In general one can distinguishes between the mapping of the connectome over multiple levels such as micro- and macroscopic scales, and the multi-scale analysis based on varying the number of regions defined over grey matter. In most studies, networks are defined over a single scale (i.e. a predefined number of regions), by using for example an atlas template. However, the resulting networks are strongly dependent on the number of nodes/regions over which they are defined, which makes comparisons across studies difficult. In particular, it has been shown that network measures vary according to the networks scale, which poses challenges for comparisons.

Region dependence of network measure

The application of nodal multi-scale approaches has become of interest. Cammoun and colleagues introduce a multi-scale framework using an atlas based segmentation of the grey matter with 66 regions and subdivided these regions further into four additional sets of about 125, 250, 500 and 1000 regions with approximately equal region size. By doing so they aimed to define representative connectivity matrices over multiple scales. Additional studies have compared networks defined over multiple scales in both structural and functional data in adults and the developing brain. These studies mainly investigate whether the conclusions based on network measures are stable across scales and emphasise that a comparison of results across studies should be made with reference to scale.

The right number of regions and their location is still undetermined in neonates (and arguably also for adults). Until the "correct" number of regions has been determined, an analysis over multiple scales is reasonable in order to avoid biases based on the position and number of nodes.