The Effect of Mutual Information on Independent Component Analysis in EEG/MEG 
Analysis: A Simulation Study

Neumann, A; Grosse-Wentrup, M; Buss, M; Gramann, K

doi:10.1080/00207450802324655

Item

ITEM ACTIONSEXPORT

Add to Basket

Local TagsRelease HistoryDetailsSummary

Released

Journal Article

The Effect of Mutual Information on Independent Component Analysis in EEG/MEG Analysis: A Simulation Study

MPS-Authors

There are no MPG-Authors in the publication available

External Resource

https://www.tandfonline.com/doi/pdf/10.1080/00207450802324655?needAccess=true
(Publisher version)

Fulltext (restricted access)

There are currently no full texts shared for your IP range.

Fulltext (public)

There are no public fulltexts stored in PuRe

Supplementary Material (public)

There is no public supplementary material available

Citation

Neumann, A., Grosse-Wentrup, M., Buss, M., & Gramann, K. (2008). The Effect of Mutual Information on Independent Component Analysis in EEG/MEG Analysis: A Simulation Study. International Journal of Neuroscience, 118(11), 1534-1546. doi:10.1080/00207450802324655.

Cite as: https://hdl.handle.net/11858/00-001M-0000-0013-C665-4

Abstract

Objective: This study investigated the influence of mutual information (MI) on temporal and dipole reconstruction based on independent components (ICs) derived from independent component analysis (ICA). Method: Artificial electroencephalogram (EEG) datasets were created by means of a neural mass model simulating cortical activity of two neural sources within a four-shell spherical head model. Mutual information between neural sources was systematicallyvaried. Results: Increasing spatial error for reconstructed locations of ICs with increasing MI was observed. By contrast, the reconstruction error for the time course of source activity was largely independent of MI but varied systematically with Gaussianity of the sources. Conclusion: Independent component analysis is a viable tool for analyzing the temporal activity of EEG/MEG (magnetoencephalography) sources even if the underlying neural sources are mutually dependent. However, if ICA is used as a preprocessing algorithm for source localization, mutual information between sources introduces a bias in the reconstructed locations of the sources. Significance: Studies using ICA-algorithms based on MI have to be aware of possible errors in the spatial reconstruction of sources if these are coupled with other neural sources.