Figure 2. 10 structural components derived from the NMF decomposition.

A) Cortical mappings of each of the spatial patterns for each of the 10 components. For each component, lateral and medial views of both left and right hemispheres are shown. Components are identified using the putative descriptors from the main text (e.g., Fronto-temporal) as well as lettering at the centre of each set of surface views (e.g., C1). Red areas indicate vertices loading heavily onto a particular component (thresholded at 25% to max value). Each component identifies a selection of vertices sharing a structural variance pattern. Together, components cover the entire brain, are largely bilateral (with exception of component 9) and are not spatially overlapping. B) Subject-specific weightings associated with each of the displayed 10 components. Each row corresponds to a specific component’s NMF weightings for each subject-metric combination, describing the CT, SA, MD, FA and RD patterns of each subject in each component. Together these two outputs describe the morphological patterns of each subject within each spatial component. Each element of the matrix is displayed as a fraction of its row mean, such that values below 0 indicate a below average weight for a given component and vice versa.