Extracting robust single-trial somatosensory evoked potentials for non-invasive brain computer interfaces

Area of Science:

• Neuroscience and Biomedical Engineering
• Signal Processing
• Rehabilitation Technology

Background:

• Reliable single-trial somatosensory evoked potentials (SEPs) are crucial for brain-computer interface (BCI) applications in post-brain injury rehabilitation.
• Current noninvasive SEP extraction methods often require extensive averaging or processing due to small and variable signals.
• Optimizing stimulation parameters and signal processing is needed to enhance the signal-to-noise ratio (SNR) for real-time BCI feedback.

Purpose of the Study:

• To describe and evaluate optimized electrical stimulation parameters for enhancing the SNR of noninvasive single-trial SEPs.
• To enable reliable, real-time extraction of SEPs for BCI applications.
• To assess the feasibility of real-time SEP detection in healthy individuals and those with central nervous system (CNS) injuries.

Main Methods:

• Recorded SEPs using scalp electroencephalography (EEG) during tibial nerve stimulation in 13 healthy participants and 2 individuals with CNS injuries.
• Evaluated three lower-than-recommended stimulation frequencies (0.2 Hz, 1 Hz, 2 Hz) with a longer pulse width (1 msec).
• Assessed single-trial SEP detectability using offline, pseudo-online, and real-time analyses, employing Laplacian filtering.

Main Results:

• The SEP N70 component was primarily observed in central scalp regions.
• Online decoding performance significantly improved with Laplacian filtering, achieving Area Under the Curve (AUC) scores from 0.78-0.90.
• Feasible SEP detection was demonstrated in individuals with incomplete spinal cord injury (AUC 0.86) and stroke (AUC 0.81), with real-time detection showing an AUC of 0.89.

Conclusions:

• The study presents a system for real-time single-trial SEP extraction using optimized electrical stimulation parameters and signal processing.
• This enhanced SNR approach is suitable for BCI-based operant conditioning for rehabilitation.
• The optimized system demonstrates potential for effective BCI applications in individuals with neurological impairments.