Extracellular recording from living neurons employing microelectrode arrays has attracted paramount attention in recent years as a way to investigate the functionality and disorders of the brain. To decipher useful information from the recorded signals, accurate and efficient neural spike activity detection and sorting becomes an essential prerequisite. Traditional approaches rely on thresholding to detect individual spikes and clustering to identify subset groups; however, these methods fail to identify temporally synchronous spikes due to neuronal synchrony. To address this challenge, we introduce a novel spike sorting algorithm incorporating both quantitative and probabilistic techniques to better approximate the ground truth information of the spike activity. A novel pre-clustering method for identifying key features that can form natural clusters and a dimension reduction technique for identifying the spiking activity are introduced. To address the temporal neuronal synchrony phenomenon leading to detection of multineural overlapped spikes, a procedure for template spike shape estimation and iterative recognition is developed employing the cross correlation methodology tailored to individual neuron's spike rate. A performance comparison between the proposed method and existing techniques in terms of the number of spikes identified and efficiency of sorting the spikes is presented. The outcome shows the effectiveness of the proposed method in identifying temporally synchronous spikes.
- Neural activity analysis
- Spike sorting
- Temporally synchronous spikes
ASJC Scopus subject areas
- Computer Science Applications
- Cognitive Neuroscience
- Artificial Intelligence