A Novel Method for Non-Proliferative Diabetic Retinopathy Disease Classification Using Image Processing Techniques

The retinal blood vessels' alterations Numerous extensive clinical investigations have looked at the structure and development of disorders such retinopathy of prematurity (ROP), diabetes, and hypertension. This study introduces an algorithm for the recognition and classification of non?proliferative diabetic retinopathy (NPDR) lesions, such as dark and bright lesions. Using a multilayered perceptron, bright and dark diseases are distinguished by the presence of exudates and cotton wool spots in bright lesions and microaneurysms and haemorrhages in dark lesions, respectively. which is broad enough to be used with high definition fundus pictures? First, we preprocessed the fundus images, and then we described the straightforward classification method for diabetic retinopathy lesions that was written in the multilayered perceptron algorithm, the database is collected from Manoj Saswade and Neha Deshpande and was utilised, and this approach obtains a true positive rate of 96%, a false positive rate of 0%, and an accuracy score of 0.99