Abnormal neural network connectivity in heart failure with reduced and mild-ranged ejection fraction patients: an independent component and dynamic functional network connectivity analysis

Figures

• 

  Figure 1. Resting-state fMRI data were preprocessed and submitted to group ICA to extract resting-state networks (RSNs). Static network connectivity was then assessed to characterize internetwork functional alterations in HF patients. For dynamic FNC (dFNC), a sliding-window approach was applied; the resulting time-resolved connectivity matrices were clustered using k-means to derive recurring connectivity states. For each state, temporal characteristics such as mean dwell time and occurrence rate were computed and further correlated with clinical measures. (a) Resting-state fMRI data were preprocessed and submitted to group ICA to extract resting-state networks (RSNs). (b) Static network connectivity was then assessed to characterize internetwork functional alterations in HF patients. (c) For dynamic FNC (dFNC), a sliding-window approach was applied; (d) The resulting time-resolved connectivity matrices were clustered using k-means to derive recurring connectivity states. For each state, (e) Temporal characteristics such as mean dwell time and occurrence rate were computed and further correlated with clinical measures.

• 

  Figure 2. Spatial maps of the 12 selected resting-state networks, assigned to 9 functional domains.

• 

  Figure 3. Results of ICA group comparison. Brain regions in HF patients exhibit decreased internal network FCs in the DAN and FPN, while showing significantly increased FC in the SN. The color bars indicate the t values of FCs. ICA, independent component analysis; HF, heart failure; FC, functional connectivity; DAN, dorsal attention network; FPN, frontoparietal network; SN, salience network.

• 

  Figure 4. The dFNC matrices for state 1–5 and the dFNC diagrams for each state. The horizontal and vertical axes are the selected RSNs, and their functional networks are depicted. The color bars indicate the z values of the dFNC. dFNC, dynamic functional network connectivity; RSNs, Resting-state networks. The dFNC matrices for state 1–5 and the dFNC diagrams for each state (A—E).

• 

  Figure 5. Temporal features of dFNC states for the HF and HC groups, and correlations between the temporal features and the clinical scales. (A) Reoccurrence fraction of time, (B) dwell time and (C) numbers of transitions were plotted using violin plots. Horizontal lines indicate group medians and interquartile ranges. The width of the violin plot bars indicates the distribution density of subjects in each group at the corresponding ordinate level. The horizontal lines above some violins indicate pairwise comparisons that demonstrated statistical significance: * p < 0.05. Cardiac function indicators. (D,E) A significantly positive relationship was observed between the mean dwell time in State 5 and the MMSE score (r = 0.360, p = 0.0037, Bonferrioni correction) and LVEF (r = 0.358, p = 0.004, Bonferrioni correction). (F,G) A significantly positive relationship was observed between the mean dwell time in State 2 and the HAMD score (r = 0.3264, p = 0.009, Bonferrioni correction) and HAMA (r = 0.467, p = 0.0001, Bonferrioni correction). dFNC, dynamic functional network connectivity; HF, heart failure; HC, healthy control; MMSE, Mini-Mental State Examination; LVEF, left ventricle ejection…