Retinopathy of prematurity (a disease of the retina in premature babies) is the most common cause of severe visual impairment or blindness in children in developed countries. Therefore, all premature babies undergo regular eye examinations, where a retinal camera is used to take pictures of the background of the eye.
"We then assess changes in the retina and, if necessary, inject a biological preparation into the eye or treat the peripheral part of the retina with a laser," says Juraj Timkovič, head of the Centre for Children with Visual Impairment at the Ostrava University Hospital.Currently, doctors use digital imaging systems to evaluate images, but they are not perfect. In order to make a faster and more reliable diagnosis, medical professionals are trying to use artificial intelligence (AI) in collaboration with scientists from the Technical University of Ostrava.
"Thanks to the extensive screening that doctors at the Ostrava University Hospital perform on premature babies, we had 9,000 images of the eye background. In retinopathy, pathological changes are manifested by abnormal curvature of blood vessels and changes in their thickness. These were the main inputs for the artificial intelligence, which learned to recognise from the images whether the blood vessels were healthy or damaged," explains Jan Kubíček from the Faculty of Electrical Engineering and Computer Science at VSB-TUO.Using advanced techniques, the researchers were able to create algorithms and develop a highly accurate system for automatically segmenting blood vessels in the background images of the eye, which will help with image classification, facilitate the assessment of retinal changes and improve the possibility of early detection of the disease."Artificial intelligence can help us not only in diagnosing the specific stage of the disease, but also in treatment. It would be able to detect if there has been a change and the blood vessel is starting to improve after the drug has been administered. It will thus help us to predict the development after a medical intervention," says Juraj Timkovič. And how will the research results be used in practice? A web application is being developed where the user uploads images of the eye background and the system analyses them. "The application will be accessible especially to the professional public so that we can share information and experience. The main motivation is the practical use of our results, which should help improve diagnosis and increase the success rate of treatment in paediatric patients," adds Jana Nowaková from the Faculty of Electrical Engineering and Computer Science.The research was part of the LERCO project, in which the University of Ostrava, VSB - Technical University of Ostrava and the University Hospital in Ostrava are collaborating.
Furthermore, the projects were Multi-Criteria Optimization with Constraints using Potential Surface Analysis and OP JAK cutting-edge research Robotics and Advanced Industrial Manufacturing.