Study Of The Brain Connectomics Based On Multi-modal Magnetic Resonance Imaging

It remains a challenging scientific issue to explore the brain and uncover the brain activity.Brain function is not only determined independently by a single neuron or a single brain region;instead,it depends on the interaction between neuron cluster and the neural circuit in the functional block or the brain region.Brain can be modeled as a complicated network,which can realize the highly effective information transmission.With the development of the complex network theory,brain network analysis has gradually become an important research direction in the neuroscience field.In recent years,techniques,such as neuroimaging and magnetoencephalography,have been utilized to construct multiple brain networks at various scales,which have further pushed the development of brain network research.Apart from investigating the normal brain working mechanism,numerous studies indicate that,many mentally ill brain injuries will frequently reflect on the structural and functional brain networks.Schizophrenia is one of the chronic and severe mental disease,which is characterized by hallucination,delusion,loss of initiativeness and cognitive impairment.Therefore,it is necessary to adopt appropriate mathematical and statistical models to construct the brain network,and to carry out in-depth research on the brain structureand function.On the one hand,the working mechanism of the normal brain should be explored;on the other hand,the pathophysiological network mechanism of mental disease should be explored.At the same time,different modal images have carried complementary information,so the development and application of the multi-modal fusion method can provide a more sensitive measurement,which can support the neuroimaging technique-based clinical diagnosis.Consequently,our research mainly focused on the following four parts based on these problems:1 We used functional magnetic resonance imaging(fMRI)to construct functional network.Using dynamic functional connectivity analysis,we aimed to delineate the role of putative ‘ resting-state instability ’ in the genetic diathesis and clinical phenotype of schizophrenia.Also,we examined the antipsychotic dose effects to study the secondary effects of treatment on putative markers of clinical expression and undertook an exploratory analysis of the relationship between clinical severity(positive and negative symptoms scores)and regional variability measured using dynamic functional connectivity.Our results showed that the time-varying instability of precuneus may be a core feature of the expression of schizophrenia phenotype rather than being a genetically mediated or illness severity or treatment-related epiphenomenon.2 The white matter abnormalities in patients with schizophrenia were studied using diffusion tensor imaging(DTI).we combined two important but previously separate approaches,voxel-based analysis(VBA)and diffusion tractography,to examine the uni-,bi-,and multi-variate properties and identify white matter abnormalities in schizophrenia.VBA showed widespread decreases in fractional anisotropy(FA)values,fiber tracking also confirmed the observed FA-value reduction,as 31 connections were impaired in patients,especially the interhemispheric connections and thalamo-cortical circuits.Our topological property analysis suggests that the structural network of patients shows a weaker global integration and that their brain network hubs are less functional.The present study provided evidence of white matter disruptions in patients with schizophrenia,using a multilevel approach in a large sample size.3 We developed a novel multivariate data fusion method to combine the resting state fMRI,structural magnetic resonance imaging(sMRI)and DTI without reducing the dimension or using the priors.By constructing the multi-index feature for each brain region,we calculated the p value of significant difference between groups for four different models(fMRI,sMRI,DTI,fusion).Our findings showed that the three-way fusion feature has the smallest p value in most of the brain regions which indicates that we can increase the difference between the groups byintegrating additional information.The results of discriminant analysis confirmed this conclusion which shows that each modality is indispensable for achieving good combination and classification.The fusion feature has the highest predictive accuracy of 86.12%.4 We utilized brain imaging data from three different modalities to construct the functional network,probability tracking structural network,and KL-based morphological network of each participant,respectively.These networks were used in combination with the machine learning method to identify more consistent biomarkers of brain connectivity.Our results showed: although each modality yielded a different connectivity biomarker,all were mostly located within the basal ganglia-thalamus-cortex loop.Furthermore,using the biomarkers of these three modalities as a feature yielded the highest classification accuracy(91.75%,relative to a single modality).