Novel Diabetic Retinopathy Disease Predictor using CNN for Healthcare Systems

Diabetic Retinopathy (DR) is a common complication of diabetes mellitus that damages the retina and leads to vision loss and potential blindness. Traditional screening using color fundus photographs is time-consuming and labor-intensive. Given the increased prevalence of diabetes, computer-aided diabetic retinopathy diagnosis has drawn scientific interest, as early diagnosis of the disease remains challenging. Deep and machine learning-based techniques have heavily contributed to enhancing medical image processing, enabling quick diagnosis and diabetic retinopathy categorization. Manual retinal vasculature segmentation is time-consuming and requires experienced supervision. This paper explores diabetic retinopathy segmentation and classification deficiencies, highlighting major issues and potential research directions. VGG-16 CNN model has been used in this paper to detect diabetic retinopathy. The accuracy of the model is 95.2%.