Case Studies to Demonstrate Real-World Applications in Ophthalmic Image Analysis

In the digital era we live in, people are able to generate and store data at an unprecedented rate. This explosion in available data for further analysis is as evident in medicine as it is elsewhere. Numerous artificial intelligence techniques been applied to different medical problems with the aim of automating time-consuming, and often subjective, manual tasks implemented by practitioners in diverse specialties. This chapter focuses on several real-world applications in ophthalmic image analysis. In this context, the objective quality assessment of retinal images plays an important role to guarantee the success of any computer-aided system. For this reason, we first present a case study on retinal image quality assessment, which uses computer vision and machine learning techniques. Additionally, we introduce two other case studies: the automatic computation of the arteriolar-to-venular index, a predictive biomarker of cerebral atrophy and cardiovascular events; and the automatic diagnosis of retinopathy of prematurity, a disease affecting low birth weight infants that shows a high amount of disagreement among experts. The experimental results presented were obtained on several datasets, both public and private, and demonstrate their suitability to be used in daily practice, both for clinical and research purposes.