I think, therefore I grasp

There is major interest in trying to “cure” paralysis and help patients overcome mobility limitations, trying everything from repairing damaged spinal cord neurons, to engineering human exoskeletons.  One branch of this research focuses on brain machine interfaces (BMI) in which a multi-electrode recording chip is implanted into the brain.  This array can “read” what the neurons in the brain are signaling.  It sends this information to a computer which decodes these signals into the intended action of the brain.  Normally these brain signals would travel down neurons in the spinal cord to the motor neurons that control muscle contraction.  During spinal cord injury, the pathways from the brain to muscles are damaged and non-functional.  With the array implanted in the brain, the BMI can bypass the damaged nerves and stimulate the intended muscles directly.  If the brain sends the signal to “flex right arm” then the implanted array and the decoding computer will interpret the intended message and stimulate the right arm muscles.  In time the patient will learn to control their own paralyzed arms by thinking about the movement, which is essentially what we all do anyway.  In the case of the BMI, there is an electrode array in the brain, an external computer and stimulating electrodes in the arm muscles to help get around the damaged spinal cord.

A paper was published online this week in Nature by Ethier et al. that used BMIs in monkeys to control hand grasp.  I haven’t read all the previous papers about BMIs, so I don’t know how much of an advance this is, but according to the authors, this is one of the first times that force of muscle contraction, as opposed to direction of movement was taken into consideration.  Their experiments worked very well and the paralyzed monkeys were able to grab balls.

Experimental set up

Multi-electrode arrays were implanted into two monkeys.  There is a region in the brain called the motor cortex which controls muscle movement.  BMIs only work because there are specialized regions in the brain devoted to different tasks.  The researchers put the arrays specifically in the area devoted to contracting the muscles found in the hands (i.e. for grasping movement).  The monkeys were totally functional at this point (not paralyzed), so they trained the monkeys to grab a ball and put it into a tube, while recording the neuronal activity in the motor cortex.  They mapped the brain activity associated with muscle contraction and movement until they were able to develop a model that could predict hand movement based only on the activity in the brain (amazing!)  While the monkey was thinking about moving, the model could predict quite accurately, in real-time, what kind of movement would happen in each muscle.

Next, they injected an anesthetic into the arm, so it was temporarily paralyzed.  When monkeys went through the trials where they had to grab a ball to get a treat, their brain activity showed that they were thinking about moving their hands, but they couldn’t actually move them because they were paralyzed.  The monkeys also had a series of stimulating electrodes implanted into their hand muscles.  Could the BMI be used to control hand grasping movements in paralyzed arms?

Brain-controlled grasping

The paralyzed monkeys were indeed able to grab the ball and place it in the tube just by thinking it.  Just to reiterate: the motor cortex of the monkeys was activated, this was recorded by the implanted array, which sent the signal to a computer, which interpreted the intended movement, which was finally achieved by implanted electrodes in the hand muscles.  

This is completely amazing.  It’s amazing that the computer model was accurate enough to predict hand movement just by recording activity of 100 neurons in the brain.  Both monkeys had about a 75% success rate of picking up the ball, which isn’t bad at all considering the fact that their arms were paralyzed and being controlled by a machine.

One of the monkeys also learned a different task.  The monkey was taught to squeeze a tube to make a cursor on a screen move to a target location.  The harder the squeeze, the more the cursor moved.  This task is all about modifying how hard to contract muscles and controlling the grasp, a fairly complex movement.  Again, they made a model to predict the amount of force based on the neuron firing pattern.  And once again, the monkey was paralyzed and was able to do the task without having voluntary control of its hand muscles.  As it completed the task, its muscles got tired (muscle fatigue), so it had to send stronger signals to them to contract.  The BMI was able to control the amount of force necessary to counteract the muscle fatigue, just like the fully functioning monkey would do. 

So where does this leave us?  It was only two monkeys, but the BMI worked so well, I imagine this would be true for others as well.  One problem is that their computer model was fine tuned by recording activity from the functioning monkey while comparing that to actual muscle movement.  This would be impossible to do in someone who is already paralyzed.  How well would they be able to adapt to a prediction model that isn’t designed specifically for their brain? 

The authors bring up an interesting point that the muscle stimulation would be helpful to maintain muscle and bone structures, and if the patients are associating that with a particular task, it might be more rewarding.  As they say, “it may be that drawing on a conscious process to restore natural movement will bring the additional benefit of improved psychological health.”  I just find it so impressive that we know enough about how the brain sends signals that we can tap into that to control machine-based movement.

UPDATE: This paper has officially been published in the print version of Nature and there is a companion article that shows that BMIs work in humans too!  Two paralyzed older people had multielectrode chips implanted into their motor cortices.  Instead of controlling their own arms, like the monkeys did, they controlled a remote operated robot arm.  The commands in their motor cortex were converted into a digital code that could tell the robot arm how to move.  One woman was even able to grab a bottle and bring it to her mouth and drink from a straw!  And this was after being paralyzed for 15 years.  It's amazing that her motor cortex is still functional after not being used for years.  Read the New York Times article about this research.