MEG and multimodal integration

Abstract

Functional brain imaging methods provide measures of various physiological processes with a range of spatial and temporal scales. Because the sensitivity properties of the imaging modalities differ, combining multimodal data is expected to provide more information about brain activity than is available by any single method alone. Data from multiple modalities can be described as complementary or supportive and, consequently, can be analyzed using symmetric or asymmetric data fusion approaches. Complementary modalities have similar physiological origin and are observed with similar experimental paradigms. In a supportive role, data from one imaging modality guides the analysis and interpretation of another modality. In this chapter, we focus on the fusion of magnetoencephalography (MEG) data with electroencephalography (EEG), structural magnetic resonance imaging (MRI), and functional MRI (fMRI) data. For example, MEG and EEG are complementary modalities because they have similar source types, i.e., both are generated by cortical primary currents, but have different spatial sensitivity characteristics. The combination of MEG and EEG data can resolve certain ambiguities that can occur when data from only one of the modalities are available. MEG and fMRI can also be considered complementary if the different types of signals are obtained from a common experimental paradigm and are analyzed using symmetric, model-based, or datadriven fusion approaches. Structural MRI can provide supportive data for MEG source estimation, e.g., by indicating allowable locations and orientations of the MEG source currents. Similarly, fMRI can be used in a supportive role to suggest a likely source distribution for MEG among multiple alternatives. This chapter describes various approaches to multimodal neuroimaging data fusion and discusses their benefits and limitations.