Five-year performance analysis of a chronic fully-implanted ECoG-based BCI in an individual with tetraplegia

Introduction: Spinal cord injuries (SCI) affect over 300 million individuals in the US. While traditional medical therapies have limited success, brain-computer interfaces (BCIs) offer promise in enabling control over external devices, such as computers, robotic limbs, or exoskeletons. Recent clinical trials have explored invasive intracortical BCIs for motor control, showing multiple degrees of freedom. However, their signal quality is known to degrade in the chronic period, and such approaches tether patients to the data acquisition hardware. To address this, we implemented a fully implanted electrocorticography (ECoG)-based BCI with two 4-contact ECoG leads over the sensorimotor cortex in a patient with cervical C5 SCI as part of a clinical trail. The patient used the system to control an external orthosis at home, decoding motor imagery using a 2-state hidden Markov process.

Methods: Over 5 years, we measured electrode contact impedance, signal quality, and decoder performance to examine long-term system utility. The signal quality was evaluated by measuring the maximum bandwidth of the signal. Motor imagery viability was assessed by differences in average power within select frequency bands and the area under the receiver operator characteristic curve measured decoder performance.

Results: Over the first 6 months, electrode impedance increased and followed a relatively stable period during the next 4 years. Similarly, average power became stable after 6 months and demonstrated stable separation between motor imagery states, ensuring that long-term detection of motor intention. The median accuracy of the decoder was 86.54% over time and AUROC was 0.959.

Conclusion : Data show that following the initial 6-month impedance changes, device signals and metrics remained stable. Importantly, the patient successfully used the device at home without errors, indicating long-term viability for external device control using implanted devices with sensing capabilities. Future research should explore larger electrode counts and more participants to confirm this stability beyond the initial 6-month post-surgery period. Understanding these trends is crucial for determining the clinical utility in broader patient populations.