Efficient estimation of a time-varying dimension parameter and its application to EEG analysis

Abstract

This paper considers the problem of estimating the dimension of nonstationary electroencephalogram (EEG) signals and describes the implementation of an efficient algorithm to calculate a time-varying dimension estimate. The algorithm allows the practical calculation of a dimension estimate and its statistical significance over large data sets with a high temporal resolution. The method is applied to EEG recordings from patients with temporal lobe epilepsy and in one case the results of the analysis are compared with those obtained from an existing method of computing the correlation density.