Skip to main content
. 2016 Nov 15;142:14–26. doi: 10.1016/j.neuroimage.2015.04.052

Fig. 5.

Fig. 5

Graphs constructed by probabilistic thresholding and correlation thresholding.

(A) Left panel shows the correlation between framewise displacement (see Section 2.6 of the Methods) and voxel-wise time series for an fMRI dataset acquired with high levels of head movement. Dark blue areas show high artifact regions. Right panel shows parcellated df maps for the subject after time series denoising by wavelet despiking.

(B) Results of a series of permutations tests designed to evaluate our method for estimating nodal df. Observed vs. expected Type I error rates for each node are shown in dark blue for the wavelet-based df estimator, and in light blue for the equivalent analysis assuming nominal df = N.

(C) The top 0.5% of edges ranked by P value (shown in green) vs. the top 0.5% of edges ranked by Pearson correlation r (shown in blue) plotted between their corresponding coordinates in the brain. The node size is proportional to node degree at this connection density.