Nonlinear denoising of transient signals with application to event-related potentials

We present a new wavelet-based method for the denoising of event-related potentials (ERPs), employing techniques recently developed for the paradigm of deterministic chaotic systems. The denoising scheme has been constructed to be appropriate for short and transient time sequences using circular state space embedding. Its effectiveness was successfully tested on simulated signals as well as on ERPs recorded from within a human brain. The method enables the study of individual ERPs against strong ongoing brain electrical activity. (C) 2000 Elsevier Science B.V. All rights reserved.