Investigation of nonlinear ECoG changes during spontaneous sleep state changes and cortical arousal in fetal sheep

We examined the processes of cortical activation and deactivation of the fetal brain during spontaneous sleep state transitions and during central nervous processing of vibroacoustic stimulations (VASs) using nonlinear analysis of the electrocorticogram (ECoG). Tests of nonlinearity and a random shuffling routine revealed deterministic and nonlinear portions in the fetal ECoG. As common nonlinear measures are not applicable to nonstationary time series, we developed an algorithm to estimate the predictability of the ECoG in its time course by means of a point prediction error (PPE). The ECoG was recorded before and during VAS from the maternal abdominal surface in seven chronically instrumented fetal sheep at 0.8 of gestation. The PPE during REM sleep was significantly higher than during NREM sleep. VAS in NREM sleep resulted in an abrupt increase of the PPE not reaching the level of REM sleep. The steep increase of the PPE at onset and its slow decrease after cessation of the stimulus were very similar to the dynamics of spontaneous sleep state transitions, suggesting the involvement of the same cortical activating mechanisms. In conclusion, the stage and the time course of fetal brain activation and deactivation patterns can be clearly shown by PPE techniques. The PPE is a useful complement to spectral analysis. Both techniques describe different properties of the ECoG.