Abstract
Purpose:
The retinal pigment epithelium (RPE) is critical for support and maintenance of photoreceptors. While dysfunction of the RPE underlies numerous retinal pathologies, biomarkers sensitive to early changes in RPE have been elusive. Because such changes start at cellular level, there has been increased interest in targeting the spatial arrangement and distribution of individual RPE cells. To do so in the living human retina is extremely challenging, owing to the lack of intrinsic contrast of RPE, optical waveguiding by the overlying photoreceptors, and blurring by ocular aberrations. In this study, we take advantage of the micron-level 3D resolution afforded by adaptive optics and optical coherence tomography (AO-OCT) to overcome these obstacles in order to visualize RPE cells and investigate their packing geometry.
Methods:
Using the Indiana AO-OCT imaging system (λc=790 nm, Δλ=42 nm), volumes of 1°×1° field of view were acquired at 3° and 10° temporal retina in two normal subjects. Volumes were registered, segmented, and RPE en face images extracted. Voronoi analysis was applied to the en face images to determine number of neighbors (NN) and center-to-center nearest neighbor distance (NND) of the RPE cells. 2D power spectra were used to provide additional information about cell spacing.
Results:
RPE cell mosaics were resolved in both subjects and retinal eccentricities. Voronoi analysis indicates hexagonal cells (with six NN) are most frequent (>50%) at 3° retinal eccentricity and are of lower frequency (<50%) at 10° retinal eccentricity. NND was 11.4±2.2 μm and 12.8±3.0 μm for subject 1 at 3° and 10° retinal eccentricities respectively, and 12.0±2.0 μm for subject 2 at 3°. Processing of 10° data for subject 2 is ongoing. NND measurements are consistent with the 2D power spectra estimations of 11.9 μm and 12.9 μm of subject 1 (3° and 10°), and 12.7 μm of subject 2 (3°).
Conclusions:
AO-OCT imaging permits visualization and quantification of the RPE packing geometry in the living human retina.