There are 2.2 billion visually impaired people in the world, and 1 billion of them have avoidable vision impairment, according to the World Health Organization's World Report on Vision 2019. Inequalities in eye care coverage, treatment, and rehabilitation are a global problem. Early diagnosis of ocular disorders can prevent vision loss. Retinal fundus scans can be very useful while diagnosing ophthalmic diseases. Using these retinal scans, it is possible to detect over 46 different diseases. This use case of retinal scans can be of great benefit to doctors. Using just one simple retinal scan doctors can diagnose multiple diseases at once. This research aims to review and summarise different deep-learning approaches that can be used to solve the problem of diagnosing diseases using retinal scans. The focus is on reviewing deep learning techniques that can be used for multi-disease detection and nerve segmentation.