Gezondheidsrisico’s in kaart brengen met AI en een netvliesfoto

Researchers have demonstrated that analyzing retinal images with artificial intelligence can identify early signs of systemic diseases, offering a non-invasive window into broader health risks through a simple eye scan.

A team from University College London and Moorfields Eye Hospital developed an AI model called RETFound that detects the risk of various diseases by examining retinal photographs. Rather than relying on large sets of labeled medical images, which are expensive and time-consuming to prepare, the researchers trained the model using self-supervised learning on 1.6 million unlabeled retinal images. This approach allowed the model to learn the fundamental structure and appearance of a healthy retina before being fine-tuned for specific disease detection.

In the loop of millions of images the model somehow learns what a retina looks like and what all the features of a retina are.

Pearse Keane, ophthalmologist involved in the project

Once the model established a baseline understanding of retinal anatomy, it could be adapted with additional training to identify patterns associated with specific health conditions. This method represents a shift from traditional supervised learning in medical imaging, reducing dependence on expert-labeled data while maintaining diagnostic potential.

It is not the first time that AI is used to analyze retinal scans, but the approach of the researchers with their RETFound model will accelerate developments in this area of medicine.

DailyAI, September 14, 2023

Retinal scans linked to cognitive decline risk

Separate research from the National University of Singapore shows that retinal imaging can also predict the risk of cognitive decline and dementia. Scientists there developed a biomarker called RetiPhenoAge, which uses deep learning to measure the biological age of the retina from a standard eye scan. The retina shares microvascular and neuronal structures closely connected to brain health, making it a useful indicator of neurological aging.

In a study involving over 500 patients from memory clinics in Singapore, individuals with an elevated retinal age had up to a 40 percent higher risk of cognitive decline within the next five years. These findings were validated in an external study using data from the UK Biobank, which followed more than 33,000 participants over twelve years.

The researchers emphasize that the required eye scans can be performed with equipment already widely available in primary care clinics and ophthalmology practices, making the technology scalable, cost-effective, and patient-friendly for early detection efforts.

Broader applications in preventive health

Other studies confirm that AI analysis of retinal images can detect signs of conditions such as diabetic retinopathy, reinforcing the eye’s role as a biomarker for systemic health. By identifying subtle changes in retinal blood vessels and neural tissue, these tools may support preventive interventions before symptoms appear.

As retinal imaging becomes more integrated into routine health assessments, AI-driven analysis offers a promising path toward earlier identification of disease risk, leveraging a widely accessible and non-invasive procedure to inform preventive care strategies.