Most people think of eye exams as a way to check whether they need glasses or to look for eye diseases. However, scientists have long known that the eyes can reveal clues about a person’s overall health.
Tiny blood vessels inside the retina often reflect changes occurring throughout the body. In some cases, eye doctors can spot signs of diabetes, high blood pressure, and other medical conditions before patients even realize they are sick.
Recent advances in imaging technology have made eye examinations even more detailed. Optical coherence tomography, commonly called OCT, allows doctors to create highly detailed three-dimensional pictures of structures inside the eye.
The scan is quick, painless, and widely used around the world. It can reveal changes that are invisible during a routine eye examination.
The challenge is that each scan generates hundreds of images. Reviewing every image carefully takes time and requires significant expertise. Even experienced specialists can face difficulties when examining large amounts of information every day. Researchers have therefore been exploring whether artificial intelligence could help.
A research team from Washington University School of Medicine in St. Louis, together with collaborators from the University of Washington and Genentech, has developed a new AI system designed specifically for these complex eye scans.
The system, called OCTCube-M, was created to interpret three-dimensional retinal images and assist doctors in identifying disease.
The study describing the technology was published in Nature Biomedical Engineering.
Unlike earlier AI systems that relied mainly on flat, two-dimensional pictures, OCTCube-M analyzes the full three-dimensional structure of the retina.
This is important because many eye diseases affect tissue beneath the surface and spread through multiple layers. Looking at the complete structure may allow the AI to find warning signs that would otherwise be missed.
To teach the system, researchers used more than 26,000 eye scans containing over 1.6 million retinal slices. This extensive training helped the AI learn the differences between healthy eyes and eyes affected by disease.
When tested, the system showed clear improvements compared with older technologies. It was more accurate in identifying several serious retinal diseases, including age-related macular degeneration, diabetic eye disease, and other conditions that can lead to vision loss.
The improvement may seem small in percentage terms, but it could translate into many additional patients receiving an earlier diagnosis.
Early diagnosis is important because many retinal diseases cause irreversible damage if left untreated. Patients often do not notice symptoms until vision has already been affected. Detecting disease earlier gives doctors a better chance to slow progression and preserve eyesight.
The researchers were particularly interested in geographic atrophy, a severe form of age-related macular degeneration.
This disease gradually destroys central vision, making activities such as reading, driving, and recognizing faces increasingly difficult. By combining different types of eye images, the AI became much better at predicting how quickly the condition would worsen.
Accurate prediction could be valuable for both doctors and researchers. Patients at higher risk may need closer monitoring, while scientists developing new treatments could design smaller and more efficient clinical trials. This could reduce costs and speed up the development of new therapies.
Perhaps the most unexpected discovery involved diseases outside the eye. The researchers found that retinal images contain information about blood vessel health throughout the body. Since similar biological processes affect blood vessels in the eye, heart, brain, and kidneys, the AI was able to identify patterns associated with serious medical conditions.
Using only eye scans, the system showed the ability to predict risks linked to heart attack, stroke, and kidney failure. Although the technology is not yet ready to replace standard medical tests, it suggests that future eye exams could become a valuable screening tool for many health problems.
The idea is appealing because eye scans are already widely available, non-invasive, and relatively quick to perform. If future studies confirm the findings, a routine visit to an eye clinic might provide information about both vision and overall health.
Researchers emphasize that OCTCube-M remains an experimental system. More data, larger studies, and additional testing are needed before it can become part of everyday medical practice. They are continuing to expand the technology by adding more diseases, more patients, and more imaging types to improve its performance.
The study demonstrates how artificial intelligence may help doctors handle increasingly large and complex medical datasets. Rather than replacing physicians, systems like OCTCube-M are designed to support them by identifying patterns, highlighting risks, and helping prioritize patient care.
If you care about kidney health, please read studies about pesticide linked to chronic kidney disease, and this drug may prevent kidney failure in people with diabetes.
For more health information, please see recent studies about drug duo that may treat kidney failure, and results showing these vegetables may protect against kidney damage.
Source: Washington University School of Medicine in St. Louis.
