Reinnervated Muscles Produce EMG Information for Real-Time Control of Artificial Arms
The idea to transfer a nerve no longer in use to a nearby area to help people who have lost an arm gain more natural control of a prosthetic arm began with a one-liner in a journal more than two decades ago — a colleague reflecting on such a far-off possibility.
Artificial arms were so cumbersome that most single-arm amputees opted not to wear them. But Todd A. Kuiken, MD, PhD, a physiatrist and biomechanical engineer at the Rehabilitation Institute of Chicago, wanted to put this idea to the test. He worked with colleagues to develop a surgical technique called targeted muscle reinnervation, mostly done by Gregory Dumanian, MD, a plastic and hand surgeon at Northwestern University.
In 2001, Jesse Sullivan became the first person to undergo targeted muscle reinnervation. The former power company worker lost both arms when he touched a high-voltage power line a year earlier. His limbs were destroyed to the shoulder. The remaining arm nerves were transferred to residual chest muscles, which meant that the brain would receive signals from the missing arm, which was re-routed through the chest.
Since then, 30 upper limb amputees have had the nerves that once controlled their elbow and hand re-routed to nerves in the chest or upper arm that take over the neural connections that once drove movement of the hand.
Now, these investigators have tested the boundaries of the technique by fitting three patients with bionic artificial arms designed to decode the EMG signals generated during a range of arm and hand movements. The findings, published in the Feb. 11 Journal of the American Medical Association (JAMA), suggest that the reinnervated muscles produce enough EMG information to enable them to control the advanced prosthetic arms to perform a variety of hand and arm movements in real time.
Five patients who received targeted muscle reinnervation (TMR) surgery between 2002 and 2006 and five control participants with normal use of their arms were enrolled in the study. Surface EMG signals were recorded while they were all asked to repeatedly perform 10 different elbow, wrist and hand movements. At the same time, a pattern-recognition algorithm was used to decode the EMG signals and then control actions of a virtual arm. The pattern-recognition algorithm was also used to operate the bionic experimental arm prostheses fitted on three of the five TMR patients.
The computer had to sort out the EMG signals at every turn to figure out how to make a movement happen. The goal is to have a person think about moving the wrist, for instance, and have the algorithm determine the patient's intent and then control the prosthesis.
The patients fitted with experimental computerized prosthetic arms had only two weeks of training, and the pattern-recognition algorithms allowed the bionic arm to interpret their thoughts to manipulate many different hand grasps.
“These muscles are biological amplifiers of motor commands from the arm nerves,” explained Dr. Kuiken. This allows physiologically appropriate EMG signals for controlling the elbow, wrist, hand, and fingers. “The technique allows intuitive, natural movements,” he said. “People don't have to think so much and it reduces the cognitive burden. What's more, they can now simultaneously control opening and closing of the hand as well as extension and flexion of the elbow.”
The scientists reported that the amputees could manipulate the virtual hand with almost the same speed as those with two hands, and could do tasks that most people take for granted: catch a checker rolling off the table, stir a spoon in a cup, pick up small blocks and stack them, and move a ring across a geometric wire. These tasks were unthinkable without this advanced technology. The prosthetic arms tested were funded by the Department of Defense and developed by Johns Hopkins Applied Research Laboratory and DEKA Integrated Solutions, Inc.
Dr. Kuiken and his colleagues have been working on designing and testing pattern-recognition algorithms based on the EMG recordings to power the next generation of upper-limb prostheses. “Our hope is that people with upper-limb amputations will be able to have a range of arm and hand movements that are more natural,” said Dr. Kuiken. “The artificial arms of today allow for limited movement and it is not intuitive. We want to allow patients to just think about moving their arm, hand, or fingers and have it happen naturally. We are showing that it can be done.”
Marc Schieber, MD, PhD, a professor of neurology and neurobiology at the University of Rochester and the Brain Injury Rehabilitation Unit at St. Mary's Hospital, said, “This TMR technique has worked surprisingly well to control prosthetic arm motion.” The beauty of it, Dr. Schieber added, “is that the patient doesn't have to train the brain to do things it hadn't done before the injury.” Now, standard upper limb prostheses are powered by learning how to use the shoulder to work the arm and hand. “I think this is a very viable approach.”
Every year, 50,000 people lose a limb and about 20 percent of these amputees lose an arm. Lower limb extremities are easier to control since there is one thing the person has to learn how to do: walk. Upper limb prostheses have been much harder to develop because hands and arms must flex in any given direction and hands used for a myriad of actions requiring knowledge of where the hand is going in space and what it needs to accomplish. It is so cumbersome, Dr. Kuiken said, that half of single arm amputees do not wear their upper-limb prostheses.
ARTICLE IN BRIEF
Three patients with bionic artificial arms designed to decode the EMG signals generated during a range of arm and hand movements were able to control their artificial arms with greater speed and skill.
WE WANT YOUR INPUT
We want to hear from you. Please feel free to respond to our stories by writing NeuroToday@LWWNY.com.
Letters, which should be no more than 400 words, will be edited for clarity and style.
When writing in, please include your phone number and e-mail for contact.