The Modeling And Evolution Of Brain Network Based On Functional Magnetic Resonance Imaging

Brain consists of billions of neurons which couple with each other, forming a complex system to complete a variety of physical and mental activities. With the vigorous development of neuroimaging, scientists are able to continue more in-depth exploration. Due to its non-invasive characteristics, high resolution, and mature technology and, functional magnetic resonance imaging(f MRI) has been widely used to investigate the brain activities in vivo experiments. Complex network theory as a powerful tool for studying the complex coupling system, becomes the theoretical basis of brain network modeling, and is used to measure topological changes in brain networks. In this paper, we used complex network theory to study the relevant research in f MRI data.First, we propose a new brain network construction method, sub-network voting with sliding window, to address the issue that the confidence in traditional methods is not satisfying. This method can improve the consistency within group and highlight the differences between groups, thus, reducing the false positive rate of discovering an edge, and making the brain networks constructed more reliable. Meanwhile, the classification between the young and the elderly groups based on this method can significantly improve the performance.Second, we proposed a dynamic evolution model of normal aging in functional brain networks. As the normal aging in brain is a dynamic process, the traditional static brain network model is not able to fully describe the topological changes. The proposed model can simulate the transition from the young group via the middle-aged group to the elderly group. It simulated the changes in topological properties, and the re-organizations of edges in the network.Third, we further proposed an evolution model of the Alzheimer's disease(AD). This model successfully simulated the re-arrangement of edges in pathological process of AD, and captured the degradation of brain network from a small-world architecture to a random one.Finally, we utilized ‘j PCA' to perform the signal reduction and data visualization for blood oxygen level dependent(BOLD) signal. The nerve signals are often complex and high-dimensional, which makes it very difficult to calculate the statistics. After signal reduction, results showed that in the resting state, the two-dimensional trajectory of BOLD signal exhibited a nearly circular. It inferred that the ‘resting state' is a basic state that consist of low frequency and long period in physiological activities.