Related Experiment Video
Updated: Aug 22, 2025
08:22
Author Spotlight: Advancing the Study of Brain-Heart Interplay with a Comprehensive EEGLAB Plugin for Multimodal Signal Analysis
Published on: April 26, 2024
2.1K
Graph Theoretical Analysis of EEG Functional Connectivity Patterns and Fusion with Physiological Signals for Emotion
Vasileios-Rafail Xefteris1, Athina Tsanousa1, Nefeli Georgakopoulou1
1Centre for Research and Technology Hellas, Information Technologies Institute, 6th Km Charilaou-Thermi, 57001 Thessaloniki, Greece.
Sensors (Basel, Switzerland)
|November 11, 2022
Summary
This study introduces graph theory analysis of electroencephalograph (EEG) functional connectivity for emotion recognition. Combining EEG with peripheral signals and advanced feature selection significantly improves accuracy, exceeding prior results.
Area of Science:
- Neuroscience
- Computer Science
- Biomedical Engineering
Background:
- Emotion recognition is crucial for human-computer interaction.
- Electroencephalograph (EEG) and other physiological signals offer non-intrusive emotion detection methods.
- Graph theory analysis of EEG functional connectivity patterns remains underexplored for emotion recognition.
Purpose of the Study:
- To propose a novel approach for emotion recognition using graph theoretical analysis of EEG functional connectivity.
- To integrate EEG functional connectivity features with peripheral physiological signals.
- To enhance emotion recognition accuracy by exploring brain network characteristics.
Main Methods:
- Extracted functional connectivity from EEG signals.
- Computed global and local graph theory features from EEG data.
- Fused graph theory features with statistical features from peripheral physiological signals.
- Utilized classifiers and a Convolutional Neural Network (CNN) for emotion recognition.
- Applied a feature selection algorithm to optimize performance.
Main Results:
- Achieved average accuracies of 55.62% (valence) and 57.38% (arousal) for subject-independent classification using CNN.
- Reached 83.94% (valence) and 83.87% (arousal) for subject-dependent classification using CNN.
- Improved subject-independent classification to 75.44% (valence) and 78.77% (arousal) with feature selection.
- Enhanced subject-dependent classification to 88.27% (valence) and 90.84% (arousal) with feature selection.
- Exceeded current state-of-the-art results in emotion recognition.
Conclusions:
- Graph theoretical analysis of EEG functional connectivity, combined with peripheral signals and feature selection, significantly advances emotion recognition.
- The proposed method demonstrates superior performance in both subject-dependent and subject-independent emotion classification tasks.
- This approach offers a promising direction for developing more accurate and robust emotion recognition systems.
