Automatic detection of slow wave sleep using two channel electro-oculography

Area of Science:

• Neuroscience
• Sleep Medicine
• Biomedical Engineering

Background:

• Accurate detection of slow wave sleep (SWS) is crucial for understanding sleep architecture and diagnosing sleep disorders.
• Traditional visual scoring methods are time-consuming and require expert interpretation.
• There is a need for automated, efficient, and reliable SWS detection techniques.

Purpose of the Study:

• To develop and validate an automatic method for detecting slow wave sleep (SWS) using electro-oculography (EOG) and electroencephalography (EEG) signals.
• To assess the performance of the automatic SWS detection against standard visual scoring criteria.
• To evaluate the feasibility of online application of the automatic method.

Main Methods:

• Developed an automatic SWS detection algorithm utilizing two-channel EOG and EEG.
• Employed cross-correlation analysis of EOG signals within a 0.5-6 Hz band to identify synchronous EEG activity.
• Utilized an amplitude criterion for slow wave detection and beta power (18-30 Hz) for artifact exclusion.
• Validated the method on 265 subjects, with thresholds optimized on 133 training subjects and applied to 132 validation subjects.

Main Results:

• The automatic method achieved substantial agreement (Cohen's Kappa = 0.70) with visual scoring for SWS detection, showing 93% epoch-by-epoch agreement.
• SWS epoch detection demonstrated a sensitivity of 75% and a specificity of 96%.
• The method also accurately estimated total slow wave time (SWT).

Conclusions:

• The developed automatic method provides a reliable and efficient means for detecting SWS.
• Its ability to be applied online using minimal electrodes (four) offers a significant advantage for real-time sleep monitoring.
• This automated approach has the potential to streamline sleep analysis and improve accessibility to sleep studies.