Advances in Diabetic Retinopathy Classification using Deep Learning: The Last 5 Years Review

Diabetic retinopathy (DR) is an eye disorder that mostly affects diabetes patients and is a primary cause of vision loss all over the world. One of the key concerns that makes DR a critical condition is that it is irreversible. This means that if individuals affected by this condition do not receive appropriate treatment during the early stages, they will permanently lose their vision. Since DR is so difficult to recognize in its early stages, early diagnosis is also essential to its effective treatment. High blood sugar levels may harm the blood vessels in the retina, causing leakage, blockages, or the development of new blood vessels, all of which can cause partial vision loss or complete blindness. Several machine learning (ML) methods have been developed to recognize and categorize the condition in retinal pictures, which is DR, in order to diagnose the disease. This analysis examines recent studies that explore the use of deep learning (DL) methods. Furthermore, we go through some of the publicly available DR datasets and the metrics used to assess how well a deep learning model performs in diagnosing DR. It also highlights key conclusions and suggests areas for future research.