Purchase this article with an account.
Hao Zhou, Zhongdi Chu, Qinqin Zhang, Yining Dai, Giovanni Gregori, Philip J Rosenfeld, Ruikang Wang; Automated three-dimensional segmentation, visualization and quantification of choroidal vasculature with swept-source OCT. Invest. Ophthalmol. Vis. Sci. 2019;60(11):023.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Automated methods for assessing choroidal vasculature in the whole 3D scanning volume are surprisingly limited. This work aims to develop an automated method that accurately segments choroidal vessels from 3D OCT scans, without the necessity of OCT angiography. We propose a depth-resolved mapping for choroid vessels and 3D metrics for choroidal vasculature quantification: choroid vessel volume (CVV), choroid stroma volume (CSV), choroidal vessel volume density (VVD) and choroidal stroma-to-vessel volume ratio (SVR).
SS-OCT (PlexElite, Carl Zeiss Meditec Inc.) was used to collect 12x12mm macular scans from subjects. The steps for automated segmentation of choroid slab include: 1) attenuation correction; 2) excluding optic disc; 3) graph search for Bruch’s membrane and the outer border of choroidal vessels. The 3D choroidal vessels were segmented on each B-scan of the segmented slab. From the binarized 3D choroidal vessels, en face sum projections were generated with color map indicating the depth of the vessel from Bruch’s membrane and montaged to large field-of-view. Metrics of mean choroid thickness (MCT), CVV, CSV, VVD and SVR were investigated in 3D (Fig. 1).
144 normal subjects were recruited. The automatic choroid segmentations assisted by attenuation correction were validated with manual segmentations and showed good agreement on all metrics (all P<0.0001). Depth-resolved mapping of segmented 3D choroidal vessels presented vessels at different depth with different color in ultra-large field-of-view (15x20mm) (Fig.2). CVV and CSV showed significant correlations with choroid thickness (all P<0.0001). VVD and SVR were constant with small variations among all subjects (68.9±1.3% and 0.45±0.03). All choroid-related metrics were significantly correlated with age (all P<0.0001) except VVD and SVR.
The proposed automated choroid assessment on thickness and vasculature was successful in 3D OCT scans, which will be clinically useful in quantitative assessment of a myriad of ocular diseases involving the choroid such as AMD, central serous chorioretinopathy, polypoidal choroidal vasculopathy and pathologic myopia.
This abstract was presented at the 2019 ARVO Imaging in the Eye Conference, held in Vancouver, Canada, April 26-27, 2019.
Fig. 1. Flow chart for automated 3D segmentation and visualization of choroidal vasculature.
Fig. 2. Depth-resolved mapping of choroidal vasculature (15x20mm ultra-large field-of-view).
This PDF is available to Subscribers Only