Tactile Feedback Achieved in Bionic Limb
ARTICLE IN BRIEF
A research team developed the prototype for a sensory-enhanced prosthetic limb that enabled the subject to experience tactile sensations.
It is still a long ways away, but a truly bionic hand may be one step closer, according to a new study that demonstrated that when tactile sensation is delivered from an artificial hand to intact peripheral nerves, motor function improves. During the course of this 30-day trial, the subject using the prosthetic reported that the new artificial hand began to feel like a part of his own body. A description of this approach was published in the Feb. 5 issue of Science Translational Medicine.
The value of sensory feedback for control of an artificial limb has been understood for a long time, Paolo Maria Rossini, MD, PhD, co-leader of the study and director of the Institute of Neurology at Catholic University of The Sacred Heart in Rome, Italy, told Neurology Today. Current state-of-the-art hand prostheses rely on so-called “open-loop” control, in which the subject must use visual feedback to control the movement and position of the fingers of the hand. Visual feedback is useful, he said, but without tactile input, the hand may operate at only 25 percent efficiency in the natural environment. Many tasks, such as those requiring a delicate and graded grasp, may be impossible to perform.
The ideal prosthesis would instead employ “closed-loop” control, in which the hand's actions generated sensory information to the brain, exploiting the pre-existing tight connection between tactile input and motor output. But that requires three key things, said study co-leader Silvestro Micera, PhD, director of the Translational Neural Engineering Laboratory at the Center for Neuroprosthetics, part of the Swiss Federal Institute of Technology in Lausanne, Switzerland.
First, he said, there need to be pressure-sensitive sensors that can be embedded in the fingers, which could record graded tactile information. Second, that raw sensory output has to be translated via computer into the neurally relevant output of amplitude, frequency, and width, and then delivered as electrical stimulation to the afferent nerves. Finally, Dr. Micera said, “the only way to make a closed loop possible is to make an almost zero delay in each step,” from recording to nerve stimulation.
FULLY ARTICULATED MECHANICAL HAND
Five years of developing those key features led to “LifeHand 2,” a fully articulated mechanical hand whose fingertips are covered in a pliable polymer. Pressure sensors are embedded in the tips of the index and little fingers. Wires lead from its base to an external computer, which in turn powers a set of multichannel electrodes. The electrodes are embedded in the remaining limb stump, where they contact the median and ulnar nerves.
The subject in this first-ever trial was a Danish man who, nine years earlier, had lost his left hand in a fireworks accident. The electrodes, thinner than a human hair, were implanted under general anesthesia, and two days later connected to the computer. Over several days, the researchers tested combinations of contacts and stimulation settings, eliminating those that registered as pain or temperature, while fine-tuning those that were sensed as pressure originating from the index and little finger.
“At the end of this mapping procedure, we retained a number of contacts, from which the subject was receiving stable sensation, very precisely localized in the missing hand,” Dr. Rossini said. “Next, we connected the hand. At this point, we had a system which was theoretically able to dispatch sensory information from distinct parts of the fingers that the subject could localize quite precisely.”
To test the ability of closed-loop tactile feedback to improve motor control, the researchers posed several grasping challenges to the subject, using a blindfold and headphones to eliminate visual and auditory clues about the hand's motions. Within a week of daily trials, the subject improved his grasp control to achieve the desired force 90 percent of the time. Control at faster speeds was more challenging than at slower speeds. Still blindfolded, he was also able to distinguish different objects based on their tactile properties: cylinders versus spheres, cotton versus wood.
“It is still only a very rough type of sensation,” Dr. Rossini noted, with no proprioception, pain, or temperature, “but it is still useful for producing efficient movement, and for recognizing some physical properties of the object.”
Perhaps as remarkable as the high level of control achieved was that the subject's brain still mapped incoming sensations as originating from a hand at all, nine years after amputation.
“Apparently, all the targets of this missing information remained available,” he said. “He was able to reconstruct a body scheme in which the missing hand was again present, because of the sensory flow coming from the artificial prosthesis. He was reporting that he felt like the sensations were coming from his real hand.”
Regulatory permission for the implant was limited to only 30 days, so it is not yet known how far this device might go to restoring daily function. Dr. Rossini referred to the subject as a “hero,” willing to undergo two major surgeries despite knowing he had nothing to gain in the end. “He worked with us, to help make a small step towards a solution for millions of people suffering from limb amputation.”
The team is already at work on a new iteration, with a goal of a fully wearable system than can be implanted for much longer periods. The safety record of similar implants is good, Dr. Rossini said. The device is likely to be limited to amputations up to the elbow. “The real challenge will be to see whether this kind of approach can be useful in larger number of patients.”
“This is clearly an important proof of principle,” commented Karunesh Ganguly, MD, PhD, assistant professor of neurology at the University of California, San Francisco, and an expert in neurorehabilitation. He noted that the many examples of visually controlled devices don't provide the naturalistic type of feedback described in this study.
Contributing to the promise of the strategy here is that it remains a peripheral device. “The barriers are a lot less, versus a central interface,” he said. “There will be challenges as they move forward, of course. Moving from 90- to 100-percent success is not trivial, but I think these sort of efforts have a higher probability of success than the brain-machine interfaces.”
Another challenge, he said, is developing a device that does not require a room full of experts to manage. In this study, there were 21 contributing authors, a good number of them likely “PhD's watching the screen looking for artifacts,” he suggested. “Moving to a fully wearable, autonomous system that doesn't require multiple engineers around is going to be a major challenge. But I don't think it is insurmountable.”