Retinal Nerve Fiber Layer Measurement Reproducibility with Varying Sampling Patterns and Locations in Rhesus Macaque Eyes.

Abstract

Purpose : To determine the reproducibility of retinal nerve fiber layer (RNFL) measurements using circular and elliptical sampling methods on optical coherence tomography (OCT) images acquired in rhesus macaques.

Methods : Repeated in vivo three-dimensional spectral-domain OCT scans (Leica, Chicago, IL) were obtained from 15 healthy adult rhesus macaques (Macaca mulatta; 8 males, 7 females; age 4.2-10.3 years). The scans were acquired as raster scans (400 x 400 x 1024 pixels) in a 5mm x 5mm standardized region centered on the optic nerve head (ONH). 17 eyes (34 scans) were analyzed using OCT segmentation software of our own design to calculate point-by-point RNFL thickness measurements. Concentric 1-pixel radius measurements, centered on the geometric ONH center and starting at the disc margin, were calculated for the following (Figure 1, B-F; respectively): (1) circle, (2) ellipse with a fixed aspect ratio of 1:1.33, (3) eye-specific disc aspect ratio ellipse, (4) fixed 1:1.33 ellipse rotated along eye-specific disc angle, (5) eye-specific disc aspect ratio ellipse rotated along disc angle. Two distances of interest were specifically compared at (a) previously reported preferred sampling location in rhesus eyes which maximizes the dynamic range of measurements and (b) human default location adjusted for the animal’s eye size. RNFL measurement reproducibility thickness globally and in quadrants was calculated using within-subject coefficient of variation (WSCoV).

Results : WSCoV ranged between 0.024-0.180 for global RNFL thickness across all sampling method and radii measures. Overall, all sampling methods had similar reproducibility per radius, with the least reproducible measurements occurring near the disc margins (Figure 2). For both sampling locations (macaque and default human), there was no difference noted between sampling patterns (circular or elliptical, adjusted or non-adjusted for disc size or rotation). Across quadrants, the inferior quadrant had the best reproducibility in these locations.

Conclusions : Despite adjusting for the shape of the primate ONH, both circular and elliptical sampling methods of global and quadrant RNFL offer robust reproducibility (< 5%) regardless of sampling location. Therefore, the conventional and more practical circular method can reliably be used for repeated non-human primate RNFL measurements.

This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.