Is sensor space analysis good enough? Spatial patterns as a tool for assessing spatial mixing of EEG/MEG rhythms
This repository provides analysis code to analyze spatial mixing in electrophysiological data through lead field and spatial pattern coefficients.
Schaworonkow N & Nikulin VV: Is sensor space analysis good enough? Spatial patterns as a tool for assessing spatial mixing of EEG/MEG rhythms. NeuroImage (2022).
The results are based on following available openly available data set: "Leipzig Cohort for Mind-Body-Emotion Interactions" (LEMON dataset), from which we used the preprocessed EEG data Additionally, we used the MOUS data set. The associated data set research articles:
- Babayan A et al.: A mind-brain-body dataset of MRI, EEG, cognition, emotion, and peripheral physiology in young and old adults. Scientific Data (2018).
- Schoffelen JM et al.: A 204-subject multimodal neuroimaging dataset to study language processing. Scientific Data (2019).
We also use the New York Head, a head model and pre-computed lead field. The asssociated research articles are:
- Huang Y, Parra LC, Haufe S: The New York Head -- A precise standardized volume conductor model for EEG source localization and tES targeting, NeuroImage (2015).
- Haufe S , Huang Y, Parra LC: A highly detailed FEM volume conductor model of the ICBM152 average head template for EEG source imaging and tCS targeting. In: Conf Proc IEEE Eng Med Biol Soc (2015).
To reproduce the results, the preprocessed EEG and MEG data and leadfield matrix (file name: sa_nyhead.mat) should be downloaded and placed into the folder data (or otherwise, the path to the data needs to be adjusted).
The provided python3 scripts are using scipy and numpy for general computation, pandas for saving intermediate results to csv-files. matplotlib for visualization. For EEG-related analysis, the mne package is used. For computation of aperiodic exponents: specparam. Specifically used versions can be seen in requirements.txt.
To reproduce the figures from the command line, navigate into the code folder and execute make all. This will run through the preprocessing steps and generate the figures. The scripts can also be executed separately in the order described in the Makefile.