Research On EEG Network Of Schizophrenia And Its Recognition

Schizophrenia(SCZ)is a chronic mental illness of unknown etiology,along with brain dysfunction in aspects of sensory perception,thinking,behavior,as well as cognitive,etc.Therefore,the investigation of the potential neurological mechanism that accounts for the disease is now urgent.SCZ is a heterogeneous syndrome with many different biological subtypes,each showing preferential response to different treatments.Therefore,the identification accuracy of SCZ subtypes is of great significance for the accurate diagnosis and treatment of this disease.Currently,SCZ can be clinically diagnosed with multiple subtypes,such as hebephrenic,paranoid,and simple SCZ,etc.,but the diagnosis mainly depends on doctors’ consultation and scale evaluation,which is largely influenced by the clinicians’ subjective judgment.Therefore,this kind of a research must find a biomarker and apply it to the reliable diagnosis and treatment of SCZ subtypes.In our present study,depending on resting-state electroencephalogram(rsEEG),we capture the neural activity of the brain by applying temporal variability analysis into dynamic functional connections of rsEEG,to reveal the physiological and pathological mechanism of SCZ,and further explore the potential value of rsEEG brain functional network in the clinical diagnosis of SCZ subtypes.The main works of this dissertation are as follows:1.The research on the temporal variability of schizophrenia.In this chapter,we first verified the feasibility of fuzzy entropy algorithm to temporal variability analysis of rsEEG,then captured the potential differences in brain neural activity between SCZ and healthy control(HC),further explored the potential relationship between temporal variability and individual cognitive task performance.Specifically,a sliding window approach combined with coherence analysis was first used to construct dynamic functional connections,whose temporal variability was quantitatively described by fuzzy entropy,we evaluated finally the Pearson correlation between temporal variability biomarker and task P300.The results showed that the connections with larger temporal variability in SCZ patients were mainly concentrated in the temporal lobe,left prefrontal lobe,left occipital lobe regions,while which in HC were concentrated in the frontal lobe and frontotemporal regions.The connections with smaller temporal variability formed the default mode network for both groups.The quantitative statistical analysis showed the temporal variability in SCZ patients was significantly lower than in HC in the whole brain area,with the most obvious on the right side of the brain.Besides,the temporal variability biomarker was significantly negatively correlated with P300 amplitude in SCZ patients,nevertheless,there was an opposite relationship in HC.Herein,from the perspective of rsEEG temporal variability,we revealed abnormal neural activity in SCZ patients,which helped to understand the physiological and pathological mechanisms of SCZ and deepened our understanding of the disease.2.The research on EEG recognition of schizophrenia subtypes.In this chapter,owing to the excellent performance of rsEEG network topology in the diagnosis of SCZ and HC,we focused on the differences of functional brain networks among different SCZ subtypes,then proposed for the first time to apply a spatial pattern of the network(SPN)to the multi-classification identification of SCZ subtypes.In specific,we firstly constructed functional brain networks under different oscillation rhythms by using phaselocking value,and calculated the differences of network topology among six SCZ subtypes.Then the SPN features extracted by the supervised learning method,to distinguish six SCZ subtypes in combination with the voting algorithm.Also,we discussed the classification performance of other rsEEG features,such as relative power spectrum density and network properties.The results showed that the classification accuracy of three SPN features was achieved 75.30% under the θ rhythm,which outperformed the other conventional rsEEG features.These findings consistently demonstrated SPN approach of rsEEG was a promising and potential clinical diagnosis tool for SCZ subtypes.To sum up,this paper developed a kind of temporal variability method based on rsEEG datasets,which was aimed to depicte accurately the brain activity,helped to investigate the flexibility of the brain region,interpret the brain dysfunction of SCZ.In the meantime,we preliminarily explored the spatial network differences among different SCZ subtypes corresponding to the resting networks,then explored the diagnosis performance of SPN features on multiple subtypes.