Stabilized linear model for neuromagnetic source localization

Abstract

In MEG studies, head movement during data recording often causes problems in event-related averaging as well as source localization accuracy. We propose in this work a stabilized linear model which can solve this head movement problem by conceptually increasing the sensor number, under the condition that the head movement is known. This method is particularly beneficial for MEG studies in which head movement commonly Occurs, such as those studies requiring long experimental duration or those when patients/children are involved. Phantom studies and studies of somatosensory evoked field clearly show the effectiveness and accuracy of the proposed method.