A robust algorithm for segmentation of blood vessels in the presence of lesions in retinal fundus images

Diabetic Retinopathy is a vascular disorder caused by the variations in blood vessels pattern of retina which appears as lesions. Segmentation of blood vessels in the presence of lesions is a challenging task. This paper proposes a robust method for accurate segmentation of blood vessels from retinal images even in the presence of lesions. The proposed method involves three main steps to cater false vessels. First step is to extract region based features from the candidate vascular regions of priori segmented vascular pattern. A discriminant feature vector for each candidate region is formulated. Then feature selection is performed to get the best features. Lastly, blood vessels classification is performed using Support Vector Machine (SVM) classifier to categorize each region as true vessel and false vessel. The proposed approach is tested on the dataset of fundus images from normal and abnormal DR patients and gives good segmentation results even in the presence of lesions.

I. Introduction

Diabetic Retinopathy (DR) is the most common eye disease which causes blindness in the adults in the age of 20–60 years old [1]. It is a progressive eye disease that is caused by the increase of insulin in blood and can cause blindness if not detected timely [2]. So, it is necessary to detect and treat DR as early as possible to reduce the risk of blindness. Diabetic Retinopathy is diagnosed by analyzing the fundus images. Digital fundus image contains blood vessels, optic disc and macula as normal components. As the disease progresses, different lesions start appearing on surface of retina and also effect the shape of blood vessels. In the advance stage of DR, abnormal blood vessels start growing known as neovascularization [3]. These vessels appear as holes and bunches of small vessels. Therefore, an automated retinal image analysis is required to screen and grade DR images. Most of the automated methods devised for retinal image analysis rely on vessels segmentation. Automated vessel segmentation serves many purposes in computer aided diagnosis systems which tend to perform timely detection of DR. Fig. 1 shows segmented blood vessels from normal and DR retinal image. The red circles on image (c) and (d) show retinal lesions detected due to presence of the false structures. These false structures appear as false positives and computer aided diagnosis systems consider these vessels as abnormal vessel and grade image as advanced DR. Hence, this leads to misdiagnosis of diseases and can degrades the automated screening systems. Fig. 1

(a) Normal retinal image, (b) segmented vascular pattern of image (a), (c) DR fundus image with lesions, (d) segmented vascular pattern of image (c)