Abstract Purpose The retinal pigment epithelium (RPE) is critical in the pathophysiology of retinal diseases, such as age‐related macular degeneration. Adaptive optics transscleral flood illumination (AO‐TFI) offers rapid, detailed morphometric characterization of the RPE layer. This study evaluated AO‐TFI's efficacy and feasibility to detect clinically relevant morphological characteristics in a clinical setting. Methods A total of 230 participants, categorized by their central retinal health, underwent comprehensive examination including SD‐OCT, fundus imaging and AO‐TFI using the Cellularis® prototype 2.0. Image quality control and RPE layer quantification were performed with an AO‐TFI‐specific automated segmentation algorithm. Density and area of hyper‐ and hyporeflective regions in the RPE layer plane, and, if detectable, RPE cell density, were quantified. We hypothesized that the RPE cell density would be lower in diseased retinas than in healthy retinas. Imaging results of healthy participants were statistically compared to those of diseased eyes. Additionally, generalized linear and logistic regression mixed‐effect models identified associations between ocular characteristics and imaging parameters. Results After quality evaluation, high‐quality images from 200 subjects (87%) were selected and segmented. The number of hyperreflective regions and their mean surface area were significantly higher in diseased than in healthy eyes (68 ± 40/mm 2 vs. 51 ± 39/mm 2 ; p < 0.001; 302 ± 196 μm 2 vs. 155 ± 55 μm 2 ; p < 0.001). The RPE distinct mosaic pattern was more often visible in healthy retinas ( n = 103 vs. n = 30, p < 0.0001). The mean RPE cell density was 6354 ± 695/mm 2 , with comparable counts for healthy and diseased, 6327 ± 687/mm 2 vs. 6532 ± 725/mm 2 ( p = 0.1). Conclusion AO‐TFI detected differences between healthy and diseased eyes, indicating its potential as a promising clinical modality providing quantitative and qualitative insights into RPE layer dynamics.
Eppenberger et al. (Tue,) studied this question.