ACTION-POTENTIAL CLASSIFIERS - A FUNCTIONAL COMPARISON OF TEMPLATE MATCHING, PRINCIPAL COMPONENTS-ANALYSIS AND AN ARTIFICIAL NEURAL-NETWORK

Multiunit neural activity occurs often in electrophysiological studies when utilizing extracellular electrodes. In order to estimate the act ivity of the individual neurons each action potential in the recording must be classified to its neuron of origin. This paper compares the a ccuracy of two traditional methods of action potential classification- template matching and principal components-against the performance of an artificial neural network (ANN). Both traditional methods use avera ges of action potential shapes to form their corresponding classifiers while the artificial neural network 'learns' a nonlinear relationship between a set of prototype action potentials and assigned classes. Th e set of prototypic action potentials and the assigned classes is term ed the training set. The training set contained action potentials from each class which exhibited the full range of amplitude variability. T he ANN provided better classification results and was more robust in a nalysis of across-animal data sets than either of the traditional acti on potential classification methods.