Group independent component analysis of resting state EEG in large normative samples

EEG (Electroencephalography) resting state was studied by means of group blind source separation (gBSS), employing a test–retest strategy in two large-sample normative databases (N = 57 and N = 84). Using a BSS method in the complex Fourier domain and a model-driven distributed inverse solution we closely replicate both the spatial distribution and spectral pattern of seven source components. Norms were then constructed for their spectral power so as to allow testing patients against the norms. As compared to existing normative databases based on scalp spectral measures, the resulting tool defines a smaller number of features with very little inter-correlation. Furthermore, these features are physiologically meaningful as they relate the activity of several brain regions, forming a total of seven patterns, each with a peculiar spatial distribution and spectral profile. This new tool, that we name normative independent component analysis (NICA), may serve as an adjunct to diagnosis and assessment of abnormal brain functioning and aid in research on normal resting state networks.