GAIA-MED: AI-based Home Monitoring of Eye Diseases

In the GAIA-MED project, the foundations are being laid for an infrastructure for the cooperative and integrative use of medical data in and between medical users, science and medical technology companies.  To provide evidence for the functionality of the GAIA-MED concept, the Institute for Medical Informatics (IMI) in close cooperation with the company Visotec GmbH is developing an AI-based service for the diagnosis and monitoring of eye diseases in the home environment using home OCT. The home OCT devices developed by Visotec have the potential to monitor the individual, progression of the disease as a 4D image sequence (space+time). This enables an individual treatment especially for severe retinal diseases like wet age-related macular degeneration (AMD) and retinopathia centralis serosa (RCS) regarding intervention method and application time. The work is performed in cooperation with the Institute for Software Engineering and Programming Languages (ISP), the Institute for Telematics (ITM) and the UniTransferKlinik Lübeck GmbH (UTK). The developed AI solutions will be made accessible to patients, physicians, scientists, and medical device manufacturers by means of GAIA-X compliant protocols (Figure 1).

Figure 1 OCT image analysis integrated as Advanced Smart Service into the GAIA-Med platform, allowing access for physicians, patients, scientists, and companies.

During the project, optimized deep neural networks for automatic AI-based analysis of three-dimensional OCT images (OCT: Optical Coherence Tomography) will be developed at IMI for improved care of patients with eye diseases. Deep neural networks achieve high accuracies provided that the target data come from the same distribution as the training data and thus have similar image properties. In practice, this assumption is often not met. In order to provide a robust system on the GAIA-MED platform for different actors, each with different image data distributions, methods will be developed that take into account the different image properties. Subsequently, a multimodal holistic assessment with additional ophthalmic image modalities such as fundus images, fluorescence angiographies and clinical information can be added.

The GAIA-MED project is funded by the state of Schleswig-Holstein from 2022-2024.


M. Sc. Helge Sudkamp
Fa. Visotec GmbH, Lübeck

Prof. Dr. Martin Leucker
Institute für Softwaretechnik und Programmiersprachen, Universität zu Lübeck
UniTransferKlink, Lübeck

Prof. Dr. Stefan Fischer
Institut für Telematik, Universität zu Lübeck