SVM based Supervised Machine Learning Framework for Glaucoma Classification using Retinal Fundus Images

Glaucoma is group of ocular conditions that damage the optical nerve. Glaucoma diagnosis in the early condition is beneficial for better vision. The available clinical instruments are nonautomated and work on the manual operating principle. In this article, we proposed an SVM with the supervised algorithm of a machine-learning framework for glaucoma classification. In this study, a 2-dimensional variational mode decomposition tool has been employed for fundus image extraction. Then, texture-based features such as Zernike moment, chip histogram, haralick features have been computed from the high-frequency modes. The Students t-test approach is applied for the chosen robust features. In the end, a multi-stage classifier (support vector machine) has been used for glaucoma prediction. The effectiveness of the deployed technique is tested using a publically available dataset. The developed automated system obtained the highest Acc of 89.45%.