Subject identification through standard EEG signals during resting states

Area of Science:

• Neuroscience
• Biometrics
• Signal Processing

Background:

• Individual identification is crucial for security and authentication.
• Traditional biometrics face challenges like spoofing and invasiveness.
• Electroencephalography (EEG) offers a non-invasive alternative for biometric identification.

Purpose of the Study:

• To evaluate the efficacy of electroencephalographic (EEG) activity for individual identification.
• To determine the optimal resting-state conditions (eyes open vs. eyes closed) for EEG-based biometrics.
• To identify specific brain regions yielding the highest accuracy in individual recognition.

Main Methods:

• Recorded high-density EEG signals from 50 healthy subjects during resting states (eyes open/closed).
• Computed Power Spectrum Density (PSD) in the 1-40 Hz range for feature extraction.
• Utilized a Naive Bayes classifier with K-fold cross-validation for recognition rate assessment.

Main Results:

• Achieved high Correct Recognition Rates (CRR) at parieto-occipital electrodes.
• CRR reached ~78% with eyes open and ~89% with eyes closed.
• Highest CRRs were observed at occipital electrodes (O2: 92%, O1: 91%) during eyes closed state.

Conclusions:

• Resting-state EEG signals, especially from occipital regions with eyes closed, demonstrate significant potential for reliable individual identification.
• EEG-based biometrics offer a promising, non-invasive method for personal authentication.
• Further research can explore advanced signal processing and machine learning techniques to enhance EEG biometric systems.