Automated Optic Disc Segmentation and Classification Model using Optimal Convolutional Neural Network for Glaucoma Diagnosis System
Narmatha Venugopal1, Kamarasan Mari2

1Narmatha Venugopal* ,Department of Computer and Information Science, Annamalai University, Chidambaram, Tamilnadu India.
2Kamarasan Mari, Department of Computer and Information Science, Annamalai University, Chidambaram, Tamilnadu India.
Manuscript received on September 23, 2019. | Revised Manuscript received on October 15, 2019. | Manuscript published on October 30, 2019. | PP: 7555-7561 | Volume-9 Issue-1, October 2019 | Retrieval Number: A1928109119/2019©BEIESP | DOI: 10.35940/ijeat.A1928.109119
Open Access | Ethics and Policies | Cite | Mendeley
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (

Abstract: In present days, Glaucoma is an important disease which affects the retinal portion of the eye. The identification of Glaucoma in a color fundus image is a difficult process and it needs high experience and knowledge. The earlier identification glaucoma could save the patient from blindness. An important way to diagnose the glaucoma is to detect and segment the optic disc (OD) area. The region of OD area finds useful to help the automated identification of abnormal functions occurs in the case of any injury or damage. This paper presented an automated OD segmentation and classification model for the detection of glaucoma. The presented model involves feature extraction using median filter, segmentation using morphological operation and classification using convolution neural network (CNN). Here, optimal parameter settings of the CNN are automatically tuned by the use of particle swarm optimization (PSO) algorithm. The presented model is validated using DRISHTI-GS dataset and a detailed quantitative analysis is made to ensure the goodness of the presented model. In addition, the extensive simulation outcome pointed out that the presented model showed outperforming results with the maximum accuracy of 97.02% in the classification of OD.
Keywords: Feature extraction; Glaucoma; Optic disk; Segmentation; particle swarm optimization.