Dynamic Functional Network Connectivity Changes Associated with fMRI Neurofeedback of Right Premotor Cortex

Figures

• 

  Figure 1. The diagram of the fMRI experiment.

• 

  Figure 2. The overview of the framework of DFNC processing. Si represents the raw fMRI data of the ith subject. Yi represents the preprocessed fMRI data of the ith subject. Ri represents the mixing matrix (time courses) and Ai represents the source matrix (independent components) of the ith subject. FCj(i) represents the FC pattern of the jth sliding window of the ith subject.

• 

  Figure 3. The median pattern of 7 FC states in motor execution tasks run clustered by K-means. The total number and percentage of occurrences is listed above each FC pattern.

• 

  Figure 4. Time measurements for the pre-training and post-training ME runs of the two groups. (a) The mean dwell time of state 2, (b) the fraction of time of state 2, and (c) the fraction of time of state 6. Pre_exp represents the pre-training run of the experimental group. Post_exp represents the post-training run of the experimental group. Pre_con represents the pre-training run of the control group. Post_con represents the post-training run of the control group. Error bars represent the standard error (* p < 0.05; ** p < 0.01).

• 

  Figure 5. The mean strength of the sub-networks showing significant group/training differences. (a) The D-D subnetwork in state 2, and (b) the C-D subnetwork in state 6. Error bars represent the standard error (* p < 0.05).

• 

  Figure 6. The relationship between tapping frequency and DFNC for post-training ME run of the experimental group. (a) The significant regression relationship between the tapping frequency and the mean dwell time of state 2, and (b) the significant regression relationship between the tapping frequency and the fraction time of state 2.

• 

  Figure 7. The relationship between the PSC of the neurofeedback runs and DFNC of the post-training run for the experimental group. (a) The significant regression relationship between the mean dwell time of state 2 and PSC of run C, (b) the significant regression relationship between the fraction time of state 2 and PSC of run C, and (c) the significant regression relationship between the mean dwell time of state 2 and PSC of run D.

• 

  Figure 8. The relationship between the PSC of the neurofeedback runs and the subnetwork strength of the post-training run. (a) The significant regression relationship between the PSC of run C and the strength of D-D subnetwork for the experimental group’s state 2. (b) The significant regression relationship between the PSC of run D and the strength of D-D subnetwork for the experimental group’s state 2. (c) The significant regression relationship between the PSC of run C and the strength of the V-C subnetwork for the control group’s state 6. (d) The significant regression relationship between the PSC of run D and the strength of the V-C subnetwork for the control group’s state 6.

• 

  Figure 9. Significant connectivity differences in the target FC states for the experimental group and control group. (a,b) The connectives showing significant differences for the post-training minus the pre-training run in state 2 of the experimental group (a) and in state 6 of the control group (b), (c) the significant regression relationship between the alteration of the tapping frequency and the alteration of the connectivity of connection 1 for the experimental group after neurofeedback training, (d) the significant regression relationship between the alteration of the tapping frequency and the alteration of the connectivity of connection 2 for the experimental group after neurofeedback training, and (e,f) the mean connectivity strength of connection 1 (e) and connection 2 (f) for the pre-training and post-training runs of the two groups. Error bars represent the standard error (* p < 0.05).