Neuroscientist Dr Edward ‘Eddie’ Chang spoke at the Falling Walls conference in Berlin about his groundbreaking work. The Brilliant’s founder Kylie Ahern caught up with him there.
A stroke victim who hasn’t been able to speak or move for 15 years thinks of what he would say if he could. Specialised sensors attached to his brain identify the word he is thinking of and feed it to a computer that displays and vocalises it for him.
No longer the stuff of science fiction, such technology could change the lives of patients with traumatic brain injury. It’s the work of Dr Edward Chang, an award-winning neuroscientist and chair of the Department of Neurological Surgery at the University of California, San Francisco (UCSF).
Chang’s work in decoding brain function, particularly in terms of speech, could help patients with injuries that prevent them from communicating efficiently, such as locked-in syndrome (characterised by complete paralysis of almost all voluntary muscles except for eye movements and blinking). “Similar technology can be used to monitor and stimulate and reduce seizures as well,” Chang told The Brilliant.
As well as such life-changing applications, research into how we process language is also helping to clarify what differentiates humans from other species.
“If you look at other non-human primates, like monkeys, there are similar parts of the brain that don’t seem to have all the same functions as what we have,” says Chang. “How they became specialised in humans is still very, very unclear.”
Decoding speech
Early in his medical studies at UCSF, Chang was hooked by neuroscience. “One of the things that excites me so much about neurosurgery and neuroscience is that there’s so much to learn about the brain – the human brain in particular,” he says. “I thought about other career options, like cardiac surgery, but there’s so much to learn about the brain that I thought this is going to be exciting.”
Chang recalls one of his neuroanatomy teachers taking him to view live neurosurgery.
“It was an awake surgery – the patient was awake,” says Chang. “The surgeons were stimulating different parts to map the brain. I said, ‘That’s what I want to do.’ I took some time off medical school and spent two years in a research lab. I used microelectrode technology in rodents to study how rodents hear sounds when they’re babies. It got me thinking a lot about how the human brain may process speech.”
Chang was given the opportunity to explore the language centres of the brain as he cared for patients with epilepsy and brain tumours. “We would implant these electrodes on the brain of patients with epilepsy to figure out where the seizures were coming from,” he says, “While the electrodes were there, patients volunteered to participate in studies where they basically spoke into a microphone as they were waiting for their seizures to happen. The electrodes that were sitting there for epilepsy actually turned out to be extremely good for also studying how the human brain works.”
Initially, Chang and his team investigated how the human brain forms phonemes – segments of speech, such as vowels or consonants. Instead, they found a code for something “even more basic”, he says: the way in which mouth parts move to create individual sounds.
“So, for example, when you round your lips to make an ‘oo’ sound, that’s a feature. Or when your tongue goes to the front of your teeth to say ‘this’ or ‘that’. Those are specific movements that are correlated with specific sound properties,” Chang explains. “There are about a dozen or so of those kind of movements that give rise to all consonants and vowels Those sub-parts that by themselves have no meaning, but in combination with other features, they give rise to all these sounds.”
Reading thought patterns
Then, Chang met Pancho, a patient in his mid-30s who had experienced a massive brain-stem stroke 15 years earlier and could only communicate by spelling out roughly three words per minute on a board using a pointer attached to his forehead. (Normal speech, by comparison, is about 150 to 180 words a minute.)
Chang and his team implanted a neuroprosthesis with 120 sensors on the ‘speech part’ of Pancho’s brain and connected it to a computer via a port through the scalp. Over time, the computer learned to recognise about 50 different words when Pancho thought about them.
“A man who had been paralysed for 15 years was able to express sentences for the first time,” says Chang, who published some of the results of his work with Pancho in The New England Journal of Medicine in 2021.
My hope is that this kind of technology will become a standard for people who have these kind of devastating injuries,” says Chang. “Now it’s about optimisation – how do we make it more accurate? How do we increase the vocabulary size? We need the final device to be fully wireless and implantable. Those are things that we’re working on.”
We are still at the “tip of the iceberg” of understanding how the brain forms and uses speech, says Chang. But some of the most fundamental questions, such as what specifically about the human brain has endowed us with the ability to process words and language, for example, are still a mystery.
Chang is also investigating the root causes of epilepsy and pursuing treatments for different types of tumours, such as the aggressive brain and spinal cord cancer glioblastoma. One option could be slicing out just enough tissue to effectively remove the cancer without affecting the patient’s brain function.
“People think that glioblastoma is rare, but it really isn’t. It is understudied and it is very different to most other cancers,” says Chang. “It’s also very difficult to treat because it’s usually not one single gene mutation, but multiple. It’s really tough. There’s something about [these cells] that makes the immune system unable to recognise them. If we can figure out that switch, and turn it off so the immune system can see it, we’re going to have a much better chance.”
Article by Kylie Ahern