Abstract
Purpose :
To investigate the choroidal vascularity index (CVI) in normal Japanese eyes using image binarization technique and to evaluate the factors affecting CVI.
Methods :
The study included 52 subjects (24 male and 28 female) with healthy eyes, the average age was 40.1 ± 10.3 years old. 12 × 12 mm Angio cube data were acquired using the Plex Elite 9000 swept-source OCT (Carl Zeiss Meditec, Dublin, CA). A 3D Median filter was applied to the cube data for noise-reduction before extracting the foveal B scan image. All images were subsequently binarized using the public domain software Image J, and the CVI was calculated as the ratio between the vascular luminal area and the total choroidal area. The results were examined with respect to gender (t-test) and age/axial length (correlation analysis of Pearson) and compared to a similar data collection with averaged EDI B Scans on an SD-OCT. In addition, the variation of CVI was investigated with respect to the eccentricity of the analysis window relative to the fovea, image quality and resolution.
Results :
The average CVI for a 3 mm wide analysis window was 64.3% ± 2.6% (Male: 64.8% ± 2.4%, females: 63.9% ± 2.8% with no statistically significant difference P = 0.237). The correlation between age or axial length and CVI, yielded no significant association between them (P = 0.82, P = 0.99). Varying the central analysis window size from 1.5 mm to 6 mm the averaged CVI value significantly increases from 64.1%-65.6% (P = 0.000). The CVI value of a 3 mm wide analysis window increases similarly from 64.3% centrally to 66.2% at 4 mm off-center (P=0.001). CVI was fond to be sensitive to image quality (3-4%) and resolution (1-2%) due pixel-based parameters of the default Niblack threshold algorithm.
Conclusions :
CVI (central 3 mm) for normal eyes was 64.3% on average, not related to sex, age, and axial length. CVI depends on eccentricity and to compare values the analysis window must be of identical size and position. Variations with image quality and resolution make a comparison between instruments or studies difficult.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.