Abstract
BACKGROUND: Dynamic functional network connectivity (dFNC) assesses temporal fluctuations in functional connectivity (FC) during magnetic resonance imaging (MRI), capturing transient changes in neural activity. Investigating dFNC may provide valuable insights into the complex clinical manifestations of Alzheimer's disease (AD). However, research on dynamic FC alterations in AD remain limited. This study aimed to comprehensively characterize dFNC patterns in patients with AD. METHODS: A total of 100 patients diagnosed with AD and 69 with healthy controls (HC) were included. Resting-state functional magnetic resonance imaging (rs-fMRI) data were analyzed using a sliding-window approach and k-means clustering based on independent component analysis to examine dFNC alterations. Correlation analyses were conducted to assess associations between dFNC variations and clinical scores in individuals with AD. Additionally, an exploratory multivariate pattern analysis was performed to classify AD across different dFNC states. RESULTS: Four recurrent connectivity states were identified. In state III, patients with AD exhibited a significantly longer mean dwell time and a higher fractional time compared to the HC group, whereas the opposite trend was observed in state IV. In state III, both fractional time and mean dwell time were negatively correlated with cognitive scores. Significant differences in FC strength were observed between states II and III. The highest classification accuracy for distinguishing AD was achieved in state II, which was characterized by intra- and inter-network dysfunction across multiple functional networks. CONCLUSION: Distinct alterations in dFNC were identified, with significant associations observed between connectivity patterns and clinical symptoms. These findings provide new insights into the pathophysiology of AD.