When the world’s first artificial retina, Argus II, was approved for use in the early 2010s, it marked a huge technological breakthrough. It provided a way to bypass photoreceptors that had been damaged by disease: in a healthy eye, these cells sense light and convert it into signals that are passed to the brain. Instead, the Argus II captured an image using a camera mounted on a pair of glasses. This information was then transmitted wirelessly to a chip implanted within the retina, where it was converted into electrical signals that ultimately reached the brain. Yet the improvements to vision it offered were modest – broadly, patients could distinguish simple shapes such as horizontal lines and sometimes large-print letters – and the device has since been phased out.
A handful of other solutions available today follow the same rationale: solid chips are implanted into the retina that receive information about an image, which is converted into electrical signals. Here, each electrode – which is typically about 50–500μm in diameter – acts as a pixel. That means that the number and size of electrodes on the chip determines the visual resolution.
But, again, these devices are far from perfect. For one, there’s limitations on how many electrodes you can pack onto a chip. It’s also challenging to ensure the electrical signals hit their target cell precisely, without accidentally stimulating others around it – thereby muddling the information that reaches the brain.
And because these implants are solid, they’re fixed in place and can’t move dynamically with the eye. That means they produce a kind of tunnel vision. “The most successful existing device has a very tiny view field,” says Chen Yang, professor of chemistry and electrical and computer engineering at Boston University. “It can mainly help patients to read… but they could only read maybe two or three letters within the view field.”
Labs like Yang’s, along with a handful of biotech companies, are working on solutions that aim to more reliably reproduce natural sight in people with degenerative retinal diseases. And within the next few years, they hope to be trialling them in humans.
Stimulating the right cells
When photoreceptors convert light into neural signals, these are passed on to other cells – bipolar cells and then ganglion cells – before they reach the brain. It’s these latter two cell types that artificial retinas largely aim to target as they mostly survive the degeneration process.
EJ Chichilnisky, professor of neurosurgery and ophthalmology at Stanford University, is focused on ganglion cells. There are about 20 types of ganglion cells that each detect different aspects of a visual and carry this information on to the brain. For instance, some convey colour while others are sensitive to flickering light.
Chichilnisky’s lab is developing an electronic implant that recreates this natural process: it stimulates the right ganglion cells according to the attributes of a visual. “You can think of it as conducting an orchestra,” he explains, where each cell type represents a different instrument. By understanding a person’s unique arrangement of ganglion cells, you can see how the orchestra is laid out and ‘conduct’ (i.e. stimulate) them to play their part in the correct timing and order to recreate the image.
To do that, the device first records the spontaneous activity within the retina to identify which ‘instruments’ are present and in which position. The next step is to figure out where these cells are in relation to an electrode grid, which is implanted in the retina. The chip receives wireless signals from a camera within a pair of glasses, which captures the visual. Then, you stimulate the grid to confirm which electrodes are activating which cell type. And after that, you’re ready to go, Chichilnisky says. “You’re ready to reproduce the patterns of activity.”
But how to avoid accidentally triggering offtarget cells and axons, which run over the surface of the retina? By using tiny needle electrodes, Chichilnisky shares, “that can go past the axons into the layer of ganglion cells and stimulate them without, hopefully, stimulating the axons.” These electrodes are about 10μm in diameter, roughly the same size as their target cells.
However, because the chip is solid, Chichilnisky acknowledges that the resulting field of vision will be small – at least, initially. “What we want to see is that we can create very high quality, tiny tunnel vision, and if we can demonstrate that in animals and humans, then we have good reason to put all our effort into developing a much larger implant.”
From light to ultrasound
What if we avoided solid chips altogether? After all, these limit how many electrodes – i.e. pixels – can be packed on. Yang’s lab is developing an alternative: a soft, photoacoustic device that aims to deliver high-resolution visuals. “It’s a piece of material that absorbs light, and the intrinsic material can actually convert the light energy into mechanical energy, in the form of a small ultrasound field,” she explains. These ultrasound waves then stimulate neurons that generate signals that reach the brain.
The material can detect very small areas of light, Yang shares. “We use infrared light that we can focus to less than one micron.” This allows much more pixels to be loaded onto the material: the ultrasound field generated is about the same size as each spot of detected light – and the neurons stimulated by the field define the pixel size. “We have the foundation to be able to incorporate pixel numbers ten or 100 times more than what’s available in the existing technology,” she says.
To put this in perspective: the human eye has roughly one million pixels, Yang explains. With “existing technology we have approved by the FDA, their pixels are between 60 and 400”. Plus, having a soft material could allow for a wider range of vision than solid devices, as it can cover more of the retina. “It can actually be made to cover the entire retina, if needed,” adds Yang. “We’re confident this will have the potential to provide a larger view field.”
Based on where their implant is placed, Yang’s team believe that it stimulates bipolar cells. In rats, they’ve shown that their device can cause changes in brain activity that indicate the visual pathway has been activated. But, notes Yang, that doesn’t tell us what the rats are actually seeing. In a few years’ time, the team hopes to trial their device in human patients.
Proton-pumping protein
Another school of thought focuses on using lightactivated proteins, like bacteriorhodopsin, to mimic the light-absorbing properties of photoreceptors. Biotech company LambaVision is developing one such implant: a scaffold coated with bacteriorhodopsin.
Bacteriorhodopsin is a proton pump: that is, it captures light energy and uses it to transport protons from one side of the protein to the other. In response to light stimulation, “the protein pumps protons, so hydrogen ions, to the bipolar and ganglion cells. With hydrogen ions, you’re changing the pH around those cells that’s going to be picked up by receptors,” explains LambaVision’s CEO, Nicole Wagner. “That’s what sends a signal to the bipolar cells, the ganglion cells and eventually the brain.”
The scaffold itself is a lattice that has alternating layers of the protein and a polymer deposited on top – a process that’s done 200 times. “Think about painting a wall: the more layers, the more light you absorb,” Wagner says. “And the greater the signal that you get to send to the back of the eye.”
The team expects that this will bring on better results than solutions currently on the market. With some of the electrode-based prosthetics out there, patients are able to sort high-contrast socks or walk along a large black line, Wagner says. “We’re expecting higher resolutions than that… we would expect a patient to be able to do the daily activities of life.” Like watching a baseball game on television, for instance.
Plus, this strategy doesn’t rely on singular connections between an electrode and neuron, as if plugging an electrical cable into a wall socket. Instead, it aims to mediate the ion environment, Wagner shares. This would allow for a wider range of signals to be collected and passed on. “The analogy is Wi-Fi… you’re actually changing that whole environment so you don’t need a one-to-one connection. You’re actually getting a greater reach than some of those electrode-based prosthetics.”
Studies on rats with retinitis pigmentosa showed that the implant could stimulate ganglion cells, though we don’t yet know what sort of visual resolution this could bring on – nor how wide the field of vision would be.
Human trials on the horizon
There’s a lot of activity and different approaches within this field, notes Wagner – after all, we still don’t have a cure for blindness. But soon, we’ll hopefully see some artificial retinas in human trials: Chichilnisky aims to do it within four years, while Wagner hopes to get there in three.
For Yang, it’s the real-world impact these devices could have that keep her team motivated. She shares an anecdote: when in Paris to visit their collaborator lab, Yang and her colleagues were having lunch in a small, cosy restaurant. A group of blind people entered, but because the space was too narrow for them to use their canes, they walked in single file with their hands touching the person in front of them. “I think that was a moment for the entire team,” she says. “The patients that, hopefully, we will be able to help in future are sitting right next to you.”
