A real-time spike classification method based on dynamic time warping for extracellular enteric neural recording with...

Area of Science:

• Computational Neuroscience
• Bioinformatics
• Neurophysiology

Background:

• Real-time analysis of extracellular neural recordings requires computationally efficient spike recognition methods.
• The enteric nervous system (ENS) is crucial for human health but remains understudied due to a lack of suitable spike recognition algorithms capable of handling its significant waveform variability.

Purpose of the Study:

• To develop a computationally efficient spike recognition method for real-time analysis of neural recordings, specifically addressing the challenges posed by the enteric nervous system's waveform variability.
• To improve spike classification accuracy and computational efficiency compared to existing methods.

Main Methods:

• A novel method based on dynamic time warping (DTW) was developed, featuring high tolerance to temporal and magnitude variations in neural waveforms.
• Adaptive temporal gridding was employed within the fastDTW algorithm to substantially reduce computational costs during similarity calculations.
• Automated threshold selection was implemented to enable real-time classification of extracellular recordings.

Main Results:

• The proposed method demonstrated improved classification accuracy and computational complexity on synthesized data compared to conventional cross-correlation based template-matching (CCTM) and PCA+k-means clustering.
• Application to mouse enteric neural recordings under mechanical and chemical stimuli successfully classified biphasic and monophasic spikes.
• Effective spike classification was achieved despite waveform variability exceeding milliseconds in width and millivolts in magnitude.

Conclusions:

• The fastDTW method offers superior computational efficiency and a high tolerance to waveform variability compared to traditional template matching and clustering techniques.
• An adaptive fastDTW algorithm was successfully developed for real-time spike classification of ENS recordings, robustly handling significant waveform variability caused by factors like colony motility, ambient changes, and cellular heterogeneity.