Within reaching distance of mind-controlled movement
Neural control of robot limbs is tantalizingly close for paralyzed patients. So what’s holding the technology back?
Madhu Venkataramanan • January 20, 2011
When Matt Nagle moved the mouse cursor on his computer screen in June 2004, he made history. When he went on to open e-mails and use a TV remote control, it sent waves of excitement through the neuroscience research community. Nagle’s accomplishments were doubly sensational: not only was he the first almost fully paralyzed person to do this, but he was also the first to do it entirely with his mind.
Desperate to regain normalcy, the 25-year-old paraplegic — a former football star from Massachusetts who was stabbed trying to help his friends in a brawl — agreed to take part in a study using the BrainGate Neural Interface System. After surgeons implanted 96 tiny electrical sensors into Nagle’s brain, a computer decoded the information in his brain cells, allowing him to use a simple detached robotic arm to manipulate a mouse cursor. Within a year, his decoded “thoughts” could make the arm move physical objects, making him BrainGate’s glowing success story.
Almost seven years later, however, the technology is still experimental, and the ultimate goal of restoring movement to deadened human limbs or attached prosthetic arms remains out of reach. To finally achieve it, researchers will need a deeper understanding of how the brain controls limb movements. They’ll also need to learn how to build and implant electrical sensors sturdy enough to last for years in our hostile brain matter. But there has been major progress overcoming these and other barriers recently, and as a result researchers are embarking on a new round of experiments in which paraplegics are using robot arms to do much more complex movements than Nagle could — perhaps even including picking up a cup of water and taking a sip.
“Progress is booming,” says John Donoghue, a neuroscientist on the BrainGate team and director of the Brown University Brain Institute. “But this [technology] takes time and patient research, and we don’t want to push ahead without doing things properly.”
Brain-machine interface research is more than a century old, having started when British and German doctors began stimulating the brain with electricity, according to Daofen Chen, a program director for neurological diseases and motor coordination at the National Institutes of Health. But it wasn’t until a decade ago that this field was transformed by technology.
To explain the high-tech robotic tools we have today, Chen draws a simple analogy: Imagine there is a building with many rooms, filled with people. You’ve been trying to eavesdrop on their conversations for a while, but you’ve never had good microphones to record their voices. Now, advances in nanotechnology allow you to stick in a bunch of small, high-quality microphones into these rooms and record everything. This is what neuroengineers have done by implanting micro-electrodes into a patient’s brain and recording its electrical impulses.
“Putting sophisticated electrodes into the brain was made possible in just the last five years,” Donoghue says. This technical leap, along with our increased knowledge of brain anatomy, is why there have been major advances in the field recently. In October, for example, University of Pittsburgh neuroscientist Andrew Schwartz and his team showed that monkeys implanted with these electrodes could accomplish unprecedentedly complex robotic arm, wrist and hand movements. “We plan to start testing in humans in six months,” says Schwartz.
But the brain is not a simple penthouse suite; it’s a sprawling, opaque building with a secret maze of interconnecting rooms, each of them filled with scores of neurons whose language we still cannot understand, even if we can now “hear” it with implanted electrodes. One of the major obstacles to commercial development of technologies like BrainGate, Chen says, is that scientists don’t know enough about how the brain works. Schwartz agrees: “We don’t understand enough to do dexterous types of movements like typing and piano-playing, buttoning a jacket or zipping a coat.”
The other key problem that goes with implanting anything into the brain is that the human body is very tough on foreign materials. Implanted electrodes become less reliable over time because, Chen says, “as soon as you stick that microphone into the building, our body’s immune system is doing everything it can to get it out.” The only solution is to seal the devices securely — a challenging task because our immune system is designed to mount a fierce and sustained attack on any foreign object.
An additional issue is that the current technology relies on a plug that goes through the subject’s skin and is hooked up to computers on the outside – a set-up that limits a patient’s mobility. “We are working on developing devices that are completely internal and transmit wireless signals to the outside,” Donoghue says.
Building on the original work with Matt Nagle, who died in 2007 due to his sustained injuries, Donoghue’s BrainGate team has a new round of human experiments under way, using more advanced technology. Their subject is a woman who suffered a brain-stem stroke, leaving her unable to speak and paralyzed from the neck down. The mechanical arm she’s manipulating is a sophisticated robot designed by the German Space Agency. The patient has already been able to pick up a glass of water and now the team is awaiting approval from the Food and Drug Administration to let her actually drink it. Because the interaction of humans and robots is still relatively new, such close contact with a robot arm still needs official approval.
“This will be the first time a [paralyzed] human has begun to manipulate the world again, under her own brain power,” says Donoghue.
All three experts — Donoghue, Chen and Schwartz — agree that commercialization is imminent, once FDA regulations are resolved, but that it is going to come in baby steps. “There will be a progression of better and better technologies, with improvement in quality and control, just like cardiac pacemakers,” says Donoghue. It will start off with less invasive and more primitive technology, where they just listen in from outside the brain, he says. These technologies are less desirable because the information they provide does not have enough detail. For advanced movement and manipulation, a more invasive tool is necessary, according to Donoghue.
Today, patients are manipulating detached robot arms, but eventually brain-machine interfaces will allow them to move prosthetic limbs, say experts. So does that mean that someday, paraplegics and amputees will be able operate their own dead or artificially re-attached limbs just by thinking about it? “To physically rewire the nervous system back to the body would be the ultimate ambitious goal,” Donoghue says, “and I believe we are on a path to doing that.”