Enhanced EEG signal classification in brain computer interfaces using hybrid deep learning models.

Area of Science:

• Neuroscience and Biomedical Engineering
• Artificial Intelligence and Machine Learning

Background:

• Brain-computer interfaces (BCIs) enable communication by decoding neural signals.
• Accurate classification of electroencephalogram (EEG) data is vital for BCI performance.
• Motor Imagery (MI) classification is a key challenge in BCI development.

Purpose of the Study:

• To enhance Motor Imagery (MI) classification accuracy in Brain-Computer Interface (BCI) systems.
• To evaluate traditional machine learning and deep learning techniques for EEG signal classification.
• To develop and validate a novel hybrid deep learning model for improved BCI performance.

Main Methods:

• EEG data from the PhysioNet EEG Motor Movement/Imagery Dataset was analyzed.
• Five traditional classifiers (KNN, SVC, LR, RF, NB) were evaluated.
• Deep learning models including CNN, LSTM, and a hybrid CNN-LSTM were implemented and compared.

Main Results:

• Random Forest achieved the highest accuracy among traditional methods at 91%.
• CNN and LSTM models achieved 88.18% and 16.13% accuracy, respectively.
• The proposed hybrid CNN-LSTM model reached a superior accuracy of 96.06%.

Conclusions:

• Hybrid deep learning models offer significant advancements for BCI systems.
• The CNN-LSTM hybrid model provides a robust and precise approach to motor imagery classification.
• This research paves the way for more sophisticated and effective BCI applications.