Brain decoder controls spinal cord stimulation

In a significant development in spinal cord injury treatment, researchers at Washington University in St. Louis have created a neurodecoder that bridges the communication gap between the brain and spine and potentially opens new avenues for recovery.

A team led by Ismael Seáñez, assistant professor of biomedical engineering at WASHU, shows that brain activity can be explained and used to deliver precisely timed electrical stimuli to the spinal cord. Their findings, published on April 25 in the Journal of Neuroengineering and Rehabilitation, are a key step towards non-invasive rehabilitation techniques for paralyzed people.

“After providing this data to the decoder, it will learn to predict based on neural activity whenever there is any movement or no movement,” Seáñez said. “We show that even if someone is not actually moving, we can predict whenever someone considers moving the legs.”

When someone suffers from spinal cord injury, the normal communication pathway between the brain and muscles is interrupted, causing paralysis, even though both the brain and lower spinal circuits remain functional. This new approach aims to reconnect these systems using external technologies.

The study invited 17 sound participants wearing specialized EEG (EEG) caps equipped with electrodes to measure brain activity. Participants were asked to perform knee stretching while sitting, and they were just thinking about the same movement without really moving. The data allow researchers to train algorithms to identify brain motor intentions.

What makes this approach particularly promising is its total non-invasiveness. Unlike other brain helical interfaces that require surgical implantation of electrodes, the system uses external EEG caps and percutaneous spinal cord stimulation—electrical pulses delivered through the skin.

During the test, the system achieved an impressive 83% accuracy only when predicting motion intentions from brain signals. Perhaps more striking is that the decoder can successfully determine when participants imagine leg movements with only 77% accuracy, suggesting that the system may have a potential to work for people who are completely paralyzed.

Business implications may be very large. Current methods of spinal cord injury usually reach the plateau after about six months. Techniques to re-establish communication between the brain and spinal circuits may go beyond this conventional limitation.

This represents a potential turning point for investors who watch the field of neurotech. In recent years, brain computer interfaces have attracted a large amount of venture capital, focusing mainly on invasive technologies that require surgery. A non-invasive alternative that exhibits clinical efficacy can greatly expand market access.

The technique is by detecting characteristic changes in brain wave patterns called “event-related pair synchronization” that occur in a specific frequency band when someone intends to move. The system then uses these patterns to accurately trigger spinal stimulation.

“One person, we are more likely to decode motion intentions than artifacts or noise, and secondly, whenever we use it on people with spinal cord injury, they will not have the ability to actually move the legs to label data for us, we can use their imagination to move the legs to train our decoder,” Seáñeez explains.

From a technical point of view, the system is relatively simple compared to other neural interfaces. It uses Linear Discriminant Analysis (LDA) algorithm – a complete machine learning technology that provides interpretability and efficiency without the need for a large number of computing resources.

While still developing early, researchers are already seeking clinical applications. They plan to test whether generalized decoders trained on data from multiple participants can execute and personalize data, which has the potential to simplify deployment in clinical settings.

The market for spinal cord injury rehabilitation technology remains underserved, with approximately 18,000 new cases per year in the United States alone. The current rehabilitation costs may exceed $500,000 in the first year after injury, bringing strong financial incentives to more effective treatments.

For policy makers, this study highlights the growing potential of non-invasive neurotechnology to address significant medical challenges in terms of the risks and costs of surgical interventions without the risk and cost. As these technologies mature, regulatory frameworks may need to be updated to illustrate medical devices that do not require physically implanted brain-controlled.

While promising, there are still some challenges ahead of clinical deployment. The system is currently best suited for precise timing motion movements rather than natural, spontaneous movements. The researchers found significant differences in brain activity patterns between cue exercise and undisclosed exercise, suggesting that practical applications will require additional refinement.

The study was funded by Washington University in St. Louis, the National Institutes of Health, and the MacDonald Center for Systems Neuroscience at the University of Washington.

As neural interface technology continues to advance, this non-invasive approach could represent an alternative to more readily obtain surgical solutions, potentially accelerating rehabilitation innovation while reducing patient risks. The question now is that the technology can be transformed from a laboratory demonstration to a clinical application and whether it can fundamentally change the recovery trajectory of people with spinal cord injuries.

Related