These studies highlight a fundamental issue in MRI graph theoretic studies, which is the extent to which an abstract network is constructed to reflect the functional network structure of the human brain. In many systems, nodes are obvious, such as stations along subway lines, or people who have coauthored papers. Nodes are, however, not obvious in the functional network of the brain. Presumably, when studying the functional network architecture of the brain, nodes should correspond to some unit of functional organization, such as neurons, columns, or functional areas. Unfortunately, the number, locations and extents of such functional units in the human brain are poorly defined at this point, and researchers therefore typically form nodes from voxels (Buckner et al., 2009; Eguiluz et al., 2005; van den Heuvel et al., 2008), pre-defined anatomical brain parcellation schemes (Achard et al., 2006; He et al., 2009; Meunier et al., 2009; Salvador et al., 2005), or pre-defined ROIs, often obtained from fMRI or fcMRI studies (Church et al., 2009; Dosenbach et al., 2007; Nelson et al., 2010). If the brain is a collection of
