Wavelet-based segmentation techniques in the detection of microarousals in the sleep EEG

The presence of undesirable frequency bursts called microarousals (MA), within any stage of sleep, causes a medical condition known as excessive daytime sleepiness (EDS). Traditionally, using the electroencephalogram (EEG) and the electromyogram (EMG), a sleep technologist detects the MAs. To reduce the time, cost and errors associated with manual scoring, a three-stage computerized automatic detection procedure is proposed. The first stage involves spectral decomposition using the discrete wavelet transform (DWT). The second stage uses three different segmentation techniques: the autocorrelation function (ACF), the nonlinear energy operator (NLEO) and the generalized likelihood ratio (GLR) methods to segment the detail function into stationary segments. The third stage scores the MAs, by comparing the power and spectral content of each stationary segment with the rules established by Rechtschaffen and Kales. The procedure is applied to two cases: to one EEG channel, in one case, and to a combination of the EEG and EMG channels, in the other. The results show that the presence of the DWT significantly improves the correct MA detection while the combined channel provides the most impressive detection results.