Carl T. Hall, Chronicle Science Writer, Monday, November 10, 2003
view original

New Orleans -- Less than a month after a widely heralded experiment showed how thought-reading implants can work in monkeys, scientists presented new findings Sunday suggesting such machines could work in people, too.

Dr. Miguel A.L. Nicolelis of Duke University said previously unreported

human experiments demonstrated success with one type of a so-called brain computer interface, or BCI.

He and others discussed their latest findings Sunday at the annual meeting

in New Orleans of the Society for Neuroscience, the world's largest

gathering of brain researchers. About 28,000 people are attending the

weeklong event.

Much of the attention on Sunday was given to technology designed to

overcome paralyzing injuries or illnesses afflicting the nervous system.

About 11,000 new cases arise every year, adding to a total estimated at

more than 200,000.

Nicolelis said the new study had been done in a few Parkinson's disease

patients while they were undergoing open-skull neurosurgery for their

disease.

Full results, he said, have been submitted for peer review to a scientific

journal and were not a formal part of the program, in which he and

colleagues reported new details from the monkey experiments already

published.

Nicolelis said the important point was that the principle had been shown

to work: People can control devices merely by thinking.

Ultimately, it may be possible to design high-tech implants that can read

and direct the muscles using the patient's own intentions and natural

sensory equipment.

For now, it's a much less grandiose business of just tuning the equipment

to the human brain's frequency.

In the Duke experiments, patients were being fitted with standard

electrical stimulator devices, which can help to control Parkinson's

symptoms.

This procedure requires the patient to be awake while the surgeon

identifies a safe route through brain tissue, taking care not to harm

brain cells needed for essential functions. As part of that process, the

surgeon periodically asks the patient to speak or move while recording

localized brain activity.

Nicolelis and his colleagues took advantage of the opportunity and

recorded the information the surgeon was obtaining. Then, for five-minute

periods while the patient was being operated on, they conducted simple

reaching-and-grasping experiments to determine whether the patient's

intentions could accurately be read -- the first essential step in

controlling a limb by computer implant.

That's a far cry from proving that a workable long-term implant would be

safe and effective. Nicolelis said it was much too soon to "even think

about" moving any particular device into full-blown clinical trials.

A competing group, however, led by founders and collaborators of a company

called Cyberkinetics Inc., has announced plans to begin a small safety

study next year of an implant designed to allow a paralyzed patient to

control a desktop computer.

That device, called "BrainGate," is based on research at Brown University,

led by scientist John Donaghue. He and other company officials described

the technology on Sunday as a "novel gateway" for people with no other

options.

"These are the opening days of a new era in neurotechnology," Donaghue

said. The competition, however, has gotten somewhat testy of late amid an

explosion of interest. Some scientists accuse Nicolelis of overreaching,

noting that his latest monkey experiment actually wasn't the first to show

a "thoughts-into-action" device could function in a primate; he was merely

the first to show that a monkey's brain firings could be harnessed to

direct complicated movement, involving both reaching and grasping.

Meanwhile, Nicolelis decried the entry of corporate interests into a field

once thought to be purely science fiction, now being taken seriously as

modern medicine at the cutting edge of technology.

"I am a university professor," Nicolelis said. "I have no interests in any

business. I am Brazilian -- I want to have fun, I don't want to make

money. What I am very afraid of is that people who really want to make a

buck out of this will be rushing into the clinical thing. I don't believe

in that. A lot of important science needs to be done, and we need to go

step by step in a very careful way."

All the labs claim to be pursuing the technology responsibly.

Donaghue and his colleagues pointed out they were also university

scientists who realized the only way to fully exploit the technology was

to form a company capable of raising the money needed to carry out very

expensive clinical studies. Cyberkinetics is proceeding with the guidance

of the U.S. Food and Drug Administration.

In the latest studies on people, Nicolelis' Duke group had to use a

simplified version of the animal study protocol to stay within the bounds

of a five-minute surgical window. But that was still enough, Nicolelis

said, to show animal and human brains can be read much in the same way.

"We are showing the same computational algorithms work, the same

technology in general works, suggesting the principle would work in a

patient that is severely handicapped," Nicolelis said. "We are able to

predict the hand position, and the hand force, while they are doing the

task during the surgery."

Before you can lift even a finger, nerves fire in the brain, along the

spinal cord and nerve pathways of the arm, then back again in a tightly

controlled feedback loop.

Douglas J. Weber, of the University of Alberta in Edmonton, reported new

research Sunday suggesting that the motion of a limb can be accurately

predicted by reading the firings of just a handful of brain cells -- only

10 or so in one case.

That means it may be simpler than once imagined to tap into the body's own sensory apparatus to keep some natural motion going with a brain implant merely as a detour around a damaged spinal cord or other problem in the brain's natural circuitry.

Dr. Jonathan Wolpaw of the New York State Department of Health's Wadsworth Center described new methods of reading signals that can be detected outside and just beneath the surface of the skull, suggesting the

possibility that some devices may not even have to be implanted into the

brain. Implants run some risk of infections and other problems.

But he and others emphasized it might be several years before the first

such devices were ready for widespread use, and they noted that the

technology worked only in individuals who might be utterly disabled and

"locked in," with no ability to move even their eyes, and yet had enough

healthy brain activity to drive the implants.

The revolution will start slowly, Wolpaw said, in a few people "who are

the most disabled and who have no other options."