EEG motor imagery decoding: a framework for comparative analysis with channel attention mechanisms

Area of Science:

• Neuroscience
• Machine Learning
• Biomedical Engineering

Background:

• Traditional spatial filters in Brain-Computer Interfaces (BCI) for motor imagery decoding have limitations.
• Channel attention mechanisms offer a novel approach to enhance spatial feature extraction in BCI.

Purpose of the Study:

• To systematically investigate and compare various channel attention mechanisms for motor imagery decoding in BCI.
• To evaluate the impact of these mechanisms within a lightweight, easily integrated architecture framework.

Main Methods:

• Development of a straightforward, lightweight baseline architecture for seamless integration of channel attention mechanisms.
• Systematic comparison of different channel attention mechanisms under consistent experimental conditions.
• Extensive testing across four diverse datasets to assess model effectiveness and generalizability.

Main Results:

• The proposed architecture framework demonstrates significant strength and generalizability across multiple datasets.
• Integration of channel attention mechanisms notably improves decoding performance.
• The framework maintains a small memory footprint and low computational complexity.

Conclusions:

• Channel attention mechanisms are effective in enhancing BCI motor imagery decoding performance.
• The lightweight and generalizable architecture provides an efficient solution for electroencephalogram (EEG) based BCIs.
• This approach offers versatility for various BCI applications requiring robust motor imagery decoding.