Screening for Eye Disease using Automated Analysis of Retinal photography
The SEDAAR project is one of the 10 winners of the Health Data Hub Grand Challenge program.
The project leader is the Fondation Adolphe de Rothschild Hospital in Paris, France.
Fondation Adolphe de Rothschild Hospital
Private health care institution with state-of-the-art technical facilities and world-renowned experts in the field of ophthalmology.
If you wish to learn more about the Fondation Adolphe de Rothschild Hospital, click here.
Software company expert in Image Processing, Analysis, Understanding and Artificial Intelligence with a world-renowned expertise in the field of health and more particularly in ophthalmology.
Objectives of the Project
To develop a powerful Artificial Intelligence application, integrated at the beginning of the ophthalmology care pathway, with the aim to:
- Provide broad access to screening for frequent retinal pathologies;
- Target patients whose pathology requires rapid medical expertise;
- Serve as a basis for public and/or private research.
The aging of the French population combined with the announced and already begun shortage of ophthalmologists have led ophthalmology professionals to reorganize their care mission. It is also one of the fields of medicine that has innovated the most to rethink patient pathways. Among the proposed solutions, enhanced cooperation with orthoptists and the use of AI feature prominently. It is precisely on these two axes that the SEDAAR project was built. In addition, SEDAAR benefits of screening facilities that are fully managed by the Fondation Adolphe de Rothschild Hospital, with advanced technology tools such as color retinograph and OCT devices.
The innovative nature of the SEDAAR project lies in its multi-pathological approach with 37 retinal pathologies detected at the completion of the project. In addition, SEDDAR will benefit of a very large image database since over 300 patients are screened on a weekly basis. AI systems already exist to diagnose Diabetic Retinopathy, glaucoma or AMD, like OphtAI, but no AI allows, from images acquired during the examination phases carried out by orthoptists, to determine with precision whether the patient presents or not an ophthalmological pathology outside these three major pathologies.
- The first phase of the project consists of collecting and annotating 45,000 images acquired using Optical Coherence Tomography (OCT) for 3D images and a retinograph for 2D images. This work could be done thanks to the ophthalmology expertise of the Fondation Adolphe de Rothschild Hospital and to the image processing and artificial intelligence expertise of ADCIS.
- The second phase of the project, supported by ADCIS alone, is the development of algorithms to automate the detection of pathologies and to provide assistance in the diagnosis of retinal diseases.
- The third phase is the testing of diagnostic assistance algorithms with orthoptists and ophthalmologists of the Fondation Adolphe de Rothschild Hospital.
The next step in the project will be to validate the reliability of the algorithms through a CE certification and a large-scale clinical study, before making the algorithms available to ophthalmology professionals and equipment manufacturers.