Figure 2
From: Statistical Modeling of the Default Mode Brain Network Reveals a Segregated Highway Structure
A flowchart describing the means of analyzing functional connectivity data. The left half depicts traditional methods, in which matrices are dichotomized based on some threshold (here shown at a threshold retaining the top 10% of connections), then analyzed by running descriptive statistics on the resulting thresholded network. The cGERGM approach is shown on the right, in which the network is first transformed to the constrained partial correlation space, and then topological features are assessed, in order to recover estimates of the parameters which characterize the generative process for the network under that model. Finally, these parameter estimates can be used to simulate new networks for the purpose of comparison. The images above each correlation matrix are the graphical representation for the thresholded (left) and continuous (right) networks. Note that while the graphical software does not show every edge for the weighted network case, no edges are censored to 0 for analysis purposes.
