DARPA To Support Brain-Machine Research

DURHAM, N.C. -- Devices including "neuroprosthetic" limbs for paralyzed
people and "neurorobots" controlled by brain signals from human
operators could be the ultimate applications of brain-machine interface
technologies developed under a $26 million contract to Duke
sponsored by the Defense Advanced Research Projects Agency (DARPA).

The DARPA support will help launch Duke's Center for Neuroengineering,
co-directed by Miguel Nicolelis, professor of neurobiology, and Craig Henriquez, the W.H. Gardner Jr. Associate Professor of Biomedical Engineering. The center's scientists and engineers will seek to pioneer a new technological era in which brain signals could control machines that augment and extend human capabilities in a way never before possible.

The Duke center will consist initially of a collaboration of separate
laboratories in the medical center's department of neurobiology and in the
Pratt School of Engineering department of biomedical engineering.
However, the researchers expect to unite the center's efforts in the new
multidisciplinary engineering building now under construction.

Nicolelis will be principal investigator for the DARPA project. Co-PIs are Henriquez, Professor of Neurosurgery Dennis Turner and Patrick Wolf, associate professor of biomedical engineering. Other center collaborators include John Chapin of the State University of New York, Brooklyn, Jose Principe of the University of Florida, Mandayam Srinivasan of Massachusetts Institute of Technology and Harvey Wiggins of Plexon Inc. in Dallas.

The contract is part of DARPA's Brain-Machine Interfaces Program, which seeks to develop new technologies for augmenting human performance by accessing the brain in real time and integrating the information into external devices.

Besides development of brain-controlled prosthetic limbs, neurosurgeons could apply brain-mapping enabled by the new technologies to aid surgeons in distinguishing healthy brain tissue from that which is part of a tumor or a focus for epileptic seizures.

"This technology can immediately increase the resolution with which
surgeons can map the extent of a tumor or a specific brain region," said
Nicolelis. "Such improved mapping can translate into a better prognosis
for the patient, since less tissue might have to be removed."

Beyond medical uses, brain-machine interfaces also could be applied to
enhance the abilities of normal humans, said the researchers. As
examples, they said, neurally controlled robots could enable remote
search-and-rescue operations or exploration of hazardous or
inaccessible environments.

As part of the DARPA support:

  • Biomedical engineer Henriquez and his colleagues will coordinate
    development of equipment and methods for visualizing and analyzing the
    massive amounts of data produced from electrode arrays in the brains of
    experimental animals. 
  •  Neurosurgeon Turner and his colleagues will investigate potential use
    of brain-machine interfaces in patients with neurological disorders. 
  •  Biomedical engineer Patrick Wolf and his colleagues will develop a
    miniaturized "neurochip" for detecting and analyzing brain signals, as well
    as optical communications links between the chip and the control
    components of the interface. 
  •  John Chapin's laboratory will develop the sensory feedback mechanism
    by which animals and humans can "feel" the actions of a neurorobotic
    arm or hand. 
  •  Jose Principe and his colleagues will develop new computer
    algorithms for translating brain-derived signals into control commands to
    operate a robot arm. 
  •  Mandayam Srinivasan's laboratory will develop new interfaces to
    provide visual and tactile feedback signals to animal subjects operating
    robot arms, and 
  •  Harvey Wiggins of Plexon Inc. in Dallas will supply hardware and
    software that will enable development and testing of brain-machine
    interfaces.

According to Nicolelis, the initial concentration of the new center will be on
neuroprosthetic arms for paralyzed people, based on the success of
initial experiments with animals.

"Last year, we reported experiments in primates showing that a
brain-machine interface could, indeed, control a robot arm," said
Nicolelis. "While this was a first-generation system, it proved to us that
there was an enormous opportunity to pursue research leading to clinical
applications. We are extremely grateful to DARPA for their vision in
establishing a program that will provide the crucial support to launch this
effort."

In 2000, Nicolelis and his colleagues tested a neural system on monkeys
that enabled the animals to use their brain signals, as detected by
implanted electrodes, to control a robot arm to reach for a piece of food.
The scientists even transmitted the brain signals over the Internet,
remotely controlling a robot arm 600 miles away. The technique they
used, called "multi-neuron population recordings" was originally
developed by center collaborator Chapin.

In the experiments, the scientists used arrays of up to 96 electrodes to
sense signals from multiple areas of the brain, including the motor cortex
from which movement is controlled. The scientists then recorded the
output of these electrodes as the animals learned "reaching tasks,"
including reaching for small pieces of food.

The scientists fed the mass of neural signal data generated during many
repetitions of these tasks into a computer, which analyzed the brain
signals to detect tell-tale patterns that would enable researchers to
predict the trajectory of the monkey's hand from the signals.

Then, by programming the computer connected to the robotic arm to
sense these signal patterns emanating from the monkey's brain, the
scientists could enable the monkey to, in effect, control the arm only via
neural signals.

This proof-of-concept experiment showed the effectiveness of recording
from multiple areas of the brain and then allowing the computer to "learn"
brain signal patterns that triggered certain movements.

In the new center, Nicolelis, Henriquez and their colleagues will aim to
increase the number of recording electrodes to more than 1,000 to enable
control of more complex actions by robotic arms and other devices. The
"neurochip" being developed by Wolf and his colleagues will greatly
reduce the size of the circuitry required for sampling and analysis of brain
signals.

"Our dream is to develop a palmtop-like device that routes the signals
either to robotic devices, computers, or even to the physician, to alert the
physician to some problem," said Nicolelis. According to Henriquez, the
greater number of recording electrodes will also enable far more
sophisticated analysis of brain signals.

"This research involves a major effort to decode how the brain manages
information," said Henriquez. "Once we are able to use computation to
decode such information, we can translate that understanding into an
algorithm that can be incorporated into hardware." Ultimately, the
researchers hope to be able to record and analyze such signals for long
periods of time without damage to brain tissue, said the researchers.
They have already shown that animals can tolerate the electrodes for
periods of years without apparent harm.

According to Nicolelis, the technology and computational methods
developed under the DARPA support also will lead to a deeper
understanding of the brain itself.

"This research will provide us with a powerful new set of experimental
tools and techniques to answer the question of how millions of brain cells
come together to generate a particular behavior," he said. "Traditionally,
the neurosciences have taken a reductionist approach, with investigators
trying to understand individual neurons, molecules and genes. We are
trying to understand the brain's function as a dynamic system."

Nicolelis, Henriquez and their colleagues are among researchers
developing a theory that neurons are not hard-wired circuit elements
permanently assigned to one computing task, like the microprocessor
inside a computer. Rather, the new theory holds that neurons are
adaptable, living entities that can participate in many processing tasks at
once. Moreover, the theory holds that those tasks may change from
millisecond to millisecond. For example, Nicolelis' experiments have
revealed that the brain signals producing a single event, such as a
monkey reaching out, are mirrored in many places in the same brain
region -- as if the neurons "vote" on such actions.

In their current experiments, the center's scientists and engineers are
developing "closed-loop" systems, in which movement of the robot arm
generates tactile feedback signals in the form of pressure on the animals'
skin. Also, they are providing visual feedback by allowing the animal to
watch the movement of the arm.

Such feedback studies could also potentially improve the ability of
paralyzed people to use such a brain-machine interface to control
prosthetic appendages, said Nicolelis. In fact, he said, the brain could
prove extraordinarily adept at using feedback to adapt to such an artificial
appendage.

"One provocative, and controversial, question is whether the brain can
actually incorporate a machine as part of the neural representation of the
body," he said. "I truly believe that it is possible. The brain is continuously
learning and adapting, and previous studies have shown that the body
representation in the brain is dynamic. So, if you created a closed
feedback loop in which the brain controls a device and the device provides
feedback to the brain, I would predict that as people or animals learn to
use the device, their brains will basically dedicate neuronal space to
represent that device."

Development of the Duke center's brain-interface technologies also will
involve collaborations with industry, said the researchers. The market for
such devices should be considerable, they said. According to a market
analysis commissioned by DARPA, the current worldwide market of about
$270 million annually is projected to be $1.5 billion by 2005.

"In our discussion with corporations, we've found that, even though these
technologies are in their infancy, the companies are emphasizing their
commercial development," said Henriquez. "We believe that the Duke
center will help propel development of the next generation of brain
interface technologies. And the opportunities for their application seem
almost boundless."

DARPA (www.darpa.mil) is the central research and development
organization for the Department of Defense. It manages and directs
selected basic and applied research and development projects for DoD,
and pursues research and technology where risk and payoff are both very
high and where success may provide dramatic advances for traditional
military roles and missions.

The DARPA sponsored contract is being managed by the Space and
Naval Warfare Systems Center in San Diego
(http://enterprise.spawar.navy.mil).