Experience of phantom limbs lets amputees control real replacements


IN THE EARLY 16th century a knight called Gottfried von Berlichingen spent decades marauding and feuding on behalf of the Holy Roman Empire. He conducted most of his career singlehandedly—the other having been blown off by a cannonball. To replace it he had a metal duplicate made, with spring-loaded fingers that could hold a sword, shield or the reins of his horse. This early prosthetic device gave him the nickname “Götz of the Iron Hand”.

Prostheses have come a long way since Götz’s day. A technique called targeted muscle re-innervation (TMR) permits surgeons to take the nerves that once controlled a missing limb and attach them to muscles in a patient’s chest or back. The redirected nerves grow into their new muscular homes. These then act as signal amplifiers: a muscle’s electrical activity reflects that of the nerves supplying it, but is far more powerful and therefore easier to detect using external electrodes. That activity, duly interpreted by computer, can be used to drive motors within the prosthesis to make it do what its wearer wants.

For this to happen, though, the patient must spend weeks, or even months, learning to twitch the re-innervated muscles in particular ways to achieve particular outcomes. That is frustrating and tedious. Nor is the re-innervating surgery itself without risk. A better way to control prosthetic limbs would be welcome. And one may now be on offer.

Some amputees feel the presence of a phantom limb where the real one was. Often, they feel that phantom to be under their control. If it were possible to use these feelings to direct the behaviour of a prosthesis, TMR might be made redundant. Nathanael Jarrassé of the Sorbonne and Jozina de Graaf of Aix Marseille university, both in France, have begun working on how to do this.

In their latest study, just published in Frontiers in Bioengineering and Biotechnology, Dr Jarrassé and Dr Graaf stuck six electrodes onto the arm stumps of two volunteer amputees who had each lost that limb above the elbow. (The loss of the elbow joint as well as the wrist greatly complicates the task of interpreting the signals and controlling the prosthesis.) These half-dozen electrodes read activity coming from the arm’s remaining muscles as the volunteer thought about moving the missing limb.

The trick was that the learning needed to manipulate the prosthesis was done not long-windedly, as in TMR, by the patient, but rapidly, by a computer algorithm. This recognised within minutes the different patterns of electrical activity that corresponded to different actions of the phantom limb as imagined by the volunteer, such as opening or closing the hand, or moving the wrist. It then directed motors to replicate such actions in the prosthetic arm. Both patients were thus able to use the device intuitively—successfully grasping, placing and releasing objects.

The new system is not perfect. At the moment the algorithm recognises only the type of movements the phantom limb is making in the patient’s mind, not their speed or their amplitude. It also takes half a second or so to process the electrodes’ signals. This delay between intention and execution means the user does not yet experience the prosthesis as if it were part of the body. These imperfections are, though, things that might be overcome in the future. And if they can be, the phenomenon of phantom limbs will have been turned from something that is often distressing to those experiencing it, into something of great benefit.