“We want to galvanize people’s imaginations,” says Miguel Nicolelis, the Brazilian neuroscientist at Duke University who is leading the Walk Again Project’s efforts to create the robotic suit. “With enough political will and investment, we could make wheelchairs obsolete.”
Mind-controlled leg armor may sound more like the movie “Iron Man” than modern medicine. But after decades of testing on rats and monkeys, neuroprosthetics are finally beginning to show promise for people. Devices plugged directly into the brain seem capable of restoring some self-reliance to stroke victims, car crash survivors, injured soldiers and others hampered by incapacitated or missing limbs.
A sip of water
Nicolelis is a pioneer in the field. In the 1990s, he helped build the first mind-controlled arm. Rats learned that they could manipulate the device to get a drink of water simply by thinking about doing so.
In that project, an electronic chip was embedded in the part of each rodent’s brain that controls voluntary muscle movements. Rows of wires that stuck out from the chip like bristles on a brush picked up electrical impulses generated by brain cells and relayed those signals to a computer.
Researchers studied the signals as the rats pushed a lever to guide the arm that gave them water, and they saw groups of neurons firing at different rates as the rats moved the lever in different directions. An algorithm was developed to decipher the patterns, discern the animal’s intention at any given moment and send commands from the brain directly to the arm instead of to the lever. Eventually, rats could move the arm without pushing the lever at all.
Using similar brain-machine interfaces, Nicolelis and his colleagues learned to translate the neural signals in primate brains. In 2000, they reported that an owl monkey connected to the Internet had controlled an arm located 600 miles away. Eight years later, the team described a rhesus monkey that was able to dictate the pace of a robot jogging on a treadmill half a world away in Japan.
Small groups of neurons, it seemed, were surprisingly capable of communicating with digital devices. Individual cells learn to communicate with computer algorithms more effectively over time by changing their firing patterns, as revealed in a study of a mouse’s brain published last year in Nature. “You can count on this plasticity when designing a prosthetic,” says Jose Carmena, a neuroscientist at the University of California at Berkeley. “You can count on the brain to learn.”
Counting on the brain
Capitalizing on that adaptability, several human quadriplegics have received implanted brain chips in FDA-approved clinical trials. One of the first was Matt Nagle, who lost the use of his extremities after being stabbed in the spine. With the aid of electrodes placed in his brain at Brown University in 2004, he learned to raise, lower and drop a piece of hard candy using a primitive jointed arm not connected to his body.
In a widely publicized demonstration of that system, now owned by a company called BrainGate, a 58-year-old woman paralyzed by a stroke sipped a cup of coffee last year using a five-fingered robotic arm — again, not attached to her body. Despite the slickness of the presentation, however, the woman actually had little control over the aesthetically pleasing arm. Her thoughts triggered preset choreography.
“What she was controlling was really simplistic, really rudimentary,” says Andrew Schwartz, a neuroscientist at the University of Pittsburgh.
His team’s robotic arm, showcased in the Feb. 16 issue of the Lancet, offers much more freedom, as well as greater agility and speed. Funded in part by the U.S. military and built at the University of Pittsburgh, the freestanding mechanical limb sports a wrist that bends and rotates. Altogether, it reproduces seven of the 20 to 30 types of motion possible with a human arm.
Jan Scheuermann, a 53-year-old with a rare degenerative disorder, named the arm “Hector” and learned in a single day to move it around like a claw in an arcade game. After 13 weeks of training, she mastered the fine control needed to grasp objects and stack cones. Fulfilling a longstanding goal, she fed herself a chocolate bar — and followed it with some string cheese and a red pepper.
“It’s just the tip of the iceberg of the many tasks that Hector and I will be doing,” she said in a videotaped interview distributed by the university.
To achieve this dexterity, Schwartz’s team had implanted two chips in Scheuermann’s brain instead of the usual one. The duo monitored about 200 neurons at once, more than ever before. More neurons communicate more information, helping to clarify the brain’s desires.
But even hundreds of cells may not be enough to allow someone to control two mechanical limbs at once — the device that scientists hope to showcase at the World Cup. “You really need to reach thousands of neurons,” says Nicolelis. That’s why his team is developing a new kind of electrode that branches like a tree, covering a larger volume of the brain. Made of a flexible plastic that conducts electricity, the electrode can monitor nearly 2,000 brain cells in a mouse.
The more neurons, the better
Upping the number of neurons might also help extend the impractically meager lifetimes of neuroprosthetics. When electrodes are left in the brain for years, the signals that they detect degenerate, possibly because cells die or move out of range. Accessing thousands of neurons might compensate for those that disappear.
Other technologies might minimize that deterioration. At Washington University in St. Louis, Daniel Moran places electrodes on the surface of the brain instead of implanting them within it. Small holes must still be drilled into the skull, but electrocorticography, or ECoG, leaves the thick membrane encasing the brain intact.
“ECoG is safer, and it’s more durable,” says Moran. “I have monkeys that have been implanted for over two years, and the signals are better now than they were at the beginning.”
Because Moran’s electrodes are positioned farther from the action, they pick up less detail; it’s like someone trying to listen to a conversation through a door. But with amplifiers turning up the volume, Moran can extract enough information to allow a monkey to move two cursors on a screen at once. In the first human test, reported Feb. 6 in PLOS One, a quadriplegic learned to move a computer cursor within 28 days.
Even as scientists explore the brain’s ability to make prosthetics move, they’re thinking ahead to the next step. Some want the brain to receive messages from prosthetics that feel.
“Our dream is to create a hand like the one Luke Skywalker had,” says Silvestro Micera of the Swiss Federal Institute of Technology in Lausanne.
His team’s artificial hand, intended to be worn by amputees, connects not to the brain itself but to peripheral nerves near the end of a severed arm. Flexible electrodes threaded into those nerves provide a two-way street for information. Brain signals traveling outward can close the false hand’s five fingers, pinch the index finger and thumb together and wiggle the little finger. Sensors on the prosthetic’s palm send information back to the brain, providing a sense of pressure.
In Nicolelis’s lab, monkeys can feel virtual objects displayed on a computer screen when areas of the brain associated with the sense of touch are stimulated. The blueprints for next summer’s soccer exoskeleton include sensors that will provide an artificial skin for its human wearer. With the world watching, Nicolelis hopes not only that his bionic teenager will be able to feel the ball but also that disabled people everywhere will feel a sense of hope.
Powell is a freelance science reporter based in the District.