Long-term Evolution of Cortical Networks Throughout Brain-controlled Neurorehabilitation After SCI

Introduction: By coupling brain-computer interface (BCI) technology with spinal cord stimulation, brain-controlled neurorehabilitation offers patients with spinal cord injury the ability to recover voluntary control of movement and improves functional outcomes over time. Despite advancements in BCI decoding methods, a significant challenge in clinical translation stems from the instability of recorded signals. Signal instability, or ‘non-stationarity’, necessitates constant recalibration of the decoder, which takes time and requires skilled technical staff. Non-stationarity likely originates from a multitude of underlying causes including physical changes in the device-tissue interface, learning-related changes, and changes related to strategy adjustments by the patient over time.

Methods: To better characterize these changes, we recorded neural activity from an electrocorticography (ECoG) array that was chronically implanted in a clinical trial participant with a cervical spine injury (NCT05665998). Resting state and motor-related activity was recorded over the course of one year, during which the patient underwent brain-controlled rehabilitation. We assessed changes in the correlational and connectivity structure over time.

Results: We found both linear and non-linear changes in activation and connectivity patterns and were able to characterize periods of stability and instability throughout the course of the experiment that were state-dependent. We quantified the relative contribution of extrinsic (interface related) and intrinsic (plasticity related) changes over time.

Conclusion : A deeper understanding of the nature of these long-term changes in cortical networks and BCIs will guide the development of novel tools for stabilization, putting us one step closer to brain-controlled neurorehabilitation in a wider patient population.