Diabetic Retinopathy Detection Using Deep Learning Methods

Abstract

Diabetes mellitus causes diabetic retinopathy (DR), which is the primary source of blindness worldwide. Initial identification and cure are required to postpone or avert visual degradation and loss. In this regard, various artificial-intelligence-powered approaches for detecting and classifying diabetic retinopathy on fundus retina pictures have been proposed by the scientific community. This paper will mostly examine existing early DR diagnostic tools to determine their merits and drawbacks. Although pictures from fluorescein angiography, colour fundus medical images or visual lucidity tomography angiography are used for early diagnosis. Only colour fundus medical images are included in this study. It is possible to categorise the early DR detection methods described in this paper as either classical image processing, traditional machine learning, or deep learning. The issues that must be addressed in creating such efficient, effective and resilient methods for initial detection of DR systems are discussed in length in this study, as is the substantial opportunity for future research in this field.