Spying on Your Brain to Move Your Hand
A spinal cord injury is bypassed to directly control hand muscles with thoughts.
Posted Jun 01, 2016
Most of us take for granted the many routine tasks of everyday life in which we grasp and manipulate objects. But it can be devastating to lose these abilities, as often happens following cervical (neck) spinal cord injury or brain damage that causes arm and leg paralysis (tetraplegia). Neurobiologists and engineers have worked for decades toward the goal of monitoring electrical signals in the brain and using them to control the hand or a robotic (prosthetic) hand, bypassing the site of injury. In this way, such patients could simply think of moving their hand in a certain way and it would happen. We are now closer to this goal, thanks to a key advance that was announced a few weeks ago.
The idea is that nerve cells (neurons) in certain parts of the brain, such as the motor cortex, normally generate electrical signals (action potentials, or spikes) that trigger particular movements. Different neurons trigger different movements. A particular sequence of brain neuron spikes causes a precise movement sequence such as we use to grasp and manipulate an object. These brain neurons normally activate neurons in the spinal cord, which in turn activate muscles in the arm and hand. This can no longer happen if the route from the brain to those spinal cord neurons is blocked (such as by a spinal cord injury). But if we could detect spikes in key brain neurons and if we knew which movement each neuron normally triggers, then we could artificially stimulate hand muscles (called functional electrical stimulation, FES) or trigger a certain movement of a prosthesis each time these neurons spiked, so that hand movements would be directly triggered by thoughts. This is often called a brain-machine interface.
The first step is to monitor spikes in brain neurons. This can be done with a microelectrode array (see photos), which is essentially a grid of tiny metal needles or wires that can be implanted on the surface of the brain. Microelectrode arrays typically allow monitoring of around 100 neurons at a time. (A noninvasive alternative is an electroencephalogram—an EEG—but it averages the electrical signals of a very large number of neurons that do different things, which dilutes the meaning of neuronal signals so much that it is hard to trigger precise movements.)
The second step has been harder: how do you decipher the meaning of the spikes in all the neurons monitored? What movement does each neuron normally trigger?
One way to find out is simply to monitor the spikes while a normal subject is making normal movements. Such research was initially done with monkeys trained to make particular movements and it built a foundation for use of brain-machine interfaces in human patients. Once researchers know in what context each neuron normally spikes, they can develop an algorithm that extracts the meaning of the whole set of neurons’ spikes in terms of movement sequences. The algorithm essentially translates the set of spikes into movement commands to the muscles or to a prosthesis.
After these methods were developed with monkeys, they were applied to a few human patients who could otherwise not perform even basic functions with their hand. For example, in work reported in the journal Nature in 2012 by Leigh Hochberg, John Donoghue and colleagues at Brown University, a tetraplegic patient was able to pick up a cup and bring a straw to her lips just by thinking about it, with spikes in motor cortex neurons controlling the prosthesis (see video).
In another approach, reported in Nature in 2008 by Chet Moritz, Steve Perlmutter and Eberhard Fetz at the University of Washington, they used the spike rates of just one or two motor cortex neurons to control electrical stimulation of a monkey’s wrist muscles (after the wrist was anesthetized). This was set up as a video game in which wrist movement caused movement of a computer cursor (see illustration). Remarkably, in about 30 minutes monkeys learned to control the spike rates of these particular neurons to make the wrist move correctly and win the video game, no matter what the neurons were doing before the game started. Presumably, people could learn to do the same. This means it might be possible to skip the step in which the meaning of brain neuron spikes is deciphered, if people could instead learn to consciously control the spike rates of whichever neurons were monitored.
In the new research, reported in Nature on May 12 by Chad Bouton, Ali Rezai, and colleagues at Battelle Memorial Institute and Ohio State University, a tetraplegic patient was able to use his own hand to grasp a jar and pour from it, as well as to pick up a stick and stir with it (see video).Spikes in motor cortex neurons were monitored with a microelectrode array and deciphered by an algorithm that triggered electrical stimulation of several specific muscles in his hand. This is the first time a person’s thoughts have been used to electrically stimulate his own muscles to accomplish such a complex task. This advance gives hope to numerous patients that they may someday be able to bypass their brain or spinal cord injury and make the ordinary movements of everyday life that we all depend on.