Simultaneous Decoding of Wrist Angles and Grasp Forces Based on Channel-Wise Cumulative Spike Trains

Area of Science:

• Neuroscience
• Biomedical Engineering
• Human-Machine Interfacing

Background:

• Simultaneous decoding of wrist angles and grasp forces from neural signals is crucial for advanced human-machine interfaces.
• Current methods face challenges in accurately capturing complex neuromuscular control signals.

Purpose of the Study:

• To propose and validate a novel scheme using channel-wise cumulative spike trains (cw-CSTs) for simultaneous decoding of wrist angles and grasp forces.
• To compare the performance of cw-CST-derived features against conventional motor unit discharge rate features.

Main Methods:

• Utilized spatial spike detection to extract cw-CSTs from surface electromyography, maximizing motor unit activity observation.
• Extracted three neural features: cw-MUdrive, DR-cwCST (derived from cw-CSTs), and DR-MUST (conventional).
• Employed Gaussian process regression for wrist- and hand-specific decoders to estimate angles and forces in ten subjects.

Main Results:

• cw-CST-based neural features demonstrated superior accuracy and stability compared to conventional DR-MUST features.
• The cw-MUdrive feature outperformed DR-cwCST in grasp force estimation.
• DR-cwCST showed comparable performance to cw-MUdrive in wrist angle estimation.

Conclusions:

• The proposed cw-CST scheme offers an effective solution for simultaneous decoding of wrist movements and grasp forces.
• This advancement promotes more natural and intuitive control in neural interface applications.
• The findings pave the way for enhanced prosthetic devices and assistive technologies.