MRI-Based Brain Networks Approaches And Clinical Applications

The human brain is so far considered to the most complex, most delicate object inthe universe, which is organized into parallel, interacting systems of anatomicallyconnected areas. Nowadays, brain network research is one of the hot topics in the fieldof neuroscience, which can investigate the connection mechanism between brainregions on the level of system, and further reveal the underlying organization of thebrain. In recent years, the invention and rapid development of brain imagingtechnologies, especially magnetic resonance imaging (MRI), provide imaging tool toacquire various brain data; while complex network analysis based on graph theoryprovides a valuable technical means for brain network research. Focusing on thedevelopment and clinical applications of brain network analysis, this dissertationcombined two popular imaging technologies based on MRI, that is, functional magneticresonance imaging (fMRI) and diffusion tensor imaging (DTI), and graph theoreticalanalysis to investigate brain functional connectivity and structural connectivitynetworks, reveal the underlying topological organization of the brain in the healthysubjects, and further detect the pathophysiological mechanism of clinical brain diseasesfrom the system level. The main works and contributions of this dissertation are asfollows:1. Combining independent component analysis (ICA) and graph theoreticalanalysis to deal with resting-state fMRI (rs-fMRI) data in healthy subjects andinvestigate the topological properties of resting-state networks (RSNs). Our resultsshowed that each RSN had robust small-world properties. More important, the networktopological properties were significantly different between higher cognitive networks(dorsal attention network, central-executive network and default mode network) andperceptual networks (visual network, auditory network and somato-motor network).These findings for the first time provide quantitative evidence for the topologicalfractionation between higher cognitive and perceptual networks. Our approach toinvestigate topological properties in RSNs may be extended to clinical research, especially to diseases that show selective abnormal connectivity in specific brainnetworks2. Using large-scale functional connectivity analysis to investigate whole-brainfunctional connectivity network in social anxiety disorder (SAD). Compared withhealthy controls, SAD patients exhibited abnormal functional connectivity, especiallyinvolving the frontal cortex and occipital cortex. In addition, correlations were detectedbetween changes in resting functional connectivity and social anxiety symptom severityin some cases. The aim of this study was to investigate the abnormal functionalconnectivity under pathologic state from the whole-brain level, and our findings couldrepresent an early imaging biomarker for SAD.3. Combining functional connectivity based on resting-state fMRI signalcorrelations and structural connectivity based on diffusion tensor imaging tractographyto investigate brain functional and structural networks in psychogenic non-epilepsyseizures (PNES). Using graph theoretical analysis, we found that PNES patients lostoptimal topological organization in functional connectivity and structural connectivitynetworks reflected by a shift towards more regular brain architecture. In addition,structural connectivity networks exhibited altered regional characteristics in some keyregions associated with attention, sensorimotor, subcortical and default-mode systems inPNES patients. Most importantly, we found that the coupling strength betweenfunctional and structural connectivity was significantly decreased, and this decrease wasmore marked in patients with a longer duration of disease. When taking couplingstrength as an index for plotting receiver operating characteristic (ROC) curves, PNESpatients can be differentiated from healthy controls with high sensitivity and specificity.These results may provide new insights into our understanding of thepathophysiological mechanisms of PNES.4. Using functional connectivity density (FCD) mapping, an ultrafast andvoxelwise data-driven approach, to measure short-and long-range FCD in PNESpatients. Compared with health controls, PNES patients showed abnormal FCD inregions associated with attention, emotion and sensorimotor systems. In addition, someregions in occipital cortex also exhibited abnormal long-range FCD, and were correlatedwith duration of disease, which could be related to generalization of greater vigilance orresponsivity to environmental stimuli in PNES patients. The findings further improve our understanding of the pathophysiological mechanisms of PNES using functionalconnectivity density mapping.