International Research Journal of Education and Technology

Diabetic retinopathy (DR) has been discovered to be the main factor contributing to avoidable blindness. Despite the fact that there are many undiagnosed and untreated cases of DR, precise and appropriate retinal screening could aid in the early detection and treatment by hybrid algorithm technique. The goal of this project is to create a reliable DR screening and detection model in order to reduce the incidence of DR-related blindness. DR-infected eyes should be sent to an ophthalmologist for additional inspection and diagnosis, which may lessen the risk of vision loss and offer quick and correct diagnostic information. In this paper, the multiclass Support Vector Machine learning technique is employed to identify and categorize images of diabetic retinopathy. Lesion and vascular analysis of the retinal images are also done. To distinguish between normal and abnormal diabetic retinopathy image data, a multi-layer deep learning algorithm is used. By recognising the proper medical DR instances, the algorithm analyses and classifies coloured fundus images as five stages of Diabetic retinopathy which are Severe DR, Proliferative DR, Moderate DR, Mild DR and No DR. Hence, this project employs a hybrid multiclass Support Machine learning classifier and Deep learning Technology with Convolution Neural Network to classify the stages of Diabetic Retinopathy.

Keywords: Diabetic Retinopathy, Support Vector Machine Learning, Multilayer Deep learning, Convolution Neural Network.