Multi-class classification algorithm for ocular diseases classification based GLCM texture features

Ocular diseases cover a broad spectrum of conditions impacting eye health and functions, potentially resulting in vision impairment or blindness. Common eye diseases include cataracts, glaucoma, and Diabetic Retinopathy. The World Health Organization estimates that there are around 94 million people globally with significant visual impairment or blindness due to cataracts; approximately 7.7 million people have glaucoma, and 3.9 million people have Diabetic Retinopathy. This work uses machine learning (ML) techniques to detect ocular disorders, including Diabetic Retinopathy, cataracts, and glaucoma. Gray-Level Co-Occurrence Matrix (GLCM) feature optimization in the images allows for feature extraction and classification. The suggested method is evaluated using 2776 categorized photos gathered from diverse sources. These extracted features are then classified using ML models, including k-nearest Neighbors (k-NN), Neural Network (NN), Gradient Boosting (GB), Decision Tree (DT), and Random Forest (RF). The classification accuracies of the ML models in training phase were DT 99%, GB 86%, k-NN 57%, ANN 60%, and RF 99%.