Denoising and Time-window selection using Wavelet-based Semblance for improving ERP detection

Carolina Saavedra (1) , Laurent Bougrain (1)

1 NEUROSYS - Analysis and modeling of neural systems by a system neuroscience approach

Wavelet denoising has been successfully applied to Event-Related Potential (ERP) detection, but it usually works using channels information independently. This paper presents an adaptive approach to denoise signals taking into account the channels correlation in the wavelet domain. Moreover, we combine phase and amplitude information to automatically select a time window which increases ERP detection. Results on the P300 speller show that our algorithm has a better accuracy with respect to the VisuShrink wavelet technique and the XDAWN algorithm among 22 healthy subjects, and a better regularity than XDAWN.

Domaines: Réseau de neurones [cs.NE]

Carolina Saavedra, Laurent Bougrain. Denoising and Time-window selection using Wavelet-based Semblance for improving ERP detection. BCI meeting, Jul 2013, Pacific grove, United States. ⟨hal-00857523⟩