Localization of Independent Components from Magnetoencephalography

Blind sour e separation (BSS) de omposes a multidimensional time series into a set of sour es, ea h with a one-dimensional time ourse and a xed spatial distribution. For EEG and MEG, the former orresponds to the simultaneously separated and temporally overlapping signals for ontinuous non-averaged data; the latter orresponds to the set of attenuations from the sour es to the sensors. These sensor proje tion ve tors give information on the spatial lo ations of the sour es. Here we use standard Neuromag dipoletting software to lo alize BSS-separated omponents of MEG data olle ted in several tasks in whi h visual, auditory, and somatosensory stimuli all play a role. We found that BSS-separated omponents with stimulusor motor-lo ked responses an be lo alized to physiologi al and anatomi ally meaningful lo ations within the brain.