Related Experiment Video
Updated: Jun 26, 2026
08:51
Statistical Modelling of Cortical Connectivity Using Non-invasive Electroencephalograms
Published on: November 1, 2019
Connectivity estimation of three parametric methods on simulated electroencephalogram signals
Hugo Vélez-Pérez1, Valérie Louis-Dorr, Radu Ranta
1Research Centre for Automatic Control, CRANUMR 7039, National Center for Scientific Research, Nancy University, Vandoeuvre les Nancy, France. hugo.velez-perez@ensem.inpl-nancy.fr
Summary
This study estimates brain connectivity in multichannel electroencephalogram (EEG) recordings using multidimensional autoregressive (AR) models. It evaluates coherence and directed transfer functions on simulated data, proposing a new performance metric for enhanced accuracy.
Area of Science:
- Neuroscience
- Signal Processing
- Computational Biology
Background:
- Accurate estimation of brain connectivity from electroencephalogram (EEG) data is crucial for understanding neural dynamics.
- Multichannel EEG recordings offer rich spatial and temporal information but require sophisticated modeling techniques.
Purpose of the Study:
- To evaluate the performance of different connectivity estimation methods in multichannel EEG.
- To introduce a novel performance criterion for assessing the accuracy of connectivity matrices.
Main Methods:
- Modeling multichannel EEG signals as multidimensional autoregressive (AR) processes.
- Evaluating coherence, directed transfer function (DTF), and partial directed coherence (PDC) on simulated EEG data.
- Calculating relative error and a new entropy-based performance criterion (eta) for method evaluation.
Main Results:
- The study systematically compared the performance of coherence, DTF, and PDC on simulated EEG data.
- The proposed eta criterion, based on connectivity matrix entropy, provides a novel way to assess estimation accuracy.
- Results indicate the potential for these methods in analyzing real-world EEG recordings.
Conclusions:
- Multidimensional AR modeling provides a robust framework for EEG connectivity analysis.
- The evaluated connectivity functions show promise for characterizing brain network dynamics.
- The novel eta criterion offers a valuable tool for future research in EEG connectivity estimation.
