Early Detection of Glaucoma Disease in Retinal Fundus Images Using Spatial FCM With Level Set Segmentation
B. Sudha1, Surjeet Dalal2, Kathiravan Srinivasan3
1B. Sudha, Research Scholar, Department of Information Technology and Engineering, VIT, Vellore (Tamil Nadu), India.
2Dr. Surjeet Dalal, Associate Professor, Department of Computer Science and Engineering, SRM University, Haryana (Punjab), India.
3Dr. Kathiravan Srinivasan, Associate Professor, Department of Information Technology and Engineering, VIT, Vellore (Tamil Nadu), India.
Manuscript received on 02 September 2019 | Revised Manuscript received on 12 September 2019 | Manuscript Published on 23 September 2019 | PP: 1342-1349 | Volume-8 Issue-5C, May 2019 | Retrieval Number: E11910585C19/19©BEIESP | DOI: 10.35940/ijeat.E1191.0585C19
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: Glaucoma is a malady of the optic nerve brought about by the expansion in the intraocular weight of the eye. It for the most part influences the optic plate by expanding the cup size. In this proposed method the clinical parameter such as vertical optic cup to disk ratio (CDR) is determined to identify the glaucomatous disease. The segmentation of optic disc (OD) and optic cup in retinal fundus image is an important step in the determination of CDR. Optic Disc is extracted from the fundus image by circular region of interest with Hough transformation. Linear regression fit is used to find the Gold standard value for the experimentally obtained CDR A Bayesian classifier is used to train the classifier set of CDR values obtained. Results produced from the classification obtain a accuracy of 94.28%, sensitivity of 94.38% and specificity of 94.11%. ROC curve is plotted to study the relation between specificity and sensitivity of the CDR and GSV. This proposed approach is robust in segmentation and the region boundaries are precise and is able to yield regions more homogeneous which can be used for objective mass screening of retinal images for early detection of Glaucoma.
Keywords: Glaucoma, CDR, Spatial Fuzzy Clustering, Level Set Method.
Scope of the Article: Clustering