randomly selected from 54 subjects (24 normal, 19 ocular hypertensive and 11 glaucomatous) were scanned using Scanning Laser Polarimetry (GDxVCC) and Optical Coherence Tomograph (OCT 3). A glaucoma diagnosis was based on repeatable SITA 24–2 visual field defects and not on optic nerve head (ONH) appearance. Axial length and corneal radius were used to correct for ocular magnification according to the device software. RNFLT values were calculated from the default measurement annuli of 3.2mm diameter for the GDxVCC and 3.4mm diameter for the OCT3. To account for the different measurement annulus size, GDx measurements were normalized to OCT3 diameter. Bland & Altman plots were calculated to assess agreement between measurements. Linear regression analysis was performed between GDx and OCT3 RNFLT measurements for global and 90° sector values, to evaluate scaling differences around the ONH circumference. Regression slopes and 95% confidence intervals (C.I.) were recorded. Results: The mean difference and standard deviation (±) of differences between global RNFLT measurements (OCT–GDx) = 39.0 ± 9.9 µm. The mean difference increased using GDx measurements normalized to the OCT annulus diameter. The difference suggests an offset bias or scaling difference. Regression analysis between OCT and GDx RNFLT gave the equation: y = 1.5 x + 17 (R2=0.56, p<0.0001) . When regression is forced through origin, the fit was almost as good (R2=0.53 p<0.0001), giving the relationship y = 1.8 x. This suggests scaling difference (1.8*GDx = OCT). GDx measurents were rescaled and agreement re–evaluated, giving a mean difference of 0.9 ± 9.8 µm. Slopes of regresion between OCT and GDx measurements were: 1.8 for global, 1.5 for superior and inferior, and 0.9 and 1.0, respectively, for nasal and temporal sectors. 95% C.I. of slope values overlaped. Conclusions: The unmodified output of the devices show poor agreement but moderate correlation, indicating a scaling difference. Agreement following correction for scaling differences is fair. The results suggest the relationship between GDx and OCT3 RNFLT measurements may vary around the ONH circumference.
This PDF is available to Subscribers Only