Decoding Multi-Class EEG Signals of Hand Movement Using Multivariate Empirical Mode Decomposition and Convolutional...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Brain-computer interfaces (BCIs) offer potential for restoring motor control in stroke patients.
• A key challenge in BCIs is achieving high classification accuracy for multi-class movement decoding using electroencephalogram (EEG) signals.

Purpose of the Study:

• To propose a novel algorithm, MECN, for enhanced decoding of EEG signals corresponding to four distinct hand movements.
• To improve the classification accuracy of multi-class movements in BCI applications.

Main Methods:

• Utilized Multivariate Empirical Mode Decomposition (MEMD) to decompose EEG signals into multivariate intrinsic empirical functions (MIMFs).
• Applied sequential forward selection for optimal MIMFs fusion.
• Fed selected MIMFs into a Convolutional Neural Network (CNN) for movement classification.

Main Results:

• Achieved an average classification accuracy of 81.14% (pre-movement onset) and 81.08% (post-movement onset) across thirteen subjects.
• Demonstrated statistically significant improvements over existing state-of-the-art methods.
• Effectively decoded four types of hand movements using the proposed MECN algorithm.

Conclusions:

• The MECN algorithm effectively decodes four hand movements from EEG signals.
• This novel approach shows significant potential for advancing BCI technology, particularly for motor control restoration.