Estimating and detecting slow wave events in EEG.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Slow-wave activity (SWA) in electroencephalogram (EEG) signals is a key indicator of deep sleep and anesthesia.
• Conventional spectral analysis methods struggle with the quasi-periodic nature of individual slow waves (ISWs).
• There is a need for robust methods to quantify SWA as a biomarker for altered brain states.

Purpose of the Study:

• To develop and validate a time-domain framework for detecting and quantifying individual slow waves (ISWs).
• To establish an instantaneous measure of SWA, termed SWA probability, using ISW detection.
• To assess the utility of this SWA probability metric in general anesthesia and deep sleep.

Main Methods:

• Utilized time-delay embedding to represent univariate EEG signals in a two-dimensional delay embedded space (DES).
• Defined candidate ISWs as segments between zero crossings and identified occurrences within an admissible region in the DES.
• Applied Kalman filtering to a binary ISW occurrence time series to estimate SWA probability.

Main Results:

• Successfully detected and quantified ISWs using the proposed time-domain framework.
• Generated a time-varying SWA probability metric reflecting the instantaneous SWA.
• Demonstrated elevated SWA probability in EEG data from both general anesthesia and deep sleep states.

Conclusions:

• The developed time-domain method provides an effective means to detect and quantify SWA via ISW occurrences.
• The SWA probability metric serves as a sensitive and instantaneous measure of SWA.
• This method holds promise as a biomarker for quantifying states of anesthesia and deep sleep.