Related Experiment Video
Updated: May 18, 2026
08:51
Statistical Modelling of Cortical Connectivity Using Non-invasive Electroencephalograms
Published on: November 1, 2019
A critical assessment of connectivity measures for EEG data: a simulation study
Stefan Haufe1, Vadim V Nikulin, Klaus-Robert Müller
1Machine Learning Group, Department of Computer Science, Berlin Institute of Technology, Franklinstr. 28/29, 10587 Berlin, Germany. stefan.haufe@tu-berlin.de
Neuroimage
|September 26, 2012
Summary
Volume conduction in electroencephalography (EEG) complicates brain connectivity analysis. This study shows a new method using time-reversed data to accurately measure neuronal interactions from EEG, improving upon traditional Granger-causal (GC) measures.
Area of Science:
- Neuroscience
- Computational Neuroscience
- Signal Processing
Background:
- Estimating brain area communication from EEG is challenging due to noise and volume conduction.
- Existing effective connectivity measures struggle with EEG data limitations, potentially yielding unreliable results.
Purpose of the Study:
- To evaluate popular effective connectivity measures and inverse source reconstruction techniques using simulated EEG data.
- To develop and validate a method to overcome volume conduction artifacts in EEG connectivity analysis.
Main Methods:
- Simulated EEG data were used to test Granger-causal (GC) and Phase-Slope Index (PSI) measures.
- A novel procedure using time-reversed data surrogates was developed to mitigate volume conduction effects.
- The performance of inverse source reconstruction methods was assessed in conjunction with connectivity analyses.
Main Results:
- Volume conduction significantly limits the neurophysiological interpretability of sensor-space connectivity.
- Standard Granger-causal analysis with typical significance testing produced spurious connectivity, even with source-reconstructed data.
- The Phase-Slope Index (PSI) demonstrated better performance due to theoretical robustness against volume conduction.
- The proposed time-reversed surrogate method successfully suppressed volume conduction artifacts, making Granger-causal measures reliable for EEG data.
Conclusions:
- Volume conduction poses a significant challenge for accurate brain connectivity estimation from EEG.
- A robustified Granger-causal analysis using time-reversed surrogates provides a reliable method for assessing brain interactions in EEG.
- The developed approach offers guidance for improved EEG-based brain connectivity research.
