Meg methods: A primer of basic meg analysis

Abstract

Methods for analysis of magnetoencephalography (MEG) data are diverse in their implementation, but traditionally involve inverse modeling of brain activity from data acquired at the sensors. Preprocessing methods vary depending on the specific type of MEG data acquired, but normally involve filtering/removal of noise signals from the data, and may also require the co-registration of coordinate spaces between MEG head positioning systems and structural data. Depending on the variables to be compared, analysis of MEG data can range from simple to complex, and computational demand for data analysis can be minor or substantial. Processing and analysis may include things such as stimulus averaging, time-frequency analysis, source analysis, or a combination thereof. Analyses can be restricted to the cortex or allowed to span the brain volume, and can be focused to predefined regions of interest or examined across all sensors/brain areas. In this chapter, we review basic MEG analysis methods, beginning with preprocessing and closing with source-level MEG analysis.