Research On The Characteristic Extraction Of Rest State EEG And TMS-EEG In Time-Frequency-Space Domain And The Assessment Of Disorders Of Consciousness

Disorders of consciousness(DOC)is a brain state of loss of consciousness caused by severe brain injury.Clinically,the consciousness assessment of DOC is mainly assessed by behavior scale,which is affected by the subjectivity of evaluators and the cooperation of patients.Electrophysiological technology can provide objective and quantifiable data for clinical diagnosis and help to improve the accuracy of DOC’s consciousness assessment;Neuromodulation technology has been applied to the rehabilitation of DOC,and the accurate and rapid assessment of its rehabilitation effect is also the main problem in the clinical practice of DOC.Assessing the neuromodulation effect through objective and quantifiable EEG characteristics is conducive to optimizing the modulation parameters,adjusting the modulation methods and improving the rehabilitation results of DOC.Aiming at the difficult problem about assessment of DOC’s consciousness states and neuromodulation effect,this paper develops EEG analysis methods from two modes:resting state EEG(rs EEG)and TMS-EEG,extracts multi-dimensional EEG characteristics to characterize the differences of DOC’s neural activities at different levels and dimensions,and provides quantitative characteristics for DOC research;Integrating multimodal EEG,a DOC’s EEG assessment framework is proposed,including operation details of feature acquisition,data preprocessing and multi-dimensional feature analysis.The main research contents are as follows:(1)Multidimensional analysis in time-frequency-space domain and optimal feature selection of DOC’s spontaneous brain activity.Aiming at the problem that the analysis dimension of DOC’s spontaneous brain activity is single and the feature analysis is not clear,the complexity in time domain,relative energy in frequency domain and connectivity in space domain are developed to extract the rs EEG characteristics of DOC.The machine learning method is used to screen the optimal characteristics for identifying DOC patients in different states,including permutation entropy,power in gamma band and permutation mutual information of frontal-parietal lobe.The value of optimal characteristics in assessment of DOC’s consciousness states and effects of spinal cord electrical stimulation is verified.The study identifies the key characteristics of spontaneous brain activity in DOC patients,which is helpful to improve the accuracy of assessment and optimize the neuromodulation scheme.(2)Efficiently quantify the complexity of DOC’s TMS-EEG data Aiming at the quantitative analysis of TMS-EEG data in time-domain,the fast perturbation complexity index(PCIst),is developed and optimized,which can effectively and quickly quantify the complexity of DOC’s evoked brain activity.Build machine learning models for DOC’s diagnosis and prognosis to verify the ability of PCIst in DOC’consciousness assessment and recovery prediction.The study provides an efficient and accurate quantitative characteristic of TMS-EEG in time-domain,which is helpful to the clinical application and popularization of TMS-EEG.(3)Accurately extract the time-frequency characteristics of DOC’s TMS-EEG data.Aiming at the problem of characteristics extraction and analysis of TMS-EEG data in time-frequency domain,a rhythm feature analysis method of instantaneous sudden change signal is proposed,which is synchrosqueezing wavelet transform(SST).The cortical natural frequency analysis method is developed,and the natural frequency differences of different DOC patients are found,which proves that spinal cord electrical stimulation can regulate the cortical characteristics of DOC patients.The research provides a characteristics extraction and analysis method of TMS-EEG data in time-frequency domain,which is helpful to deeply study the impact of consciousness injury and neuromodulation on cortical characteristics.(4)Detection method of cortical network connectivity in DOC patients by TMS-EEG.Aiming at the quantitative characteristics of evoked brain activity in space domain,the TMS-evoked synchronization index(TES)is proposed to characterize the connectivity degree of cortical network.The differences of cortical network connectivity in different DOC patients are compared and verified,and the effect of deep brain stimulation in DOC is studied.The study quantifies the DOC’s cortical connectivity,verifies that brain injury led to the reduction of cortical connectivity,and provides important biomarkers for DOC’s assessment.(5)Establish an efficient and accurate assessment scheme for DOC patients.Aiming at the problems of electrophysiological technology in DOC’s consciousness assessment,such as low application rate,high threshold and poor repeatability,an efficient and accurate EEG assessment scheme is developed.The multimodal EEG data acquisition scheme is optimized,a scientific and standardized data preprocessing process is formulated,and a quantitative index to measure the consciousness level of DOC patients is proposed by integrating multiple characteristics.The study provides a complete solution for DOC’s consciousness assessment.This paper studies the DOC’s rs EEG and TMS-EEG characteristics in the time-frequency-space domain,and constructs the EEG assessment scheme for DOC.On the one hand,it provides solutions for the DOC’s diagnosis and treatment problems;On the other hand,taking DOC as the research model,it improves the characteristics extraction technology of neural activities related to consciousness and promotes the studying of the neural mechanism of consciousness.