Paralyzed patient speaks with power of thoughts using brain-computer interface

Researchers at Tel Aviv University and Tel Aviv Sourasky Medical Center succeeded in reading a patient’s thoughts with a speech-brain-computer interface, a groundbreaking new study.

A silent participant imagined saying a syllable. Depth electrodes, according to a press release, implanted deep in his brain transmitted the thoughts as electrical signals to a computer which then vocalized them.

Not only does this experiment provide a rare glimpse into the depths of the human brain, but it may also ultimately help completely paralyzed individuals express themselves once again.

The set-up for the experiment: An epileptic patient

Firstly, the study was a collaboration between Dr. Ariel Tankus of TAU’s School of Medical and Health Sciences, Tel Aviv Sourasky Medical Center (Ichilov Hospital), Dr. Ido Strauss of TAU’s School of Medical and Health Sciences, and the director of the Functional Neurosurgery Unit at Ichilov Hospital.

They worked with an epileptic patient because they belonged to a subset of epileptics who don’t respond well to medication and thus require neurosurgical intervention, study leader Tankus explained.

Furthermore, he was also part of an even smaller subset within this group, where the focus seems to be deeper in the brain, rather than on the surface of the cortex. The focus refers to “the source of the ‘short’ that’s sending powerful electrical waves through the brain,” he continued.

To find the exact location of the source of the problem, doctors have to implant electrodes into the deep structures of their brains.

So, the patient used in the study already had these “brain readers” in place and was in the hospital waiting to have another seizure, as this type of epileptic must do. It is only during the seizure that doctors can discern where the focus is, so they can operate.

The speech neuroprosthesis experiment

With electrodes already implanted and consent given, researchers asked the patient to repeat the syllables /a/ and /e/ out loud. They recorded the brain activity associated with these syllables using deep learning and machine learning and trained these AI models to pinpoint specific brain cells.

Specifically, they were able to identify the electrical activity that indicated the desire to say /a/ and /e/ as per the press release. The computer was able to recognize these electrical patterns.

Next, they asked him to imagine that he was saying them. The computer effectively could vocalize the syllables from the desire to say them AKA thought, as per Medical X Press .

The study is in its earliest stages, so they’ve begun with two building blocks of speech, or two syllables, as per the press release . The ultimate objective is to reach complete speech but as Tankus said, “even two syllables can enable a fully paralyzed person to signal yes or no.”

Strikingly, this machine learning technique can be used preemptively, for an ALS patient in the early stages, for example, when they can still talk. The machine can learn their thought and speech patterns, so when they lose the physical capacity to speak, the computer can still help them communicate.

“Our study is a significant step toward developing a brain-computer interface that can replace the brain’s control pathways for speech production, allowing completely paralyzed individuals to communicate voluntarily with their surroundings once again,” Tankus concluded, Tel Aviv University reports .

The study was published in the official publication of the Congress of Neurological Surgeons: Neurosurgery .