Classifying neuronal spikes from multiunit recording by using a triode

A method of classifying neuronal spikes was proposed and realized by multichannel multiunit recording with a triode and considering spatial damping vectors. Each element of a damping vector is the covariance between a template and a spike waveform observed by each channel. In the plot of damping vectors in 3-D space, the distribution showed some clusters. Assuming that a cluster corresponds to a neuron, the spikes were classified by referring to the 3-D distribution and the waveforms of the spikes in each class were well superimposed. The authors discussed the stability of the damping vectors and concluded that this method is promising for precise classification of neuronal spikes.