Using neural activity recorded from a sheet of electrodes laid directly on the surface of a patient’s brain, scientists can predict the movement of fingers, as well as which of several sounds the patient is imagining. Eventually, researchers hope to use the findings to develop intuitive neural prostheses, such as a robotic hand that moves its fingers with as little mental effort as it takes to move real ones, or a computer interface that detects imagined words. To realize this vision, scientists are also developing smaller, more flexible technology, which could be more easily implanted and make better contact with the brain. Details of the latest brain-computer interface technology were presented this week at the Society for Neurosciences conference in Washington, DC.

“It could create the basis for a brain-computer interface that is very intuitive, and a recording platform that is very robust,” says Gerwin Schalk, a research scientist at the Wadsworth Center, in Albany, NY, who led one of the projects.

Schalk and his colleagues studied epilepsy patients undergoing a procedure known as electrocorticography (ECoG), in which a flat array of electrodes is laid over an exposed section of cortex to record electrical activity. Normally, surgeons use this information to pinpoint the source of seizures and to map the location of specific brain functions, which must be avoided during surgery. The technique generates a better spatial resolution than electroencephalography (EEG), a noninvasive approach that records activity through the scalp. ECoG is now being explored for use in brain-computer interfaces. “There’s a growing interest in use of ECoG signals because nothing penetrates into the brain, and that appeals to people more than penetrating electrodes,” says Marc Schieber, a physician and scientist at the University of Rochester Medical School, who was not involved in the research.

Schalk and his collaborators recorded electrical activity from the motor cortex and Broca’s area, a part of the brain involved in speech, in five patients as they moved their hands and fingers in specific ways and vocalized or imagined specific sounds. The researchers then used specially developed algorithms to search the neural activity for patterns relating to a certain movement or sound. “We can tell you how they are flexing each of their fingers,” says Schalk. What’s more, the researchers could determine in real time which of two sounds a patient was imagining. This kind of information could be used to control a brain-computer interface, providing a lifeline for people with severe paralysis, such as that associated with end-stage amyotrophic lateral sclerosis, a neurodegenerative disease, or locked-in syndrome, the result of a specific kind of stroke that leaves the patient unable to move or communicate.

“If you’re paralyzed and can’t speak but your cortex is still okay, the ability to transmit a few words like ‘yes’ or ‘no,’ ‘food’ and ‘water,’ could be very useful,” says Schieber. “But the question is, will we be able to decode all the phonemes of human language from ECoG signals? Can you get enough specific information to distinguish different kinds of grasps, like a pinch versus how you hold a hammer?”

In order to use the information to control a prosthesis or computer, scientists will also need to be able to extract the relevant information in real time. (In the current project, the analysis was done after the neural recording.)

Schalk and others are studying ECoG as a possible alternative to electrodes implanted into brain tissue. Scientists have made rapid progress using the latter as an interface for prosthetic devices, recently showing that monkeys can feed themselves with a robotic arm controlled by a brain-computer interface, and paralyzed patients can move a cursor on a computer screen using similar equipment. It’s not yet clear that ECoG, which records extracellular electrical activity and thus averages information coming from different cells, will be able to provide the same accuracy as implanted electrodes, which record activity from single cells. “As far as limb control, I think it will be somewhat basic,” says Andrew Schwartz, a neuroscientist at the University of Pittsburgh.

However, ECoG possesses some significant advantages. With implanted electrodes, the quality of the recorded signals degrades over time, and the stiff electrodes can sometimes move within the squishy brain, thus requiring recalibration of the system. ECoG devices are less sensitive to movement. And because they lie on the surface of the brain, they may be less susceptible to the immune reaction thought to impair implanted electrodes. “Surface electrodes are more likely to be fit for long-term use,” says Schalk.

Miniaturized ECoG devices now under development may make this technology even more appealing. With the current procedure, a surgeon must remove a large piece of skull to insert the electrode array. But Justin Williams, a biological engineer at the University of Wisconsin-Madison, is developing a miniature ECoG device that could be fed through a small hole in the skull and then unfurl to cover a larger area of the cortical surface. Made of platinum wires embedded in a flexible polymer called polyimide, which is frequently used in electronics, the electrode array is flexible and sticks to the wet brain. That means it moves as the brain moves, capturing a better signal. “It acts like Saran wrap on a Jell-O mold,” says Williams.