Visual analytics of heterogeneous brain networks using multilevel methods

Abstract

Networks have been used extensively for modeling functional, structural, and dynamic connectivity of the human brain. This representation can modeling (and integrate) heterogeneous data from various sources, as different types of neuroimaging and information about diseases and genes from text collections. This research field helps the understand and enriches our understanding of global and local functional and structural aspects of the neural architecture, and it has been demonstrating clinical relevance and high scientific impact in the area. In particular, the visual analytics of this type of networks is used to assist specialists to iteratively inspect the structure of the human brain with ease and speed, which helps to find patterns or disorders that define the presence or absence of diseases, furthermore, it can help in understanding their dynamics. In this context, some problems are inherent, as high computational loads due to the massive data, time and complexity of the layout and mining algorithms; or the high number of graphic elements and overlapping, which impair the readability and interaction. These issues are aggravated if the network has a heterogeneous structure. Although a large number of strategies for iteratively visual analytics of these networks have been successfully proposed and applied, there are still many challenging questions that need to be scientifically explored. The project aims to advance the state-of-the-art by investigating and propose methods for efficient and effective iterative visualization of brain networks modeled from heterogeneous data. Furthermore, the project aims to compare networks in different conditions, as networks in resting-state and pre- and post-crisis periods in the Epilepsy context. The multilevel method will be used as a theoretical basis, which describes a scalable strategy that creates a hierarchy of reduced (or simplified) versions of the original network. The theoretical and technological developed resources will be used to design different applications, focusing on Epilepsy, and can be useful to complement the current and well-known tools. It is hoped that this research will generate relevant contributions for a better understanding of the mechanism of some neurological diseases through the analysis and visualization of functional and structural brain networks using the multilevel method. (AU)