Artificial intelligence helps eye doctors

By Nana Olejank Hansen, nolejank@health.sdu.dk, 12/15/2023

In 2022, over 700,000 individuals in Denmark sought the services of an eye doctor. Simultaneously, there is a shortage of eye doctors, necessitating support, which is now achievable through the application of artificial intelligence.

- We aim to use artificial intelligence to improve diagnostic accuracy and save health care resources, which is beneficial as it can reduce waiting times for patients, notes Dr. Jakob Grauslund, an ophthalmologist and Clinical Professor affiliated with the Research Unit of Ophthalmology & Steno Diabetes Center Odense, Department of Clinical Research, University of Southern Denmark and Odense University Hospital.

He leads the research unit at the Eye Department of OUH. The Eye Department at OUH is among the 13 eye departments in hospitals in Denmark. These departments provide care to patients with more critical eye conditions than those attended care at the 157 privately practicing Danish eye doctors.

Individuals with diabetes require regular eye examinations, and artificial intelligence provides valuable support for busy eye doctors. Allowing artificial intelligence to manage certain tasks enables them to save time, affording them more opportunities to focus on patient care.

The eye undergoes examination using images and scans

Typically, eye examinations involve doctors reviewing images of the eye, searching for indicators such as bleeding or fluid. This underscores the importance of images and scans in evaluating and monitoring the ocular conditions.

Up to this point, it has been the responsibility of an eye doctor to examine the eye images, but now artificial intelligence can handle this task. The automatic image analysis with artificial intelligence is so advanced that it can now equal or even exceed human capabilities.

This is attributed, for instance, to Deep Learning, the cutting-edge technology in artificial intelligence. Deep Learning does not necessitate programming but instead relies on an extensive collection of images with the information that artificial intelligence seeks. This involves presenting the computer with a sequence of images indicating what the doctor is searching for—such as determining whether there is bleeding in the retina, a simple yes or no.

On the same level as a specialist

By presenting the computer with a substantial number of images, it eventually gains the ability to ascertain whether there is bleeding in the retina. This parallels the process of human learning. The more we encounter something, the more proficient we become over time in evaluating it.

- Some of the examinations we can conduct with the assistance of artificial intelligence are truly of such high quality that they match the expertise of a specialist in eye diseases, states Jakob Grauslund.

Interdisciplinary collaboration

While the analyses achievable with artificial intelligence are noteworthy, Jakob Grauslund and his colleagues persist in refining the methods.

- We maintain a strong research collaboration with researchers at the Maersk Mc-Kinney Moller Institute within the Faculty of Engineering. They possess a comprehensive understanding of all technical aspects, while we bring our healthcare perspective into the partnership, states Jakob Grauslund.

If you are interested in learning more, you can explore Jakob Grauslund's comprehensive article on the topic of Artificial Intelligence in Eye Diseases in “Ugeskrift for Læger” (In Danish).