Automated EEG signal processing: A comprehensive investigation into preprocessing techniques and sub-band extraction...

Area of Science:

• Neuroscience and Biomedical Engineering
• Signal Processing and Machine Learning

Background:

• Electroencephalogram (EEG) signals are crucial for monitoring brain activity, neurological conditions, and cognitive states.
• Manual analysis of EEG data is inefficient, driving the need for automated processing and classification.
• EEG sub-band components (delta, theta, alpha, beta, gamma) are vital for understanding brain responses.

Purpose of the Study:

• To comprehensively review and evaluate EEG preprocessing techniques, particularly sub-band extraction methods.
• To assess the practical applicability and performance of various signal processing techniques for EEG analysis.
• To demonstrate the utility of these methods in a real-world application, such as drowsiness detection using Brain-Computer Interface (BCI).

Main Methods:

• Evaluation of signal processing techniques including Fast Fourier Transform (FFT), Short-Time Fourier Transform (STFT), Finite Impulse Response (FIR) and Infinite Impulse Response (IIR) filters.
• Analysis of wavelet transforms such as Discrete Wavelet Transform (DWT) and Wavelet Packet Transform (WPT).
• Qualitative and quantitative parametric analysis of selected EEG preprocessing methods.

Main Results:

• Comparison of the effectiveness of different sub-band extraction techniques for EEG signal processing.
• Assessment of the practical performance and suitability of various methods for Brain-Computer Interface (BCI) applications.
• Demonstration of improved EEG classification accuracy through optimized preprocessing in a drowsiness detection task.

Conclusions:

• Effective EEG preprocessing, especially sub-band extraction, is essential for accurate automated analysis and classification.
• The choice of preprocessing method significantly impacts the performance of Brain-Computer Interface (BCI) systems.
• The evaluated methods provide a foundation for developing more robust and efficient EEG-based applications.