A novel motor imagery EEG decoding method based on feature separation
Lie Yang1, Yonghao Song1, Ke Ma2
1Shien-Ming Wu School of Intelligent Engineering, South China University of Technology, Guangzhou, People's Republic of China.
Journal of Neural Engineering
|February 5, 2021
Summary
This study introduces a novel feature separation network for electroencephalography (EEG) decoding in brain-computer interfaces (BCI). The method enhances motor imagery decoding accuracy by filtering out irrelevant signal information.
Area of Science:
- Neuroscience
- Biomedical Engineering
- Machine Learning
Background:
- Motor imagery electroencephalography (EEG) decoding is crucial for brain-computer interface (BCI) systems.
- Existing decoding methods struggle with class-independent information in EEG signals, limiting accuracy.
Purpose of the Study:
- To propose a novel motor imagery EEG decoding method that overcomes interference from class-independent information.
- To improve the decoding accuracy of BCI systems.
Main Methods:
- A feature separation network based on adversarial learning (FSNAL) was designed.
- FSNAL separates class-related and class-independent features from raw EEG data.
- Motor imagery decoding is performed using only the extracted class-related features.
Main Results:
- The proposed method was validated on two public EEG datasets (BCI competition IV 2a and 2b).
- Experimental results demonstrated superior performance compared to state-of-the-art methods.
- The motor imagery EEG decoding method significantly outperformed all compared approaches.
Conclusions:
- The developed motor imagery EEG decoding method effectively mitigates interference from class-independent features.
- This approach holds significant potential for enhancing the performance of future motor imagery BCI systems.
Keywords:
adversarial learningbrain–computer interface (BCI) systemsfeature separationmotor imagery electroencephalography (EEG) decoding
