Optimal EEG feature selection from average distance between events and non-events.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Neuroscience

Background:

• Biosignal classification faces challenges like extraneous features, imbalanced datasets, and low signal-to-noise ratios (SNRs).
• Effective feature selection/reduction is critical for robust biosignal analysis, particularly in applications like electroencephalogram (EEG) based microsleep detection.

Purpose of the Study:

• To evaluate a prototype EEG-based microsleep detection system using artificial data.
• To investigate the efficacy of a novel feature selection method, Average Distance Between Events and Non-events (ADEN), compared to traditional methods like Principal Component Analysis (PCA).

Main Methods:

• Generated artificial EEG data with varying SNRs (16 to 0.03) and class imbalances (down to 2% events).
• Extracted 544 spectral features from 16 EEG channels.
• Compared ADEN with PCA and other configurations for feature reduction and classification.

Main Results:

• ADEN achieved a phi correlation of 0.94 on imbalanced data with SNR 0.3, significantly outperforming PCA (phi=0.00).
• ADEN demonstrated consistent superior performance across different classifiers and data conditions.
• The proposed method showed strong potential for feature selection in challenging biosignal classification tasks.

Conclusions:

• ADEN presents a promising feature selection/reduction method for EEG-based microsleep detection.
• The simulation highlights ADEN's robustness, especially in scenarios with low SNR and imbalanced datasets.
• Further validation on real-world data is warranted to confirm ADEN's effectiveness.