AI-Enhanced Retinal Imaging 100X Faster Than Humans, and Better Too, NIH Says

The new technology provides a better tool to evaluate age-related macular degeneration and other retinal diseases.

Another day, another reminder that artificial intelligence in healthcare is gathering momentum.

A new study from the National Institutes of Health finds that AI takes high-resolution images of the cells in the back of the eye that are processed 100 times faster than when done manually and with a 3.5-fold improvement in image contrast.

Ultimately, researchers say, this new technology will provide a better tool to evaluate age-related macular degeneration and other retinal diseases.

"Artificial intelligence helps overcome a key limitation of imaging cells in the retina, which is time," Johnny Tam, PhD, leader of the Clinical and Translational Imaging Section at NIH's National Eye Institute, says in an NIH media release.

Tam and his team are developing adaptive optics (AO) to improve imaging using new optical coherence tomography (OCT) that is noninvasive, fast, painless, and available in most eye clinics.

"Adaptive optics takes OCT-based imaging to the next level," Tam says. "It's like moving from a balcony seat to a front row seat to image the retina. With AO, we can reveal 3D retinal structures at cellular-scale resolution, enabling us to zoom in on very early signs."

Tam's work targets the retinal pigment epithelium (RPE), a layer of tissue behind the retina that is of particular interest to researchers because many diseases of the retina occur when the RPE breaks down.

Overcoming the Speckle

Imaging RPE cells with AO-OCT are susceptible to a complication called speckle, which Tam says interferes with AO-OCT much like clouds interfere with aerial photography. Currently, clinicians must repeatedly take images until the speckle shifts and allows different parts of the cells to become visible. Clinicians must then piece together the images to create an image of the RPE cells that is speckle-free, a long and laborious process when done manually.

Tam and his team created an AI-based deep learning algorithm called parallel discriminator generative adverbial network (P-GAN) that processed 6,000 manually analyzed AO-OCT-acquired RPE images, each paired with its corresponding speckled original. The network was trained to identify and recover speckle-obscured cellular features. 

When tested on new images, P-GAN successfully de-speckled the RPE images, recovering cellular details. With one image capture, it generated results comparable to the manual method, which required the acquisition and averaging of 120 images.

With performance metrics that assess things like cell shape and structure, P-GAN outperformed other AI techniques, and NIH researchers say P-GAN reduced the processing time 100-fold, while producing images with contrast that was 3.5 greater than before.

By integrating AI with AO-OCT, Tam believes that a major obstacle for routine clinical imaging using AO-OCT has been overcome, especially for diseases that affect the RPE, which has traditionally been difficult to image.

"Our results suggest that AI can fundamentally change how images are captured," Tam says. "Our P-GAN artificial intelligence will make AO imaging more accessible for routine clinical applications and for studies aimed at understanding the structure, function, and pathophysiology of blinding retinal diseases."

"Thinking about AI as a part of the overall imaging system, as opposed to a tool that is only applied after images have been captured, is a paradigm shift for the field of AI."