![]()
A new brain implant has successfully monitored the activity of individual neurons in human patients. Researchers report in Science Advances that the early proof-of-concept could someday help improve decades-old technology for epilepsy treatment.
The CDC’s 2013 estimates state that about 2.9 million people have active epilepsy—a seizure-inducing brain disorder that can occur with physical trauma such as strokes or tumors. About one third of people with epilepsy develop refectory epilepsy—a condition untreatable with drugs—according to the John Hopkins Medicine Health Library. In these cases, doctors can only try to cut out parts of the brain where seizures appear, but it’s challenging to pinpoint the targets.
Historically, doctors have monitored suspected parts of the patient’s brain to select where to perform operations. Often, this process requires implanting an uncomfortable, rigid electrode grid into the brain for one or two weeks.
It’s “very expensive,” says György Buzsáki, a neuroscientist at New York University Langone Medical Center. In 2015 research published in Nature Neuroscience, he and his colleagues created something that could theoretically be more comfortable yet more accurate: a conformable, non-inflammation-causing electrode grid that can monitor the activity of single neurons instead of only groups of neurons as happens with existing technologies. The catch? Then and now, researchers have only tested grids like this in animals such as mice.
There’s a big difference between animals and humans, Buzsáki says. Among the brains of species, “electricity is electricity,” but safety and reliability can have diverse meanings. “It doesn’t matter how sure you are in anything—you still have to test it in a patient,” he says.
Researchers have previously monitored the brain activity in humans by either placing electrodes on the scalp and looking at electroencephalogram signals or draping electrodes as a mesh over the cortex’s surface to view electrocorticogram records. Buzsáki says “each of these methods have their own advantages,” in terms of non-invasiveness, but the downside is that the coverage area is too large. Getting to more precise coverage usually requires invasive electrode grids that irritate the brain tissue.
In the new research, Buzsáki and his colleagues created a biocompatible, comfortable brain implant that can monitor the firing activity of individual neurons at a much higher resolution. He compares the advantage of detecting individual neurons to picking an individual instrument out of an orchestra—without good spatial resolution, “all you get is an envelope of the music.”
They call the technology “NeuroGrid.” It’s a 4-micrometer-thick parylene grid—parylene is FDA approved—with 120 conductive polymer PEDOT/PSS electrodes, reaching a coverage area of about 420 mm2 of neocortex. It has very similar properties to cellophane wrapping paper, Buzsáki says, because it sticks to wet surfaces such as the brain.
To test the grid, the researchers used human patients undergoing brain surgeries. Surgeons temporarily inserted the grid on the brain’s surface, and the researchers recorded what happened.
For every 10 electrodes in the NeuroGrid there is a single wire that connects to the lone silicon chip sitting outside the brain acting as an amplifier. The chip compresses the signals and sends them via wire to a computer for decompression and analysis. With this setup, researchers found that they could successfully detect when individual neurons fired.
New York University Langone Medical Center neuroengineer colleague Dion Khodagholy says highly localized performance comes from a combination of how nicely parylene interfaces with the brain, how nicely the conductive polymer tracks electricity, and the silicon-based microfabrication process.
“Hopefully recognizing any pathological activity will be easier,” Buzsáki says.
Mikhail Lebedev, a neuroscientist at Duke University who was not involved in the study but who tracks brain activity for controlling prosthetics, agrees that the tech could “allow [us] to localize the epileptic focus more accurately.” He also sees potential applications in the control of neural prosthetics or sensory feedback with electrical stimulation.
Areas for improvement could be making the power supply and any amplifiers “fully implantable” inside the skull. The team could also try improving the power supply’s efficiency and the electrical signal amplification—the more electrodes used, the more wires required, and the more that noise can weaken signals.
“It’s not instant panacea for a clinical practice,” cautions Nitish Thakor, a neuroengineer at John Hopkins University who was also not involved in the study. He also studies ways to track neuron firing activity for such applications as prosthetic control.
First, there are the usual multi-year regulatory requirements for getting such a technology approved for clinical use. Then, although the benefits from recording individual neuron firing are clear, he says the tradeoff of the new technology is that it would require even more electrodes, data processing, and computer system changes to get the same spatial resolution as existing grids. This could raise costs. “Clinically—who’d pay for it?” Thakor asks.
Buzsáki says the next goals are to decrease the NeuroGrid’s size and the number of wires coming out so the researchers could use more grids to gain focused surgical areas while maintaining patient comfort. The researchers also hope to eventually monitor the human brain’s electrical activity over days or even weeks.
“It will pave the way for clinical studies,” Thakor says, “but also offers new challenges.”
