OHBM 2022 | 113 | Talk | Victor Férat | Pycrostates, a python library to study EEG microstates…

Title: Pycrostates, a python library to study EEG microstates.
Session: Talk
Speaker: Victor Férat
Pycrostates is a free and open source project for EEG microstates analysis. Microstates analysis is a method allowing investigation of spatiotemporal characteristics of EEG recordings. It consists in 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. Pycrostates implements core functions: preprocessing tools (global field power peaks extraction, resampling), clustering algorithm (modified version of the kmeans algorithm), clustering quality evaluations tools (silhouette, Dunn) needed to perform such analysis. It is aimed at researchers wishing to design their own microstate study, and complements graphical tools such as Cartool or the Koenig toolbox. Pycrostates was built to fit as best as it can in the python scientific environment (numpy, scipy, pandas) and more particularly scikit learn and MNE-python from which it is inspired in its philosophy and its implementation. Pycrostates takes advantage of continuous integration and delivery tools such as auto-test , automatic code review and is tested against several python versions and operating system to. In addition, this library comes with extensive documentation including descriptions of all its algorithms and functions as well as several tutorials to help researcher to get started. Finally, pycrostates is provided under the new BSD license allowing code reuse, even in commercial products.