Early Detection of Diabetic Retinopathy and Glaucoma Using Machine Learning

Glaucoma is one of the critical conditions of the eye which, along with cataract, forms the reasons for majority cases of vision impairment issues. Diabetic Retinopathy (DR) is a major contributing factor for Glaucoma, whose early identification poses a significant challenge. Additionally, considering the cost of the initial laboratory examinations required for the identification of the DR, the work employs Convolution Neural Network (CNN) for the early identification of DR through fundus image, thereby helping in the effective screening of the DR and Glaucoma in its initial stage. The experiment was conducted on a set of data containing the labeled images for training the neural networks, using which it effectively discriminated between the normal and the pathological fundus images. The proposed model exhibited 95% of accuracy in comparison with the existing models, thereby emphasizing the effectiveness of the deep learning methods in the computer aided diagnosis.