Automatic tortuosity detection and measurement of retinal blood vessel network

Abstract

Increased dilation and tortuosity of the retinal blood vessels causes the infant disease, retinopathy of prematurity (ROP). Automatic tortuosity evaluation is a very useful technique to prevent childhood blindness. It helps an ophthalmologist in the ROP screening. This work describes a method for automatically detection of a retinal image into low, medium or highly tortuous. The proposed method first extracts a skeleton retinal image from the original retinal image to get the overall structure of all the terminal and branching nodes of the blood vessel based on morphological operations. Then, it separates each branch and rotates it so that partitioning process is easier which follows a recursive way. Finally, from the partitioned vessel segments, the tortuosity is calculated and the tortuous symptom of the image is detected. The results have been compared with the eye specialist's analysis on twenty five images which gives good result. © Springer International Publishing Switzerland 2014.