Fig. 2
Image processing to extract the conjunctival microvascular network in OCTA and FSLB. a The raw OCTA image of the temporal bulbar conjunctiva (304 × 304 pixels, 8.775 × 8.775 mm2); b The raw FSLB image of the temporal bulbar conjunctiva (5184 × 3456 pixels, 15.74 × 10.50 mm2); c A custom software was used to segment the vessels from the raw image of FSLB to create the microvascular network by a series of image processing procedures. The image was resized to 1024 × 683 pixels in grayscale; d The segmented OCTA vessel image (skeletonized) was merged into the FSLB image; e The resized OCTA image with 768 × 768 pixels and the FOV was 6.581 × 6.581 mm2; f The resized FSLB image with 768 × 768 pixels and a FOV of 6.581 × 6.581 mm2; g Equalized OCTA image; h Equalized FSLB image; i OCTA image after Gaussian Blurring (Sigma = 4); j FSLB image after Gaussian Blurring (Sigma = 4); k OCTA image after bandpass filtering; l FSLB image after bandpass filtering; m Binary OCTA image after Otsu Thresholding; n Binary FSLB image after Otsu Thresholding; o Skeletonized OCTA image after Otsu Thresholding; p Skeletonized FSLB image after Otsu Thresholding; q Merged skeletonized images obtained using OCTA and FSLB. Yellow: vessels extracted from both devices; red: vessels only extracted from OCTA, and green: vessels only extracted from FSLB