Network-based features for retinal fundus vessel structure analysis

• DOI: 10.1371/journal.pone.0220132
• Publication Date: 2019-07-25T17:39:45Z
• Authors: Pablo Amil; Cesar F. Reyes-Ma...; Lev Guzmán-Vargas; Irene Sendiña-Nadal; Cristina Masoller
• Keywords: Fundus (uterus)

Retinal fundus imaging is a non-invasive method that allows visualizing the structure of blood vessels in retina whose features may indicate presence diseases such as diabetic retinopathy (DR) and glaucoma. Here we present novel to analyze quantify changes retinal vessel patients diagnosed with glaucoma or DR. First, use an automatic unsupervised segmentation algorithm extract tree-like graph from structure. The nodes represent branching (bifurcation) points endpoints, while links segments connect nodes. Then, structural differences between graphs extracted groups healthy non-healthy patients. We also fractal analysis characterize graphs. Applying these techniques three image databases find significant (p-values lower than 0.005 0.001 depending on database). results are sensitive (manual automatic) resolution images.