Prompt Title: DR Detection with Deep Learning

This article is a review of recent deep learning-based methods for detecting and classifying diabetic retinopathy (DR) from fundus images. DR is a serious public health problem affecting millions of people worldwide and is the leading cause of blindness among working age adults. Regular eye screening for patients with diabetes is essential for early diagnosis and appropriate treatment, as the only preventive strategy is the control of hyperglycemia, hyperlipidemia, and hypertension. However, diagnosing DR from fundus images demands a high level of expertise and effort from a professional ophthalmologist, especially in densely populated or remote areas where the number of ophthalmologists is disproportionally low. Recent advancements in Artificial Intelligence (AI) and the increase of computational resources and capabilities have created the opportunity to develop Deep Learning (DL) applications for accurate DR detection and classification. This review article presents a novel holistic overview of the analysis pipeline, including a thorough analysis of publicly available datasets, the commonly used preprocessing pipelines, comparative technical information regarding the development of published DL models for the classification and segmentation of fundus images, and a presentation of models that have been applied in real clinical settings.