EEG classification of emotions using emotion-specific brain functional network

Area of Science:

• Neuroscience
• Brain-Computer Interfaces
• Computational Psychiatry

Background:

• Understanding brain function relies on analyzing its complex network mechanisms.
• Human emotions are associated with distinct patterns of neural activity and connectivity.

Purpose of the Study:

• To formulate emotion-specific brain functional networks using electroencephalography (EEG).
• To identify subject-specific and dynamic network patterns related to emotional states.
• To classify emotions based on these identified network patterns.

Main Methods:

• Analysis of brain functional networks using the phase-locking value (PLV) synchronization measure in EEG.
• Identification of reactive channel pairs and frequency bands associated with emotional tasks compared to rest.
• Formation of subject-specific functional networks based on identified reactive pairs.
• Utilizing dynamic network patterns for emotion classification.

Main Results:

• Emotion-specific brain functional networks were successfully formulated.
• Significant variations in synchrony were observed in subject-specific and emotion-specific dynamic networks.
• An average classification accuracy of 62% was achieved for emotion recognition using the proposed technique across 4 subjects.

Conclusions:

• The study demonstrates the feasibility of using EEG-based functional network analysis for emotion recognition.
• Phase-locking value and network dissimilarities are effective in identifying emotion-related brain dynamics.
• The developed subject-specific and emotion-specific dynamic network patterns show promise for BCI applications in emotion detection.