The Research Of Brain Death Determination Based On EEG Feature Analysis

Brain death refers to the irreversible loss of whole brain functions,including brainstem function,which is the most reliable criterion for death determination.Although most countries have made corresponding standards for the brain death determination,there are urgent problems to be solved,such as time consuming,high risk and poor robustness.Electroencephalogram(EEG)has become one of the most effective aided methods for clinical diagnosis of brain state because of its advantages such as low acquisition cost,high time resolution and non-invasive acquisition.However,at present,the method of brain death determination based on EEG is still immature.In this thesis,the EEG signals of deep coma patients(13)and brain death patients(17)are analyzed,and it is found that there are significant differences in various EEG feature between the two kinds of patients,and the validity and feasibility of brain death diagnosis based on EEG are verified.The results of this study can provide reference for doctors and help doctors make clinical decisions.The research work in this thesis includes:(1)Single variable-based EEG feature analysis for the two kinds of patients:the approximate entropy(ApEn)EEG features in different rhythms are calculated,and then the ApEn value for the denoised EEG using empirical modal decomposition are compared and finally the EEG energy feature is analyzed.Results show that the EEG complexity of brain death patients is significantly greater than deep coma patients in alpha,beta and gamma bands and the ApEn values are decreased after EMD processing and there is significant difference in theta frequency band.The EEG energy in deep coma patients is significantly higher than brain death patients across the whole frequency band.(2)Brain connectivity feature analysis for the two kinds of patients:a method for calculating the synchronization index brain regions is proposed,and three indicators of the inter-hemispheric phase synchronization(IHPS)index,the left hemispheric phase synchronization(LHPS)index and the right hemispheric phase synchronization(RHPS)index are defined.1)The phase locked values of patients with deep coma in the delta,theta,alpha,and beta bands are significantly stronger than those in the brain death patients;2)In the low frequency bands(delta,theta and alpha),the IHPS index of all patients with deep coma is significantly stronger than that of their LHPS and RHPS indices,and there exists significant difference,but this phenomenon does not occur in patients witli brain death(only 1 exception);3)It is found that the above phenomenon is verified in almost all trials,indicating that this phenomenon may provide a robust neuron-marker for brain death diagnosis.Finally,using the above phenomenon to construct a real-time detection system for brain death diagnosis,the results show the reliability of the method.Therefore,based on the above results,it is expected to develop a visualization tool to aided clinical brain death diagnosis.(3)Effective connectivity feature analysis for the two kinds of patients:Granger causality(GC)is applied to study the brain effective connectivity for the patients.Firstly,the signal must be smoothed,then the EEG with the stability condition are selected for GC calculation,and finally the brain effective network of two kinds of patients is established on the validation GC values.The results show that the brain network indicator(out-degree and in-degree)of deep coma patients is larger than brain death patients,and the information flow effectiveness in frontal lobe of deep coma patients is richer and tighter causality.