Anatomically distributed areas are dynamically linked to form functional networks for processing and integrating the different modalities of information in the human brain. A part of such networks is considered to be realized with synchronization of neuronal activities, which can generate correlated neural oscillation at the same and/or different frequency bands. To investigate the networks with the synchronization, analysis of connectivity between not only same frequency oscillation but also different frequency (i.e. cross-frequency) is needed. For source estimation with electroencephalogram (EEG) or magneto-encephalogram (MEG) signals, a spatial filtering technique is recently applied as an alternative method for equivalent current dipole (ECD) estimation technique. Non-adaptive type of spatial filtering technique, such as the Standardized low-resolution brain electromagnetic tomography (sLORETA), is reported to discriminate correlated sources. However, it may lead to inaccurate results due to its low spatial resolution. In the present study, we proposed a new systematic approach for localizing the sources of correlated cross-frequency oscillations. The method we propose can overcome the limitation of the non-adaptive spatial filtering technique by proactively using identified information in sensor level analysis (e.g. cross-correlation map and correlation topography), which allow us to focus on target sources. The performance of our proposed method is evaluated with simulated EEG signals, and is compared with traditional method.
PMID: 22255954

A new method for localizing the sources of correlated cross-frequency oscillations in human brains.