[Spike waves pattern recognition in EEG long term registrations using a variably programmed laboratory computer].

Abstract

The application of automatic approaches in EEG pattern recognition essentially has two aims: to increase the level of objectivity (by means of quantification) and to decrease the amount of diagnostic work (by means of data reduction). In this study a computerized spike-wave detection method is described, realized on a small laboratory computer (PDP 11/03) and especially designed for the analysis of long term registrations. The program is written in FORTRAN in order to provide a high level of flexibility. The user may easily modify it for his own purposes. Pattern recognition is based on the configuration of maxima and minima. Different degrees of digital filtering is used for spike analysis and wave analysis. The time codes of the events are registered and stored on magnetic disk. In a subsequent computer run the results may graphically be displayed or submitted to further analysis. An example is given testing in which way the results depend on the choice of the parameter set (duration and amplitude of spike and wave). The time course of the frequency of SW-events (or regular SW-sequences, respectively) are compared to the results of conventional analysis.