Diabetic Retinopathy From Retinal Imaging

Retinopathy is a sort of ocular disease that caused by diabetes damages the retina’s blood vessels, which holds irreversible impairment or blindness unless treated. The usual mode used for identifying Diabetic retinopathy takes a lot of time and is prone to mistakes made by people. because of the intricacy involved in the anatomy of the eye. Numerous automated techniques have been put forth to identify diabetic retinopathy. from fundus images. However, these methods have limitations with regard to the detection of the subtle features characterizing diabetic retinopathy, particularly in its early phases. In this study, we present a contemporary method for identifying diabetic retinopathy using the CNN-based approach. The proposed network extracts feature by using two different DL architectures, namely Resnet50 and Inceptionv3, and then combines the features. To classify, feed these into the CNN. A dataset of fundus images that are accessible to the public is used to illustrate the effectiveness of the suggested model. The findings of the trial indicate that the suggested CNN model outperforms various models with accuracy 9 5.84 %, sensitivity 98.18%, specificity 97.82%, precision 95.44%, and f1 score of 99.65%.