Smart Technology Improves Control of Robotic Leg

Pauline Anderson

September 25, 2013

A new report establishes the feasibility of using a state-of-the-art pattern recognition system to improve control of a robotic leg.

With the new system, which uses electromyographic (EMG) signals from re-innervated residual thigh muscles, a patient who had undergone an above-the-knee amputation could seamlessly maneuver his robotic leg, move with a near-normal gait, and climb stairs with relative ease, researchers write.

The robotic leg is safer than a regular prosthetic leg because the chances of falling are lessened, the paper's lead author, Levi J. Hargrove, PhD, an electrical engineer and research scientist at the Center for Bionic Medicine, Rehabilitation Institute of Chicago, Illinois, told Medscape Medical News.

Amputees with currently available above-the-knee prosthetic legs must press buttons on a key fob or perform a set of predefined exaggerated motions and manually reposition the device when seated.

Over a million Americans who have lower-extremity amputations could benefit from the improved prosthetic leg when it comes on the market in what may be as little as 3 years, said Dr. Hargrove.

The brief report is published in the September 26 issue of the New England Journal of Medicine.

Redirected Nerves

Under normal circumstances, signals from the brain are sent down the spinal cord through peripheral nerves to instruct muscles to contract, which causes joints to move as intended. After a leg amputation, the signals are still generated and still travel down the spinal cord and out through peripheral nerves, but there is no longer a muscle to carry out the instruction.

But that is now changing with new smart technology.

About 36 hours after a motorcycle accident in 2009, a 31-year-old man underwent knee disarticulation amputation (the bottom part of the femur was left but he has no knee joint). During targeted muscle re-innervation surgery, surgeons redirected nerves onto a different muscle that was still healthy after the amputation.

"We kind of unplugged one little tiny nerve that went into the hamstring muscle and we plugged in a bigger nerve, the tibial nerve, that would have gone down and connected to muscle that would have lifted his foot up into the air," explained Dr. Hargrove.

Inside the socket, which acts as the interface between the prosthetic leg and the patient's skin, researchers inserted electrodes that act like antennae to detect tiny signals generated by the re-innervated muscle. The system uses pattern recognition to decode signals from the muscle when it contracts, explained Dr. Hargrove.

"The patient now just thinks about moving his knee or moving his ankle," he said."The microprocessor on the leg has been programmed to be incredibly intelligent and has learned what his muscle signals look like, so whenever he tries to do something, it instantaneously decodes it and instructs the leg how it should move."

This EMG pattern recognition system technology is analogous to a voice recognition system used in fields of communication, said Dr. Hargrove.

The researchers found that the re-innervated hamstring muscle generated robust EMG signals, especially during contractions corresponding to ankle movement.

Zac Vawter, recipient of the robotic leg, and Dr. Levi J. Hargrove. Rehabilitation Institute of Chicago

"When the patient thinks about pushing his toes into the ground, called plantar flexing or lifting, and pointing his foot up, called dorsal flexing — essentially moving his ankle — instead of causing muscle that is below his knee to contract, those nerve signals have been redirected and his hamstring contracts," he said.

Error Reduction

The error rate was reduced from 12.9% to 2.7% when considering EMG signals pertaining to muscles that would have controlled the knee and hip, but there was an additional error reduction — to 1.8% — when muscle information on ankle movements was also included.

"We were quite surprised with the amount of error reduction we got without even considering ankle information," said Dr. Hargrove.

The reduced error rate allowed the patient to move confidently and transition seamlessly with near-normal gait kinematics, said the authors. The patient was especially encouraged by being able to enter buildings without difficulty and without using handicapped accessible entrances, they note. And when not standing, he could more easily reposition the prosthesis, which made dressing and getting in and out of vehicles easier.

With a standard prosthetic leg, wearers have to "kind of drag the limb up behind them, and this doesn't allow for seamless transition between activities," said Dr. Hargrove.

"If you are walking along and want to go up set of stairs, with existing devices you have to stop at the bottom of stairs, perhaps use a remote control or use exaggerated movement like rocking back and forth, to signal to the leg that you want to climb up a set of stairs. With this new device you just walk up to the stair and take a step as you would have before your amputation."

Similar EMG decoding technology has been used in the past for robotic arms, but controlling the error rate for artificial legs is more important from a safety perspective, said Dr. Hargrove. "When controlling arms, if you make a small mistake the elbow might flex little more than you intended it to, but when controlling legs, if you make even a little mistake, it could cause a patient to fall."

There are many more leg amputations than arm amputations because of complications of diabetes.

Several challenges still remain before the control system will be clinically viable. The electrodes must remain in full contact with the residual limb during walking, and patients may experience chafing or pressure sores at electrode contact points after prolonged use. However, this is a concern that the team constantly monitors for, and has yet to note, said Dr. Hargrove.

Improvements are also needed in the pattern-recognition classification algorithms and the mechanical sensor system, said Dr. Hargrove. As well, he said, the robotic prosthetic leg must be made "more reliable, quieter, smaller and lighter." It currently weighs about 10 pounds.

The research was funded by the US Army. More than 1200 soldiers have had leg amputations after being wounded in recent conflicts. Dr. Hargrove credits this research accomplishment to the diverse team of scientists, engineers, and clinicians, who each provide important input.

An animation of the leg can be found here.

The authors have disclosed no relevant financial relationships.

N Engl J Med. 2013;369:1237-1242. Abstract


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.