Figure 1.
Summary of network reconstruction stages applied to structural and diffusion images for P1 and P2.
The pipeline stages are shown on the left and the alternative implementations of the methods are shown inside the boxes. Arrows indicate the passage of merged (dark arrows) and native (light arrows) atlases through the pipeline stages (red and blue refer to Hammers and Desikan-Killiany atlases, respectively). Nodes were defined by registration of the cortical parcels to diffusion space. Edges were defined by performing tractography from the parcel boundary through the fiber orientations. Note that the whole-brain probabilistic tractography methods differed only in relation to the recommended settings for the software used to track through the fiber orientations. The network construction stage calculated the connecting fiber density between all cortical parcel pairs across the entire cerebral cortex and was identical for both pipelines. Applying these stagesto the merged and native atlases resulted in comparisons between pipelines at three node scales; the merged atlas scale (34 nodes, dark arrows), Hammers atlas scale (44 nodes, light red arrows) and Desikan-Killiany scale (68 nodes, light blue arrows). We also applied the registration and whole-brain tractography pipelines to the AAL atlas (not shown).
Figure 2.
Merging cortical parcels of P2 parcellations.
The native scale P2 parcellation (68 parcels) is shown on the left and the merged P2 parcellation (34 parcels) is shown on the right. The merging pattern was identical for both hemispheres and therefore only the left hemisphere is shown. The colour scheme of brain regions is as in Fig. 2. Lines represent merging of native scale parcels (left) to their equivalent common scale parcels (right). Coloured vertical lines correspond to regions in the temporal (purple), frontal (green), parietal (blue), occipital (red), insula (light-blue) or limbic (yellow) lobes. Native scale P1 parcellations (44 nodes) were merged to the common scale parcellation by merging all temporal lobe parcels.
Figure 3.
Representative cortical parcellations of P1 and P2 at the native and common node scale.
Temporal lobe regions in P1 native scale parcellations (P1-44, far left) were merged, resulting in a lower scale parcellation (P1-34, middle right). Selected regions across the entire cerebral cortex in P2 native scale parcellations (P2-68, middle left) were merged (P2-34, see Fig. 1). This resulted in a common and anatomically equivalent parcellation scale of 34 nodes for both P1 and P2 networks.
Figure 4.
Similarity between P1 and P2 networks across density thresholds using atlases at three node scales.
Significance of similarity was calculated by comparing the distribution of within-subject DC to the expected DC by chance, given the density of the networks. Shown is the mean negative -value of the DC between binary networks thresholded at a given density. The within-subject DC's were bootstrapped to obtain a standard error on the mean (dashed lines). A global peak similarity was found at a density of 0.196, 0.161, 0.106 and 0.142 for the Common (34 nodes), Hammers (44 nodes), Desikan-Killiany (68 nodes) and AAL (78 nodes) atlases, respectively.
Figure 5.
P1 fibers underlying convergent connections in the left hemisphere of a repre-sentative subject.
Fibers are coloured by their network connection. (a) Inter-lobe fibers viewed from the medial aspect. (b) Intra-lobe fibers viewed from the medial aspect. (c) Inter-hemispheric fibers shown from the coronal aspect. The paths of fibers underlying convergent inter-lobe connections agrees with that of major anatomical tracts, such as the ILF (orange) and cingulum (green). Convergent intra-lobular connections were mostly represented by short-range cortical U-fibers. Convergent inter-hemispheric fibers travel via the corpus callosum and connected homotopic cortical regions, such as the superior, middle and inferior frontal gyri (green). For visual clarity, a maximum of 200, 50 and 100 fibers from the subset of whole-brain tractography fibers are shown per connection for (a), (b) and (c), respectively. Also, only fibers greater than 7 cm are shown for (a) and (b) and greater than 10 cm for (c).
Figure 6.
Prevalance of convergent connections across subjects at the peak convergence density.
The prevalance is shown for the common (left), Hammers (middle-left), Desikan-Killiany (right-left) and AAL (right) atlases. Convergent connections were defined as the intersection of subject networks thresholded at the peak convergence density obtained from our bootstrap statistical analysis. The node lobe memberships are indicated by the adjacent colour bars, as in Figure 2. Colours represent the temporal (purple), frontal (green), parietal (blue), occipital (red), insula (turquoise) and cingulate (brown) lobes.
Table 1.
Graph theoretical properties of peak convergent networks.