Artificial intelligence and glaucoma progression

Abstract

Detection of progression in glaucoma is crucial to avoid visual impairment and blindness. Throughout the clinical course of the disease, glaucoma patients can present very different trajectories, as some patients may remain stable using single eye drops whereas other patients may require surgical procedures to control the disease. Thus, the decision of intensifying a treatment by adding new eye drops or performing a glaucoma surgery need to rely on precise data of true progression ofthe disease. In addition, assessing the velocity of progression can help to identify rapid progressors that are more prone to develop functional impairment. In clinical practice, we use both structural (retinography and optical coherence tomography) and functional (visual field) measurements, along with clinic-demographical data to evaluate if the patient is progressing. However, in some patients the correlation between structural and functional exams makes the detection of progression a challenge. Currently we are facing a growing use of artificial intelligence in medicine with the application of complex algorithms such as deep learning models. In this review, we summarize the findings from recent studies that investigated the use of artificial intelligence in detecting glaucoma progression.