Reconstructing Hand Movements Using Brain Signals

news story image

Bioengineering graduate student Trent Bradberry prepares a volunteer to use the EEG sensor cap that relays signals generated by the brain.

Fischell Department of Bioengineering graduate student Trent Bradberry, advised by Associate Professor José Contreras-Vidal (kinesiology, neuroscience and cognitive science program, and affiliate professor, Graduate Program in Bioengineering), published research in the March 3 issue of The Journal of Neuroscience which for the first time demonstrates that it is possible to decode and reconstruct 3-D hand movements from brain signals recorded through the use of noninvasive electroencephalography (EEG) technology. The new technique could someday enable those who have lost motor function to operate brain-controlled prostheses, computers or wheelchairs using a headset with scalp sensors that sends signals from the brain to the device. Bradberry's collaborators and co-authors on the project are Contreras-Vidal and Rodolphe Gentili (assistant research professor, kinesiology).

"Until now," says Contreras-Vidal, "this was not thought possible—people assumed EEG data was too limited."

The findings are significant because they open the door to the development of safe, portable, brain-controlled assistive devices for the neurologically-impaired or physically disabled. Prior to this study, researchers have used non-invasive but non-portable magnetoencephalography (MEG) technology and invasive methods that implanted sensors in the brain to reconstruct hand motions.

The reaching apparatus used to study the finger paths from a center button to eight other buttons in random order. Image courtesy of the Journal of Neuroscience.

The reaching apparatus used to study the finger paths from a center button to eight other buttons in random order. While volunteers touched the buttons, researchers recorded their brain signals and hand motions. Afterward, the researchers attempted to decode the signals and reconstruct the 3-D hand movements. (Image courtesy of the Journal of Neuroscience.)

The team placed an array of 34 EEG sensors on the scalps of five participants to record their brains' electrical activity. Volunteers were asked to reach from a center button and touch eight other target buttons in random order at least ten times each. Bradberry and his colleagues recorded their brain signals and hand motions in order to interpret the brain activity that occurs when a person decides how to move. They found that one sensor in particular provided the most accurate information. It was located over a part of the brain called the primary sensorimotor cortex, a region associated with voluntary movement. Useful signals were also recorded from another region of the brain called the inferior parietal lobule, which is known to help guide limb movement.

"Our results showed that electrical brain activity acquired from the scalp surface carries enough information to reconstruct continuous, unconstrained hand movements," says Contreras-Vidal. "We are currently working with [controlling] robotic arms and wearable upper limb exoskeletons, but our findings could also lead to improvements in existing EEG-based systems that are designed to allow people to control a computer cursor with their thoughts."

The information gathered by the ongoing study, he adds, could also help doctors understand how the progression of neurological disorders such as Parkinson's disease affect the brain's ability to control the body's movements.

Bradberry's groundbreaking research has recently been covered by Scientific American, NPR’s Science Friday, and TV Globo News (Brazil; in Portuguese).

The research was supported by the Paris-based La Fondation Motrice.

Story adapted from the original University of Maryland press release, courtesy of Leon Tune.

For More Information:

See "Reconstructing Three-Dimensional Hand Movements from Noninvasive Electroencephalographic Signals," Trent J. Bradberry, Rodolphe J. Gentili, and José L. Contreras-Vidal. The Journal of Neuroscience, 30(9):3432-3437 »

Learn more about Trent Bradberry and his work »

Watch a demonstration at the School of Public Health's Healthy Turtle blog »

Published March 30, 2010