Fused Multi-modal Analysis On Dynamic Characteristics Of Brain Connectivity

Recently,dynamic property of brain activity and functional interaction of different brain regions has become hotspot in resting-state functional Magnetic Resonance Imaging(rs-fMRI)analysis.As an effective method to characterize neural activity and functional interaction within human brain,dynamic functional connectivity analysis has been widely applied.Moreover,the development of simultaneous data acquisition and preprocessing algorithm promoted the research of simultaneous EEG-fMRI analysis and made it possible to probe dynamic feature of brain under better space-time resolution.In this dissertation we performed dynamic functional connectivity analysis on a set of singlemodality rs-fMRI data and achieved some interesting results.In addition,we applied fused analysis on multi-modality neuroimaging data based on a set of simultaneous EEGfMRI data.The dissertation mainly includes the following contents:Analysis and application of dynamic functional connectivity.We applied dynamic functional connectivity analysis to a set of resting-state fMRI data including schizophrenia patients,patients' siblings and healthy control and investigated the heredity characteristics of schizophrenia.We identified five shared aberrant functional connectivity among patients and their siblings compared with normal control,which are as follow: 1)right anterior prefrontal cortex-right precuneus;2)left fusiform gyrus – left inferior temporal gyrus;3)left anterior insular – left inferior temporal gyrus;4)left anterior insula – right angular gyrus;5)left ventral medial prefrontal cortex – right medial occipital lobe.These abnormal functional connectivity revealed the heredity characteristics of schizophrenia and indicated the higher risks for patients' siblings to develop schizophrenia than normal control from a neuroimaging perspective.EEG correlated fMRI analysis based on supervised sparse learning.In this dissertation,we performed traditional statistical parametric mapping(SPM)and supervised sparse learning on simultaneous EEG-fMRI data of familial myoclonic epilepsy respectively and analysed brain regions related to interictal epileptic discharge(IED).It is shown that the supervised sparse learning algorithm can effectively extract brain regions that are responsible for IED,suggesting its promising future application on fused analysis of simultaneous EEG-fMRI data.The contribution of this dissertation lies on: 1)we performed dynamic functional connectivity analysis to research the heredity characteristics of schizophrenia and detected five shared aberrant functional connectivity among schizophrenia patients and their siblings compared to healthy control.This result explained the higher risks for schizophrenia patients to develop schizophrenia than healthy controls in a neuroimaging perspective;2)we applied supervised sparse learning algorithm on EEG-correlated fMRI analysis and utilized simultaneous EEG-fMRI data of familial myoclonic epilepsy patients as verification.The algorithm achieved reliable results and indicated its promising future application on fused EEG-fMRI analysis.