Resting State Functional Brain Network Analysis In Major Depressive Disorder And Construction Of An Anatomical Image Template For Rat Brain

Studies in this dissertation could be classified into two separate parts. Part One, the resting state functional brain network analysis in major depressive disorder (MDD); Part Two, the construction of an anatomical image template for rat brain.Chapter One is a major introduction of the backgrounds, including the concepts and principles of functional and structural MRI, the resting state networks and associated analysis methods, MDD and recent studies, and voxel-based morphometry.Part One is composed of Chapter Two to Chapter Four. Resting state functional magnetic resonance imaging (rs-fMRI) has been widely used for exploring the intrinsic brain functional architectures. One essential application is involved in discovering the altered functional patters of mental disorders, which is important for the scientific research and clinical diagnosis and treatment. Progression in data analysis methods could provide more information implied in fMRI data. In this part, we analyzed the resting state networks of MDD from different points of view, including the extraction of networks, functional connectivity, functional transfer and sub-regional functions. Various methods were employed, mainly a new established seed-based iterative cross-correlation analysis (siCCA), volume and connectivity analysis, spectral clustering and functional specificity analysis.In Chapter Two, to solve the disadvantages of traditional seed-based correlation analysis (sCCA) and independent component analysis, we investigated the reliability and stability of the new methods siCCA in analysis of resting state networks. The flowchat of siCCA was established and used to extract default mode network (DMN) and stable task control network (STCN) in two independent datasets. The resulting networks were compared with those extracted by traditional correlation analysis and independent component analysis. The results revealed that, siCCA is a seed-independent method which could obtain stable and intact resting state networks, only if the seed was in the expected network. Inter-dataset similarity could also improved by siCCA, which is important for the disease research; that is, the alterations would be more likely to be induced by the disease but not inter-dataset variability.In Chapter Three, after extracting the DMN and STCN networks of MDD patients and normal controls, we studied the resting state brain connectivities and networks in MDD patients, and the correlation between networks’ characteristics and clinical scores. Four principal conclusions were achieved:1), functional centers were changed within both DMN and STCN of depression patients, according to the results of volume and functional connectivity anlaysis of siCCA results;2), in depression, the fronto-insular cortex transferred from DMN to STCN, and the limbic and basal ganglia structures had either increased connections with DMN or decreased connections with STCN;3), the volume of DMN was found to be negatively correlated with the patients’ social disability screening schedule (SDSS) scores;4), the connectivity between DMN and STCN was positively correlated with the depression (HDRS) and anxiety (HARS) levels, and this tendency would exist whether the regression of global signal in the preprocessing steps was performed or not. Most of the conclusions and viewpoints above were novel in MDD research, and we could find some correlations between these results and published studies. Some results in the functional connectivity analysis were consistent with previous researches. In Chapter Four, we focused on the amygdala, which is associated with negative moods and was proved to be associated with MDD brain dysfunctions both by our results and previous studies. Amygdala is composed of some structurally and functionally separated subregions. Thinking over the impact of inter-subject variability, we first parcellated the individual amygdala into three parts based on the Correlation matrix among the voxels in amygdala and spectral clustering analysis. Then we investigated the functional connectivity, specially the connectional specificity of each subregions in MDD and controls. According to the results, the three subregions revealed different functional patters; and the connectional characteristics of centromedial and laterobasal subregions were changed in depression.Chapter Five constitutes Part Two of the dissertation. Voxel-based morphometry (VBM) is a reliable and objective methods for discovering anatomical differences based on structural MRI images. Data analysis toolbox such as SPM has been widely used for the VBM analysis of human brain. One constraint of applying these tools for VBM analysis of rat’s brain is that, the methods needs apriori tissue probability maps (TPMs) for the normalization and segment in the procedure. Several studies has applied their own TPMs, but the resolution is just not enough for our subsequent researches.To meet the demands, we collected38MRI structural images of rat brain acquired from the7.0T magnetic scanner, and constructed a high-resolution MRI template and corresponding TPMs. The work was accomplished using fuzzy c-means clustering and other methods. Application in VBM analysis using SPM of these maps was also tested during the construction. The template and TPMs have been applied to several studies until now, and will be largely helpful for subsequent anatomical analysis of rats in the lab.Studies in this dissertation proved the feasibility of multiple methods in analysis of resting state functional MRI and structural MRI data, analyzed the dysfunction of brain connectivity and networks in depression, found a correlation between resting state networks and clinical scores, and provided a basis for analysis of rat’s brain structures. The results enriched our experience and knowledge in MRI data analysis methodology and depression researches.