Thought-Controlled Prosthetics

Ponca City, We Love You writes "Physiatrist Todd A. Kuiken, M.D., Ph.D. has pioneered a technique known as targeted muscle reinnervation (TMR), that allows a prosthetic arm to respond directly to the brain's signals, allowing wearers to open and close their artificial hands and bend and straighten their artificial elbows nearly as naturally as their own arms. Doctors first perform nerve transfer surgery to redirect nerves that go to the amputated arm to the patient's chest muscles. Then when the chest muscle contracts, an electromyogram picks up the electrical signal to move the prosthetic arm. So when the patient thinks 'close hand,"' the hand closes. Now the team wants to see if they can extract more information from the electrical signals produced by the nerves to provide a greater number of hand and arm movements. Theyd have been able to identify unique EMG patterns with 95% accuracy for 16 different elbow, wrist, hand, thumb, and finger movements. 'We've been able to demonstrate remarkable control of artificial limbs and it's an exciting neural machine interface that provides a lot of hope,' says Dr. Kuiken."

Comments

And as a bonus...

[ChrisMounce]

After a couple months of using the hand, you get rock-hard abs!

nerve signals / muscle signals

[xristo70]

Obviously I'm not a neurosurgeon. I look forward to posts from the experts.

But what is the difference between the electrical signals from the nerves and those given off by contracting muscles? Since the nerves which carry the signals are known, why can't those nerve signals be read straight away? Is it a case of much easier signal patterns to identify with the electrical signals of muscles or just a question of signal strength or something much more complicated?

Interesting as well that they should say that when the muscles are touched, for the patient is seems like the prosthetic arm is touched. Too bad they don't mention the perceived sensitivity to temperature and pressure with this effect. Put sensors on the tip of the hand and a little device on his chest and you might give the patient movement and "feeling" as well.

Re:nerve signals / muscle signals

[WarlockD]

From what I understand, the communication with nerves not only have to be bi-directional but also we arn't sure what other signals are sent (chemical, etc). We can detect the messages to muscles as those are VERY simple and only require small electrical detectors. Decoding a straight nerve seems still seems beyond our reach.

Mind you, this is much better than before. Previous robotic arms are built this same way, but it takes months and months of training to use your chest muscles to move your arm. Now it looks like you don't need that much therapy since they rout your arm nerves to your chest.

PS - I am no expert, I just looked into it a bit ago when I met someone with a claw hand.

Re:nerve signals / muscle signals

[Alrua]

There is some research being done into bidirectional prosthetics. Kevin Warwick from Reading University in the UK has successfully implanted a chip in his own arm allowing him to control an external robotic arm and receive sensory input from it.

Some of Warwick's work is pretty controversial (see e.g. various articles from The Register), but he does do some solid research.

Wikipedia has more details

Re:nerve signals / muscle signals

[cortex]

The signals in the nerve can be detected directly, but the are very small, and it is harder to get a micro-electrode array in the nerve. The muscle acts like a bio-amplifier, so that the small impulses from the nerve are measured as larger electro-myographic signals (EMG). I am a neural engineer on f the team at the Univ. of Utah that is working on using the signals in the nerve directly. We can already decode the movement signals from the nerve directly and are investigating how to provide sensory feedback. We have been discussing with Todd Kuiken using our array to map out the sensory and motor fibers in the nerve prior to his surgery, so that he can achieve better separation of the signals. That is, he'll know which nerve fibers carry which signals prior to implanting them on the muscle. We and other universities (Caltech, Brown, U. Pitt...) are also looking into using signals straight from the cerebral cortex to control prosthetic limbs.

Re:Ok now for a non stupid post

[aussie_a]

I'm ambidextrous you insensitive clod!

Muscle vs Nerve signals

[spineboy]

The nerves in the body are usually buried somewhat deep, and are insulated usually by a layer of fat, and by their insulation(fatty Schwann cells). Since the nerves innervate the muscles, the signal becomes amplified, thus making it much easier to pick up the signal (stronger, and just under the skin, not insulated).

Another reason is that many different nerve fibers run together in a nerve, especially up in the brachial plexus (shoulder are). If this prosthesis is meant for people who have lost their are high up, then the nerves in this location, are somewhat big (between a pencil and strand of linguinni thickness), and contain many different fibers. There are about 30 different muscles in the forearm/hand, and another 20 in the shoulder and arm (and don't forget all the sensory fibers too). It might be just too hard to pick out usable signals from that mess, If some of the fibers are re-routed to a superficial muscle (chest wall pectoralis major), then it's much easier for the person to choose discrete movements, and have control over the prosthesis.

I am an orthopaedic surgeon, so I'm just posting this part to squelch any criticism about the facts above.