Spike sorting based on PCA and improved fuzzy c-means

Proper classification of spikes from extracellular recordings is essential for the study of neuronal behavior. A lot of algorithms have been presented in the technical literature. Combining with subtractive clustering and fuzzy c-means, we present a new algorithm named improved fuzzy c-means. Compared with fuzzy c-means, the dependence on the initial centers of improved fuzzy c-means is reduced. Not only do the new algorithm improve the accuracy of classification, but also the results of classification are more stable. Three types of classifier were employed in this paper to assess the performance of the spike sorting algorithm. When the noise level rises gradually, the accuracy of k-means of fuzzy c-means decreased quickly.