Adaptive GCN and Bi-GRU-Based Dual Branch for Motor Imagery EEG Decoding

Area of Science:

• Neuroscience
• Signal Processing
• Machine Learning

Background:

• Decoding motor imagery electroencephalography (MI-EEG) signals is challenging due to complex channel connectivity and temporal signal dependencies.
• Low spatial resolution and high signal redundancy in EEG hinder traditional linear model performance.
• Existing methods struggle to effectively capture both spatial and temporal features of MI-EEG signals.

Purpose of the Study:

• To propose a novel dual-branch framework for enhanced MI-EEG signal decoding.
• To effectively model channel correlations and temporal dependencies in EEG signals.
• To improve the accuracy and robustness of brain-computer interface (BCI) systems.

Main Methods:

• A dual-branch framework integrating an adaptive graph convolutional network (Adaptive GCN) and bidirectional gated recurrent units (Bi-GRUs).
• Chebyshev Type II filter for sub-band decomposition and frequency domain analysis.
• Adaptive GCN for spatial-spectral feature extraction and Bi-GRU with Multi-Head Attention (MHA) for deep time-spectral feature extraction.
• Feature fusion for final prediction generation.

Main Results:

• Achieved an average classification accuracy of 80.38% on the BCI-IV Dataset 2a.
• Achieved an average classification accuracy of 87.49% on the BCI-I Dataset 3a.
• Outperformed existing state-of-the-art MI-EEG decoding approaches.

Conclusions:

• The proposed dual-branch framework effectively decodes MI-EEG signals by modeling spatial-spectral and temporal dependencies.
• This approach offers a foundation for developing personalized and adaptive BCI systems.
• The method demonstrates superior performance compared to current state-of-the-art techniques.