ARTICLE IN BRIEF
Investigators linked two types of technology involved in brain-machine interfaces between two monkeys, enabling the recording of signals from the premotor cortex of one monkey to another, which helped provide functional electrical stimulation to activate muscles to produce movement.
After a spinal cord injury, brain signals directing a muscle to move can no longer reach their destination. If those brain signals could be sent to a computer that interpreted them and then transmitted them to the spinal cord below the lesion, the target muscles — in theory at least — might be induced to move.
In an effort to demonstrate how this could be done, researchers linked brain signals from one monkey trained to move a cursor on a computer screen to another monkey anesthetized to mimic paralysis. The second monkey, known as the avatar, despite deep unconsciousness, then moved a joystick and guided the cursor in the same direction.
Both monkeys had a recording chip implanted in the premotor cortex to detect signals from a small number of neurons activated by the intention to move the arm, and stimulating electrodes implanted in the cervical spinal cord that could receive those signals from a computer and stimulate the appropriate muscles. This allowed the monkeys to take turns as the master, who would send the signals, and the avatar, who would receive them. Sedating the avatar mimicked paralysis, which allowed the researchers to test their cortical-to-spinal bypass without actually paralyzing a monkey, according to lead author Ziv Williams, MD, assistant professor of neurosurgery at Harvard Medical School.
“The general goal of the project was to find ways of treating people with cervical spinal cord injuries, or people who are quadriplegics from brainstem strokes,” said Dr. Williams.
The researchers, who published their findings in the Feb. 18 issue of Nature Communications, linked two types of technology involved in brain-machine interfaces. One technology enables the recording of signals from the premotor cortex that reveal what the subject intends to do. The other involves functional electrical stimulation (FES) to activate muscles to produce movement.
“The trick was to connect those two technologies in real time so we could record what one monkey wanted to do, and then use that information to control a functional stimulator in the other monkey,” Dr. Williams said. “The ability to generate movement in a paralyzed limb is a big advantage. While brain-machine interfaces have been used to control a mechanical device like a robotic arm, another goal would be to control your own paralyzed arm so you wouldn't have to rely on a robotic arm or other mechanical device.”
But the research, which was reportedly inspired by the movie “Avatar” and widely reported in the popular press, drew a decidedly indifferent response from experts in the field who are doing similar work, including several who would not comment on the record.
Calling it “less of a scientific advance than an apparently dramatic demonstration that will spark popular interest,” Eberhard Fetz, PhD, a professor in the department of physiology and biophysics at the University of Washington, criticized the paper for implying that the study accomplished more than it actually did.
“The way this article is written could be misleading,” said Dr. Fetz, who has conducted similar work using a monkey that is actually paretic due to a cervical spinal cord lesion. “The paper implies that whatever the master monkey intended could be evoked in the avatar. That is simply not true.”
The researchers reverse-designed the experiment so the master would be presented with one of the two targets that were previously acquired by stimulation of the avatar's spinal cord, according to Dr. Fetz.
“In fact, the neural activity does not directly stimulate the spinal cord,” he said. “Neural activity is processed by a computer that determines whether the pattern of neural activity is more consistent with movements toward one target or the other. The computer decodes the intent of the master monkey and converts it into a binary decision that determines which of two electrodes should be stimulated. Decoding neural activity and evoking movements from spinal stimulation have both been well documented. Simply combining these in a restricted context does not represent a significant advance toward achieving prosthetic control.”
Dr. Fetz and his colleagues published an article last April in Frontiers in Neural Circuits that described work with a monkey that had been trained to use its hand to move a cursor on a computer screen into a target area. After the monkey became paretic it could no longer generate sufficient force to perform this task, but it learned that by increasing its own neural activity it could trigger a spinal stimulator that helped it move the cursor into the target area.
“In our case the stimulation was delivered to the spinal cord of the same animal,” he said. “Although our experiment also involved specific movements, using a single awake animal probably comes closer to the circumstances that would be relevant for prosthetic applications. In their case, the avatar is anesthetized and totally unaware of what's going on. It's like pulling the strings of a marionette.”
V. Reggie Edgerton, PhD, a distinguished professor in the departments of neurobiology and of integrative biology and physiology at the University of California, Los Angeles, has been conducting spinal stimulation of human paraplegics, and he too had reservations about the research reported in the Nature Communications paper.
“I think the probability of this approach getting to the point of being clinically important is very low,” he said. “Even the authors say that it will take a much more complex system to reach more than two targets, but the inherent limitations to this approach are pretty significant.”
The fundamental problem, according to Dr. Edgerton, is that neurons receive thousands of synaptic inputs second by second. “That's one of the reasons why all brain interface experiments are limited in the accuracy they can achieve,” he said. “The other limitation is the difficulty of implanting electrodes in the spinal cord so you don't cause damage. Technically it's a huge problem because the spinal cord moves so much.”
In short, Dr. Edgerton is not optimistic about the Nature Communications paper advancing the development of brain-machine interfaces for helping paralyzed people.
“Each of the procedures they use has been tried and tested to some degree already,” he said. “Using a readout from the brain to determine a binary decision has been done. That's not new. And years ago there was a lot of excitement about using computer-controlled direct stimulation of muscles to help a (paralyzed) person stand and walk, but it didn't go anywhere for a variety of reasons. The logic of this experiment from a biological perspective is not obvious.”
Marc Slutzky, MD, PhD, assistant professor of neurology, physiology and physical medicine at Northwestern University's Feinberg School of Medicine, said many of the ideas described in the Nature Communications paper have already been demonstrated by other researchers, so the paper does little to advance the field of brain-machine interfaces.
“The primary novelty here is that they used brain signals from one monkey to control the arm movement of an avatar monkey,” said Dr. Slutzky, whose own research focuses on brain-machine interfaces that employ a computer to decode signals from the cerebral cortex. “I think that has gotten a lot of attention because of the movie ‘Avatar.’”
Using signals from the premotor cortex, which indicate the intended target of a reach, rather than signals from the motor cortex, which represent the trajectory of a reach, might make more sense in a paralyzed patient, Dr. Slutzky said, since intent would remain reliable despite their injury, even though execution may be impaired.
“That way, even if they don't succeed at reaching a set of objects, you know they were trying to reach them,” Dr. Slutzky said. “You know that they know what the targets should have been, and you can assume that's the goal they intended to hit.”
Another interesting aspect of the experiment was the use of intraspinal stimulation to get the arm of the avatar to move, although “others have done much more sophisticated versions that stimulated the arm in more sophisticated ways,” Dr. Slutzky added.