Sleep Spindle Detection Based on Complex Demodulation

摘要

In this paper, we investigated the characteristic waveform of sleep spindle by using complex demodulation method (CDM). The ultimate purpose is to develop the automatic sleep spindle detection algorithm for overnight sleep data inspection. The main method includes four procedures. Firstly, the influences of disturbance of alpha waves, eye movement and muscle artifacts are analyzed and eliminated. Secondly, CDM is adopted to obtain the changing amplitude for sleep spindles. After CDM analysis, a set of parameters are calculated to find the candidate waveforms. Finally, the judgments of sleep spindle are made according to the defined thresholds of parameters. The overnight sleep recordings of three subjects were analyzed. Compared with the visual inspection of sleep stages by clinician, the obtained results showed that the detected sleep spindle were mostly distributed at sleep stage 2 and deep sleep stages. It is testified that CDM can well depict the instantaneous character of sleep spindle. The presented method can be an assistant tool for sleep spindle detection and sleep stage determination.