Diabetic retinopathy is a prevalent ocular condition characterized by the deterioration of retinal blood vessels, leading to visual impairment. The increasing prevalence of this condition can be attributed to its strong association with diabetes. Diabetic retinopathy, a medical condition, possesses the capacity to lead to total vision loss, thereby emphasizing the utmost significance of timely detection for the purpose of minimizing the likelihood of visual impairment and its related complications. Currently, the assessment and diagnosis of ocular conditions heavily rely on the visual inspection of fundus images. The implementation of this methodology requires the utilisation of expensive ophthalmic imaging technology and involves a meticulous analytical procedure. The objective of this project is to facilitate a substantial overhaul of the screening procedure for diabetic retinopathy. The objective of this study is to develop a machine learning model that exhibits intuitive characteristics and demonstrates consistent accuracy in predicting the presence of diabetic retinopathy. This prediction will be based on the analysis of pre-recorded digital fundus images. The methodology employed in this study involves the retrieval of annotated fundus photographs from publicly available repositories. Two powerful machine learning techniques, namely support vector machine (SVM) and deep neural network (DNN), are utilised in the analysis. The support vector machine (SVM) model exhibited a noteworthy average area under the receiver operating characteristic curve (AUC) of 97.11% during the evaluation process conducted on the test dataset. In contrast, the deep neural network (DNN) model demonstrated significantly higher performance, as evidenced by an average area under the curve (AUC) of 99.15%. The results of this study demonstrate a promising methodology for the screening of diabetic retinopathy that exhibits notable attributes of efficiency, accuracy, and cost-effectiveness.

Index Terms—GLCM, Diabetic Retinopathy, Support Vector Machine and Deep Neural Network.