Source localization of EEG/MEG data by correlating columns of ICA solution with lead field matrix

Independent Components Analysis is increasingly being used on EEG/MEG data as a preprocessor to source localization methods such as beamforming and dipole fitting. Here we show how ICA can be used to perform source localization directly. The proposed method consists of dimension reduction preprocessing, estimating and inverting the reduced-dimension demixing matrix, and then correlating the columns of the resulting mixing matrix estimate with the columns of the forward model. The results section shows a comparison of the proposed method and a scalar minimum variance beamformer, where it is shown that the proposed ICA-based method has less localization error.