Identification of Glaucoma Using Deep Learning: A Systematic Literature Review

Retinal abnormalities impact millions of individuals globally. Timely identification and intervention in these abnormalities have the potential to halt their progression, thereby preventing countless individuals from experiencing avoidable blindness. Glaucoma is a condition characterized by the damage to the optic nerve in the eye. Typically, this occurs when there is an accumulation of fluid in the front portion of the eye, leading to increased intraocular pressure. The elevated pressure adversely affects the optic nerve, resulting in damage. Glaucoma emerges as a primary cause of blindness among individuals aged 60 and above. However, timely intervention can frequently prevent blindness associated with glaucoma through early treatment. The manual detection of ocular diseases is known for being time-consuming, & lacking in repeatability. In response, there have been initiatives to automate the detection of such diseases, leveraging the success of applying Deep Convolutional Neural Networks (DCNNs) for Computer-Aided Diagnosis (CAD). While these models have demonstrated effectiveness, challenges persist due to the intricate nature of retinal lesions. This study provides an introduction of popular imaging modalities, critically examines existing deep-learning techniques for glaucoma detection & grading, and focuses on the Systematic Literature Review (SLR) of the most prominent retinal diseases.