A Methodology to Segment Retinal Vessels Using Region-Based Features
Analysis of the retinal blood vessels has become remarkable area of research in biomedical field. This paper presents fundus image blood vessel segmentation approach using region-based features. In the pre-processing phase, the input fundus image is segme
- PDF / 455,441 Bytes
- 11 Pages / 439.37 x 666.142 pts Page_size
- 78 Downloads / 207 Views
Abstract Analysis of the retinal blood vessels has become remarkable area of research in biomedical field. This paper presents fundus image blood vessel segmentation approach using region-based features. In the pre-processing phase, the input fundus image is segmented as major vessel and minor vessel region. Further, to enhance segmentation accuracy region-based features are extracted from minor vessels by applying morphological operations. Fuzzy entropy measure is used to select the relevant features and for classification, a k-NN classifier is employed. The proposed algorithm is evaluated using two openly available data sets DRIVE and CHASE_DB1. The method presented is independent of training samples and achieves 96.75% of classification accuracy. Keywords Diabetic retinopathy ⋅ Morphology ⋅ Region based feature Fundus images ⋅ Fuzzy entropy measure ⋅ k-NN classifier
1 Introduction Diabetic eye disease is the leading pathological cause of blindness in adults. Abnormal blood glucose level is the main cause of diabetic eye disease, that upsurges permeability of the vessel which later, leads to retinal rupture. Diseased patients reckons no symptoms until loss of optical vision. Study of the blood vessels from fundus images of retina has been extensively used by the medical masses for detecting complications instigated owing arteriosclerosis, hypertension, glaucoma, cardiovascular disease, diabetic retinopathy (DR) and stroke [13]. Blood vessel segmentation is essential to successfully detect the optical diseases that manifests in eye. V.P. Gangraj (✉) ⋅ G.K. Birajdar Department of Electronics & Telecommunication, Pillai HOC College of Engineering & Technology, Raigad 410206, Maharashtra, India e-mail: [email protected] G.K. Birajdar e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2018 D.K. Mishra et al. (eds.), Information and Communication Technology for Sustainable Development, Lecture Notes in Networks and Systems 10, https://doi.org/10.1007/978-981-10-3920-1_37
353
354
V.P. Gangraj and G.K. Birajdar
This paper presents a two phase methodology for segmentation of blood vessel. In the first phase, the total count of pixels relayed for classification are distinctly reduced by extracting the major essential vessels. In the second phase, the minor part of the vessel sub-image is analysed instead of classifying all the vessel pixels. This radically reduces the intricacies of segmentation time. For classification of vessel and non-vessel pixel in vessel sub-image, 50-dimensional feature vector is extracted using region-based morphological descriptors. Fuzzy entropy measure-based feature selection technique is used to choose important features. Finally, k-NN classifier is used for vessel and non-vessel pixel classification. This technique is computationally less complex and is independent of training samples. The proposed approach achieves 96.75% classification accuracy. The organization of the paper is formulated as follows: Sect. 2 reflects literature survey of existing algorithms. Proposed algorithm
Data Loading...
