Limits...
Connectomic Insights into Topologically Centralized Network Edges and Relevant Motifs in the Human Brain.

Xia M, Lin Q, Bi Y, He Y - Front Hum Neurosci (2016)

Bottom Line: We found that the pivotal WM connections with highly topological-edge centrality were primarily distributed in several long-range cortico-cortical connections (including the corpus callosum, cingulum and inferior fronto-occipital fasciculus) and some projection tracts linking subcortical regions.Computational simulation models indicated the sharp decrease of global network integrity when attacking these highly centralized edges.Together, our results demonstrated high building-cost consumption and substantial communication capacity contributions for pivotal WM connections, which deepens our understanding of the topological mechanisms that govern the organization of human connectomes.

View Article: PubMed Central - PubMed

Affiliation: State Key Laboratory of Cognitive Neuroscience and Learning and IDG/McGovern Institute for Brain Research, Beijing Normal University Beijing, China.

ABSTRACT
White matter (WM) tracts serve as important material substrates for information transfer across brain regions. However, the topological roles of WM tracts in global brain communications and their underlying microstructural basis remain poorly understood. Here, we employed diffusion magnetic resonance imaging and graph-theoretical approaches to identify the pivotal WM connections in human whole-brain networks and further investigated their wiring substrates (including WM microstructural organization and physical consumption) and topological contributions to the brain's network backbone. We found that the pivotal WM connections with highly topological-edge centrality were primarily distributed in several long-range cortico-cortical connections (including the corpus callosum, cingulum and inferior fronto-occipital fasciculus) and some projection tracts linking subcortical regions. These pivotal WM connections exhibited high levels of microstructural organization indicated by diffusion measures (the fractional anisotropy, the mean diffusivity and the axial diffusivity) and greater physical consumption indicated by streamline lengths, and contributed significantly to the brain's hubs and the rich-club structure. Network motif analysis further revealed their heavy participations in the organization of communication blocks, especially in routes involving inter-hemispheric heterotopic and extremely remote intra-hemispheric systems. Computational simulation models indicated the sharp decrease of global network integrity when attacking these highly centralized edges. Together, our results demonstrated high building-cost consumption and substantial communication capacity contributions for pivotal WM connections, which deepens our understanding of the topological mechanisms that govern the organization of human connectomes.

No MeSH data available.


Related in: MedlinePlus

Pivotal edges and the rich-club structure. (A) The normalized rich-club coefficient Φnorm(k) of the group-level WM network was above 1 for a range of k from 9 to 16. The peak at k = 14 was selected as the hub threshold for further analysis. (B) The network hubs were mainly located in the medial line of the brain and the connections of the brain network can be further classified into three categories: rich-club (red), feeder (yellow) and local (blue) connections. (C) The edge betweenness centrality values were significantly different among rich-club, feeder, and local connections. (D) The pivotal edges had significantly different building contribution (indicated by the proportion of number) to three categories of connections. (E) The communication contribution (indicated by the proportion of edge betweenness centrality), of the pivotal edges was also significantly different among three categories of connections. The center pie illustrates the building/communication percentage of the three types of connections, and the surrounding pies show the building/communication percentage in each category of the connections. SFGdor, superior frontal gyrus, dorsolateral; CAL, Calcarine fissure and surrounding cortex; L, left; R, right. For other abbreviations, see Figure 2.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4835491&req=5

Figure 4: Pivotal edges and the rich-club structure. (A) The normalized rich-club coefficient Φnorm(k) of the group-level WM network was above 1 for a range of k from 9 to 16. The peak at k = 14 was selected as the hub threshold for further analysis. (B) The network hubs were mainly located in the medial line of the brain and the connections of the brain network can be further classified into three categories: rich-club (red), feeder (yellow) and local (blue) connections. (C) The edge betweenness centrality values were significantly different among rich-club, feeder, and local connections. (D) The pivotal edges had significantly different building contribution (indicated by the proportion of number) to three categories of connections. (E) The communication contribution (indicated by the proportion of edge betweenness centrality), of the pivotal edges was also significantly different among three categories of connections. The center pie illustrates the building/communication percentage of the three types of connections, and the surrounding pies show the building/communication percentage in each category of the connections. SFGdor, superior frontal gyrus, dorsolateral; CAL, Calcarine fissure and surrounding cortex; L, left; R, right. For other abbreviations, see Figure 2.

Mentions: We examined the contribution of pivotal WM edges to the rich-club architecture of the WM network. Figure 4A illustrates the curve of the normalized rich-club coefficient, Φnorm(k), over a range of nodal degree, k. The Φnorm(k) were larger than 1 at k > 8, indicating the existence of a rich-club structure in the WM network (van den Heuvel and Sporns, 2011). Here, we chose the peak Φnorm(k), where the nodes with k > 14 were considered the brain hubs (see Table S2 for results of other thresholds). We identified 11 network hubs that were primarily located in the bilateral precuneus, the bilateral orbital part of superior frontal gyrus, the right dorsolateral superior frontal gyrus, the bilateral calcarine sulcus, the left middle occipital gyrus, the right superior occipital gyrus and the bilateral putamen (Figure 4B). Most of these hubs (72.7%) were structurally connected with the pivotal edges. Further, we divided the whole-brain WM connections into three categories according to the types of nodes they linked (van den Heuvel et al., 2012): rich-club connections between rich-club nodes (n = 21), feeder connections between rich-club and non-rich-club nodes (n = 142) and local connections between two non-rich-club nodes (n = 268). Significant differences in EBC were observed among these three edge categories [F(2, 428) = 44.9, p = 2.0 × 10−18], with a descending order of the rich-club, feeder and local connections (post hoc comparisons, permutation tests, all ps < 0.001) (Figure 4C). Importantly, the proportions of the number of pivotal WM edges among the three categories, which represents the network building contribution, were significantly different [, p = 1.1 × 10−16]: 57.1% (12/21) within the rich-club connections, 19.7% (28/142) within the feeder connections and 3.0% (8/268) within the local connections (Figure 4D). The proportion of EBC of the pivotal edges among the three categories, which represents the network communication contribution, was also significantly different [ = 2490.3, p < 1.0 × 10−64]: 83.8% within the rich-club connections, 41.4% within the feeder connections and 11.3% within the local connections (Figure 4E). These results suggest that the rich-club architecture of the brain networks was topologically supported by the pivotal WM edges.


Connectomic Insights into Topologically Centralized Network Edges and Relevant Motifs in the Human Brain.

Xia M, Lin Q, Bi Y, He Y - Front Hum Neurosci (2016)

Pivotal edges and the rich-club structure. (A) The normalized rich-club coefficient Φnorm(k) of the group-level WM network was above 1 for a range of k from 9 to 16. The peak at k = 14 was selected as the hub threshold for further analysis. (B) The network hubs were mainly located in the medial line of the brain and the connections of the brain network can be further classified into three categories: rich-club (red), feeder (yellow) and local (blue) connections. (C) The edge betweenness centrality values were significantly different among rich-club, feeder, and local connections. (D) The pivotal edges had significantly different building contribution (indicated by the proportion of number) to three categories of connections. (E) The communication contribution (indicated by the proportion of edge betweenness centrality), of the pivotal edges was also significantly different among three categories of connections. The center pie illustrates the building/communication percentage of the three types of connections, and the surrounding pies show the building/communication percentage in each category of the connections. SFGdor, superior frontal gyrus, dorsolateral; CAL, Calcarine fissure and surrounding cortex; L, left; R, right. For other abbreviations, see Figure 2.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4835491&req=5

Figure 4: Pivotal edges and the rich-club structure. (A) The normalized rich-club coefficient Φnorm(k) of the group-level WM network was above 1 for a range of k from 9 to 16. The peak at k = 14 was selected as the hub threshold for further analysis. (B) The network hubs were mainly located in the medial line of the brain and the connections of the brain network can be further classified into three categories: rich-club (red), feeder (yellow) and local (blue) connections. (C) The edge betweenness centrality values were significantly different among rich-club, feeder, and local connections. (D) The pivotal edges had significantly different building contribution (indicated by the proportion of number) to three categories of connections. (E) The communication contribution (indicated by the proportion of edge betweenness centrality), of the pivotal edges was also significantly different among three categories of connections. The center pie illustrates the building/communication percentage of the three types of connections, and the surrounding pies show the building/communication percentage in each category of the connections. SFGdor, superior frontal gyrus, dorsolateral; CAL, Calcarine fissure and surrounding cortex; L, left; R, right. For other abbreviations, see Figure 2.
Mentions: We examined the contribution of pivotal WM edges to the rich-club architecture of the WM network. Figure 4A illustrates the curve of the normalized rich-club coefficient, Φnorm(k), over a range of nodal degree, k. The Φnorm(k) were larger than 1 at k > 8, indicating the existence of a rich-club structure in the WM network (van den Heuvel and Sporns, 2011). Here, we chose the peak Φnorm(k), where the nodes with k > 14 were considered the brain hubs (see Table S2 for results of other thresholds). We identified 11 network hubs that were primarily located in the bilateral precuneus, the bilateral orbital part of superior frontal gyrus, the right dorsolateral superior frontal gyrus, the bilateral calcarine sulcus, the left middle occipital gyrus, the right superior occipital gyrus and the bilateral putamen (Figure 4B). Most of these hubs (72.7%) were structurally connected with the pivotal edges. Further, we divided the whole-brain WM connections into three categories according to the types of nodes they linked (van den Heuvel et al., 2012): rich-club connections between rich-club nodes (n = 21), feeder connections between rich-club and non-rich-club nodes (n = 142) and local connections between two non-rich-club nodes (n = 268). Significant differences in EBC were observed among these three edge categories [F(2, 428) = 44.9, p = 2.0 × 10−18], with a descending order of the rich-club, feeder and local connections (post hoc comparisons, permutation tests, all ps < 0.001) (Figure 4C). Importantly, the proportions of the number of pivotal WM edges among the three categories, which represents the network building contribution, were significantly different [, p = 1.1 × 10−16]: 57.1% (12/21) within the rich-club connections, 19.7% (28/142) within the feeder connections and 3.0% (8/268) within the local connections (Figure 4D). The proportion of EBC of the pivotal edges among the three categories, which represents the network communication contribution, was also significantly different [ = 2490.3, p < 1.0 × 10−64]: 83.8% within the rich-club connections, 41.4% within the feeder connections and 11.3% within the local connections (Figure 4E). These results suggest that the rich-club architecture of the brain networks was topologically supported by the pivotal WM edges.

Bottom Line: We found that the pivotal WM connections with highly topological-edge centrality were primarily distributed in several long-range cortico-cortical connections (including the corpus callosum, cingulum and inferior fronto-occipital fasciculus) and some projection tracts linking subcortical regions.Computational simulation models indicated the sharp decrease of global network integrity when attacking these highly centralized edges.Together, our results demonstrated high building-cost consumption and substantial communication capacity contributions for pivotal WM connections, which deepens our understanding of the topological mechanisms that govern the organization of human connectomes.

View Article: PubMed Central - PubMed

Affiliation: State Key Laboratory of Cognitive Neuroscience and Learning and IDG/McGovern Institute for Brain Research, Beijing Normal University Beijing, China.

ABSTRACT
White matter (WM) tracts serve as important material substrates for information transfer across brain regions. However, the topological roles of WM tracts in global brain communications and their underlying microstructural basis remain poorly understood. Here, we employed diffusion magnetic resonance imaging and graph-theoretical approaches to identify the pivotal WM connections in human whole-brain networks and further investigated their wiring substrates (including WM microstructural organization and physical consumption) and topological contributions to the brain's network backbone. We found that the pivotal WM connections with highly topological-edge centrality were primarily distributed in several long-range cortico-cortical connections (including the corpus callosum, cingulum and inferior fronto-occipital fasciculus) and some projection tracts linking subcortical regions. These pivotal WM connections exhibited high levels of microstructural organization indicated by diffusion measures (the fractional anisotropy, the mean diffusivity and the axial diffusivity) and greater physical consumption indicated by streamline lengths, and contributed significantly to the brain's hubs and the rich-club structure. Network motif analysis further revealed their heavy participations in the organization of communication blocks, especially in routes involving inter-hemispheric heterotopic and extremely remote intra-hemispheric systems. Computational simulation models indicated the sharp decrease of global network integrity when attacking these highly centralized edges. Together, our results demonstrated high building-cost consumption and substantial communication capacity contributions for pivotal WM connections, which deepens our understanding of the topological mechanisms that govern the organization of human connectomes.

No MeSH data available.


Related in: MedlinePlus