Classification of naturally evoked compound action potentials in peripheral nerve spatiotemporal recordings

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Peripheral neural signals are crucial for neuroprosthetic and neuromodulation applications.
• Extracting meaningful information from these signals remains a significant challenge.

Purpose of the Study:

• To develop and validate a method for classifying naturally evoked compound action potentials (CAPs) using spatiotemporal signatures.
• To enhance the precision of control signals for neuroprosthetic devices.

Main Methods:

• Implanted 56-channel nerve cuff electrodes on the sciatic nerve of 9 rats.
• Evoked afferent activity in specific nerve fascicles using mechano-sensory stimuli.
• Extracted spatiotemporal signatures from recorded CAPs to train classifiers.

Main Results:

• Achieved a mean classification accuracy of 0.686 ± 0.126 for a 3-class problem using the best classifier.
• Obtained a mean F1-score of 0.605 ± 0.212.
• Demonstrated a mean Pearson correlation coefficient of 0.728 ± 0.276 between original and estimated neural firing rates.

Conclusions:

• The proposed method successfully classifies individual naturally evoked CAPs from peripheral neural signals recorded via extraneural electrodes.
• This technique holds promise for improving control in neuroprosthetic applications by providing more precise neural decoding.