Creating Biomedical Technologies to Improve Health


Science Highlight: August 21, 2015

Researchers use computing power to see individual cells in the eye

New technique could lead to earlier diagnoses, better treatments of degenerative and neurological eye diseases

Many diseases affecting vision begin as changes to microscopic structures such as cells, nerve fibers, and blood vessels at the back of the eye. The ability to visualize these early changes could lead to quicker diagnoses and better assessment of treatments for degenerative and neurological eye diseases.

Currently eye doctors use a technology called optical coherence tomography (OCT) to image the back of the eye. OCT generates images by sending near-infrared light into the eye and measuring the intensity of the light that is reflected back, similar to the way an ultrasound imaging system measures the reflection of sound off tissues. While sufficient for visualizing large structures in the eye such as the retinal layers, including the layer of nerves on the retina’s surface, OCT alone can’t see down to the level of individual cells. This is because tiny imperfections exist in each person’s eye that cause the light entering it to become distorted, resulting in a less clear OCT image.

In this cartoon representaiton of optical spectroscopy blue light waves illuminate thousands of red photoreceptors

Cartoon representation showing individual photoreceptors in the retina being probed by near-infrared light as part of a new technique that could help eye doctors see cells in the back of the eye more clearly. Graphic by Alex Jerez Roman. 

About fifteen years ago, researchers borrowed a technique from astronomers—called adaptive optics—to increase the resolution of OCT images of the eye. Astronomers struggle to see the stars clearly because light travelling from the stars becomes distorted as it travels through the Earth’s atmosphere; this is why stars appear to twinkle. Using special deformable mirrors to correct for the atmospheric effects, astronomers have been able to obtain a clearer picture of the stars.

When researchers applied adaptive optics to OCT imaging of the eye, they were able to visualize—for the first time—individual photoreceptors, the light-sensitive cells in the retina that enable us to see, but which are in a layer below the surface of the retina. However, the slow, complex, and costly setups required for adaptive optics have prevented this technique from being implemented by eye doctors for use in the clinic.

“These systems are complicated and they’re expensive,” says Stephen Boppart, M.D., Ph.D., Director of Imaging at the University of Illinois, Urbana-Champaign, who has been developing OCT technologies for the past twenty years. “They’re really only used in research labs or research-focused ophthalmology practices, not for routine clinical practice.”

Now, with support from NIBIB, Boppart and his team have pioneered an approach that achieves an imaging resolution similar to adaptive optics but relies on computing power instead of additional hardware. The approach, coined computational adaptive optics, applies additional algorithms to OCT data to correct for the eye’s aberrations as well as its constant motion.

In the June 22, 2015 online issue of Nature Photonics, Boppart and his team demonstrated, for the first time, the ability to use OCT to image individual photoreceptors without the use of additional adaptive optics hardware.

“This work represents an innovative medical adaptation of a technology originally developed in an entirely different area of science,” says Behrouz Shabestari, Ph.D., program director of Optical Imaging and Spectroscopy at NIBIB. “By distinguishing the appearance of individual cells in the back of the eye, medical experts could provide better diagnoses and treatments for eye diseases and improve tracking of progressive eye and neurological diseases.”

A difference between hardware-based adaptive optics and Boppart’s computational technique is that the former requires manipulation of hardware during the OCT exam whenever a new area of the retina is being imaged, whereas the latter enables corrections to be carried out after the OCT exam has been completed. This lets doctors focus on new areas of the retina without having to repeat the exam.

This image consists of three black and white pictures. On the left is a close up of a human retina. The middle compares images of photoreceptors before and after using computational adaptive optics. The right compares images of the retinal nerve fiber layer before and after using computational adaptive optics.

Left: A human retina. Middle: Compares images of photoreceptors before and after using computational adaptive optics. Right: Compares images of the retinal nerve fiber layer before and after using computational adaptive optics. Photo courtesy of Stephen Boppart, University of Illinois at Urbana-Champaign.

“With our computational technique, we capture all of the data at once, and then we can go back later and make corrections and adjustments. This allows us to custom tailor the process towards each individual patient,” says Boppart.

Boppart believes computational adaptive optics will be instrumental for earlier detection of diseases that affect the eye and for monitoring their progression. He is particularly interested in using the technique to study age-related macular degeneration, the leading cause of vision loss in individuals over 50. A hallmark of macular degeneration is bright patches on the retina, which can be picked up by current OCT and other retinal imaging techniques. These patches are believed to be a result of photoreceptor death and correspond to vision loss; the areas also expand over time as the disease progresses.

“With computational adaptive optics, we can begin to look directly at photoreceptors in patients with macular degeneration,” says Boppart. “In particular, it will be interesting to determine the health of photoreceptors in the periphery of these patches. As new treatments are developed, we may be able to use this to determine whether we are changing the rate of disease progression.”

The new technique could also be useful for assessing patients with multiple sclerosis, a neurological disease in which nerve cells lose the sheath that surrounds them, called myelin. Because the retina contains a layer of nerve fibers on its surface, the eye is considered a window into the health of patients with multiple sclerosis.

“In multiple sclerosis, the thickness of the nerve fiber layer starts decreasing due to demyelination and that can be picked up today with standard OCT. Doctors measure the thickness of the nerve fiber layer and try to correlate it with disease severity or disability status,” says Boppart. “What we’re starting to see with our computational adaptive optics are the individual nerve fibers within the nerve fiber layer. This could give us a more sensitive indicator of changes in disease status.”

Looking forward, Boppart and his team are creating new algorithms that could automate the computational corrections so that they don’t have to be tweaked manually. They are also looking at changes in blood vessels that could also be indicators of disease.

This research was supported in part by NIH grants EB013723 and EB012479

Shemonski ND, South FA,  Liu Y, Adie SG, Carney PS, Boppart SA.
Computational high-resolution optical imaging of the living human retina.  
Nature Photonics. 2015 Jun 22; 9: doi:10.1038/nphoton.2015.102


Health Terms: