Machine Learning-Based Automated Detection of Diabetic Retinopathy Using Retinal fundus images.

Diabetic Retinopathy (DR) is the most familiar complication of diabetes. It exists in patients with diabetes and affects the human eyes. DR patients have damaged blood veins in their retina, the hypersensitive layer in the back of the eyes. Initially, DR may not cause indications or only cause mild vision problems, but if left untreated, it might cause blindness. This research compares texture analysis-based retinal classification and different stages of DR, namely mild, moderate, non-proliferative, proliferative, and regular human eye. DR stages show misconception in its physical appearance. So, it is difficult for the physician to diagnose the stage of DR a patient is going through. This research introduces the automated framework that diagnoses and classifies the DR stages using the image processing (IP) and machine learning (ML) approaches. The m has been generated for texture analysis by applying a data fusion approach. An ML classifier has been employed (using cross-validation 10) on a multi-feature dataset to build the model. The multi-layer perceptron (MLP) has shown considerably high classification accuracy, 98.53%, respectively.