Methods of EEG signal features extraction using linear analysis in frequency and time-frequency domains

Area of Science:

• Biomedical Engineering
• Signal Processing
• Neuroscience

Background:

• Feature extraction is crucial for simplifying complex data, minimizing information loss, and reducing computational costs.
• Electroencephalography (EEG) signal analysis is vital for brain-computer interfaces (BCIs), medical diagnosis, and rehabilitation engineering.
• Various methods exist for EEG feature extraction, each with unique strengths and weaknesses.

Purpose of the Study:

• To discuss conventional electroencephalography (EEG) feature extraction methods.
• To compare the performance of different EEG feature extraction techniques for specific tasks.
• To recommend the most suitable EEG feature extraction method based on performance.

Main Methods:

• Time Frequency Distributions (TFD)
• Fast Fourier Transform (FFT)
• Eigenvector Methods (EM)
• Wavelet Transform (WT)
• Auto Regressive Method (ARM)

Main Results:

• Performance comparison of various EEG feature extraction methods is presented.
• The study identifies the most effective methods for specific EEG analysis tasks.
• Detailed performance metrics for each method are discussed.

Conclusions:

• The optimal EEG feature extraction method depends on the specific application and performance requirements.
• This research provides a comparative analysis to guide the selection of EEG feature extraction techniques.
• Informed selection of feature extraction methods can enhance the efficiency and accuracy of EEG-based systems.