Introducing SPoC: a multivariate analysis framework for the analysis of cross-frequency power coupling as well as for multimodal integration of EEG/MEG power with hemodynamics
-
1
Berlin Institute of Technology, Machine Learning, Germany
-
2
City College of New York, Biomedical Engineering, United States
-
3
Charité University Medicine Berlin, Department of Neurology, Germany
-
4
National Research University Higher School of Economics, Department of Psychology, Russia
-
5
Korea University, Department of Brain and Cognitive Engineering, Korea
Introduction: We present the SPoC (Source Power Comodulation) framework for source separation of multivariate EEG/MEG/LFP recordings, which focuses particularly on envelope/power dynamics of oscillatory sources. In the SPoC framework, the source separation is based directly on the quantity of interest (co-modulation of power dynamics), rather than on auxiliary assumptions (e.g. mutual independence). Methods: The SPoC family of algorithms optimizes spatial filters such that the power of extracted EEG/MEG oscillations maximally correlate with: - a univariate signal such as a behavioral- or stimulus variable (RT, ratings, intensity, etc.) (SPoC, Dähne et al. 2014a) - a signal which is simultaneously extracted from a multivariate input such as concurrent hemodynamic measurements (e.g. NIRS, fMRI) (mSPoC, Dähne et al. 2013) - the envelope of simultaneously extracted oscillatory source signals within or across frequency bands and within or across multiple subjects (cSPoC, Dähne et al. 2014b) The SPoC algorithms have been benchmarked in extensive simulations as well as on real-world data, where they were found to outperform state-of-the-art analysis techniques such as Independent Component Analysis (ICA) and Canonical Correlation Analysis (CCA). Results: In a study on loudness-modulated steady-state auditory evoked potentials (SSAEP), SPoC achieves statically significantly larger correlation between stimulus intensity (loudness) and power at the SSAEP frequency than ICA or Regression (Dähne et al. 2014a). mSPoC outperforms CCA in integrating concurrently measured EEG and NIRS (Dähne et al. 2013). cSPoC extracts maximally envelope correlated alpha and beta sources on resting state data in an unsupervised manner (Dähne et al. 2014b). Conclusions: The SPoC algorithms extract brain oscillations, the power dynamics of which co-modulate with parameters of interest. Thus, the framework will be a valuable tool for understanding the role of neural oscillations within and between subjects.
Keywords:
EEG,
MEG,
oscillations,
envelope,
Cross-frequency coupling,
multimodal,
SPOC,
band-power
Conference:
XII International Conference on Cognitive Neuroscience (ICON-XII), Brisbane, Queensland, Australia, 27 Jul - 31 Jul, 2014.
Presentation Type:
Poster
Topic:
Methods Development
Citation:
Dähne
S,
Haufe
S,
Nikulin
V and
Müller
K
(2015). Introducing SPoC: a multivariate analysis framework for the analysis of cross-frequency power coupling as well as for multimodal integration of EEG/MEG power with hemodynamics.
Conference Abstract:
XII International Conference on Cognitive Neuroscience (ICON-XII).
doi: 10.3389/conf.fnhum.2015.217.00124
Copyright:
The abstracts in this collection have not been subject to any Frontiers peer review or checks, and are not endorsed by Frontiers.
They are made available through the Frontiers publishing platform as a service to conference organizers and presenters.
The copyright in the individual abstracts is owned by the author of each abstract or his/her employer unless otherwise stated.
Each abstract, as well as the collection of abstracts, are published under a Creative Commons CC-BY 4.0 (attribution) licence (https://creativecommons.org/licenses/by/4.0/) and may thus be reproduced, translated, adapted and be the subject of derivative works provided the authors and Frontiers are attributed.
For Frontiers’ terms and conditions please see https://www.frontiersin.org/legal/terms-and-conditions.
Received:
19 Feb 2015;
Published Online:
24 Apr 2015.
*
Correspondence:
Mr. Sven Dähne, Berlin Institute of Technology, Machine Learning, Berlin, Germany, sven.daehne@tu-berlin.de