Reliability of motor and sensory neural decoding by threshold crossings for intracortical brain-machine interface

Area of Science:

• Neuroscience
• Computational Neuroscience
• Neural Engineering

Background:

• Spike sorting is crucial for isolating single neural units in intracortical neurophysiological studies.
• The necessity of spike sorting for neural decoding remains debated, with some studies suggesting threshold crossings (TC) suffice.
• Previous comparisons of TC and spike sorting were limited in scope, focusing on specific tasks and brain regions.

Purpose of the Study:

• To comprehensively compare the performance of threshold crossings (TC) and spike sorting in neural decoding.
• To evaluate these methods across diverse tasks, cortical regions, and decoding parameters.
• To determine the stability and robustness of TC versus spike sorting for neural decoding applications.

Main Methods:

• Two macaques with Utah or floating microelectrode arrays (FMAs) in motor/sensory cortices performed motor/sensory tasks.
• Neural data were preprocessed using threshold crossings (TC), automatic sorting (AS), and manual sorting (MS).
• A support vector machine decoder assessed decoding accuracy, stability, and robustness across varied parameters and signal sources.

Main Results:

• Threshold crossings (TC) achieved similar or better decoding performance than spike sorting methods across motor and sensory tasks.
• TC performance remained strong despite variations in neural signal sources and decoder parameters.
• TC demonstrated significantly higher stability in neural decoding across sessions and robustness to threshold changes.

Conclusions:

• Spike-firing patterns can be reliably extracted using TC from multiple cortical areas for both motor and sensory decoding.
• The enhanced stability and robustness of TC suggest it may be a more suitable preprocessing method for neural decoding.
• These findings challenge the universal necessity of spike sorting, advocating for TC in specific neural decoding contexts.