Spike sorting based upon PCA over DWT frequency band selection

Abstract

When analyzing the neurobiological data many of its aspects have to be carefully looked upon. Data coming from the MRI, EMG or microrecording all have its special properties that have to be extracted during the process analysis. In case of recordings coming from the microrecording procedure i.e. from microelectrodes placed within the neuronal tissue signal can be analyzed in at least two ways. First approach focuses on the background noise present in such recordings. Second one, looks upon the presence of the spikes - electrical signs of the bioelectrical neurophysiological activity of neuron cells. In a given recording one may often find many spikes with different shapes. For further analytical reasons it is frequently desired that spikes are to be grouped according to their shape. Such grouping / shape clustering is called spike sorting and there are many known approaches to that problem. Still, before spikes are detected and sorted the raw recorded signal is almost always filtered and altered in various DSP processes. This preliminary DSP operations may significantly hamper the spike sorting efficiency. Analysis presented in this paper provides answer as to which frequency bands are alone sufficient for proper and successful spike sorting. © 2014 Springer International Publishing.