Neural activity clusterization for estimation of firing pattern

Abstract

Background: The patterns of neuronal activity may be considered one of the most important features to describe the state of the neuron and its alterations under particular circumstances. However, most of the proposed methods in this area rely on values or parameter boundaries that have been chosen arbitrarily.

New method: In this paper, we propose a method for analyzing neural patterns based on a spike density histogram and hierarchical clustering of real datasets.

Results: We used recordings of single unit activity obtained from pallidum of dystonic patients during DBS surgeries. We grouped spike trains into four main clusters based on similarities of the spike density histograms, and we estimated the underlying distribution parameters for each cluster.

Comparison with existing methods: The proposed method performs better than analogous approach that was based on spike density histogram shapes proposed by Labarre (Labarre et al., 2008) when applying to simulated data set described in original paper.

Conclusions: In the present paper, we proposed a method for defining various numbers of patterns depending on particular tasks. The method may be effective both for rough and comprehensive descriptions of neuronal activity patterns.

Keywords: Basal ganglia; Cluster analysis; Neuronal activity; Neuronal patterns; Spike density histogram.