Analyzing The EEG Of Global Hypoxic-Ischemic Brain Injury By Nonlinear Methods

Both stroke and cardiac arrest can lead to hypoxic- ischemic state. Brain is very sensitive to hypoxic-ischemic state because a large amount of energy is needed to perform well. Clinically, the real time monitoring system is in need, which can help to diagnose the states of the brain and reduce the brain injury. In this thesis, we tried to investigate the neural physiological mechanisms by analyzing the long-range correlation in EEG.Traditional methods for EEG signal processing are based on linear time and frequency analysis. Recently, as the development of the nonlinear dynamics, more attentions are paid to the nonlinearity in EEG. At present many nonlinear methods have been used to analyze the EEG, such as Lyaponov exponent and correlation dimension. In this thesis, we analyzed the long-range correlation in EEG by Detrended Fluctuation Analysis (DFA), in order to find a more suitable parameter to monitor the hypoxic-ischemic state.From the raw EEG, we extracted data corresponding to EEG energy, differentiated EEG, instantaneous frequency in EEG. We wish to determine which of these four scaling exponents best coordinates with previous histological findings: (1) subjects within a given group produce similar results; (2) tissue damage associated with high dose NAALDase inhibitor is less than that associated with low dose, which itself is less than that associated with saline solution.The results showed that the original EEG, differentiated EEG, and instantaneous frequency in EEG might not be in accord with previous histological findings. When comparing differentiated EEG and instantaneous frequency in EEG data within a given group, the expected similarities were not observed; when comparing between groups, varying NAALDase inhibitor dosage yielded results that were not significantly different. Original EEG data showed strong similarity between subjects in each group, but failed to support the known relationship between tissue damage and NAALDase inhibitor dosage. The data corresponding to EEG energy fits best with past histological findings, but does not provide an apt description of physiological changes throughout the entire process. These results suggest that evaluating physiological states with only a single parameter is unreliable, and encourage the study of fundamental physiological mechanisms in greater detail by the integration of linear and nonlinear signal processing methods.