Related Experiment Video
Updated: May 29, 2025
11:28
Concurrent EEG and Functional MRI Recording and Integration Analysis for Dynamic Cortical Activity Imaging
Published on: June 30, 2018
11.6K
Generalized Coupled Matrix Tensor Factorization Method Based on Normalized Mutual Information for Simultaneous
Zahra Rabiei1, Hussain Montazery Kordy2
1Faculty of Electrical and Computer Engineering, Babol Noshirvani University of Technology, Babol, Iran.
Neuroinformatics
|February 6, 2025
Summary
This study introduces a generalized coupled matrix tensor factorization (GCMTF) method for fusing electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) data. GCMTF effectively identifies shared brain activity components, outperforming existing methods in accuracy and identifying more brain regions.
Area of Science:
- Neuroscience
- Biomedical Engineering
- Data Science
Background:
- Electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) offer complementary insights into brain activity.
- Joint analysis of EEG and fMRI data is crucial for a comprehensive understanding of neural processes.
- Existing methods often assume restrictive equality for shared components, limiting their applicability.
Purpose of the Study:
- To develop a novel method for joint EEG-fMRI analysis that overcomes limitations of existing techniques.
- To introduce a generalized coupled matrix tensor factorization (GCMTF) approach utilizing normalized mutual information (NMI).
- To enhance the identification of shared and unshared brain activity components between EEG and fMRI.
Main Methods:
- Implementation of the generalized coupled matrix tensor factorization (GCMTF) method.
- Utilizing normalized mutual information (NMI) to define component similarity, accommodating nonlinear relationships.
- Application to simulated data with nonlinear component relationships and real EEG-fMRI data from an auditory oddball paradigm.
Main Results:
- The GCMTF method demonstrated a 23.46% increase in average match score compared to the advanced coupled matrix tensor factorization (ACMTF) model on simulated data.
- GCMTF effectively identified shared components with both linear and nonlinear relationships, unlike ACMTF.
- Analysis of real data revealed three shared components in alpha and theta bands during an auditory oddball task, identifying more active brain areas than ACMTF.
Conclusions:
- The proposed GCMTF method offers a robust and generalized approach for joint EEG-fMRI data analysis.
- GCMTF accurately identifies shared neural components, even with nonlinear relationships and varying noise levels.
- This method enhances the discovery of brain activity patterns associated with cognitive tasks, improving upon existing fusion techniques.
