Flow Void Analysis Using Different Thresholding Methods on a Choriocapillaris Optical Coherence Tomography Angiography Image Complemented with a Structural En Face Image

• Published in: Korean journal of ophthalmology Vol. 38; no. 1; pp. 34 - 41
• Main Authors: Lee, Young Joo, Hwang, Soon-Young, Yun, Cheolmin
• Format: Journal Article
• Language: English
• Published: Korea (South) 대한안과학회 01.02.2024
• Subjects: Capillaries; Choroid; Fluorescein Angiography - methods; Humans; Retinal Vessels; Tomography, Optical Coherence - methods; 안과학; Optical coherence tomography angiography; Flow void; Choriocapillaris
• ISSN: 1011-8942; 2092-9382; 2092-9382
• DOI: 10.3341/kjo.2023.0110

Purpose: To investigate the flow characteristics using different thresholding methods on a choriocapillaris optical coherence tomography angiography (OCTA) image complemented with a structural En Face image.Methods: The 42 choriocapillaris OCTA images from healthy subjects were obtained with swept-source OCTA device and the 3 × 3-mm area OCTA images were processed with ImageJ. Using a raw choriocapillaris OCTA image and structural En Face image, we adjusted the different structural signal intensity. Then, the raw images and adjusted images were binarized with a global threshold and an auto local threshold using the Phansalkar method at 1- or 2-intercapillary distance. Then, the mean area, number, and size of the flow void, were compared among the images using different thresholding methods.Results: Mean flow void area, number, and size were different according to the different binarization method both in raw and adjusted images (all p < 0.001). The mean flow void area analyzed with global threshold method were well correlated with those with auto local threshold method both in raw and adjusted images (all intraclass correlations, >0.929).Conclusions: Flow void features varied according to the different binarization methods but showed good correlation. The flow void characteristics according to the different binarization methods should be considered for the analysis of the choriocapillaris OCTA images complemented with a structural En Face image.