The brain-computer interface that works with imagined handwriting

People with "locked in" syndromes resulting from strokes, injuries or disease may soon have a new and fast way of communicating with the world. Researchers at Stanford University have developed a system that enabled a man with a spinal cord injury to communicate directly with a computer using his brain. How? By attempting handwriting.

Despite his paralysis, simply trying to write by hand produced distinctive neural patterns that allowed a machine learning algorithm to distinguish individual letters and display them on a screen.

The underlying technology is not new. Brain-computer interfaces (CBIs) are used in a number of ways, including helping paralysed or locked-in people to communicate without the need for speech or any kind of bodily movement. Usually, the user attempts a more simple movement – pointing to letters on a screen or grid.

"This new handwriting approach doubles the speed at which a person with paralysis can use a brain-computer interface to type text," lead author Dr Frank Willett told Science Focus. "Using this system, our participant could type at 18 words per minute, which is comparable to normal handwriting speeds, or typing on a smartphone if you are in his age group (65)."

The system involves an electrode array implanted in the motor cortex of the man's brain. This recorded the electrical activity of individual neurons as the subject attempted to physically move his arm and hand to write sentences. The machine learning algorithm then analysed the neural patterns associated with individual letters.

The biggest advantage of the handwriting approach is the speed it enables, Willett said. "Even though [the participant] had not used his hand to write in more than 10 years, simply by trying to make his hand write, his brain still generated highly-interpretable neural activity that encoded the intended motion of his hand," he said.

By comparison, a point-and-click typing BCI is slower and less accurate because moving a cursor to nearby keys on an on-screen keyboard evokes similar patterns of neural activity, because the motions involved are themselves similar.

While this is a research demonstration and not a medical product, Willett believes the technology can help people outside the laboratory. It even retains some sense of the user's actual handwriting.

"It’s perhaps similar to what it’s like to write letters without looking at your hand. Nevertheless, the letter shapes that we were able to reconstruct are indeed our best guess at the pen trajectory that he was trying to make."