Choroidal vascularity across aging and the spectrum of age-related macular degeneration: an AI-based OCT study.

PURPOSE: To characterize the choroidal morphology across a spectrum of participants affected by normal aging as well as age-related macular degeneration (AMD) using various metrics. METHODS: Four cohorts were analyzed: healthy young (n = 42), healthy elderly (n = 19), iAMD (n = 20), and treatment-naive nAMD eyes (n = 79). Quantification of choroidal volume (CV) and choroidal vessel volume (VV) resulting in choroidal vascularity index (CVI) as well as mean choroidal thickness (CT), vessel thickness (VT) and foveal vessel area (FVA) and vessel count (FVC) were derived from spectral-domain optical coherence tomography using an artificial intelligence (AI)-based semantic segmentation model with manual validation. Cohort differences were statistically assessed. RESULTS: All choroidal parameters were reduced in healthy elderly, iAMD and nAMD eyes compared with young controls. While most of these reduced choroidal measurements were largely similar between healthy elderly, iAMD and nAMD eyes, only CVI and FVA showed a specific reduction in nAMD eyes. CONCLUSIONS: This study demonstrates a decline in several choroidal parameters with age, which can be quantified using AI-based segmentations. While age-related reductions were comparable between elderly and iAMD eyes, a significantly higher reduction of the CVI and FVA was observed in eyes with nAMD which may indicate a more pronounced alteration of choroidal structure in this condition.