Spectral analysis method for sleep-state cycle based on the cortico-thalamo-cortical loop strength estimation

The spectral feature, especially spectral entropy, of sleep state has been studied in relation to various phenomena. While these studies show the importance of the spectral feature of sleep, their approaches have “Staging” or “Categorizing” process that make the analysis subjective or arbitrary. It is also unsatisfactory that physiological interpretation of the entropies estimated by their methods has not yet been established. In this study, we provide an alternative way to estimate the spectral feature of sleep state, without using sleep staging or categorizing. The method consists of three parts: calculation of the c
₂
(the cortico-thalamo-cortical loop strength) time series from sleep electroencephalography (EEG), Fourier transform of the c
₂
time series, and estimation of spectral entropy based on the Fourier power spectrum. Correlation analysis between the newly defined spectral entropy (CSE) and the previously defined Walsh spectral entropies (WSEs) shows that correlation between CSE and WSEs are relatively high when the direct current (DC), that is, the zero-frequency component of each spectrum, is excluded. By including the DC component, however, the correlation becomes even negative in some cases. These findings not only confirm the availability of the newly defined spectral entropy based on the cortico-thalamo-cortical model, but also cultivate a better understanding for sleep-state dynamics, especially in terms of “entropy,” which tends to be interpreted as an “irregularity” or “complexity” naively.