Purchase this article with an account.
M. L. Gabriele, H. Ishikawa, R. A. Bilonick, G. Wollstein, L. E. Kagemann, J. G. Fujimoto, J. S. Schuman; A Method for Reducing Optical Coherence Tomography (OCT) Retinal Nerve Fiber Layer (RNFL)Thickness Normative Data Variability. Invest. Ophthalmol. Vis. Sci. 2009;50(13):2258.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Superior and inferior RNFL thickness profile peak locations vary across individuals and may partially explain the wide variability observed in OCT RNFL thickness normative databases. Our goal was to investigate the effect of normalizing superior and inferior RNFL peak locations on the variability of normative RNFL data.
A 6.0x6.0x2.0 mm volume centered on the optic nerve head (ONH) was acquired from healthy subjects using spectral-domain OCT (Cirrus HD-OCT; Carl Zeiss Meditec, Inc., Dublin, CA) with RNFL thickness measured at each point using custom software. Circumpapillary RNFL thickness was measured under two conditions: 1.Baseline (B): resampling the data volume along a 3.4mm diameter circle centered on the ONH (256 samplings) similar to the conventional commercially available analysis; 2. Peak adjustment (PA): normalizing the location of the major RNFL thickness profile peaks (superior and inferior) of baseline data to the population average locations. Low pass filtering of each RNFL thickness trace facilitated detection of superior and inferior RNFL peak index locations. Subject RNFL thickness profiles were then normalized to the average location of the corresponding major peaks for the PA. A variance analysis test for dependent data was used to investigate the differences in variance of RNFL thicknesses at each sampling point and across subjects for (B - PA).
Forty-six right eyes from 46 healthy subjects (mean age 46.8±17.9 years) were included. The difference in variance can be seen in the Figure, with significant differences indicated by squares. 194 locations (75.8%) showed smaller variance after PA, and the differences were statistically significant in 46 out of 194 locations (23.5%).
Peak location normalization may reduce variance within the SD-OCT RNFL thickness normative data.
This PDF is available to Subscribers Only