- By Thomas K. Grose

Artifical limbs are taking on new life thanks to advances in embedded-chip technology that are giving wearers a freedom of movement that would have been impossible a few years ago.

A common theme in popular science fiction of the late 20th century was the notion that horribly injured humans could be repaired using cyber-replacements for lost or badly damaged body parts. From the ‘70s television show “The Six Million Dollar Man” to the 1987 film RoboCop, heroes run on legs and throw punches with arms and hands crammed with electronic gadgetry that's hardwired to their brains. But what was once fiction is rapidly becoming a reality within the grasp of researchers in these early days of the 21st century. Because of embedded chip technology, artificial limbs permanently attached to amputees and controlled by their brains could be available some time within the next decade or two. And while it's unlikely that such prostheses could ever improve upon a real arm or leg, bioengineering is nevertheless offering America's 1.3 million amputees the real possibility of mechanical limbs that closely mimic nature's. “Based on recent results at Caltech and elsewhere, I can confidently say at this point it is a matter of when, and not if, such technology will be ultimately realized,” says Joel W. Burdick, deputy director of the California Institute of Technology's Center for Neuromorphic Systems Engineering.

Until that day, however, advances in embedded systems are already revolutionizing the design of prosthetic devices, giving wearers a freedom of movement that would otherwise have been impossible a few short years ago. For instance, Paul H. Chappell, an electrical engineering professor at Southampton University in England, is overseeing the design of an artificial hand that will move beyond a pincer grip to one that incorporates all five digits. Says Chappell: “We would not be able to do this without chip technology.”

In the early 1980s, long after “The Six Million Dollar Man” had been relegated to late-night reruns, “we were still giving patients a block of wood for a foot, they were still basically walking on a peg leg,” notes Stewart Atkinson at Seattle Systems, a leading manufacturer of prosthetics. The company's origins, in fact, were based on the marketing of an energy-returning foot that harnessed the energy of the foot striking the ground to ease the workload on wearers and help propel them forward. Other advances in recent years include a rolling-joint foot that uses progressive resistance to give users a more natural gait. In upper body prosthetics, the most popular designs use a patient's own back, shoulder, or neck muscles to control steel cables that move artificial elbows and hands—a technology that evolved after the Second World War. More recently, prosthetic arms, wrists, and hands have relied on sensors that pick up the electrical signals that emanate from an amputee's residual muscles. These electromyogram (EMG) signals are then amplified and drive actuators in the replacement limb. But as great and useful as these advances have been, they've all fallen short of closely replicating natural movements. That's because to make them more complex would mean making them heavier—and weight is inimical to prosthetics. Lightweight digital technology, however, is ushering in an era where the gap between natural and mechanical will be nearly, if not entirely, bridged.

For the vast majority of us, simple physical tasks like walking or picking up a coffee cup are no-brainers. We just do them, unconscious of the amazing amount of communication and mechanical dexterity our body is experiencing every nanosecond. “All the walking that you and I do, we never have to think about it,” says C. Michael Schuch, director of the Center for Orthotics and Prosthetics Care at the Duke University Medical Center. For every step we take, however, our nerves are shooting vital bits of information to our brains about such things as terrain, speed, and obstacles. And our brains just as quickly use the nerves to tell our muscles how to react or compensate. For wearers of prosthetic devices, that two-way communication is lost. Until recently, the most advanced prostheses required users to clearly think about every movement. But the chips installed into the latest generation of devices have introduced some degree of thought-free movement. These microprocessors, constantly fed data from sensors, allow the prostheses themselves to take over some of the automatic tasks that were once the province of brains and nerves. This devolution of control from person to machine is a significant improvement. It makes, for example, an artificial hand “more like a real hand, autonomous,” says Michael E. Tompkins, founder and president of Animated Prosthetics, of Greensboro, N.C., which has—with the help of the Duke Medical Center—created two products based on chip technology that improve upper body prostheses.

Myoelectrics give artificial arms and hands advantages over prosthetic legs because they can receive some information from the brain by way of electric impulses shooting through muscles. But they don't work for all amputees. Old analogue versions that run off circuit boards are not easily adjustable and often don't work if a patient's EMG signal is too weak or scattered. “If you don't have the right signal, you're stuffed,” Schuch says. But two digital products that Animated Prosthetics has developed with Duke solve that problem, and they can be adapted to any prosthetic. The Animation Control System (ACS) relies on sensors and a microprocessor to better read and amplify EMG signals and to manage power and movement. The Prosthesis Configuration Unit (PCU) is a remote diagnostic tool that “reads” what's going on inside the arm and sends the information to a handheld computer. “Now we can fit (a prosthesis) to any patient, no matter what the EMG is,” says Schuch, who uses the PCU to read the signals. Meanwhile, because the onboard computer reacts to the data from the sensors and the EMG signals, “we take unnecessary energy and wasted motion away from the patient. Once a hand is closed, for instance, it won't try to close it again,” Tompkins says. This makes things more natural for the wearer. Using a conventional myoelectric arm, a patient must tense a muscle to open a hand and tense it again to close it. A chip-enhanced arm will close a hand once a muscle relaxes, which is how a normal hand operates.

The most popular myoelectric arm is the Utah Arm, developed about 20 years ago at the University of Utah and manufactured by Motion Control, which is run by Harold Sears, a former University of Utah researcher. Motion Control released a second version of the arm in 1997, which updated its electronics but kept it an analogue-based product. It will soon, however, launch a third version that will incorporate embedded chips—a technology that it has already placed into its ProControl 2 wrist and hand products. The current Utah Arm allows only one degree of freedom at a time. The amputee must position and lock the elbow into place before using the hand, for example. Chip technology “will allow two functions at once, moving the elbow and hand simultaneously,” Sears explains.

And researchers and manufacturers expect embedded technology to add further refinements to upper body prosthetics. Today's artificial hands make use of only the thumb, forefinger, and index finger. The ring finger and pinky are cosmetic, and the thumb moves only back to front, not side to side. This enables a pincer grip, but not a full-hand grip that one would use for, say, holding a baseball. At England's Southampton University, Chappell is overseeing work on a hand that would allow use of all four fingers and give the thumb a two-way motion. It would also increase sensitivity to slip, so that the hand would automatically put a firmer grasp on an object if it began to slip. Tompkins, who still works with Duke researchers and patients, is also working along similar lines. And he wants to develop better processing. “Extracting information from a patient is not easy to do with all the electrical activity in the arm, so we need better signal processing,” he says, which will likely mean smaller processors moved closer to the residual arm. Tompkins also hints at adopting wireless technology to send signals to and from the processor.

Key Part

Most of the focus in lower-body prosthetics is on the knee, because when it comes to walking, it is the most critical joint. It must lock in place to give us stability and it must flex to allow for the swing phase. The first artificial leg to make use of a computer chip was based on Japanese technology and manufactured by Britain's Chas. A. Blatchford & Sons (Endolite North America in the United States). It was introduced about a decade ago and used a processor to control the swing phase to give amputees more control over speed. But Schuch says it was never marketed well and, subsequently, not many were sold.

In 1999, the U.S. Food and Drug Administration approved the C-Leg, made by German manufacturer Otto Bock, which is now the industry standard. The original concept for the C-Leg came from Kelvin James, a biomechanical engineer in the Division of Neuroscience at the University of Alberta in Canada. It has two processors and four sensors—which relay data at a rate of 50 times a second—to operate its mechanical and hydraulic systems. Two piezoelectric strain gauges measure pressures on the leg and how often the heel strikes, while magnetic sensors keep tabs on the angle of the knee. Schuch explains that before the C-Leg, artificial knees sacrificed speed for stability. Each time a heel strikes the ground, a human knee must flex a tiny bit, just 5 to 7 degrees, then relock before opening to allow for the swing. All that maneuvering is done within split seconds, and it's a bit too much for a conventional prosthesis.

Traditional artificial knees had to lock fully in place between swings so they could accommodate the patient's weight, and all that locking wasn't helpful in negotiating anything but a clear, flat surface. “All it takes is a rock under the heel to trip someone up,” Schuch notes. And because an artificial leg is just so much weight affixed to a stump, descending stairs meant going slowly, one step at a time. But the C-Leg's microprocessor compensates for those nuances, and users can walk naturally—and with little forethought—over all sorts of terrain, and they can dash down stairs. That last advantage turned into a life-saver for Curtis Grimsley, a computer programmer for the Port Authority of New York. He worked in the World Trade Center, and when terrorists smashed two commercial jetliners into the twin towers, eventually collapsing them both, Grimsley escaped because his C-Leg enabled him to vault down 70 flights of stairs—an accomplishment that earned both him and his German-made leg worldwide attention and admiration. “It's really beautiful technology,” he told the New York Times.

There is competition afoot, however. At the Massachusetts Institute of Technology's Leg Laboratory, researchers are developing a knee that will keep adjusting itself to the wearer. Co-director Hugh Herr declined to talk about the MIT project, saying he was waiting until it is commercially available (from Ossur, a manufacturer in Iceland). But previous accounts and discussions with other researchers indicate that the MIT knee will use a magnetorheological fluid, rather than oil-based hydraulics, to shift metal plates and control resistance within the knee.

Seattle Systems, in conjunction with the U.S. Department of Energy's Sandia National Laboratories, is working on a “smart leg” it has christened SILL, for Smart Integrated Lower Limb, and much of the research is being carried out in Russia by scientists at a former nuclear weapons facility. Seattle System's Atkinson says a working prototype is about a year away, but if all goes according to plan, the SILL will incorporate about a dozen sensors feeding data to just one microprocessor that will control not only the knee but the ankle, foot, and socket, as well. Its ankle, knee, and socket will be powered by piezoelectric motors and hydraulic joints, and kinetic energy from knee and foot motion will be recaptured and used to aid propulsion. Atkinson admits that despite nearly $5 million in research money, it's been a difficult project that's behind schedule. “It's not easy having all those components communicating with one another,” he says. “This is research, so there are no guarantees.”

Mark R. Pitkin, an assistant professor of bioengineering at the Tufts University School of Medicine who designed the rolling-joint foot, says he's not convinced that digital technology is particularly needed for below-the-knee prosthetics. Advanced mechanics that attempt to mimic the human skeleton can provide all the needed stability and propulsion, he says. His foot, for instance, uses the natural limitations of progressive resistance to move smoothly. Pitkin likens progressive resistance to the feeling one gets when rocking back and forth in a rocking chair—the roll starts easily but gradually slows to a stop, then a reverse motion begins. Schuch, of Duke, agrees with Pitkin: “We are able to do so much now (below the knee) with simple technology, that (an embedded system) is not necessary.”

While the C-Leg is clearly a major accomplishment, there is one thing it can't help a patient do: walk up stairs easily. The current state of battery power isn't enough to let the knee lift a wearer's weight. Indeed, C-Leg originator James says further breakthroughs in prosthetic devices will need to come from better power sources. “The limiting factor is carrying enough power battery storage. The limiting technology is not chip technology, but battery technology,” James explains. That's not to say there haven't been improvements in power sources. An internal lithium ion battery keeps the C-Leg moving for up to 35 hours. Animated Prosthetics' ACS also uses a lithium ion battery that Schuch calls a vast improvement over the nicad batteries that powered previous myloelectric arms. The nicad batteries needed very regular recharging, had low memory, and lasted just a year. Improved power plants will, however, be needed to make permanently attached prosthetics a reality. One potential solution may be miniature fuel cells, which use a clean chemical reaction to generate energy. Says James: “We're all waiting on the chemical engineers.”

Throughout history, prostheses have been detachable devices. But in Sweden, one researcher is pioneering the permanent attachment of above-knee artificial legs to the femur bone using a titanium implant. Known as osseointegration, it's a technique developed by dentists to attach false teeth. Schuch calls the procedure “a moderate success” and says the main problem is bacterial infections—always a risk when the skin is penetrated. Nonetheless, he and other experts expect that problem will eventually be overcome. And that will be a triumph, because a direct skeletal fit would eliminate the need for a socket. That solves an ongoing problem with all artificial limbs: obtaining a smooth interface between the socket and the residual limb.

Straight to the Source

The Holy Grail of prosthetics, however, is a direct link between a human-made limb and the wearer's brain. Truly the stuff of sci-fi. But, as noted by Caltech's Burdick, it is a future that will be realized relatively soon—though many hurdles must first be overcome. “The human body is an incredible piece of engineering, so artificial replacements aren't that easy to do,” says Chappell of Southampton. Nevertheless, within the last two years there have been impressive strides in neural prosthetics. Caltech and three other research groups have used signals extracted from monkey brains to control an external mechanical device and images on a computer screen. The research at Caltech looks to implant custom-designed chips into the brain as close as possible to the neurons. So much for the hard wiring. Transmission of the signals from the implants to the prosthetic would likely be wireless, thus minimizing the risk of infection that a wire into the head would run. Powering the chips would have to be done wirelessly, as well. But powering them is a delicate task. If they're too powerful and emit too much heat, surrounding brain tissue could be damaged.

Other researchers are looking at ways to connect a semiconductor chip to nerve cells. This is also a difficult area because there is a conductive cleft between the neurons and the chip that acts as a barrier. Get the chip too close to the cells' membrane, and they disconnect. Peter Fromherz, of Germany's Max Planck Institute for Biochemistry, recently presented a paper that described the creation of a pulsating hybrid circuit whose signals can be read both ways across the cleft. Fromherz has said the technique could open the door to neural connections for prosthetics.

Beyond limbs, bionics may some day give vision to the blind. In 1996, Wentai Liu, an electrical engineering professor at North Carolina State University, oversaw a team that created an Artificial Retinal Component Chip, a microchip that could be implanted in an eye to replace a damaged retina. Light hitting the chip's sensors is converted into electrical signals, read by optic nerves, and translated by the brain. Liu's team has since worked on improving the chip while awaiting FDA and Federal Communications Commission approval for the first implant, which could finally happen in early 2003. Since its creation, the chip has gone from containing an array of 25 pixels to 64, and Liu says 1,024, or even 10,000 pixels, may ultimately be needed. That would require a big boost in power, and like Caltech's electrodes, that means finding a way of controlling heat. And the chip must be properly sealed so as not to be corrupted by body fluids. Sighs Liu, “It's a continuing process.”

After considering the pros and cons of neural connections and osseintegration, Schuch is convinced that “somewhere down the road, those technologies will hook up.” But these solutions will almost certainly be very expensive. Nothing on the market now, for instance, is particularly cheap. The C-Leg costs $40,000, a tidy annual salary for many people. And even the chip-less current version of the Utah Arm sells for $60,000. Because the market for prosthetics is limited and rather small, economies of scale never kick in. So while permanent cyber-limbs might be good value for amputees who are young adults or children, that may not be the case for older patients. And, the sad fact is, the majority of amputees are elderly and often suffering from diabetes. Nonetheless, embedded chip technology will soon give us the chance to fuse human beings and machines. And that's an opportunity we're unlikely to walk away from.

Thomas C. Grose is a freelance writer based in Great Britain. He can be reached by e-mail at tgrose@asee.org.