Multiresolution source localization using the wavelet transform

Abstract

The use of the wavelet transform to localize the current dipole sources from the multichannel electroencephalogram (EEG) is discussed. The wavelet approach automatically computes the critical time-slices at which the dipole sources are localized. Unlike the traditional approaches, where visually selected time-slices are used which represent only part of the information available in the data, the automatically computed time-slices are information-preserving. As a result, the EEG can be closely reconstructed using the parameters at each computed time-slice. In addition, the multiresolution framework of the wavelet transform provides a mathematical zoom lens which enables one to select major electrical sources at courser scale levels, and to observe the details at finer scale levels.<>