Graph Neural Networks on SPD Manifolds for Motor Imagery Classification: A Perspective from the Time-Frequency Analysis

TL;DR

This paper introduces Graph-CSPNet, a novel geometric deep learning architecture on SPD manifolds for motor imagery EEG classification, leveraging time-frequency analysis to improve accuracy across multiple datasets.

Contribution

The paper proposes Graph-CSPNet, a new manifold-valued graph convolutional network architecture that enhances MI-EEG classification by integrating differential geometry and time-frequency analysis.

Findings

Achieved near-optimal accuracy in 9 of 11 MI-EEG scenarios.

Utilized novel manifold-valued graph convolution techniques.

Validated effectiveness on five public datasets.

Abstract

The motor imagery (MI) classification has been a prominent research topic in brain-computer interfaces based on electroencephalography (EEG). Over the past few decades, the performance of MI-EEG classifiers has seen gradual enhancement. In this study, we amplify the geometric deep learning-based MI-EEG classifiers from the perspective of time-frequency analysis, introducing a new architecture called Graph-CSPNet. We refer to this category of classifiers as Geometric Classifiers, highlighting their foundation in differential geometry stemming from EEG spatial covariance matrices. Graph-CSPNet utilizes novel manifold-valued graph convolutional techniques to capture the EEG features in the time-frequency domain, offering heightened flexibility in signal segmentation for capturing localized fluctuations. To evaluate the effectiveness of Graph-CSPNet, we employ five commonly-used publicly available MI-EEG datasets, achieving near-optimal classification accuracies in nine out of eleven scenarios. The Python repository can be found at https://github.com/GeometricBCI/Tensor-CSPNet-and-Graph-CSPNet.