The system decodes his brain activity and uses it to control his arm muscles, bypassing his injured spine.
On June 13, 2010, college freshman Ian Burkhart was goofing off in the ocean with his friends, when he dove into the wrong wave. It pushed him down onto a shallow sandbar, breaking his neck at the fifth cervical vertebra and instantly paralyzing him. He couldn’t feel his arms or legs. It would be four years before he moved his hand again.
After his condition stabilized, Burkhart moved back home with his family in Columbus, Ohio, and started doing rehabilitation therapy at Ohio State University.
“I mentioned that I was interested in taking part in research because I knew that with the way science and technology were progressing, something would come along during my lifetime that would improve my quality of life,” he says. “I got lucky because it happened right in my backyard.”
Burkhart learned that a local team, led by engineer Chad Bouton and neurosurgeon Ali Rezai, were developing a technological bypass for injured spinal cords. Their “neuroprosthetic” would directly connect the brain to muscles in the arm, allowing paralyzed people to regain control of their own limbs. They needed test subjects, and Burkhart fit the bill perfectly.
In April 2014, the team surgically implanted an array of microscopic electrodes in Burkhart’s motor cortex—the part of the brain that governs his movements. When Burkhart thought about moving his arm, the implant would decode the activity in his neurons and feed the signals to a sleeve of electrodes on his forearm. In theory, his arm would move.
“Surprisingly, it worked pretty much right away,” says Burkhart. In June, after some preliminary tests, he thought about opening and closing his hand. And after years of inaction, it obeyed.
“For years, I had thought about it many, many times, and nothing had happened,” he says. “I wasn’t really able to feel my hand moving but I could see it.” So could the assembled on-lookers: Bouton, Rezai, a few other team members, and Burkhart’s father and sister. “It was a pretty special day,” Burkhart says.
The neuroprosthetic is still limited. It relies on a couple of engineers and a room of computers; “I can’t take it outside the lab or anything like that,” says Burkhart. The system also needs to be carefully calibrated every time it’s used, and it can only be used for a few hours every week.
And yet, in those few hours, Burkhart could move his once-paralyzed hand, rotate his wrist, and pinch his fingers. He could even pick up a bottle, pour its contents into another jar, and stir it with a small stick—a complicated sequence that involves two kinds of grasps (a wide-handed one, and a delicate pinch), and several coordinated movements of hand, wrist, and arm.
“It’s not 100 percent natural but it feels right,” he says.
The field of neuroprosthetics has made huge strides of late. Just ten years ago, the team behind BrainGate (a neuroprosthetic, not a neuroscandal) demonstrated that paralyzed people could used implanted electrodes to control a virtual cursor and a rudimentary robotic hand. A quadriplegic woman named Cathy Hutchinson would later use BrainGate to drink from a cup of coffee, brought to her lips with a mentally controlled robotic arm.
But that approach doesn't restore movement to a volunteer’s own limbs. In Burkhart’s case, “we not only decoded brain activity in the motor area of someone who’s paralyzed, but we also linked those signals back to their body,” says Bouton.
“This is the first time that it’s been done in humans. We allowed Ian to regain movement in real-time, all through his thoughts.”
“It’s an absolutely wonderful step forward in the field of neuroprosthetics,” adds Elizabeth Tyler-Kabara, a neurosurgeon at the University of Pittsburgh.
After his initial success, Burkhart went through 15 months of training, showing up at the lab for three to four hours at a time, three times a week. He’d think about the movements he was trying to achieve, and the team would record the corresponding activity in his motor cortex.
“You really don’t think about moving your hand,” Burkhart says. “Now, I had to focus on it and concentrate. What muscles do I use?”
Bouton’s team used machine-learning algorithms to decode the patterns of brain activity that signified each movement. The electrodes on Burkhart’s arm then recreated those movements by stimulating his muscles in preset sequences. By June, he was moving his hand and wrist. Over the following months, he learned to move individual fingers.
“That exceeded all of our expectations.” says Bouton. “We certainly hoped for that and didn’t know when it would happen.”
More complicated movements, like picking up a bottle, were harder.
“Once he actually started to move, his brain activity changed in a dramatic way,” says Bouton. That’s because Burkhart was adding residual movements in his shoulder to the regained movements in his hands, and those were throwing off the prosthetic. He’d pick objects up, and then drop them again. “We had to develop new methods to learn that brain activity and adapt to it in real-time.”
They are still working to improve every aspect of the neuroprosthetic. The brain implant, for example, uses just 96 electrodes to record from a small number of neurons, out of the millions that are involved in movement.
“We need to develop better sensing tech to listen in on a lot more neurons and improve the signal quality,” says Bouton.
The algorithm that decodes these signals needs to get faster; currently, there’s a short delay between thought and action. “They also had to do calibration on each day of testing,” adds Tyler-Kabara. “Hopefully [the field] will start coming up with ways of making these signals stable enough so that they’re usable over weeks.”
The stimulating sleeve needs work, too. It can’t make Burkhart splay his fingers apart, or bring his thumb to any finger other than the index one, or twist his wrist while pinching something, as if turning a screw.
“It may sound like those are very specific movements but we use our hands in very intricate ways that we take for granted.”
The solution might be to implant electrodes in the arm, rather than using a removable sleeve. The sleeve “is non-surgical, so it makes good sense as a starting point,” says Lee Miller from Northwestern University, but the surface electrodes have to be calibrated each time a patient puts them on. And since they stimulate through skin, they use higher voltages and lose accuracy.
A older neuroprosthesis called Freehand, first tested in the 1980s and since withdrawn from the market, used implanted muscle stimulators controlled by surface electrodes—essentially the opposite of Bouton’s system.
“The control signal from the brain is of higher quality, which is a good move in the right direction,” says Miller, “but they need to improve on their ability to stimulate muscles.”
For Burkhart, now 24, the study is a finite one. The implant is still unproven and unregulated, so it will have to come out at some point. It may then be another decade, perhaps longer, before he can move his hand again.
“I knew going in that it wouldn’t be a lifelong thing,” he says. “But I know that with a lot of people working on correcting these problems, hopefully there’ll be something I can use in my everyday life.”
“Ian has an incredible work ethic and attitude,” says Bouton. “We asked him early on, ‘Why did you decide to enroll?’ He said, ‘To help develop the tech that could help people in the future.’ He’s an inspiration to the entire team.”