Examination of temporal characteristic of sleep EEG subbands based on the local min-max

In this paper, the temporal characteristic of spectral subbands of sleep EEG associated with different sleep stages is examined using a simple computational analysis technique based on the local minima and maxima. The average range of local min-max R_λ defined as the average differences between amplitudes of consecutive local minima and maxima is used as a temporal characteristic measure. From the computational results, it is observed that the sleep EEG epochs associated with different sleep stages manifest distinguishable characteristics of the average range of local min-max R_λ for various subbands. The computational results also show that the lower frequency subbands of sleep EEG associated with various sleep stages exhibit opposite temporal characteristic (the average range of local min-max R_λ) compared to the higher frequency subbands of sleep EEG. Furthermore, this suggests that the average range of local min-max R_λ may be a useful feature for sleep stage classification.