An Adaptive EEG Feature Extraction Method Based on Stacked Denoising Autoencoder for Mental Fatigue Connectivity.

Area of Science:

• Neuroscience
• Cognitive Science
• Biomedical Engineering

Background:

• Mental fatigue arises from sustained cognitive effort, impacting performance.
• Brain connectivity alterations during mental fatigue remain poorly understood.
• Existing electroencephalogram (EEG) analysis methods struggle with noise, limiting connectivity insights.

Purpose of the Study:

• To develop an advanced method for analyzing brain connectivity during mental fatigue.
• To investigate the impact of fatigue and sleep deprivation on brain network dynamics.
• To improve the signal-to-noise ratio in EEG-based connectivity analysis.

Main Methods:

• Proposed an adaptive feature extraction model using a stacked denoising autoencoder.
• Analyzed the signal-to-noise ratio improvement compared to principal component analysis.
• Applied the model to analyze causal brain connectivity across frontal, motor, parietal, and visual areas under different conditions.

Main Results:

• The stacked denoising autoencoder significantly improved signal-to-noise ratio and suppressed noise.
• Distinct brain connectivity patterns were identified for awake, fatigue, and sleep deprivation states.
• Connectivity direction reversed between awake and sleep deprivation conditions.
• Fatigue exhibited complex, bidirectional connectivity between anterior and posterior brain regions.

Conclusions:

• The proposed method offers an effective approach for EEG analysis in studying mental fatigue.
• Brain connectivity patterns differ significantly across awake, fatigue, and sleep deprivation states.
• Findings provide insights into the neural mechanisms underlying mental fatigue.