Robo-suit and virtual reality reverse some paralysis in people with spinal cord injuries

Two years ago, just before leading a controversial demonstration at the World Cup in which a paralyzed man in a futuristic-looking exoskeleton controlled by his brain appeared to kick a soccer ball, neuroscientist Miguel Nicolelis hinted in an interview that his research team would aim for something equally impressive to grab the world's attention during the 2016 Summer Olympics in his native Brazil.

In a study published today, Nicolelis makes his bid. He and colleagues reveal that eight Brazilians paralyzed because of spinal cord injuries regained some small but significant sensation and muscle control in their lower limbs after many months of training with the robotic exoskeleton, and by a virtual reality avatar also controlled by brain signals. None of the study participants is close to walking again unassisted, but some can voluntarily move leg muscles and have improved bladder or bowel control, significant quality of life issues for those with severe paralysis.

Nicolelis, who is based at Duke University in Durham, North Carolina, calls the results “unprecedented” and suggests that a similar training regimen could one day help many other people with spinal cord injuries or those impaired by stroke. Other researchers cautiously agree, but stress that more meaningful recoveries may demand a combination of approaches. The new work is “an advance. I wouldn’t characterize it as a dramatic breakthrough,” says University of Toronto in Canada neurosurgeon Michael Fehlings, who specializes in spinal cord injuries.  “The improvements seen are modest. It’s more exciting from a proof of concept standpoint.”

The new work, published in Scientific Reports, comes out of the Walk Again Project, a large international research consortium funded in part by the Brazilian government and based in São Paulo. A celebrity in his home country who has clashed with some in the Brazilian science community, Nicolelis organized the project. Its original goal: developing brain-machine interfaces, which use implanted electrodes or noninvasive electroencephalogram (EEG) sensors to detect brain signals and translate them into commands for prosthesis or other devices, to the point that paralyzed people could control a motorized exoskeleton—essentially a robotic suit that keeps the body upright and allows it to move.  

The internationally televised World Cup demo failed to impress many in the scientific community, who noted that it was hard to discern how much the man actually controlled his exoskeleton. But it was far from the end of the project. The consortium ultimately recruited eight people whose lower limbs had been paralyzed for 3 to 15 years because of spinal cord injuries and guided them through three levels of training, which involved hour-long sessions twice a week. The study participants started by donning a virtual reality headset  and trying to mentally control the walking of an avatar with their brain signals, as detected by EEG sensors in a mesh cap. Next, they graduated to a treadmill system in which their body weight was supported and their brain signals controlled a robotic walker system that moved their legs. Finally, they practiced with the custom-built robotic exoskeleton shown off at the 2014 World Cup. One apparent key to the regimen’s success, Nicolelis says, is that each level of training provided neurological feedback to the brain through a “tactile shirt” that made each participant’s arms vibrate whenever they landed a step.

The neurological improvements came gradually. Some of the trainees noticed changes after more than 6 months of training, but it wasn’t until neurological exams conducted after 1 year that the research team verified some restoration of muscle control and sensation in all eight patients. At that point, the team upgraded the status of four from complete paralysis in their lower limbs to partial. Nicolelis says that since then, three of the others have also improved enough to be upgraded similarly; the only one who didn’t had to move away and discontinue training. Nicolelis notes that his team is preparing another publication on the subsequent 18 months of training and ongoing improvements they observed. “The recovery has not ended,” he says.

Although some spinal cord injuries completely transect the nerve bundle, most involve a crushing of the cord, which may leave some viable, through nonfunctioning, nerve connections. Physicians and researchers have long sought drugs or therapies, ranging from electrical stimulation to whole body harness walkers, that could rehabilitate those connections, but they have achieved few signs of success in people. As a result, Nicolelis says, his team had not expected to see much neurological  recovery from the brain-machine interface training given how long ago the injuries took place. “Something may have survived the original trauma. We may have rekindled these remaining neurons,” he said during a call with reporters. The team also documented changes in participants' brain EEG patterns that they interpret as cortical regions reorganizing to reassume some control of lower limbs.

Earlier this year, another team reported using a brain machine interface to allow a quadriplegic to move his finger and arms, but that system directly activated arm muscles with brain signals detected by implanted electrodes, bypassing the damaged spinal cord. Nicolelis’s training strategy is less invasive but would likely be hard to fully extend to a large number of paralyzed people given the expense of the custom exoskeleton. The first two stages—the virtual reality and treadmill systems—could be more easily adopted, although it’s unclear how much recovery they promoted. Fehlings suggests that brain-machine interfaces could one day be combined with several other treatment strategies—such as stem cells and drugs—to produce much greater overall benefit for people with damaged spinal cords. No single strategy will be enough to overcome severe paralysis, he emphasizes, so “it’s appropriate to be cautious.”