Artificial Intelligence Quantified Outer Retinal Disruption On Optical Coherence Tomography As Clinical Endpoints For Geographic Atrophy

Abstract

Purpose : Manual grading of outer retinal disruption (ORD) on Optical Coherence Tomography (OCT) volumes is labor-intensive and impractical for large studies. Fundus Autofluorescence (FAF) has been alternatively used as a clinical endpoint in Geographic Atrophy (GA) trials. This study introduces an Artificial Intelligence (AI) powered OCT segmentation model to automatically label the extent of ORD, and compare with manually graded GA areas on FAF.

Methods : A retinal layer segmentation model, EyeNotate, with DeepLabv3+ architecture, was trained on 6718 annotated OCT B-scans from participants involved in intermediate AMD and GA trials (NCT01790802 and NCT02399072; n=189). The model was applied to 736 OCT volumes from study eyes of a separate GA study (NCT02479386; n=227, 1-7 visits per patient. Example B-scan in Fig. 1). ORD per volume (i.e. external limiting membrane (ELM) loss, ellipsoid zone (EZ) loss, and retinal pigment epithelium (RPE) loss) were derived from en face layer thickness maps, identified as a layer thickness of zero (Fig. 2). Correlations were assessed between OCT-defined retinal layer loss and manually graded FAF-defined GA area at the baseline visit. Correlations between annualized OCT-defined retinal layer loss rates and FAF-defined GA growth rates were calculated for 205 patients with at least two visits.

Results : Baseline GA area defined by FAF, and RPE loss, ELM loss, and EZ loss on OCT were assessed (mean±SD) as 7.8±3.9, 6.0±3.3, 8.5±4.2, and 12.4±4.2 mm2 respectively. EZ loss was significantly larger than both ELM and RPE loss (p<0.01). There was a strong correlation between OCT-defined retinal layer loss and FAF-defined GA area, with correlation coefficients of 0.91 for RPE loss, 0.92 for ELM loss, and 0.76 for EZ loss (all p<0.01). When evaluating the annualized GA lesion growth rate as defined by OCT and FAF, the correlation remained significant: 0.77 for RPE loss rate, 0.78 for ELM loss rate, and 0.46 for EZ loss rate (all p<0.01). 91% of regions that developed new RPE loss within a year had EZ loss at baseline.

Conclusions : AI-segmented RPE and ELM loss on OCT strongly correlated with FAF-defined GA area, both cross-sectionally and longitudinally. Hence these OCT layers may serve as potential clinical endpoints in GA studies.

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

 

Fig. 1. Example OCT B-scan and model output.

 

Fig. 2. Example FAF and OCT en face layer thickness maps.