Bionics: Implants & Interface

On February 1, 2010, in Bionics

Cyborgs, also known as bionic people, have been among us for some time. A heart patient with a pacemaker implant, a Parkinson’s patient with a deep-brain stimulator implant, an amputee fitted with a C-Leg… these and many more are machine-enhanced humans—cyborgs, or bionic people.

Most bionic devices need to be activated and deactivated automatically, for obvious reasons. You don’t want to start searching for the “On” switch on a box in your pocket or on your belt after your heart starts fibrillating out of control, or your Parkinson’s hands twitch uncontrollably, or your epileptic seizure has you writhing on the floor, or your C-leg stays rooted to the spot when you try to gat out of the way of an approaching car.

That means bionic devices need to receive, understand, and respond appropriately and automatically to signals the nervous system sends out to the appropriate, but missing or disabled, muscles. That in turn means there has to be an interface—a connection or usually a whole set of connections—between the nervous system and the bionic device. Such interfaces have come to be known generically as the “brain–machine interface” or BMI.

Wired vs. Wireless BMI

BMI devices come in two types: One with the interface wired into the brain itself, and one interfacing wirelessly through sensors placed on the scalp. The direct-wired interface is obviously invasive: it requires opening up the skull. It is therefore risky and expensive, but it captures and produces higher quality signals than the wireless interface.

A wired BMI consisting of implanted electrodes has been available for some time to monitor seizures in epilepsy patients. Recently, that technology was adapted to enable two patients to type letters on a screen by simply concentrating their thoughts on each letter in turn. This has been done before using a wireless BMI, but the new approach should provide a faster communication rate and be much better at reading complex brain signals.

Brain Implant for Seizures

Similar technology—a brain implant designed to detect and block the onset of seizures—can significantly reduce the frequency of seizures in the 30 to 50 percent of epilepsy patients for whom medication does not work. Regulatory approval for the “Responsive Neurostimulator” is to be sought this year. It joins a handful of similar electrical stimulation devices approved or in development for epilepsy patients, but unlike those other devices, which stimulate the nervous system continuously or in a predetermined pattern, the Responsive Neurostimulator only fires when it detects abnormal electrical activity signaling the onset of a seizure.


In a trial of nearly 200 patients, the device reduced the frequency of seizures by 29 percent on average, compared to a 14 percent reduction in patients given an inactive implant. Nearly half the active implant patients had their seizures reduced in frequency by more than 50 percent. Although this response is similar to that of some other devices approved or in trial, the patients in the Responsive Neurostimulator trial had suffered especially severe epilepsy over many years. A third had tried vagus nerve stimulation and a third had undergone surgery without success.

BMI For Real-time Thought Expression

Another wired BMI has enabled a patient with locked-in syndrome (resulting from a stroke) to communicate. The patient’s cognitive abilities are intact but he is paralyzed except for slow vertical movement of the eyes. His BMI implant—essentially, an electrode and FM transmitter—was installed five years ago near the boundary between the speech-related pre-motor and primary motor cortex. Neurites have since grown into the electrode. Their signals are amplified and transmitted by the implant to an antenna temporarily glued to the patient’s head. The implanted electrode is powered by induction from a power coil also attached to the head.

A laptop computer system translates the received neural signals, that were basically intended for the vocal cords, and converts them to sounds as quickly as a non-paralyzed person would say them, although so far only three vowel sounds have been tested. After 25 sessions over a five-month period, the vowels were correctly reproduced 89 percent of the time.

The next version of the system will produce consonants and more vowels, and will record ten times as many neurons. The system has the potential eventually to enable real-time conversation and thereby minimize the social isolation that accompanies profound paralysis.

Neuron-Prosthetic Interfaces

For patients who lack not just control of a limb but the limb itself, there are ways to interface the brain, through the nervous system, with a robotic prosthetic limb. To give you a sense for what is involved, let us briefly describe one new method of doing so:

  1. Harvest cells from healthy muscle in the patient’s body.
  2. Place the cells on a scaffold on the inside of a microscopic cup made from an electrically conductive polymer.
  3. Position the cup on the ends of a motor nerve at the point where the limb was severed.
  4. Do the same for a sensory nerve.
  5. Wait for the muscle cells to proliferate in the cups, and for the nerve endings to grow into the cups and attach to the muscle cells.
  6. Do this for as many nerve endings as possible—the more, the better.
  7. Take the electrical signals coming from the motor neurons into the muscle and passing through the electroactive plastic, and broadcast them to a computer controlling a robotic prosthetic limb attached to the patient.
  8. Heat sensors, pressure sensors, and other sensors in the prosthetic send signals the other way, via the sensory neurons, to inform the patient’s brain that the prosthetic is touching something hot or cold, hard or soft, etc.

Most of this has been done in rats with a severed peripheral nerve, using just two cups and one motor neuron and one sensory neuron. This particular research project is still in its early stages but something approaching this fine level of interface must have already been used to build the state-of-the-art “SmartHand” developed by a multinational team in the EU. This artificial hand has been successfully wired to existing nerve endings in the stump of a Swedish patient’s severed arm. It resembles a real hand in function, sensitivity, and appearance.

The patient is able not only to complete extremely complicated tasks such as eating and writing but also to feel his artificial fingers. He told a television interviewer: “I am using muscles which I haven’t used for years. I grab something hard, and then I can feel it in the fingertips, which is strange, as I don’t have them anymore. It’s amazing.”

The SmartHand contains four electric motors and 40 touch sensors. Future versions will have artificial skin providing even more tactile feedback.

* * *

Where a limb is still attached but paralyzed as a result of spinal cord injury, it has proved possible to re-wire individual neurons to bypass damaged areas and connect directly to muscles in the paralyzed limb. This has been achieved in monkeys whose arms were temporarily paralyzed for the purpose, leaving them unable to use their wrists. Brain signals were re-routed around the blocked nerve pathway via a wire running between a randomly selected neuron in the motor cortex, a computer, an electric shock generator, and a muscle in the arm. Whenever the neuron fired above a certain rate, the computer told the generator to shock the arm muscle, causing the muscle to contract.

The monkeys quickly learned to move their paralyzed wrists to play a video game.

The researchers involved in this project were surprised to find that “nearly every neuron” they tested in the brain could be used to control this type of stimulation. “Even neurons which were unrelated to the movement of the wrist before the nerve block could be brought under control and co-opted,” they said. They also showed that a single neuron could work two different muscles: a high firing rate triggered the wrist to flex, while a low firing rate caused it to extend. (Still, using more than just one neuron would be better.)

In similar research conducted elsewhere, a monkey was able to pick up a ball via a wired BMI implant known as a functional electrical stimulation (FES) device. FES is already used in US Food and Drug Administration-approved implants that can restore hand function and bladder control to some paralyzed patients. One existing FES implant enables the patient to shrug a shoulder to cause their hand to grasp an object, though it cannot yet enable the patient to control the strength of the grasp.

The FES implant in the monkey brain records signals from the motor cortex, which a computer then decodes into specific muscle movements in five flexor muscles in the arm, enabling the monkeys to grasp objects and move their wrists in different directions. In a test involving putting a ball into a hole, which the monkeys could do every time before their nerves were blocked, the paralyzed monkeys succeeded only about 10 percent of the time until the FES system was turned on, when the success rate rose to 77 percent.

A similar system is already working in a human paralyzed patient who is now able to control a computer model of an arm (not yet an actual prosthetic.) Human trials of this system, with an actual prosthetic, could begin within a year or two.

Shrinking Brain Implants

The Responsive Neurostimulator we talked about earlier is the size of a small deck of cards and a cavity has to be created inside the skull to make room for it—a risky business. The large size of this device, we would hazard a guess, is probably due to its batteries, since implant chips themselves can be made very small. An experimental sensor platform called NeuralWISP fits multiple electronic components onto a circuit board currently just over two centimeters long and a future version will put them all into a single chip one by two millimeters in size.

With reduced size comes reduced power consumption, which makes it feasible to draw power wirelessly from a radio source up to a meter away, instead of from bulky batteries inside the implant. So far, the device has only been tested in moths. Getting signal and power from/to an implant centimeters deep in the human brain will no doubt be more difficult than getting it a few millimeters into a moth, but nanotechnology might eventually solve that problem.


The devices we have discussed so far have not been nanoscale devices. At best, they are “MEMS” (micro-electromechanical systems)-scale devices. MEMS engineering was an early enabler of bionics, facilitating the development of complex devices only a few centimeters (like the present NeuralWISP) or even millimeters (like the future NeuralWISP) in size. MEMS devices are clearly functional and useful at a gross level within the human body, but for the ultimate in utility and control it is necessary—certainly at the interface, where we are talking about individual neurons with diameters measuring mere billionths of a meter—to move several orders of magnitude down to the nanoscale.

The electrodes used in brain implants today typically record the electrical activity of masses of neurons per electrode. Yet as long ago as 2006, Harvard researchers used nanowire transistors to measure electrical signals not just in a single neuron but at 50 points within a single neuron. The same group has now developed a nanowire recording system and has used it “to capture some of the most precise, high-quality electrical recordings ever made from heart cells.”

The system clearly has potential application in neural prostheses and other medical devices and indeed the Harvard researchers have already begun to make recordings from neural tissue. Furthermore, they are working on nanowire devices that can simultaneously record electrical signals as well as hormone, neurotransmitter, and other chemical signals which “would give a more integrated picture of biological functions.”

Lipid-coated Nanowires

Similar (possibly connected—we don’t know) research at the Lawrence Livermore National Laboratory has resulted in an experimental artificial neuron (that’s our term for it) made of silicon-nanowire transistors encased in a fatty acid membrane. Because the membrane contains ion channels and can therefore also respond to proteins and other biomolecules, electrodes connecting a prosthetic device with the nervous system could read chemical as well as electrical signals, thus providing finer and more natural control over the prosthetic.

The first prototypes have only one type of ion channel, which limits their functions. The next generation will have more. The researchers will also begin testing the artificial neuron’s interactions with living cells.

Bare Nanowires to Electrify the Body

Nanotechnology is also able to solve a problem with current medical devices, namely, that their electronics have to be isolated from the body and are built on a rigid silicon platform. The solution is nanoscale silicon circuits built on a thin film of silk. These are biocompatible, conformable to tissue shape, and are partially biodegradable—the silk dissolves over time and the circuits left behind are too thin to irritate the tissue.

Silk–silicon LED devices under development could act as “photonic tattoos” showing (for example) blood-sugar readings, and as arrays of conformable electrodes that could interface with the nervous system. Such devices have apparently been implanted in animals with no adverse effects to the body or to the devices’ performance. The group undertaking this research is currently designing silk-based electrodes that could be wrapped around individual peripheral nerves to help control prostheses. The conformability of an array of silk electrodes would enable it to be used in otherwise inaccessible crevices deep inside the brain.


Our ability to interface at the neural level—and increasingly at the individual neuron level—with computers and robotic machines is accelerating. While this has obvious implications for healthcare in the sense of being able to offer treatments for people with disabilities, it has longer-term deeper implications because it promises to enhance the cognitive and motor capacities of perfectly healthy and able-bodied people as well. It is the dawn of the age of Superman and Superwoman. We kid you not.


Leave a Reply

Your email address will not be published. Required fields are marked *