Non-invasive decoders may help to recover movement after spinal cord injury

When a person suffers spinal cord injury, normal communication between the brain and the spinal circuits below the injury is interrupted, causing paralysis. Because the brain is functioning normally, just like the spinal cord under the injury, researchers have been working to enable rehabilitation and reestablish communication to potentially restore movement.

Members of his lab, including Ismael Señes, an assistant professor of biomedical engineering at McElbay’s School of Engineering at Washington University in St. Louis, and Carolyn Atkinson, a professor of biomedical engineering at Wash Medical’s Neurosurgery and a doctoral student, developed a type of decoder to restore that communication. Laboratory experiments with 17 human subjects without spinal cord injury allowed lower limb movement to be cued with percutaneous spinal cord stimulation or non-invasive external electrical pulses.

The results of the study were published online on April 25, 2025 in the Neuroengineering and Rehabilitation Journal.

The team used a special cap equipped with non-invasive electrodes that measure brain activity via electroencephalography (EEG). While wearing the hat, the sitting volunteers were asked to stretch their knees on their knees, and then, while stretching their legs, they were asked to allow researchers to record brain waves with both exercises.

The team provided neural activity to the decoder or algorithm, allowing them to learn how brain waves work in both situations. They found that actual and imagined movements use similar neural strategies.

After giving this data to the decoder, we learn to make predictions based on neural activity whenever there is or no movement. Whenever someone is thinking about moving their legs, it shows that they can be predicted even if their feet don’t actually move. ”

Ismael Señes, Assistant Professor of Biomedical Engineering at McKelby School of Engineering at Washington University in St. Louis and Department of Neurosurgery in Wash Medicine.

The team used controls to ensure that the volunteers truly imagined they weren’t actually moving.

“Every time people move, this can introduce signal noise and I want to make sure that signal noise is not something they’re learning to predict,” Ceañes said. “What we want to predict is movement intention or brain activity. So we imagine people are using the same algorithms that they were trained on people who move to extend their legs and predict whether they’re imagining them.”

Ceañes said this would reveal two things.

“One thing is that you are likely deciphering the intentions of movement, not artifacts or noise. Second, every time you use this on someone with a spinal cord injury, you can train your decoder using the imagination of moving your legs, as you don’t have the ability to actually move your legs and label the data.”

Seáñez said the proof-of-concept study is the first step to developing a non-invasive cerebrospinal interface that provides percutaneous spinal stimulation using real-time prediction to enhance the spontaneous movement of a single joint in rehabilitation patients with spinal cord injury.

Going forward, the team will determine whether universal decoders can run similarly to personalized decoders and test generalized decoders trained with data from all participants that can simplify use in clinical settings.