The New Reanimators Drive Brain and Machine Merger

Posted by Sam Maddox in Research News on April 18, 2016 # Research, Technology

Quad uses brain signals to play Guitar Hero. Anyone with a spinal cord injury has heard about this, even those who try not to pay attention to “cure” news. Friends and family see this and pass it along, hoping the bionic promise won’t leave any quads behind.

Wait a minute ... did anyone see the other cool reanimation paper that just came out? Two monkeys were trained to pilot little wireless wheelchairs with only brain signals. Not sure how come that didn’t make the papers, especially since the work came from the same publicity-savvy team that orchestrated the 2014 FIFA World Cup stunt – enabling a Brazilian paraplegic in an exoskeleton to “kick” the ceremonial first ball using brainwaves alone.

OK, so we’ve been hearing about brain-computer interface (BMI) studies for a number of years now, and we’ve seen quads high-fiving, turning keys, kicking balls, sipping beer – all just using just brain signals running through computers. You may be wondering, what’s really new?

First, the human study. Recall that we met BMI hero Ian Burkhart two years ago when he was presented to the media after having his skull opened to put a 96-electrode sensor chip in an area of his brain related to hand movement. Yes, the chip picked up signals from his cortex that went via wires in his skull to a computer and thus, when he wanted to, he moved his hand.

The research came from Ohio State University and Battelle Memorial Institute, a big nonprofit research and development group based in Columbus, Ian’s hometown. “We’re basically creating what we believe is going to be a virtual spinal cord,” a Battelle scientist said. But at the time they were not inclined to publish their study.

Now, same guy, same brain chip. Same team. Only this time the work has been published, in Nature, and they’ve got a whole lot more to show. Two years ago the research project was called Neurobridge. Now they call it NeuroLife.

I asked a senior member of the Battelle group to describe the real takeaway from Ian’s story. Here is bioengineer Gurauv Sharma, one of the authors of the study:

Our study is the first demonstration that a paralyzed person can regain voluntary functional control of his own hand using his thoughts alone in real-time. As you will see from the paper he was not only able to control up to six individual finger and hand movements but was also able to combine several movements to perform complex functional tasks that required coordination from both paralyzed and able muscles. To the best of our knowledge this kind of individual movements as well as combination of different functional grips have not been demonstrated before.

The videos, which accompanied many of the news stories, really show what Ian can do. His movements are voluntary but complex; he executes them fluidly. He can pick up and pour a bottle, stir his tea, run a credit card through a scanner, and indeed, he can finger the plastic keyboard of a Guitar Hero video game.

From the paper:

With use of the investigational system, our C5/C6 participant gained wrist and hand function consistent with a C7–T1 level of injury. This improvement in function is meaningful for reducing the burden of care in patients with SCI as most C5 and C6 patients require assistance for activities of daily living, while C7–T1 level patients can live more independently.

Although invasive, the neural bypass system (NBS) provides an advantage over existing functional electrical stimulation systems that utilize low dimensional control signals such as EEG or EMG like the Freehand system. These devices typically allow control over fewer movements than those demonstrated in this study, because of the relatively low information content of their control signal sources compared with intracortically recorded signals.

To allow transfer of this work to other patients, further improvements will be required on the microelectrode technology, algorithms, and NMES. However, the electronic neural bypass presented here demonstrates what is possible in the future and can offer hope for movement restoration to people living with paralysis worldwide.

Ian worked hard for this, submitting to brain surgery, coming to Ohio State three times a week for 15 months to train his mind and code the computer programs to coordinate brain and machine. Alas, it’s a shame that as this experiment winds down this summer, doctors will have to remove his chip, leaving Ian with nothing but the satisfaction of having done his part to move things forward. He says he’s going to miss that right hand.

I’d really like to be able to take the system home with me. Not being able to walk doesn’t really matter to me because you can do a lot from a wheelchair, but if I was able to use my hands I would be a lot more independent than I currently am. But even if it’s something that I can never take home in my lifetime, I’m glad I’ve had the opportunity to take part in this study. I’ve had lots of fun with it. I know that I’ve done a lot of work to help other people as well.

The research group recognizes the limitations of the study. Implanted brain chips are obviously quite invasive, and they don’t last long. Ideally, brainwave activity could be picked up (EEG) without an implant but that’s not optimal for picking up the fine signals for complex motor control -- like trying to hear a conversation from the next room, that’ how the research guys put it.

There are wires attached to Ian’s head when he’s in the BMI system. Wireless has to be worked out. Not being able to get sensory feedback to the fingers and hand is a drawback too. Ian can’t tell between an egg and a marshmallow when he’s ready to grab.

Said Sharma: "Our next goal is to build upon our experience from this study to develop a system that is not only portable, but also something that a patient can take home to assist with their activities of daily living or rehabilitation," he said.

I recommended a narrative on Ian and the Columbus group, with video, that ran in Nature. The NY Times had a good piece, as did the LA Times.

Hey Hey It’s the Monkeys

This study comes from the Miguel Nicolelis group at Duke University. He a Brazilian M.D., Ph.D., and one of the pioneers in BMI; for more than 25 years he has been sampling animal brain signals to move prosthetics, and limbs. In 1990 he and his group got a monkey hooked to the Internet to move a mechanical arm 600 miles away. Nicolelis also has data to address the sensory part of reanimation missing in Ian’s study. So what about these little primates driving wheelchairs?

Two 17-pound rhesus macaques, K and M, were trained to drive little wheelchairs using a typical four-directional joystick to navigate to a reward (a grape). They were implanted with several nerve chip arrays in both sides of their cortex. Brain activity was recorded and coded, fed via a wireless system to a computer and back to the wheelchair. After some practice, K and M were able to pilot their chairs with only brain input, and get the grapes.

Here’s the paper, available in full, with supporting video: Wireless Cortical Brain-Machine Interface for Whole-Body Navigation in Primates.

From the paper:

These BMI designs are similar to those previously used to reproduce arm movements, only replacing the end actuator with a robotic wheelchair. .... Here we hypothesized that kinematics of whole-body movements could be extracted directly from sensorimotor cortical ensembles, and utilized to control a BMI for wheelchair control. As a first step towards the development of a clinically relevant device of this type, we utilized large scale recordings from multiple cortical areas obtained by our recently developed multichannel wireless recording system to enable BMI control over whole-body navigation in a robotic wheelchair.

It is important to emphasize that, different from previous BMI studies in which animals were trained by simply observing movements of a virtual actuator on a computer screen (e.g. virtual arm, legs, etc.), here the entire animal’s body was passively transported by the actuator (wheelchair) during training of the BMI decoder. This implies that an integration of vestibular, proprioceptive, visual and auditory inputs influenced the choice of our optimal decoder settings, something that had not been tried before in the BMI field. Given the nature of our task, it is very likely that multimodal sensory inputs continued to influence neuronal ensemble activity during BMI control.

Nicolelis thinks that BMI may be doing more than just providing on-off switching to motor commands. He hypothesizes that the nervous system itself is adaptive and that BMI brings about “widespread cortical plasticity:”

... the current results promise much more than a mere demonstration that monkeys can control whole body navigation. Indeed, the finding that cortical ensembles can adapt to a whole-body navigation task propels BMIs to a new dimension and creates innovative avenues for exploration of this approach’s clinical relevance in the future. In this context, the present results support our recent clinical observation that when paraplegic patients are subjected to intense BMI training, they not only became capable of regaining walking, using a robotic exoskeleton, but they also can exhibit signs of partial neurological recovery of sensorimotor and autonomic functions.

Based on our experimental and clinical observations, we raise the hypothesis that BMIs can lead to partial neurological recovery or even augment brain function because their chronic and continuous use may trigger widespread cortical plasticity and the emergence of new cortical representations. As such, BMIs will likely have a profound clinical impact in the future.

On the topic of implanted sensors versus the kind of EEG-based sensors on a skullcap, Nicolelis thinks the inside-the-brain technology is necessary, that EEG will stick around, but that intracranial should evolve clinically:

The potential of intracranial BMIs for whole-body navigation, as demonstrated in our study, lies on the generation of continuous kinematic control signals, which are needed for subjects to navigate through complex and unknown spaces. Since typical EEG based wheelchair design only supports discrete control, it is clearly insufficient to generate continuous trajectories that can change at a moment’s notice. Yet, because of their low risk, EEG based systems will probably remain the dominant clinical BMI approach for a while, and even expand to include new applications, like, EEG controlled exoskeletons. However, as intracranial recording systems improve in efficiency and safety, they will likely become more attractive to the clinicians and patients, particularly those suffering from devastating levels of body paralysis, in the future.