Complexity measures of event related potential surface Laplacian data calculated using the wavelet packet transform.

We describe a method to obtain estimates of EEG signal complexity using the well-established wavelet packet transform with best basis selection. In particular, we use the two-dimensional wavelet packet transform to obtain estimates of the complexity of two-dimensional images. This allows us to calculate complexity estimates of high-resolution brain potential maps generated from 61 scalp electrode Visual Oddball paradigm, grand-mean data. A significant reduction in the complexity of the surface Laplacian time-slices is observed during and after the Visual Potential 300 (P3) event for the target case, possibly as a result of increased spatial synchrony associated with visual-related tasks. We also present the results of a statistical analysis of the largest principal component of the time-varying complexity curves, for control, high-risk, and alcoholic groups of male subjects. Parametric and non-parametric analyses show differences in the complexity data which are significant between the control group and the alcoholic and high-risk groups.