Computer Detection, Classification, And Analysis of Neuronal Spike Sequences*
James A. Feldman, FERNAND A. ROBERGE · INFOR Information Systems and Operational Research · 1971
In the course of experimental work with animals or clinical investigation of neurological diseases, it is useful to have a description of neuronal spike activity from a localized region of the brain. The signal is normally picked up by a single microelectrode and consists of intermixed discharges from two or more neighbouring nerve cells. The problem is to identify the spikes produced by an individual neuron and to decompose the signal into a number of spike trains, each spike train representing the activity of a single neuron. Various characteristics of the individual spike trains may be of interest to the experimenter; mean rate of discharge, distribution of interspike intervals, serial dependence between interspike intervals, cross-correlation between two neuronal spike trains, etc. With this information, possible neural networks involving closely adjacent nerve cells could be studied in vivo and compared with proposal models. In this paper, a computer system for the detection, classifi.catipn, and analysis of neuroelectric spike activity is described. The system is operational in a semiautomatic mode only since a delay of several minutes may be incurred between the actual recording of brain activity and the display of the results to the experimenter. Work is in progress to reduce this delay to the order of a few seconds to allow experimental work to be performed in real time.