Applications of second order blind identification to high-density EEG-based brain imaging: A review

Abstract

In the context of relating specific brain functions to specific brain structures, second-order blind identification (SOBI) is one of the blind source separation algorithms that have been validated extensively in the data domain of human high-density EEG. Here we provide a review of empirical data that (1) validate the claim that SOBI is capable of separating correlated neuronal sources from each other and from typical noise sources present during an EEG experiment; (2) demonstrating the range of experimental conditions under which SOBI is able to recover functionally and neuroanatomically meaningful sources; (3) demonstrating cross- as well as within-subjects (cross-time) reliability of SOBI-recovered sources; (4) demonstrating efficiency of SOBI separation of neuronal sources. We conclude that SOBI may offer neuroscientists as well as clinicians a cost-effective way to image the dynamics of brain activity in terms of signals originating from specific brain regions using the widely available EEG recording technique. © 2010 Springer-Verlag.