Computational Model Of Brain Network Features Based On Magnetic Resonance And Its Application In Motor Disorders

Magnetic resonance imaging(MRI)is a non-invasive medical imaging technique that can produce high-resolution and high-contrast 3D/4D images.It is widely used for research on brain structure and function as well as clinical diagnosis and treatment.The MRI-based brain network feature computation model is used to compute and describe the network features of brain structure and function.In this model,the brain is viewed as a complex network system,with each region representing a node and the connections between regions representing different functional or structural relationships.The goal of this model is to analyze the connection patterns between different regions of the brain,the relationships between multimodal connections,and to determine the topological structure and connection strength of the brain network.It is applied in clinical research to identify neuroimaging biomarkers for neuropsychiatric diseases.This thesis establishes a brain network feature computation model that integrates structural and functional information of the brain,and investigates the abnormal changes in brain network connections in movement disorders such as spinal cord injury and Parkinson’s disease.This provides effective assistance in understanding the physiological mechanisms of the disease and designing new treatment strategies.The main work is as follows:Firstly,based on the conditional Granger causality analysis method,we explore the causal connection brain network patterns in the emotional-motivational-motor circuit related to motor rehabilitation in spinal cord injury patients.The results showed that compared with the good recoverers(GR)and the healthy control(HC),the poor recoverers(PR)exhibited an imbalanced emotion-motor causal network.Meanwhile,post-hoc analysis results showed that compared with GR,significant inhibitory effects were observed from the left amygdala to the left supplementary motor area and from the right precentral gyrus to the right nucleus accumbens in PR.Further exploration of the relationship among depressive symptoms,abnormal causal connections,and motor recovery was carried out.Mediation analysis revealed the mediating effect of depression in the functional feedback of motor-motivation and motor recovery.The results emphasized the importance of emotional intervention in the functional recovery of motor after SCI.Secondly,due to the limitation of brain structure on brain function,this thesis constructed a calculation model of brain functional network feature incorporating structural prior information using a regression model and used this model to research the brain network features of non-motor symptoms in Parkinson’s disease(PD).First,based on the clinical scale of non-motor symptoms in PD,the subtypes were classified.Then,a brain functional network feature calculation model was established.The results of statistical tests showed that adding structural information provided more PD’s network abnormal information.The prediction model showed that the classification accuracy of PD subtype 1 and HC was 89% and the classification accuracy of PD subtype 2 and HC was 78%,and the classification accuracy of PD subtype 1 and PD subtype 2 was 80% after incorporating structural information.Compared with not incorporating structural information,the combination of structural and functional information improved the performance of different subtype classification.Finally,the structure-function coupling reflects the functional communication supported by the structure.This thesis proposes a new structural-functional coupling brain network feature computation model based on the idea of conditional probability,and analyzes the global structure-function coupling of PD and HC.The results indicate a significant reduction in the global structure-function coupling of PD,suggesting that the decrease in global structure-function coupling may be the cause of their cognitive impairment.