Leveraging Network Structure in Changepoint Detection for Dynamic Functional Connectivity in Resting-State fMRI Data

Elucidating the differences in brain function between children with and without psychopathology is necessary for growing our understanding of these disorders and of the brain more fully. Measuring dynamic functional connectivity (dFC) provides a glimpse of brain function, but a relative lack of statistical methods are available for characterizing dFC that leverage the structure of functional brain networks. We implement a changepoint detection method, PELT-LMEC-UB, which combines an existing mixed model framework for scalable covariance estimation and an accurate changepoint detection algorithm to locate structural changes in large covariance matrices. We show via simulation that PELT-LMEC-UB outperforms other changepoint methods at detecting the number and the location of the changepoints in short time series. We apply PELT-LMEC-UB to resting-state functional magnetic resonance imaging data from the Adolescent Brain Cognitive Development Study (n=4424). We use PELT-LMEC-UB to extract the number of changepoints in four brain networks, and examine the associations between the number of changepoints and measures of psychopathology and neurodevelopmental disorders in children ages 9–11. Our results show that brain networks exhibit distinct distributions of the number of changepoints and that the number of changepoints in the dorsal corticostriatal network may be positively associated with greater ADHD symptomatology. Supplementary materials for this article are available online.