Early detection of glaucoma integrated with deep learning models over medical devices

Abstract

The early detection of some diseases can be a decisive factor in postponing or stabilizing their most adverse effects on the people who suffer from them. In the case of glaucoma, which is an ocular pathology that is the second leading cause of blindness in the world, early detection can make the difference between a patient’s complete losses of vision, or preserve their sight, as well as improve their subsequent treatment. It is for this reason that there are currently medical campaigns for the early detection of pathologies with these characteristics in a certain study population, called screening, which have shown very good results. In addition, the application of telemedicine to these processes has allowed remote evaluation of cases by clinical experts and numerous initiatives have emerged for its use in new screening strategies. On the other hand, biomedical image processing techniques based on deep learning have undergone great development in recent years, and there are several works that have demonstrated their possible application in the automatic detection of glaucoma with fundus images. The article has consisted of the development of a web platform that integrates both scenarios: on the one hand, the remote evaluation of fundus images by medical specialists, and on the other, the application of a tool based on Deep Learning for the automatic detection of glaucoma in the case studies.