Sequence of Morphological Changes Preceding Atrophy in Intermediate AMD Using Deep Learning

Abstract

Purpose: Investigating the sequence of morphological changes preceding outer plexiform layer (OPL) subsidence, a marker preceding geographic atrophy, in intermediate AMD (iAMD) using high-precision artificial intelligence (AI) quantifications on optical coherence tomography imaging.

Methods: In this longitudinal observational study, individuals with bilateral iAMD participating in a multicenter clinical trial were screened for OPL subsidence and RPE and outer retinal atrophy. OPL subsidence was segmented on an A-scan basis in optical coherence tomography volumes, obtained 6-monthly with 36 months follow-up. AI-based quantification of photoreceptor (PR) and outer nuclear layer (ONL) thickness, drusen height and choroidal hypertransmission (HT) was performed. Changes were compared between topographic areas of OPL subsidence (AS), drusen (AD), and reference (AR).

Results: Of 280 eyes of 140 individuals, OPL subsidence occurred in 53 eyes from 43 individuals. Thirty-six eyes developed RPE and outer retinal atrophy subsequently. In the cohort of 53 eyes showing OPL subsidence, PR and ONL thicknesses were significantly decreased in AS compared with AD and AR 12 and 18 months before OPL subsidence occurred, respectively (PR: 20 µm vs. 23 µm and 27 µm [P < 0.009]; ONL, 84 µm vs. 94 µm and 98 µm [P < 0.008]). Accelerated thinning of PR (0.6 µm/month; P < 0.001) and ONL (0.8 µm/month; P < 0.001) was observed in AS compared with AD and AR. Concomitant drusen regression and hypertransmission increase at the occurrence of OPL subsidence underline the atrophic progress in areas affected by OPL subsidence.

Conclusions: PR and ONL thinning are early subclinical features associated with subsequent OPL subsidence, an indicator of progression toward geographic atrophy. AI algorithms are able to predict and quantify morphological precursors of iAMD conversion and allow personalized risk stratification.

Figure 1.

Three example cases of progressive morphological changes and OPL subsidence occurring during the 36-month follow-up period. Six-monthly representative B-scans of the OCT volumes (A, C, E) and corresponding colored overlays (B, D, F), indicating different topographic areas, defined at the time point of OPL subsidence occurrence: OPL subsidence area (A_S, blue); drusen area (A_D, yellow) and reference area (A_R, green). The occurrence of OPL subsidence (vertical red arrow) precedes topographically corresponding drusen regression (yellow asterisk) and various degrees of choroidal HT (pink arrowheads), indicative of progressive RORA. Thinned ONL and PR in A_S compared with A_D is evident in visits leading up to OPL subsidence occurrence.

Figure 2.

En face topographic distribution of OPL subsidence in the RORA subgroup. En face map showing the cumulative topographic distribution of OPL subsidence lesions over all patients. Pseudo heat map colors correspond to the percentage of eyes affected by a lesion at a specific location. Concentric rings indicate the fovea-centered 3 and 6 mm diameters.

Figure 3.

The course of OCT feature development during progression of iAMD. Changes over time in PR thickness, ONL thickness, drusen height and intensity of choroidal HT in the OPLS (dashed line), OPLS + RORA (solid line) and NOPLS + NRORA (alternating dashed/solid line) subgroups. Number of eyes included in the respective groups is indicated by n. Differentiated topographic zones are compared, as each time course is plotted with respect to mean values within OPL subsidence area (A_S, red), drusen area (A_D, yellow) and reference area (A_R, blue). The time axes of the plots are relative to the time point of OPL subsidence occurrence (in the OPLS and OPLS + RORA subgroup) or the last available visit (in the NOPLS + NRORA subgroup), denoted by the dotted vertical line at month 0, respectively. Bars indicate 95% CIs.