A novel method for EOG features extraction from the forehead.

Area of Science:

• Biomedical Engineering
• Neuroscience
• Human-Computer Interaction

Background:

• Electrooculogram (EOG) signals, particularly slow eye movements, are established indicators of human vigilance.
• Traditional EOG signal recording requires electrodes placed near the eyes, limiting user convenience in real-world scenarios.

Purpose of the Study:

• To investigate a more practical method for acquiring EOG signals for vigilance estimation.
• To evaluate the feasibility of using forehead-placed electrodes for EOG signal acquisition.

Main Methods:

• EOG signals were recorded from electrodes placed on the forehead.
• Independent Component Analysis (ICA) and Support Vector Machines (SVM) were employed to extract EOG features.
• Performance was assessed by correlating forehead EOG signals with traditional, eye-near EOG signals.

Main Results:

• A high correlation coefficient of 0.84 was achieved between forehead and traditional EOG signals during 14 distinct facial movements.
• For eye movements alone, the correlation reached an impressive 0.93.

Conclusions:

• Forehead EOG signal acquisition is a viable and practical alternative for human vigilance estimation.
• The proposed method maintains high signal fidelity compared to traditional approaches, enhancing user convenience.