Search

To develop an automated image recognition software for the objective quantification of choroidal vascularity index (CVI) and choroidal thickness (CT) at different choroidal locations on images obtained from enhanced depth imaging optical coherence tomography (EDI-OCT), and to validate its reliability and investigate the difference and correlation between measurements made by manual and software.A total of 390 EDI-OCT scans, captured from 130 eligible emmetropic or myopic subjects, were categorized into four grades in terms of their accessibility to identify the choroidal-scleral interface (CSI) and were further assessed for CT and CVI at five locations (subfoveal, nasal, temporal, superior and inferior) by the newly developed Choroidal Vascularity Index Software (CVIS) and three ophthalmologists. Choroidal parameters acquired from CVIS were evaluated for its reliability and correlation with ocular factors, in comparison to manual measurements. Distribution of difference and correlation coefficient between CVIS and manual measurements were also analysed.Choroidal Vascularity Index Software (CVIS) demonstrated excellent intra-session reliability for CT (ICC: 0.992) and CVI (ICC: 0.978) measurements, compared to the relatively lower intra- and inter-observer reliability of manual measurements. Choroidal Vascularity Index Software (CVIS) and manual assessments had the highest correlation at nasal choroid (CT: r = 0.829, p 0.001; CVI: r = 0.665, p 0.001). Choroidal parameters identified with CVIS showed stronger correlations with axial length than manual measurements.This automated software, CVIS, exhibited excellent reliability compared to manual measurements, which are subject to image quality and clinical experience. With its validated clinical relevance, CVIS holds promise to serve as a flexible and robust tool in future vitreoretinal and chorioretinal studies.