Recent techniques and trends for retinal blood vessel extraction and tortuosity evaluation: A comprehensive review

Abstract

Retinal blood vessel segmentation plays an important part in the early diagnosis and treatment of eye disease. It is a tool for ophthalmologists. Many diseases can be identified by examining manifestations and images of blood vessels, including diabetic retinopathy, retinopathy of prematurity, age-related macular degeneration, retinopathy due to hypertension, glaucoma and others. Early detection allows physicians to provide patients with effective treatment, while in the opposite case, the late detection of retinal disease can ultimately lead to blindness. One of the indices when examining the retina is an evaluation of blood vessels based on tortuosity, i.e. the degree of curvature of blood vessels. This article presents a comprehensive overview of all segmentation techniques for retinal blood vessel extraction from images taken with a fundus camera in adults and older children or with a RetCam fundus camera in new-borns and younger children over the last 10 years. An integral part of this review is a comprehensive overview with information on all available public and private databases with retinal images. The review includes an evaluation of segmentation techniques based on objectivization parameters, including information on all objectivization parameters used in this article. As already mentioned, the degree of curvature of retinal blood vessels is used to classify severity of blood vessels tortuosity. There is no uniform metric for determining tortuosity, but this review presents a comprehensive overview of all metrics and calculations used to determine the degree of tortuosity of retinal blood vessels.