Prediction of Glaucoma based on CNN and SVM classifier

Abstract

Glaucoma is a metaphorically called gruesome thief who steals away the most precious eyesight slowly, silently forever. The ophthalmologists broadly utilize retinal images to distinguish glaucoma, which is the second purpose for visual deficiency around the world.Considering the harm effect of glaucoma, we propose a novel intelligent diagnostic framework for discrimination of glaucoma using Convolutional Neural Network (CNN) with Support Vector Machine (SVM) from colour fundus images. This method focuses on deep feature extraction by CNN. The extracted intense features are used by a SVM to categorizethe fundus imageries as either normal or glaucomatous. To explore the distinctive ability of the suggested method,we used 781 fundus images from three publicly available HRF, Origa and Drishti_GS1   database and obtained accuracy of 97.37%. Test result reveals that the proposed system can be a significant choice for the recognition of glaucoma.