Analysis of Dynamic Evolution of Complex Brain Networks Based on Persistent Homology

Abstract

Some studies have shown that the cognitive state in Alzheimer's disease (AD) patients is related to the changes of the temporal characteristics of dynamic functional connection. Persistent homology index analysis method provides a deeper insight into the property of dynamic brain networks. However, current research mainly focused on spatial evolution, while there is little research on time-varying brain network evolution. Based on resting state-functional magnetic resonance imaging (rs-fMRI) data, this paper analyzed the static functional connectivity network in AD patients and normal controls (NC), as well as the dynamic brain networks constructed by sliding windows. The analysis method based on persistent homology was compared with the ones based on graph theory, and the temporal characteristics were analyzed with k-means clustering. The results demonstrated that there were more significant differences between the AD patients and NC subjects, when the index based on persistent homology was used. Compared with the analysis of static brain networks, the dynamic brain network evolution analysis based on persistent homology provides new ideas for the detection of functional brain network biomarkers.

Keywords

• functional magnetic resonance imaging
• dynamic functional connectivity
• persistent homology
• sliding windows