| Working Memory is a form of memory within a short period of time.It is responsible for the short-term maintenance and processing of new memory information and stored memory information.It is the basis of cognitive activity,such as learning,calculation,reasoning,decision making and language understanding.The study found that the incidence of cognitive defects in patients with mental illness is as high as 75%.Many patients experience cognitive impairments such as selective attention,information processing,working memory,learning,among which working memory impairment is a major cognitive dysfunction of mental illness.With the rapid development of cognitive neuroscience,brain imaging technology provides technical support for the study of brain structure and neurophysiology of working memory.Studies have shown that the brain network is closely linked to various cognitive functions,studying brain networks under resting state and cognitive tasks,exploring the connection patterns of brain networks,and understanding the characteristics of brain networks in patients with mental illness when working memory is of great significance.This thesis combines advanced brain imaging techniques with working memory behavioral experiments—analysis of the brain functional network under the working memory task,the study of the working memory brain network defects in mental patients,and the analysis of mental illness patients with significant differences from normal people during working memory from the perspective of brain functional networks.The mainly contents in this research work as follows:(1)The brain functional network was constructed based on EEG signals.The EEG data was processed in stages and bands before the construction.The scalp electrode was selected as the node,and the edge was determined by the PLV functional connection estimation method and threshold selection.This results in WM brain functional networks at various stages,frequency bands,and multiple sparsity levels.(2)Calculating the global and local attributes of the WM brain functional network,testing the brain network attribute values of the patient group and the the normal group in the same stage and at the same frequency band,extracting effective features and classifying the differences.The results manifested active bands and corresponding brain areas during working memory,and the classification results of local attributes are superior to global attributes.The clustering coefficients and local efficiencies can be used as biomarkers for classifying patients and normal person.(3)Analyzing the connection characteristics of brain network between patients and normal person,looking for the network topology that is different from normal person in the process of working memory,and introducing a index-graph kernel for measuring the similarity between networks in the EEG application area.With the help of graph kernels,we find the commonality between the two groups and the feature network topology of the patients and normal person in the working memory.The classification results shown that the classification results of the graph kernel values are obviously better than the classification results of the attribute values,and the classification sensitivity is up to 97%,which is of great significance for the disease clinical diagnosis.The two group have obvious differences in the subnetwork topology structure: subnet formation area,subnet size,and the connection between subnets and subnets. |
