Construction and analysis of voxel-level functional connectivity graphs. Starting with the preprocessed fMRI data, all gray matter voxels are defined as graphical nodes (1). Using their associated time series, pairwise internodal functional connectivity is measured in terms of linear correlation. Typically, this is done using Pearson’s r. Alternatively, one can derive binary time series via median-based dichotomization and employ tetrachoric correlation estimation (r
). In both cases, the result is a correlation matrix (2) representing the pairwise functional connectivity between nodes. A binary undirected graph, represented by a binary adjacency matrix (3), is obtained via thresholding. Based on the adjacency matrix, graph-theoretical metrics, such as the node degree k, are computed (4).