Imagined Speech Classification Using EEG and Deep Learning.

Area of Science:

• Neuroscience
• Machine Learning
• Biomedical Engineering

Background:

• Brain-computer interfaces (BCIs) offer alternative communication and control methods.
• Imagined speech recognition from electroencephalography (EEG) signals is a challenging but promising area.
• Efficient feature extraction and dimensionality reduction are crucial for BCI performance.

Purpose of the Study:

• To develop and evaluate a deep learning model for imagined speech-based brain wave pattern recognition.
• To investigate the effectiveness of wavelet scattering transformation for EEG feature extraction.
• To assess the feasibility of a low-cost, multi-channel EEG system for BCI applications.

Main Methods:

• Utilized an 8-channel EEG headset and MATLAB 2023a for data acquisition.
• Applied wavelet scattering transformation to reduce EEG data dimensionality and complexity.
• Employed a long-short term memory recurrent neural network (LSTM-RNN) for classification of four imagined speech commands (up, down, left, right).

Main Results:

• Achieved an overall classification accuracy of 92.50% for imagined speech recognition.
• Demonstrated high precision (92.74%), recall (92.50%), and F1-score (92.62%).
• The proposed method shows promise for real-time BCI systems.

Conclusions:

• The developed deep learning approach effectively decodes imagined speech from EEG signals.
• Wavelet scattering transformation is a valuable technique for enhancing EEG feature extraction.
• The findings support the development of reliable and accessible imagined speech-based BCIs.