A New Window into the Living Brain

Jesse Schallek, Ph.D.
Assistant Professor of Ophthalmology
University of Rochester Medical Center
Dana Grantee: 2017-19

When you visit the ophthalmologist or optometrist to check your vision, the doctor will likely use an imaging tool called an ophthalmoscope to get a detailed look at the back of your eye, where the retina, optic nerve, and vasculature reside. The resulting orange-ish image, accentuated with meandering red veins and blood vessels, helps the doctor determine whether your eyes are healthy.

But it’s possible such images could do even more. The retina is the only portion of the central nervous system (CNS) that can be observed from outside the body. Your eye doctor, in fact, can get a glimpse of it simply using a bright light and that ophthalmoscope. Jesse Schallek, Ph.D., has always been excited by the idea of finding a way to take images of living biology in its natural state. Today, using adaptive optics technology, a method first used to optimize long-range telescopes, he and his team have transformed the traditional ophthalmoscope so that it can now visualize the inner workings of the eye at the level of a single cell. The resulting technology offers a unique window into the CNS, which may help ophthalmologists understand the pathology of debilitating visual conditions like diabetic retinopathy and other forms of neurodegeneration and brain-related disease. Here, Schallek discusses how astronomy inspired his imaging advances, the challenges of imaging translucent cells, and how artificial intelligence can help give this kind of imaging approach even more clinical value.

What first interested you in studying imaging and the eye?

My initial interest in ophthalmology came when I took a keen interest in neuroscience as a college student. I had an opportunity to work in a lab that was studying retinal vision in horseshoe crabs, of all creatures. My early work there just fascinated me, so after I received my bachelor’s degree in bioengineering, I pursued a Ph.D. in systems neuroscience with a focus on retinal vision. My post-doctoral work involved vision research in terms of developing these new imaging devices. Today, I use my skillsets in bioengineering, retinal imaging, and optical design to build these creative instruments that allow you to see the motion of blood cells, neurons, and immune cells inside the living eye.

What is the value of being able to make images at the single cell level?

To start, we can learn about the eye itself. The eye is this beautiful window in which you can peer into the human body. It’s the only place you can non-invasively image, using only light, tissues that are more than skin deep. More than that, the eye is an extension of the central nervous system – the back of the eye is actually an extension of the brain. These tools give neuroscientists and ophthalmologists a fantastic view of what different cells are doing inside a living body without ever needing a scalpel. You can get high-resolution images of this part of the brain by building on the ophthalmoscope, a state-of-the-art instrument that eye doctors use every day to image the back of the eye.

How so?

We are using is something called adaptive optics. It has its origins in astronomy. To image distant objects extremely far away, telescopes have to look through the Earth’s atmosphere, which has aberrations that blur the image. Adaptive optics can take into account those effects, giving astronomers a much clearer image. In the late 1990s, researchers at the University of Rochester recognized that similar phenomena that blur objects far away like stars and galaxies also work to blur the image at the back of the eye. We harness adaptive optics technology to measure and correct for the blur at the front part of the eye so we can clearly see the cells at the back of the eye at single cell resolution.

What are some of the biggest challenges of trying to apply this technology to eye imaging?

To start, the eye is constantly in motion. If you pay attention to someone’s eyes—really look at what they are doing—you’ll recognize that the eyes never stop moving. They even have tremors and micro-saccade movements that are almost imperceptibly small. Imagine if you had a microscope and someone just kept moving the slide around as you were trying to take an image of it. It’s a big challenge, and even more so in people who have degeneration in their vision because these motions are often exacerbated. My laboratory has worked on a number of ways to mitigate that. One is to image very, very fast so we can overcome that motion blur.

The other challenge we face is that, thanks to millions of years of evolution, the cells at the back of the eye are naturally clear. It makes sense: The photons have to travel through the front of the eye and go all the way through the retina before they strike the photoreceptors. Having translucent retinal cells helps the light get to where it needs to go. Those cells are kind of like jellyfish in the ocean; light can pass readily through them, but makes them challenging to image. To manage that, my lab has developed the ability to image those translucent cells with high contrast using something called phase contrast. This work was actually one of the publications that resulted from our Dana Foundation grant.

The last problem may not be as apparent but, even when we do get cool images of immune cells working in the back of the eye, the next question is how to make sense of that information. The ability to see these things is a major innovation but what does it mean for the ophthalmologists and optometrists who are looking at that data? What does it mean for the patient sitting in front of them with a disease? To that end, we are building tools to automate measurements of things like how fast the blood cells are moving and how many immune cells are in the picture. Having that kind of tool available can help clinicians make sense of what is in the images and what it may mean for the patient.

Demonstration of image post-processing: Top row, cluster of infiltrated immune cells 6 hours after injection. Bottom row, tissue resident cell in healthy retina adjacent to a retinal capillary. (1) Raw adaptive optics scanning light ophthalmoscope (AOSLO) corrected 796 nm phase contrast imaging of C57BL/6J mouse retina. Real-time video obtained at 25 fps acquisition, showing movement from respiration and cardiac motion. (2) Custom frame-registration. (3) Application of 25 frame temporal averaging and accelerated time-lapse. Scale bars = 10 μm. (video from Joseph, et al. eLife 2020)

Much of your work focuses on using these images to better understand and diagnose diabetic retinopathy. Can it be used for other eye diseases? Beyond the eye?

First, we’ll focus on the eye. But I think the most exciting direction for this research is actually beyond the eye. Most of the tools we’ve developed can image blood cells and the smallest of capillaries. These are the microscopic vessels that are a tenth to a hundredth of the thickness of a human hair. Those vessels, of course, are where a lot of retinal disease first originates. They’re of high importance to clinicians because they want to study what’s happening at the capillary level—these structures that deliver the metabolites, glucose, and oxygen—but they also are the superhighway for getting all the waste products out of the tissue. The focus of my group, so far, has been on studying diabetic retinopathy, which is a vascular-associated disease that affects the eye. But it’s just one of many diseases that can rob a person of vision.

When you consider the major forms of retinal disease that have a vascular component: glaucoma, age-related macular degeneration, stroke, and diabetic retinopathy, we quickly realize that imaging and understanding vascular perfusion is essential to understanding not only the disease at hand but also better understand its initiation, progression, and, hopefully, its favorable response to therapy. We’re very excited to have built a tool that was created to assess one disease but may have applications for many others.

Those applications may go far beyond eye disease. Nowhere else in the human body can you peer in non-invasively to get a glimpse of the cells of the central nervous system. As a neuroscientist, I’m clearly biased, but this tissue is potentially the most important tissue in the human body. You can replenish skin cells and liver cells. You can even regenerate portions of kidney and pancreas. But neurons are precious tissue we need to preserve. The eye could give us a way to image those tissues so we can understand how they change as we age.

For example, the eye could be a clever place to look for signs and symptoms of a stroke. The strokes that take you to the hospital are often big events due to an clogged or bleeding blood vessel. They can lead to life-changing outcomes. But there is some belief that, as we age, there’s an accumulation of little mini-strokes. You can imagine that this accumulation over a lifetime may contribute to cognitive decline. We could use these adaptive optics to study these things  because we can study those single capillaries and see what is happening in them as we get older.

How do you plan to follow this work?

There’s that old adage: a picture is worth a thousand words. A video of what’s happening at the single cell level is worth a million times that. You don’t have to be a mathematician, engineer, or scientist to understand and appreciate what these images are showing us. But the next steps, for us, will be taking these images into a place where we can easily use that data. We want to go beyond, “Gee, this looks really cool!” and explain what the images actually mean. Right now, we are looking at fairly basic questions: How fast are the cells moving? What is the velocity of the cells as they are flowing through the vessels? What is the shape and identity of different immune cells? We need to be able to do that before we can use this data to understand the progression and accumulation of these cells and how these different cells may be talking to each other inside living tissue.

To do this, we will use deep learning and computer vision techniques. We think, using these kinds of algorithms, we can start to understand the behavior of these cells and get an idea of whether they are doing a helpful job or doing something hurtful to the tissue. By studying inflammatory cells, for example, we will have a powerful tool to help us not only understand what is happening in the eye and central nervous system in health and disease – but also something that can help us test cutting-edge treatments for different diseases in the future.  Immunotherapy, for example, which requires the study of how immune cell behave to attack cancer cells, could be an exciting avenue for this research.  Studying the behaviors of immune cells attacking cancer inside the living body would give this burgeoning field an extra boost.