A Novel Approach to Glaucoma Identification Using Deep Convolutional Neural Network

Glaucoma is a major cause of the irreversible blindness of specific eye, making premature detection essential for effective treatment and management. In this study, we suggest an approach for glaucoma identification utilizing Deep Convolutional Neural Networks (DCNN), improving upon traditional CNN-based methods. We utilize a publicly available Kaggle dataset consisting of retinal images for glaucoma detection. The proposed method follows a structured pipeline: image pre-processing with a median filter to reduce noise, segmentation is done in accordance with the Fuzzy C-Means (FCM) algorithm for enhanced region isolation, and feature extraction via the Gray-Level-Co-occurrence-Matrix (GLCM) to capture critical texture features. The deep convolutional neural network (DCNN) then performs classification, leveraging its ability to learn hierarchical features for improved accuracy. Performance evaluation using standard metrics— accuracy, sensitivity, specificity, and precision—demonstrates superior results compared to traditional CNN approaches. The proposed method achieves an accuracy of 94.53% for malignant cases and 94.93% for benign cases, outperforming conventional CNN models in all measured parameters. These findings highlight the potential of DCNNs for more reliable and accurate glaucoma detection, contributing to improved diagnostic precision and early intervention.