Altered dynamic effective connectivity of the default mode network in type 2 diabetes

Figures

• 

  Figure 1. Flow chart of analysis in the study. fMRI, functional magnetic resonance imaging. (1) The DM components acquired from independent components analysis. (2) Sliding windows method with the rectangle window size of 18 TRs and step length of 2 TRs was used to obtain causal impact matrices. (3) K-means clustering categorized all matrices to different clusters. (4) Measuring the difference of dynamic effective connectivity, casual information flow and temporal properties between T2DM and HCs. DM components, default mode components.

• 

  Figure 2. The distribution of DM components. DM components, default mode components.

• 

  Figure 3. (A) The averaged static effective connectivity (EC) between DM component pairs was computed using an entire scan. (B) Decreased dynamic EC from IC20 to IC51 and IC99. (C) Centroid matrices for two states and the occurrences and the percentage of the two states. (D) Differences of mean effective connectivity between two states and between two groups in each state. **p < 0.01 and ****p < 0.0001.

• 

  Figure 4. Causal influence flows of dynamic effective connectivity. The in-weighted degree (A), out-weighted degree (B), and in-out weighted degree (C) in two states for T2DM patients (red) and HCs (blue). Horizontal solid and dashed lines indicate group means and interquartile range respectively. **p < 0.01 and *p < 0.05 with FDR corrected.

• 

  Figure 5. (A) The relationship within the casual information flow and clinical and cognitive characteristics. (B) The difference of temporal properties between two groups and the association of temporal properties and clinical index. HbA1c, glycosylated hemoglobin type A1C; MoCA, the Montreal Cognitive Assessment; DBP, diastolic blood pressure; F, fractional windows; MDT, mean dwell time; and NT, number of transitions. ***p < 0.001 and *p < 0.05.