Glaucoma Detection Using CNN Ensemble Classifier and OTSU Threshold Algorithm

Glaucoma is an eye disorder that damages the optic nerve (or retina), impairs vision, and finally renders a person blind. In the field of medicine, glaucoma early diagnosis has taken on significant importance. This work develops, trains, and tests a diagnostic tool for glaucoma identification based on imagery of the eye fundus. Segmentation and feature extraction techniques are used to detect glaucoma using retinal images, while other techniques focus on convolutional neural network (CNN) training by brute force. The Glaucoma Fundus Image Recuperation (GFIR) technique was developed with the aim of classifying the patient's fundus pictures using the OTSU algorithm in order to identify glaucoma sickness. The success of the GFIR technique is assessed using the glaucoma database from Kaggle in terms of sensitivity rate, specificity rate, ranking efficiency, and feature selection time. A VCG-16 based Feature Clustering technique is developed in order to accomplish exact vessel segment location in fundus images and to boost performance rate with human graded orientation standard. The Convolutional Neural Network Ensemble Classifier (CNN-ENSEMBLE) framework has been developed to help identify glaucoma problems early, and the system's accuracy was 98%.