EEG-based mental fatigue detection using linear prediction cepstral coefficients and Riemann spatial covariance...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Mental fatigue monitoring is crucial due to severe consequences of fatigue.
• Electroencephalogram (EEG) offers high temporal resolution for fatigue detection.
• Existing EEG methods often lack comprehensive feature extraction.

Purpose of the Study:

• To develop a novel algorithm for mental fatigue detection using multi-domain EEG feature fusion.
• To improve the accuracy and comprehensiveness of EEG-based mental fatigue monitoring.

Main Methods:

• Utilized linear prediction to calculate linear prediction cepstral coefficients (LPCCs) for time-domain features.
• Employed Riemannian geometry to project spatial covariance matrices into the Riemannian tangent space for spatial-domain features.
• Fused multi-domain features to capture comprehensive spatio-temporal information.

Main Results:

• Achieved 87.10% classification accuracy on the SEED-VIG dataset (three categories).
• Attained 97.40% classification accuracy on a self-designed dataset (two categories).
• Demonstrated superior performance compared to existing state-of-the-art methods.

Conclusions:

• The proposed multi-domain feature fusion strategy is effective for mental fatigue detection using EEG.
• This approach offers a more comprehensive analysis of EEG signals for fatigue monitoring.
• The findings support the advancement of reliable mental fatigue detection systems.