Exploring the effective connectivity of resting state networks in Mild Cognitive Impairment: An fMRI study combining ICA and multivariate Granger causality analysis

Mild cognitive impairment (MCI) was recognized as the prodromal stage of Alzheimer's disease (AD). Recent neuroimaging studies have shown that the cognitive and memory decline in AD and MCI patients is coupled with abnormal functions of focal brain regions and disrupted functional connectivity between distinct brain regions, as well as losses of small-world attributes. However, the causal interactions among the spatially isolated but function-related resting state networks (RSNs) are still largely unexplored in MCI patients. In this study, we first identified eight RSNs by independent components analysis (ICA) from resting state functional MRI data of 16 MCI patients and 18 age-matched healthy subjects respectively. Then, we performed a multivariate Granger causality analysis (mGCA) to evaluate the effective connectivity among the RSNs. We found that MCI patients exhibited decreased causal interactions among the RSNs in both intensity and quantity compared with normal controls. Results from mGCA indicated that the causal interactions involving the default mode network (DMN) became weaker in MCI patients, while stronger causal connectivity emerged related to the memory network and executive control network. Our findings suggested that the DMN played a less important role in MCI patients. Increased causal connectivity of the memory network and executive control network may elucidate the dysfunctional and compensatory processes in the brain networks of MCI patients. These preliminary findings may be helpful for further understanding the pathological mechanisms of MCI and provide a new clue to explore the neurophysiological mechanisms of MCI.