Feature extraction from image for automated retinal vessel detection

• Main Author: Wang, Jiaru
• Other Authors: Jiang Xudong
• Format: Final Year Project
• Language: English
• Published: 2018
• Subjects: DRNTU::Engineering
• Online Access: http://hdl.handle.net/10356/75009
• Institution: Nanyang Technological University

Blood vessel that covers all parts of the body, especially the fundus of the eye, containing a large number of arterial vessels. With the development of modern medical science and technology and the renewal of medical equipment, more and more medical images have been applied to medical diagnosis and scientific research. Retinal image is a kind of medical image rising in recent years. At present, many diseases can cause changes in fundus blood vessels, such as leukemia, hypertension, coronary heart disease and so on, which will cause varying degrees of fundus vascular diameter. Hence, retinal vessel segmentation becomes one of most important task for disease diagnose and treatment. The existing methods of retinal vascular segmentation are divided into two main categories: supervision method and unsupervised method. The unsupervised method does not require manual labeling, and the supervision method requires manual marking. The basic idea of the supervision method is divided into two steps. The first step is the definition and extraction of the features, the second step is the selection and application of the classifier. At the present stage, most of the methods used to solve the problem of vascular segmentation in the pathological retina are unsupervised and have some limitations. In this paper, the supervision method will be used to solve the problem of pathological retinal segmentation. In this paper, a supervised method had been applied on STARE and DRIVE database respectively. Classification performance based on the region is higher than the pixel based classification, and can be more accurate and faster to classify the images.