Analyzing the Data

Abstract This chapter discusses the strengths, pitfalls, and practicalities of MEG and EEG data analysis methods and visualization strategies. Data-set segmentation, signal-to-noise considerations, signal levels, and power are examined as these may drive the chosen data analysis strategy. After basic analyses of averaged and unaveraged data, brain microstates, event-related desynchronization/synchronization, temporal spectral evolution, and time-frequency analyses, phase synchronization, and cross-frequency coupling are discussed. Measures of the introduced association and functional/effective connectivity, as studied in the time or frequency domains, include correlation, coherence, phase-locking factor, phase-locking value, phase-lag index and their variants, mutual information, transfer entropy, cross-correlation, Granger causality, dynamic causal modeling, and graph-theoretical analysis. The MEG/EEG source modeling section covers forward and inverse problems, head models, single- and multidipole models, distributed models, and beamformers. After discussion of spatial resolution, source extent, and effects of synchrony complete the topics, the chapter ends with statistical considerations regarding signal detectability in individual and group-level data.