AUTHOR=Veerabhadrappa Rakesh , Ul Hassan Masood , Zhang James , Bhatti Asim 

TITLE=Compatibility Evaluation of Clustering Algorithms for Contemporary Extracellular Neural Spike Sorting

JOURNAL=Frontiers in Systems Neuroscience

VOLUME=Volume 14 - 2020

YEAR=2020

URL=https://www.frontiersin.org/journals/systems-neuroscience/articles/10.3389/fnsys.2020.00034

DOI=10.3389/fnsys.2020.00034

ISSN=1662-5137

ABSTRACT=Deciphering useful information from electrophysiological data recoded from the brain, in-vivo or in-vitro, is dependent on the capability to analyse spike patterns efficiently and accurately. The spike analysis mechanisms are heavily dependent on the clustering algorithms that enable separation of spike trends based on their spatio-temporal behaviours. Number of clustering algorithms are reported in the literature over several decades focused on different applications. Although a small subset of clustering algorithms is adopted for spike analysis, however, not much work has been reported on the compliance and suitability of such clustering algorithms for spike analysis. In this work, we have attempted to comment on the suitability of available clustering algorithms and how do they actually perform when exposed to spike analysis. In this regard, we have performed a compatibility assessment on algorithms already employed in spike sorting as well as the algorithms yet to be investigated for application in sorting neural spikes. The performance of the algorithms is compared in terms of their accuracy, confusion matrix and well accepted internal and external indices. Three data sets comprising of easy (low noise), difficult (noisy) and a real data with known ground truth are chosen for performance comparison, ensuring a uniform test bed. The procedure also employs two feature sets, PCA and wavelets. A statistical score scheme is provided to evaluate the performance individually and overall. The open nature of the datasets, the clustering algorithms and the evaluation criteria make the proposed evaluation framework widely accessible to the research community. This is an attempt to provide a reference guide for emerging neuroscientists to select the most suitable algorithms for their spike analysis requirements.