Adaptive Optics-enhanced dual retinal layer imaging and region reflectivity-based segmentation: unveiling variances in healthy and diseased retina

Abstract

Purpose : Adaptive Optics Transscleral Flood Illumination (AO-TFI) is a novel technology for detecting in vivo retinal pigment epithelium (RPE) cells, recently combined with transpupillary illumination to observe photoreceptors (PR) simultaneously. Standard AO-metrics using mosaic (Voronoi) segmentation lack precision in diseased retina. We hypothesized that a segmentation method based on hypo-/hyper-reflective zones could differentiate healthy and diseased profiles.

Methods : We conducted a cross-sectional study with 14 participants (6 healthy, 8 dry age-related macular degeneration (AMD) who had macular 6.7°x6.7° images acquired with an eccentricity between 2° and 8.6° using Cellularis Discovery AO-TFI camera (EarlySight SA, Geneva, Switzerland). Morphometric characteristics were extracted using 2 methods: mosaic and region reflectivity-based segmentation (hyper-reflectivity in PR and hypo-reflectivity in RPE). Descriptive statistics and a Mann Whitney test were used to compare PR and RPE characteristics between healthy and diseased groups.

Results : 42 images (16 healthy, 26 dry AMD) were selected for analysis. Post-mosaic segmentation, healthy subject mean PR density was 9868±964cell/mm², RPE density 4112±1240cell/mm². Neighboring cells were approximately 6 for both PR (5.98±0.01) and RPE (5.97±0.02), as expected. Using region reflectivity-based segmentation, in the diseased group, mean region density was lower (646±458 regions/mm² PR, 796±400 regions/mm² RPE) than in healthy subjects ((8299±2123 regions/mm² PR, 2750±728 regions/mm² RPE), p<0.0001 in both cases), consistent with cell loss. Diseased group had larger mean region areas (81.1±51.7μm² vs 9.2±1.5μm², p<0.0001 PR; 69.0±39.5μm² vs 28.2±3.5μm², p<0.0001 RPE), indicating structural alterations.

Conclusions : Our findings exhibit comparability of healthy retina metrics using the newly released AO-TFI device to those obtained at similar eccentricities using other AO-based technologies such as AO optical coherence tomography (AO-OCT) or AO scanning laser ophthalmoscopy (AO-SLO). The novel segmentation method, superior to Voronoi in mapping diseased profiles, enables fair comparison between healthy and diseased groups. Future research should longitudinally investigate dynamic changes and correlate hypo-/hyper-reflective zones with ultra-high-resolution OCT (UHR-OCT) to determine their origin.

This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.