Time-Dependence of Graph Theory Metrics in Functional Connectivity Analysis (I7.001)

OBJECTIVE: To estimate the dynamic nature of graph theory measures of whole-brain functional connectivity (FC), in order to (1) improve sensitivity of connectomic investigations in epilepsy and (2) improve discriminatory power of imaging biomarkers based on graph theory methods. BACKGROUND: Connectomic analysis of temporal lobe epilepsy (TLE) using graph-theoretical methods is increasingly found to be a powerful quantitative method for investigating epileptic brain networks. Studies increasingly demonstrate the utility of graph measures of FC for identifying network abnormalities and serving as diagnostic markers for localization or disease extent. The majority of graph theory investigations of FC have relied on the assumption of temporal stationarity. However, recent evidence increasingly suggests that FC fluctuates dynamically over time. METHODS: Interictal resting-state fMRI was performed in 32 TLE patients and 24 healthy controls. Sliding window analysis was used to extract time-varying graph metrics across the length of the scan. Dynamic changes in graph metrics was quantified through Bayesian hidden Markov modeling. Temporal stability was estimated for graph measures of network connectivity, including small-world index, global integration measures (global efficiency, characteristic path length), local segregation measures (clustering coefficient, local efficiency), and centrality measures (betweenness centrality, eigenvector centrality). RESULTS: Small-world index, betweenness centrality, and global integration measures exhibited greater temporal stationarity than other network characteristics. The exception was clustering coefficient for TLE patients, which exhibited the least temporal stationarity for healthy controls but greatest for TLE. We further show that imaging markers that account for subject-level differences in network dynamics obtain better discriminatory power as a marker for TLE. CONCLUSIONS: Our results suggest that the robustness of static FC analysis depends on the graph measure investigated. Temporal stability of network topology may itself serve as a marker for TLE. Incorporating network dynamics into imaging biomarkers may improve the sensitivity of connectomic investigations in TLE. Disclosure: Dr. Chiang has nothing to disclose. Dr. Cassese has nothing to disclose. Dr. Guindani has nothing to disclose. Dr. Vannucci has nothing to disclose. Dr. Yeh has nothing to disclose. Dr. Haneef has nothing to disclose. Dr. Stern received personal compensation for activities with UCB Pharma, Sunovian, Lundbeck, Eisai Inc., and Cyberonics as an advisor and/or speaker. Dr. Stern has received personal compensation in an editorial capacity for MedLink Neurology as editor.