Blood Vessel Density and Distribution in Fundus Images

Our goal is to analyze and visualize the distribution of blood vessels in fundus images to support the medical diagnosis by providing quantitative measurements and detecting diseases and abnormalities like diabetes. Methods: Data •Our images are taken by a CANON CF-60UVi fundus camera •All the images have a resolution of 3504x2336 pixels •15 color fundus images of healthy subjects •15 color fundus images of diabetic retinopathy patients •Manual segmentation done by experts •Automatic segmentation results [1] (a) (b) Figure 1: An example input fundus image of a healthy subject(a) and its input manual vessel segmentation(b) All data used are free to use for research purposes: www5.informatik.uni-erlangen.de/ research/data/fundus-images Methods: Generated Maps Three maps are generated for both the manual and au-tomatic vessel segmentations. These maps are visual-ized using color coded images to aid the medical diag-nosis: 1.Vessel density map: generated by counting the number of vessel pixels in a large neighborhood (radius is 100 pixels) (see Figure 2) 2.Vessel distance map: encodes distance of each pixel to the closest seg-mented vessel (see Figure 3) 3.Vessel thickness map