NIH Scientists Use AI to Sharpen Eye Imaging with Standard Clinic Equipment

Researchers at the National Institutes of Health have developed a technique using artificial intelligence to dramatically enhance images of the retina taken with standard eye clinic equipment. This new method could allow more clinics to detect vision-threatening diseases earlier by making cellular-level detail visible without the need for specialized or expensive imaging technology.

The technology transforms images taken by traditional scanning laser ophthalmoscopes—tools widely used in clinics to view the retina—into higher-resolution images that approach the quality of those produced by adaptive optics, a cutting-edge method that remains largely experimental. By integrating artificial intelligence, the upgraded system identifies and enhances features previously too blurred to be of diagnostic use, enabling visibility of individual cells like those in the retinal pigmented epithelium (RPE), a critical layer of tissue beneath the retina.

To train the AI system, researchers provided over 1,400 retinal images categorized by quality and matched them with images from the same areas captured using adaptive optics. Once trained, the AI successfully enhanced standard images to reveal RPE cells with an eightfold increase in clarity. These improvements were further boosted through the use of indocyanine green dye, which enhances contrast and is already approved for use in eye clinics.

The new method addresses a long-standing challenge in ophthalmology: routine imaging of RPE cells, which play a key role in nourishing the retina and are often affected early in diseases such as age-related macular degeneration and Stargardt disease. Previously, these cells could not be reliably imaged in typical clinical settings. Now, with this AI-enhanced approach, clear images of RPE cells can be captured in seconds using common clinical instruments, potentially transforming early diagnosis and monitoring of degenerative eye conditions.

Article by multiple RFHC contributors, based upon information from a press release by the National Institutes of Health (NIH)

---