Reading brain signals for decoding speech With elderly people set to form an ever-greater proportion of the population in future, more of us will live with the consequences of diseases that may paralyse some muscles and lead to function loss. Professor Nick Ramsey tells us about his team’s work in developing a braincomputer interface designed to restore function and help paralysed people communicate. A stroke in the brain stem is one of the main causes of total paralysis, damaging the connections between the brain and the muscles, and so impairing an individual’s ability to communicate. A second major cause is amyotrophic lateral sclerosis (ALS), a neurological disease that affects motor neurons, nerve cells which control voluntary muscle movement. “The signals are intact in the brain, and it generates the impulses, but basically the wires to the body are no longer working,” explains Nick Ramsey, a Neuroscience Professor at the Brain Center of the University Medical Center of Utrecht. As the Principal Investigator of the iCONNECT project, Professor Ramsey is working to develop a brain-computer interface (BCI) that helps paralysed people communicate, building on earlier research into the brain. “We have been working with epileptic patients, who have electrodes implanted for their diagnosis. This gives us the opportunity to pursue basic research into how the brain works,” he outlines.
Brain signals This research led to important insights into how to interpret the brain’s signals, from which the idea of working with implants to decode inner speech was developed. The foundation of this work is a detailed understanding of how signals are transmitted between the brain and muscles. “There are many muscles in your body, and they are all stimulated by a particular part of the brain, the primary motor cortex. That’s where the neurons reside, that get the pulses to the muscles,” says Professor Ramsey. Different parts of the body are organised in an orderly fashion, and their movements can be related to signals from specific parts of the brain. “The cortical homunculus starts in the middle at the top of the head, and it goes to either side of the body – where the left part of your brain is connected to the right side of your body, and the other way round,” continues Professor Ramsey. “If we look at one side of the sensorimotor cortex, we can delineate which part of the brain maps onto which part of the body.”
22
Implantation of the Utrecht Neural Prosthesis in a Locked in patient. (www.neuroprosthesis.eu)
Researchers in the iConnect project have produced evidence that the movement of different muscles in the face leads to different patterns on the cortex. So for example if an individual purses their lips or clenches their jaws, then researchers monitoring their brain activity would see a pattern, where very small patches of
cortex become active. “That pattern allows us to identify what kind of movement you’re making, we can even identify different spoken letters such as ‘p’ or ‘ah’. We’ve also proved that if you cannot make a movement – but try to – then you still get the same patterns on the cortex. This supports the idea that in cases of paralysis the brain is intact and the pulses are still generated, but they don’t actually arrive at the muscles,” says Professor Ramsey. This is central to the project’s work in developing an intracranial BCI, designed for use in the home. “We decided to first try and accomplish something relatively simple, but which really helps patients who are locked-in, who are unable to communicate,” outlines Professor Ramsey. The long-term, ambitious goal in this research is to interpret brain signals so accurately that it becomes possible to develop implants that translate attempted speech to a speech computer in real-time, and implants that make muscle movements possible again. “We aim to offer a system to people with locked-in syndrome that will help them to communicate again,” explains Professor Ramsey. A core part of this work centres around developing an implant that will record brain signals, interpret them, and send the
On the left the Utrecht Neural Prosthesis is shown. On the top right a 7 Tesla functional MRI scan of the 5 fingers of the right hand (thumb/orange to little finger/red). On the bottom right the electrode grids for the next generation BCI for decoding speech and gestures.
EU Research