Effective extraction of high-frequency oscillations by blind source separation

High-frequency oscillations (HFOs) in somatosensory evoked potentials (SEP) have been observed in both humans and animals. In magnetoencephalography (MEG), the HFOs are observed as somatosensory evoked field (SEF) with advantage of higher reliability of current source localization. On the other hand, to extract the HFOs from MEG data, the conventional method requires the averaging of 2000-10,000 of evoked responses which causes difficulty of measurements. In this study, we used blind source separation (BSS) method to extract waveform of HFOs with smaller number of averaging times. As a result, we could reduce the number of averaging times to about one-fifth to get HFOs. Scalp distributions of the magnetic field of the HFOs components can be also obtained by the method.