Could there be a new, easier way to identify Alzheimer’s disease? Possibly, as researchers have created an artificial intelligence model that has been able to successfully detect Alzheimer’s patients through retinal images.
The research, conducted by Duke University, involved new computer software that looked at retinal structure and blood vessel images of the inside of the eye that have been linked with cognitive changes. The findings were published in the British Journal of Ophthalmology.
Researchers say this could lead to a less invasive way to predict Alzheimer’s once a person has developed symptoms.
Dr. Sharon Fekrat, a retina specialist at the Duke Eye Center, explains, “Diagnosing Alzheimer’s disease often relies on symptoms and cognitive testing. Additional tests to confirm the diagnosis are invasive, expensive, and carry some risk. Having a more accessible method to identify Alzheimer’s could help patients in many ways, including improving diagnostic precision, allowing entry into clinical trials earlier in the disease course, and planning for necessary lifestyle adjustments.”
Fekrat worked on the project with Duke colleagues in neurology, electrical and computer engineering, and biostatics and bioinformatics. They expanded on earlier work in which they identified changes in retinal blood vessel density associated with changes in cognition. They found decreased density of the capillary network around the center of the macula in patients with Alzheimer’s disease.
With that information, they trained a machine learning model, or a convolutional neural network, to detect differences within images. They input four types of retina scans to accomplish this.
Overall, 159 sets of eyes were examined, 36 belonging to those known to have Alzheimer’s. Researchers said they tried different approaches, but the best-performing model combined retinal images with clinical patient data. In an independent test group, the model was able to differentiate between cognitively healthy participants and those with symptomatic Alzheimer’s.
Where to Go From Here
Researchers say future studies will need to include a more diverse group of people to ensure machines can predict Alzheimer’s in all racial groups, as well as in people with glaucoma, diabetes and other conditions that change retinal and vascular structure.
Dr. Dilraj S. Grewal, a Duke retinal specialist and one of the study’s authors, explains, “We believe additional training using images from a larger, more diverse population with known confounders will improve the model’s performance.”
They also say they’ll need to compare this approach with other methods currently available to detect Alzheimer’s, which can include expensive and invasive neuroimaging and cerebral spinal fluid tests.
This news comes as the first blood test to detect Alzheimer’s has gone on sale. However, that test has not been approved by the FDA and independent experts are concerned that key test results haven’t been published.