Analysis of Retinal Blood Vessel Segmentation in Different Types of Diabetic Retinopathy
R.Manjula Sri1, J.Jyothirmai2, D.Swetha3
1R.Manjula Sri, VNR Vignana Jyothi Institute of Engineering and Technology, Hyderabad (Telangana), India.
2J.Jyothirmai, VNR Vignana Jyothi Institute of Engineering and Technology, Hyderabad (Telangana), India.
3D.Swetha, VNR Vignana Jyothi Institute of Engineering and Technology, Hyderabad (Telangana), India.
Manuscript received on 10 January 2019 | Revised Manuscript received on 20 January 2019 | Manuscript Published on 30 January 2019 | PP: 52-55 | Volume-8 Issue-2S2, January 2019 | Retrieval Number: 100.1/ijeat.B10120182S219/19©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Abstract: The extraction of retinal blood vessels from a fundus image is one of the solutions to detect number of diseases such as diabetes, hypertension and arteriosclerosis. Dimensions of the vessels is significant for detection of retinal diseases. Blood vessel thickness(diameter) in Different types (stages) of Diabetic Retinopathy (DR) are analyzed in this work. In Normal and Proliferative DR thick vessels are affected and in Hypertensive and Non-Proliferative DR the thin vessels are affected. The objective of this paper is to employ image processing techniques to enhance and measure the dimensions of the retinal blood vessels. Segmentation is implemented through three techniques namely Gaussian method, mathematical morphology method and multi-scale analysis method. Gaussian method uses a Gaussian resolution hierarchy to detect thin as well as thick vessels. It is a faster technique but presents noise, hence suitable only for detecting thick vessels. Mathematical morphology method is suitable to detect the fine details of thin vessels more precisely. The third technique detects the thick and thin vessels without noise and is preferable for its invariant analysis with transformation of images. To employ image processing techniques and measure the vessel diameter LabVIEW software is used. A comparative study on these three techniques has been carried out on different retinal images with vessel related eye diseases. The work was carried out under the guidance of senior eye care doctors.
Keywords: Gaussian Method, Mathematical Morphology, Multi-Scale Representation, Vessel Enhancement.
Scope of the Article: Predictive Analysis