A man who had been unable to speak or move may now produce words and phrases on a computer using only his thoughts.
The skill is derived from an experimental implanted device that decodes signals in the man’s brain, as reported by researchers at The New England Journal of Medicine today.
The man, who only knows 50 words and talks at a rate of about 15 per minute, is currently restricted to a vocabulary of just 50 words.
“This implies that it’s feasible,” Edward Chang, a neurosurgeon at the University of California, San Francisco adds. I believe there’s quite a lot of room for improvement over time.
Chethan Pandarinath, an assistant professor at Emory University and Georgia Tech, says a device that used brain pathways previously used for speech would be “more natural, and hopefully effortless” than current assistive technologies.
People with paralysis who have lost the ability to speak often use devices that utilize eye or head movements to spell out words letter by letter, or communication boards that allow them to spell out words one letter at a time.
Dr. Chang and his team placed a device called an electrode in the speech centre of this patient’s brain, which allowed him to speak for the first time since suffering from a stroke in 1995.
Dr. Pandarinath says they used transcranial magnetic stimulation to find the right spot, he says. “They are able to deliver magnetic pulses over this part of the brain and essentially stimulate it without having to open up the skull.”
To protect the man’s anonymity, Chang’s team sought for a better solution. The name refers to his status as the first patient in a study known as BRAVO, or Brain-Computer Interface Restoration of Arm and Voice.
Since he incurred a stroke 15 years ago, BRAVO1 has been paralyzed and unable to speak. “The stroke left him unable to speak or move any limbs,” Chang says, “except for one finger on his right hand.”
Dr. Chang and his team implanted a device in BRAVO1’s brain that detects signals from the speech centre and uses them to control a synthesized voice.
The implant is called NeuroLife, produced by NeuroLutions. It consists of a 4 by 4 millimeter grid of microelectrodes that penetrate the brain, recording electrical activity from about 40 neurons.
The electrode picks up thoughts that originate in the speech center and relays them to a computer, where they are translated into words using software created by the research team. The patient can produce approximately two words per minute. ”
Communication using dormant brain signals
Previously, Dr. Chang’s team developed a system to identify the brain signals associated with the desire to speak specific words. The technology proved successful in individuals who were still able to walk and converse after testing.
However, success was far from certain for BRAVO1, Dr. Chang points out. “We didn’t know if the speech commands in this patient would be strong enough to drive the voice synthesizer.”
The technology involved in decoding speech from brain signals is known as functional electrical stimulation, or FES. It’s similar to a technique used widely by physicians to help patients regain limb function after paralysis, Dr. Pandarinath says.
“So what you want to do is not just stimulate any part of the brain, you want to stimulate the area that controls your limb,” he says.
To figure out, the researchers installed sensors on the man’s brain surface. Then a computer studied the patterns of electrical activity generated when he tried to speak 50 distinct words.
The procedure took months. After BRAVO1 could consistently create words on a computer display, the team began having him construct sentences. To assist with accuracy, the team developed an analysis of the acoustic structure of different words.
“We needed to understand how each word is formed acoustically,” Dr. Chang says, “in order to infer what sound parameters are associated with that particular speech command.”
The system works somewhat like the texting software on most smartphones. “So, for example, if one word is just not decoded correctly, this auto correct feature may fix it,” Chang adds.
After months of fine-tuning, the man was able to produce a word each four seconds, or about 15 words per minute. “There’s still a lot of work that needs to be done,” Dr. Chang says, “but it’s very exciting.”
“High-performance neuroprostheses that can restore communication in people who are severely paralyzed represent the ultimate goal of neural engineering,” says lead author Chethan Pandarinath, MS, PhD candidate at Case Western Reserve University. ”
Krishna Shenoy, a professor in the School of Engineering at Stanford University, says a device that can interpret words in the brain might one day assist thousands of individuals who’ve had a stroke or traumatic brain injury.
Such a gadget may also help those with amyotrophic lateral sclerosis (ALS), a progressive condition that eventually renders speech impossible.
“I think the technology is still quite a few years away,” Shenoy says. “But Stanford and other institutions are putting massive effort into this.”
Unless artificial intelligence advances further, devices that can recognize whole words will be restricted to basic interaction.
Keeping some thoughts private and others public
Meanwhile, brain implants that allow people to spell out words are improving all the time. Last year, he and a team published findings showing that individuals could type quickly and correctly utilizing a device that decoded brain signals typically used for handwriting.
“We know considerably more today than we did 20 years ago about how to listen in on those thoughts in the brain and how to decode them and understand what they mean.”
While the technology may one day be used for practical purposes, such as typing an email or texting a friend, Dr. Buxton doesn’t anticipate that we’ll soon be communicating with others telepathically.
“I think most people would feel uncomfortable voicing their innermost thoughts to anyone except those closest to them,” he says. “I think it would be sort of like yelling out random thoughts in a crowded public park.”
However, listening in on someone’s conversations may infringe their privacy, according to Pandarinath. Because devices linked straight to the brain might make it difficult for individuals to discern personal thoughts from those they choose to share publicly, he claims.
“We want to ensure that the technologies we develop allow people to think privately without anything being shown around the world.”
Currently, the device still requires the patients to be hooked into a computer via their skull.
And while that might sound somewhat uncomfortable, Dr. Pandarinath says, they are only attached for four hours each day during which no more than two electrodes are being used at any one time. “It’s not intrusive,” he adds.
It may be simpler to use machines that employ brain signals to control muscles, according to on him. Because these signals aren’t usually sent unless a person makes a conscious effort to move, it’s easier with devices like this.