Article In Brief
Hugh Herr, PhD, a biophysicist and engineer, describes advances in the development of bionic limbs that feel and move like natural limbs.
While twenty-first century prosthetic limbs have become highly sophisticated electro nic marvels, the surgical amputation procedure on which their use depends has remained essentially unchanged since the US Civil War. The result, according to Hugh Herr, PhD, is that many patients who have lost a limb are left with little or no proprioception in the amputation residuum, an absence that prevents patients from making full use of the new generation of prosthetics and is a key contributor to phantom limb pain and muscle atrophy.
Dr. Herr, who himself lost both legs below the knee in a climbing accident, wants to change that. How he's doing that was the theme of his talk at the Advancing Medicine: Inspiration and Innovation forum at the AAN Annual Meeting in May.
As professor of media arts and sciences and co-founder of the Center for Extreme Bionics at Massachusetts Institute of Technology, where he helps create the highest of high-tech prostheses, Dr. Herr has spearheaded the development of a new surgical procedure that retains proprioceptive feedback in the remaining muscles. That feedback can then be harnessed by the sophisticated electronics of the newest generation of prostheses to generate highly accurate and largely unconscious control of prosthetic movement.
With that level of control, the prosthetic becomes “neurologically embodied,” Dr. Herr said. While a cane is a useful extension of the body, its user will never describe it as a part of his body. “But when you have bidirectionality between the device and the nervous system, and the human can feel and control the device, if you ask that person—‘What is your body?’—he will include the synthetic device. This phenomenon, which I call neurological embodiment, represents a truly profound shift in the way humans interact with designed machines. This level of embodiment will provide our patients with incredible ownership over their prosthetic devices, and that will enable extraordinary capabilities,” Dr. Herr said in his plenary session at the Advancing Medicine: Inspiration and Innovation forum at the AAN Annual Meeting.
In a normal biologic limb, the muscles are constantly being stretched, contracted, and moved through space. Muscle spindles, Golgi tendons, and other systems send constant feedback to the central nervous system to generate a dynamic sense of where the limb is, where it is going, and how fast it is going there—the elements underlying proprioception.
A key element of proprioceptive integration is the paired input from agonist and antagonist muscles, Dr. Herr explained, in which contraction of one muscle agonist is necessarily accompanied by the stretch of the other antagonist muscle acting about a joint. But in a standard amputation, the residual muscles are fixed in place, no longer dynamically paired with their antagonists. The brain often retains a phantom image of the missing limb, but without sufficient proprioceptive input, the phantom limb remains, in the brain's image, immobile.
“If you blindfold a person with an amputation, they have no proprioceptive sense of where their foot is in space, or the speed and load of their prosthesis,” Dr. Herr said, who stands on two robotic prostheses. “Unfortunately, I received the Civil War-era amputation. I feel my feet as a phantom awareness. I feel my toes, everything is very clear, but it feels as though my foot is stuck in a rigid ski boot. My joints are locked in space forever because of that amputation style.”
In contrast, with the availability of proprioceptive feedback, even without a prosthetic in place, the brain's phantom limb is experienced as a far more dynamic and natural extension of the body, with less atrophy and less pain, Dr. Herr said.
A Post-Amputation Surgical Procedure
To deliver that feedback, Dr. Herr and colleagues developed a new post-amputation surgical procedure, which they call an agonist-antagonist myoneural interface, or AMI (pronounced “Amy”). “For every degree of freedom you want to control in the prosthesis, for every motor muscle you have, you surgically link one muscle pair, or one AMI, an agonist linked to an antagonist,” Dr. Herr explained. “Because they are mechanically linked, when the agonist contracts upon electrical activation, it stretches the antagonist, and there is rich information from the spindles and Golgi organs sent to the nervous system telling the brain about the position, speed, and load on the joint.”
Preclinical experiments conducted at MIT confirmed that the surgery could create sufficient stretch in the antagonist to generate detectable and proportional electrical nerve signals, and there was no loss over time to scarring. But how practical was this for humans? The opportunity to find out came when a friend of Dr. Herr's, “a rock-climbing legend,” took a fall. While his leg was saved through multiple surgeries, he was in so much pain that he chose amputation instead, and volunteered to be the first human to receive an AMI. The surgery, a unilateral transtibial amputation, was performed by Dr. Herr's collaborator Matthew Carty, MD, of Brigham and Women's Hospital in Boston.
In the surgery, the patient's residual tibialis posterior and peroneus longus were joined to form an AMI that controlled the subtalar joint of the prosthesis for inversion and eversion, while the lateral gastrocnemius and the tibialis anterior were joined to form an AMI that controlled the prosthetic ankle joint for dorsiflexion and plantarflexion.
“Our hypothesis was that when a person moves their phantom limb, they would actually experience the full dynamic range of motion with input from the AMI,” Dr. Herr said. “And we were hoping that we could then attach a prosthesis to the residual limb, and then have the patient control the prosthesis using small computer chips to decode the AMI muscles' electrical signals.”
During a training period, the prosthesis used bionic control algorithms to close the gap between where it was and how it was moving, and where the patient sensed it was and how it was moving. The patient quickly learned how to control both the ankle and subtalar prosthetic joints, and in a test against subjects with the standard amputation, he wasted less motion in moving the prosthesis to a target position in free space.
“Then what we observed was truly remarkable,” Dr. Herr said. The patient “started walking up and down slopes and stairs, and all the natural biomechanics that you typically see in the foot-ankle emerged through the synthetic limb before our very eyes. It turns out that the movements were involuntary—he wasn't trying to do them. We feel that because his brain is getting the muscle tendon proprioception information, his brain knows exactly how to control the synthetic limb.”
“I didn't feel like a cyborg,” the patient said in a video the MIT team made. “I felt like I had my leg. The robot became part of me. It became my leg.” As an example both trivial and astonishing, Dr. Herr showed the patient sitting, involved in casual conversation, with the prosthetic limb crossed over the real one. As his attention is engaged elsewhere, the patient begins to absent-mindedly and entirely unconsciously jiggle his bionic foot, gesturing with his prosthesis while continuing to talk.
To date, 17 patients have undergone the AMI surgery, and Dr. Carty has begun training other surgeons to perform it. The bionic prostheses are expensive, and only a small subset of patients are eligible for reimbursement at the moment. “This is the greatest challenge, Dr. Herr said.
The addition of proprioception to control advanced prosthetics is a major goal, commented Gregory Clark, PhD, associate professor of biomedical engineering at the University of Utah, who is interfacing bionic upper limbs to users after hand amputation.
“In open-loop mode,” without sensory feedback, “every movement has to be done by dead reckoning, and even then it is inadequate—you can't see the back of a soda can, so you don't know when you are touching it,” Dr. Clark pointed out. Integration of proprioceptive feedback closes the control loop, and Dr. Herr's use of the AMI provides a simple and powerful way to do that.
“In Dr. Herr's approach, mechanically connecting two muscles together, you are getting a biologically realistic signal from both of them, and that's what the body is expecting.”
Drs. Herr and Clark disclosed no conflicts of interest.