Artificial intelligence in ophthalmology I: retinal diseases

Many retinal diseases threaten the vision of hundreds of millions of people worldwide. These retinal diseases impose significant demand on eye-care services and socioeconomic costs. Artificial intelligence (AI) may impact on the management of retinal diseases by image enhancement, segmentation, classification for screening, and automated diagnosis. In practice, AI systems have already demonstrated performances at or above expert levels in diabetic retinopathy grading, age-related macular degeneration grading, and general retinal diagnosis. Despite the advances in ophthalmology, there are limitations that need to be overcome before AI can be fully applied to clinical practice, such as applying the model to data collected by different devices, with varying quality, from different ethnic groups and for rare diseases.