Spike train clustering using a Lempel-Ziv distance measure

Multi-electrode array recordings reveal complex structures in the firing of spatially distributed neurons. The analysis of this neuronal network activity demands a classification of neurons according to similarities in their firing behavior. If similar spike patterns do not occur synchronously, but have unknown delays within spike trains, this processing step is difficult. To solve this problem, we introduce a Lempel-Ziv complexity-based distance measure. Using our distance measure as the input for a superparamagnetic clustering algorithm, we achieve an efficient classification of spike trains.