Magnetoencephalography: Basic theory and estimation techniques of working brain activity

It is widely accepted that magnetoencephalography (MEG) is a promising tool for investigating human brain activity in good temporal and spatial resolution. However, the use of MEG is currently quite limited, mostly due to excessive diversity in localization techniques to estimate regional brain activity using MEG data. Because source localization in MEG is an ill-posed problem, an adequate localization technique varies depending on the task design and the timing of interest in MEG signals, which sometimes confuses investigators when choosing an analysis technique or comparing the results with those obtained with other modalities, such as functional magnetic resonance imaging. This chapter reviews the introductory theories and applications of currently available MEG source localization techniques as well as principles of MEG signals and its measurement for beginners and possible future MEG users. The physiological and mathematical backgrounds of cerebral MEG source are briefly introduced followed by the technical requirements for MEG data acquisition. Modern localization techniques for inverse problem solving with MEG, from a simple dipole model to an underdetermined type of norm estimation or spatial filter technique, are thoroughly described.