To control the prosthetic, the patient has to think like they are controlling a phantom arm and imagine some simple manoeuvres, such as pinching two fingers together. The sensor technology interprets the electrical signals sent from spinal motor neurons and uses them as commands.
Robotic arm prosthetics currently on the market are controlled by the user twitching the remnant muscles in their shoulder or arm, which are often damaged. This technology is basic in its functionality, performing one or two grasping commands. This drawback means that globally around 40 to 50% of users discard this type of robotic prosthetic.
The researchers say detecting signals from spinal motor neurons in parts of the body undamaged by amputation, instead of remnant muscle fibre, means that more signals can be detected by the sensors connected to the prosthetic. This means that ultimately more commands could be programmed into the robotic prosthetic, making it more functional.
Dr Dario Farina, who now works in the Department of Bioengineering at Imperial College London, carried out much of the research while at the University Medical Centre Gottingen. The research was conducted in conjunction with Dr Farina’s co-authors in Europe, Canada and the USA.
“When an arm is amputated the nerve fibres and muscles are also severed, which means that it is very difficult to get meaningful signals from them to operate a prosthetic,” explained Dr Farina. “We’ve tried a new approach, moving the focus from muscles to the nervous system. This means that our technology can detect and decode signals more clearly, opening up the possibility of robotic prosthetics that could be far more intuitive and useful for patients.”
The researchers carried out lab-based experiments with six volunteers who were either amputees from the shoulder down or just above the elbow. After some physiotherapy training, the amputees were able to make a more extensive range of movements than would be possible using a classic muscle-controlled robotic prosthetic.
Further refinements are needed to make the technology more robust, but the researchers suggest the current model could be on the market in the next three years.
