Report: AI makes retinal imaging 100 times faster than manual method
Scientists at the National Institutes of Health use artificial intelligence called ‘P-GAN’ to improve next-generation imaging of cells in the back of the eye.
(Image credit: AdobeStock/cwiela_CH)
A team of researchers at the National Institutes of Health have applied artificial intelligence (AI) to a technique that produces high-resolution images of cells in the eye, which ultimately could speed the imaging process for ophthalmologists.
The researchers have found that using AI, imaging is 100 times faster and improves image contrast 3.5-fold. The advance, according to researchers, will provide them with a better tool to evaluate age-related macular degeneration (AMD) and other retinal diseases.
“Artificial intelligence helps overcome a key limitation of imaging cells in the retina, which is time,” said Johnny Tam, Ph., leader of the Clinical and Translational Imaging Section at NIH's National Eye Institute.1
According to the NIH news release, Tam has been involved in the development of a technology known as adaptive optics (AO) to build upon existing imaging devices based on optical coherence tomography (OCT).
“Adaptive optics takes OCT-based imaging to the next level,” Tam said in the news release. “It’s like moving from a balcony seat to a front row seat to image the retina.”
He noted that with AO, researchers can reveal 3D retinal structures at cellular-scale resolution.
Moreover, the researchers noted in the news release what the addition of AO to OCT can result in an improved view of cells, processing AO-OCT images after they’ve been captured takes much longer than OCT without AO.
As part of his current effort, Tam has targeted the retinal pigment epithelium (RPE), a layer of tissue behind the light-sensing retina that supports the metabolically active retinal neurons, including the photoreceptors. The researchers are focusing their attention on the RPE because many retinal diseases occur when the RPE breaks down.
However, the NIH noted in its news release that imaging RPE cells with AO-OCT isn’t without its challenges, which include a phenomenon that is known as speckle, which interferes with AO-OCT in much the same way that clouds interfere with aerial photography.
At any point during the process some portion of the image could be obscured, and managing the phenomenon can prove to be as simple as managing cloud cover. Researchers will image cells repeatedly over an extended period, and the speckle may shift, which allows different parts of the cell to be seen in the images. The researchers then assemble the images one piece at a time to build an image of the RPE cells that isn’t obscured by speckle.
According to the news release, Tam and his colleagues have created a novel AI-based method called parallel discriminator generative adverbial network (P-GAN)—a deep learning algorithm. By suppplying the P-GAN network with about 6000 manually reviewed AO-OCT-acquired images of human RPE, each paired with its corresponding speckled original, the team trained the network to identify and recover speckle-obscured cellular features.1
The researchers tested P-GAN on a new image and it de-speckled the RPE images, making cellular details visible with just one image capture. The old method could require an average of 120 images assembled to create a clear image.1
According to the NIH news release, Vineeta Das, PhD, a postdoctoral fellow in the Clinical and Translational Imaging Section at NEI, estimates that P-GAN reduced imaging acquisition and processing time by about 100-fold. P-GAN also yielded greater contrast, about 3.5 greater than before.2
Tam noted in the news release that by combining AI with AO-OCT, a key hurdle for routine clinical imaging using AO-OCT has been removed, particularly for issues impacting the RPE, which has historically been a challenge to image.
“Our results suggest that AI can fundamentally change how images are captured,” Tam concluded in the NIH news release. “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.”
