Four scientists who pioneered the development of adaptive optics (AO) technologies for imaging the human retina have been awarded the 2024 Rank Prize for Optoelectronics. The winners – Junzhong Liang, Donald Miller, Austin Roorda and David Williams – invented instruments that use AO to capture high-resolution images of the living retina and provide new insight into the structure and function of the human eye.
AO was originally developed for use in astronomy, to eliminate atmosphere-induced blur in images from ground-based telescopes. It works by measuring distortions in a reflected wavefront using a wavefront sensor, and then compensating for these distortions with a wavefront corrector, which is often a deformable mirror.
In 1997, Liang, Williams and Miller demonstrated that AO can also be used to correct for distortions caused by imperfect optics within the human eye. Using AO, they created a retinal imaging camera with unprecedented resolution, enabling clear imaging of individual photoreceptor cells in the living human retina. Two years later, Roorda and Williams used this instrument to produce the first-ever images showing the distribution of the three types of cones in the human retina.
According to Donal Bradley, chair of the Rank Prize Optoelectronics Committee, the prize recognizes the winners “seminal contribution to imaging within the eye that opens new opportunities to understand this complex optical instrument and to improve eyesight through precise interventions”. Tami Freeman spoke to two of the winners to find out more.
Since its invention, how has AO impacted the field of eye imaging?
Donald Miller AO is the only technology that allows the visualization of individual retinal cells in a living eye. And because disease and pathology start at this cellular level, that’s the level we ultimately want clinicians to operate at, for earlier diagnosis and more effective treatments.
As one example from my own lab, we’ve recently been looking at the impact of glaucoma, one of the leading causes of irreversible blindness in the world, on retinal ganglion cells – the primary cell type that dies in this disease and which line the top of the retina. While effective treatments exist, the disease is unfortunately hard to diagnose early until significant damage has occurred. With AO, we can now, for the first time, monitor individual retinal ganglion cells and track them over time in these patients.
Using AO combined with optical coherence tomography (AO-OCT), we have found that, even in eyes under treatment, we see subclinical loss of cells. That’s important because clinicians can now use these cellular-level measurements to better establish whether or not their treatment is working. It also offers considerable potential for testing the efficacy and safety of new neuroprotective and regenerative strategies. The visualization of retinal ganglion cells in human subjects has only become possible within the last few years – we are entering a really exciting time.
Austin Roorda As treatments become available for the major blinding eye diseases, like diabetes, glaucoma and macular degeneration, we can now use AO to assess how effective they are. But there are other inherited retinal diseases due to gene mutations for which very little is known. In those rare diseases, previously the only way to see what was happening on a cellular scale was to wait for a donor eye and look at it under a microscope. AO has opened up the ability to examine the retina on a microscopic scale in these patients. Treatments such as gene therapy are on the horizon that could potentially cure or halt these inherited diseases. AO is poised to play a key role in that process – to understand how the mutation affects the retina, assess the state of the retina, predict the prognosis if the patient undergoes gene therapy, and then measure the effectiveness of that therapy.
How has AO technology progressed over the last 25 years?
AR AO was originally constrained by the technology that was available, which was largely developed for the field of astronomy. So the deformable mirror was big and wasn’t suited to the eye. Over the years, when companies started recognising the potential of AO in other fields, including ophthalmoscopy, they started to build wavefront sensing devices and wavefront correctors (the deformable mirror) that were a lot better suited to applications in the human eye.
DM When we first developed the AO system, we made a lot of guesses: what type of wavefront correction to use, what wavefront sensor, the loop speed and so on. In the next five to 10 years there were a lot of improvements in our understanding of the spatial properties and temporal dynamics of ocular aberrations. These then defined the AO components: how many actuators you need in your wavefront corrector, what the stroke [actuator displacement] should be, how many sampling points you need across the pupil, and how fast the AO system should go. Those have all been optimized over the years.
For example, the wavefront corrector we used in 1997 had 37 actuators that push and pull on the back surface of the mirror to warp its shape, and it would give four microns of stroke. The ones used today have close to 100 actuators and give an order of magnitude more stroke, which is important because the eyes have severe aberrations; that’s made a big difference.
AR Now, when you use AO, you push a button and it runs automatically at anywhere from tens to hundreds of hertz. Before, we had to take a picture, a map of the of the eye’s aberrations, and scrutinize it to make sure there weren’t any errors in the initial image analysis. Then you would push the next button to apply that shape to the mirror. So the user was an integral part of the closed-loop AO system. It was fun, but it was slow.
Initially, Don, David and Junzhong built a standard flood-illumination camera that would look at the retina through an AO system to reveal the microscopic structure. Later, I incorporated AO into a scanning system to create an AO scanning laser ophthalmoscope (AOSLO) that can record video of the retina and perform depth sectioning. That’s an entirely new AO imaging platform. Other researchers have incorporated a type of phase contrast imaging that can visualize otherwise transparent cells in the retina, and in David’s group they are performing fluorescence imaging in animal eyes.
What’s your current main area of research?
AR If there was a theme for what I’ve been doing for the last 15 years or so, it’s structure and function. It turns out that our AOSLO imager is also the world’s best eye tracker. You can track eye motion very quickly and accurately because you can see the movement of single cells in the back of the eye. We took this a step further, using the scanning laser system not only to image the retina, but to control the placement of images onto the retina on the scale of a single cone.
We’ve been measuring functional properties in living humans. If you were in the device, I could deliver flashes of light into individual cones and ask whether you could see them or what colour you see. Early on, we mapped the cone mosaic, that was one of the big AO-enabled discoveries. Now we can take that cone mosaic and start asking questions about basic retinal circuits or the fundamental properties of human colour vision. We’re doing the same in eye disease. If we look at an array of cells in a patient and it doesn’t look normal, we’re interested in the functional consequences – not just seeing the structure of that diseased retina but asking about the visual outcomes.
DM We’re also focused on structure and function, but using AO-OCT. The big advantage of OCT is its axial resolution, which lets you section out whatever depth in the retina layer you want to visualize. Cones are very bright and high in contrast, but other cells tend to be much harder to image as they reflect a lot less light back. We’ve made quite a bit of headway using AO-OCT to image these other neurons in the retina at different depths. It was a big step to be able to image retinal ganglion cells, as they are highly transparent and have very low contrast.
We’ve also been using AO-OCT to look at function within photoreceptors. In 2000, Austin and David had developed their pioneering AO retinal densitometry method for cone classification. Twenty years later, we can use the phase information provided by AO-OCT to measure subtle changes in the elongation of these photoreceptor cells when stimulated by different colours of light. That turned out to be a much more accurate and far less time-consuming way to do cone classification and is a good example of the evolution of AO imaging technology.
How do you see the field of AO evolving in the future?
AR In my lab, we focus a lot on subjective measures of function, such as eye movements, acuity and colour vision. But I would envision that as AO techniques evolve, we’ll be able to measure functional properties of most cell classes in the retina. Right now, Don has generated beautiful images of ganglion cells using AO-OCT. These are the last cells before the signals from the retina reach the brain, so it’s a class of neurons whose function we’re very interested in. Using phase methods, or methods we can’t even conceive of right now, we may be able to measure the functional properties of those and other neurons in the retina.
Sub-diffraction imaging of the human eye visualizes photoreceptors with unprecedented detail
David, Don and I are immersed in basic research, but there are a lot of other people thinking about how to get these systems into the clinic. AO is not easy and it’s not cheap, it’s a complicated technology so the path to the clinic is not easy. There are a few companies now that will sell AO imaging devices, but they’re not used routinely by any stretch.
DM The field of AO waxes and wanes between trying to improve AO performance versus making AO more accessible and commercially viable. In our labs, we’re trying to achieve the very best performance, correcting aberrations and getting sharper images for research or clinical purposes. But there’s a whole other side pushing this technology to make it more compact, cheaper and more automated. The real potential is marrying AO with SLO and OCT for commercial use. I think that it’s just a matter of time.
- Founded in 1972 by the British industrialist and philanthropist Lord J Arthur Rank, the Rank Prize is awarded biennially in the fields of nutrition and optoelectronics. The Prize will be awarded formally on 1 July 2024.