Microstates#
What is EEG microstates ?#
Microstates analysis is a method allowing investigation of spatiotemporal characteristics of EEG recordings. It consists of breaking down the multichannel EEG signal into a succession of quasi-stable state, each state being characterized by a spatial distribution of its scalp potentials also called microstate map or microstate topography.
How to compute EEG microstates ?#
This decomposition is based on two consecutive steps: the clustering which allows to define topographies that best represent the studied data and the backfitting than consist on assigning one of the previously extracted topographies to each timepoint of one or several EEG recordings.
The methods relies on assigning timepoints to the most similar microstate map, which is why it is important to define how distance between two topographies is computed. For microstate analysis, the inverse of the absolute value of the spatial correlation is used as a measure of distance to carry out all computations. The absolute value is used in order to ignore the topography polarity.
Pycrostates implements a convenient class
pycrostates.cluster.ModKMeans to perform clustering through
the fit() method and backfitting through
the predict() method. It also implements
other methods to facilitate the analysis and display of results.