Brain Functional Networks Visualization Algorithms Design And Implementation Of A Visualization Tool

In recent decades, research on human brain functional networks has become a major topic in biology. And visualization of complex networks, as an important method of data analysis, plays a significant role in the study of brain science. Visualization is capable of directly displaying and vividly describing network structure, as a result, people could be able to easily obtain network information, which consequently facilitates the research of human brain structure and mechanism.This thesis discusses the current research status of complex network visualization techniques from two perspectives, layout algorithms and network scale compressing measures as well as the issues that one may encounter when utilizing existed methods to visualize brain functional networks. Secondly, an improved layout algorithm is introduced. Experimental results show the improved algorithm can steadily produce network visualization layout with a small number of edge crossing, which makes it suitable for comparing multiple networks, a common scenario in the analysis of brain functional networks. Thirdly, a new community detection algorithm is presented, with which one can iteratively find and merge communities, and consequently the original network could be turned into a hierarchical structure that has multiple network layers with various size. And visualizing such a structure can benefit from maintaining details while downsizing the network. And the experiment implies that the hierarchical structure constructed by this new community detection algorithm shows better layering and more reasonable size transition. Lastly, by integrating related techniques, a specialized visualization tool for brain functional networks is designed and developed with Python modules Networks and graph-tool. The tool is equipped with a simple but friendly UI, and is able to load network data that user processes with Matlab. Noticeably, the tool doesn’t only provide frequently used layout algorithms, but also posses a bunch of visualization optimizations at several network measures that are generally described by statistical means, such as node degree and hubs in order to reveal network information as more as possible by visualizing. Besides, this tool has many interactive functions so that users can explore and modify visualization results. As a result, the visualization tool has good practical applicability.