Deep Learning in Retinal Diseases Diagnosis: A Review

Abstract

Retinal diseases are among the leading causes of blindness and severe vision loss at the global level. Early diagnosis of retinal disease is of great importance in order to prevent irreversible damage to the eye. In recent years, deep learning methods have been widely used to diagnose retinal diseases. These models are developed for diagnosing a particular retinal disease. Most of these models detect and analyze disease features from the retinal image. Thanks to systems that can predict the detection of retinal diseases with high accuracy, it has allowed ophthalmologists to reduce their workload and reach more patients. This extensive literature review presents a comparative study of deep learning methods used to detect retinal diseases. For this purpose, the studies conducted on the subject between the years 2015-2022 will be examined. The related studies will be analyzed according to (1) rates of studies according to publishing years, (2) the data sets used, (3) the deep learning methods used, and (4) diagnosed retinal diseases.

Key Terms in this Chapter

Retina: A light-sensing tissue made up of nerves at the back of our eyes.

Deep Learning: A type of an artificial neural network with high degree of layers.

Ophthalmology: A branch of science that studies the structure of the eye and the diseases that occur in the eye.

Ophthalmologist: A doctor ofmedicine specializing in ophthalmology.

Image Processing: A process of transforming an image into a digital form and performing certain operations to get some useful information.

Retinal Image: A digital picture of the back of the eye that shows the retina, the optic disk, and blood vessels.

Retinal Disease: A disease occuring on the retina of the eye.