Improving dense conditional random field for retinal vessel segmentation by discriminative feature learning and thin-vessel enhancement

Color fundus images are acquired by making photographs of the back of the eye, where blood vessels of humans can be directly visualized in such a non-invasive way [1]. As one of the main features in fundus images, the appearance of retinal blood vessels can provide useful information for the early diagnosis of various diseases, including diabetes, glaucoma, hypertension, arteriosclerosis and cardiovascular diseases [2]. Accurate delineation and measurement of retinal vessels is therefore an important prerequisite for a number of clinical applications.
Source: Computer Methods and Programs in Biomedicine - Category: Bioinformatics Authors: Source Type: research