Limits...
Graph theoretical analysis of developmental patterns of the white matter network.

Chen Z, Liu M, Gross DW, Beaulieu C - Front Hum Neurosci (2013)

Bottom Line: During late childhood period, the structural brain network showed significant increase in the global efficiency but decrease in modularity, suggesting a shift of topological organization toward a more randomized configuration.However, while preserving most topological features, there was a significant increase in the local efficiency at adolescence, suggesting the dynamic process of rewiring and rebalancing brain connections at different growth stages.Finally, a stable and functionally/anatomically related modular organization was identified throughout the development of the WM network.

View Article: PubMed Central - PubMed

Affiliation: Department of Biomedical Engineering, Faculty of Medicine and Dentistry, University of Alberta Edmonton, AB, Canada.

ABSTRACT
Understanding the development of human brain organization is critical for gaining insight into how the enhancement of cognitive processes is related to the fine-tuning of the brain network. However, the developmental trajectory of the large-scale white matter (WM) network is not fully understood. Here, using graph theory, we examine developmental changes in the organization of WM networks in 180 typically-developing participants. WM networks were constructed using whole brain tractography and 78 cortical regions of interest were extracted from each participant. The subjects were first divided into 5 equal sample size (n = 36) groups (early childhood: 6.0-9.7 years; late childhood: 9.8-12.7 years; adolescence: 12.9-17.5 years; young adult: 17.6-21.8 years; adult: 21.9-29.6 years). Most prominent changes in the topological properties of developing brain networks occur at late childhood and adolescence. During late childhood period, the structural brain network showed significant increase in the global efficiency but decrease in modularity, suggesting a shift of topological organization toward a more randomized configuration. However, while preserving most topological features, there was a significant increase in the local efficiency at adolescence, suggesting the dynamic process of rewiring and rebalancing brain connections at different growth stages. In addition, several pivotal hubs were identified that are vital for the global coordination of information flow over the whole brain network across all age groups. Significant increases of nodal efficiency were present in several regions such as precuneus at late childhood. Finally, a stable and functionally/anatomically related modular organization was identified throughout the development of the WM network. This study used network analysis to elucidate the topological changes in brain maturation, paving the way for developing novel methods for analyzing disrupted brain connectivity in neurodevelopmental disorders.

No MeSH data available.


Related in: MedlinePlus

Flowchart for the construction of the DTI white matter (WM) network for each subject. The T1-weighted image of each subject (B) was first coregistered into DTI native space (A) using rigid transformation to the b0 image (not shown). The resultant T1 image was then non-linearly registered to the ICBM 152 template (C) in the MNI space to obtain transformation matrix T. The AAL template (E) was then inversely warped back to the individual DTI space (F) using the inverse transformation (T−1). Whole brain white matter fibers were reconstructed using a deterministic tractography method in native DTI space (D). The WM fibers connecting any pair of regions were located and the edge weight between the two regions was calculated from the FA, fiber number (FN) and average volume of the two cortical regions. (G) A sample white matter network for one subject.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3814848&req=5

Figure 1: Flowchart for the construction of the DTI white matter (WM) network for each subject. The T1-weighted image of each subject (B) was first coregistered into DTI native space (A) using rigid transformation to the b0 image (not shown). The resultant T1 image was then non-linearly registered to the ICBM 152 template (C) in the MNI space to obtain transformation matrix T. The AAL template (E) was then inversely warped back to the individual DTI space (F) using the inverse transformation (T−1). Whole brain white matter fibers were reconstructed using a deterministic tractography method in native DTI space (D). The WM fibers connecting any pair of regions were located and the edge weight between the two regions was calculated from the FA, fiber number (FN) and average volume of the two cortical regions. (G) A sample white matter network for one subject.

Mentions: Image preprocessing steps including motion and eddy current corrections were performed using FSL 5.0 for all DTI images (http://fsl.fmrib.ox.ac.uk/fsl/fslwiki). The T1-weighted (MPRAGE) image of each subject was first linearly coregistered (Figures 1A,B) to its corresponding b0 image. Each transformed T1 image was then non-linearly registered to a pre-segmented and validated volumetric template, the automated anatomical labeling (AAL) atlas (Tzourio-Mazoyer et al., 2002) as shown in Figures 1B,C. This parcellation divided the cortical surface into 78 regions (39 per hemisphere). See Table 2 for the name of the regions and their corresponding abbreviations. The resulting inverse deformation map (T−1) for each subject was then applied to warp the AAL template to the DTI native space of each subject using nearest neighbor interpolation method (Figures 1E,F) as each AAL region was defined as a brain network node. Whole brain WM tractography was performed using a brute-force streamline-tracking method (Basser et al., 2000) with a FA threshold of 0.2 and primary eigenvector turning angle of 45 degrees (Figures 1A,D). To limit false positive connections, two cortical regions were deemed connected if at least 10 connecting fibers with two end points were located between them; the same threshold was also applied in a recent brain network study (van den Heuvel et al., 2012). The effects of different node-connecting fiber number (FN) thresholds ranging from 3 to 10 were determined for our network analysis. We quantified the weight of each valid connection between two cortical regions (i and j) as the product of the connecting FN and mean FA of the connecting fiber, normalized by dividing the average volume of the two connecting regions to counteract the bias where larger cortical regions inherently project/receive more “virtual” fibers (wij = FN*FA/Volume). Several diffusion brain network studies have applied this weighting function (Lo et al., 2010; Brown et al., 2011). As a result, the structural brain network of each participant was represented by a symmetric 78 × 78 matrix (Figure 1G).


Graph theoretical analysis of developmental patterns of the white matter network.

Chen Z, Liu M, Gross DW, Beaulieu C - Front Hum Neurosci (2013)

Flowchart for the construction of the DTI white matter (WM) network for each subject. The T1-weighted image of each subject (B) was first coregistered into DTI native space (A) using rigid transformation to the b0 image (not shown). The resultant T1 image was then non-linearly registered to the ICBM 152 template (C) in the MNI space to obtain transformation matrix T. The AAL template (E) was then inversely warped back to the individual DTI space (F) using the inverse transformation (T−1). Whole brain white matter fibers were reconstructed using a deterministic tractography method in native DTI space (D). The WM fibers connecting any pair of regions were located and the edge weight between the two regions was calculated from the FA, fiber number (FN) and average volume of the two cortical regions. (G) A sample white matter network for one subject.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3814848&req=5

Figure 1: Flowchart for the construction of the DTI white matter (WM) network for each subject. The T1-weighted image of each subject (B) was first coregistered into DTI native space (A) using rigid transformation to the b0 image (not shown). The resultant T1 image was then non-linearly registered to the ICBM 152 template (C) in the MNI space to obtain transformation matrix T. The AAL template (E) was then inversely warped back to the individual DTI space (F) using the inverse transformation (T−1). Whole brain white matter fibers were reconstructed using a deterministic tractography method in native DTI space (D). The WM fibers connecting any pair of regions were located and the edge weight between the two regions was calculated from the FA, fiber number (FN) and average volume of the two cortical regions. (G) A sample white matter network for one subject.
Mentions: Image preprocessing steps including motion and eddy current corrections were performed using FSL 5.0 for all DTI images (http://fsl.fmrib.ox.ac.uk/fsl/fslwiki). The T1-weighted (MPRAGE) image of each subject was first linearly coregistered (Figures 1A,B) to its corresponding b0 image. Each transformed T1 image was then non-linearly registered to a pre-segmented and validated volumetric template, the automated anatomical labeling (AAL) atlas (Tzourio-Mazoyer et al., 2002) as shown in Figures 1B,C. This parcellation divided the cortical surface into 78 regions (39 per hemisphere). See Table 2 for the name of the regions and their corresponding abbreviations. The resulting inverse deformation map (T−1) for each subject was then applied to warp the AAL template to the DTI native space of each subject using nearest neighbor interpolation method (Figures 1E,F) as each AAL region was defined as a brain network node. Whole brain WM tractography was performed using a brute-force streamline-tracking method (Basser et al., 2000) with a FA threshold of 0.2 and primary eigenvector turning angle of 45 degrees (Figures 1A,D). To limit false positive connections, two cortical regions were deemed connected if at least 10 connecting fibers with two end points were located between them; the same threshold was also applied in a recent brain network study (van den Heuvel et al., 2012). The effects of different node-connecting fiber number (FN) thresholds ranging from 3 to 10 were determined for our network analysis. We quantified the weight of each valid connection between two cortical regions (i and j) as the product of the connecting FN and mean FA of the connecting fiber, normalized by dividing the average volume of the two connecting regions to counteract the bias where larger cortical regions inherently project/receive more “virtual” fibers (wij = FN*FA/Volume). Several diffusion brain network studies have applied this weighting function (Lo et al., 2010; Brown et al., 2011). As a result, the structural brain network of each participant was represented by a symmetric 78 × 78 matrix (Figure 1G).

Bottom Line: During late childhood period, the structural brain network showed significant increase in the global efficiency but decrease in modularity, suggesting a shift of topological organization toward a more randomized configuration.However, while preserving most topological features, there was a significant increase in the local efficiency at adolescence, suggesting the dynamic process of rewiring and rebalancing brain connections at different growth stages.Finally, a stable and functionally/anatomically related modular organization was identified throughout the development of the WM network.

View Article: PubMed Central - PubMed

Affiliation: Department of Biomedical Engineering, Faculty of Medicine and Dentistry, University of Alberta Edmonton, AB, Canada.

ABSTRACT
Understanding the development of human brain organization is critical for gaining insight into how the enhancement of cognitive processes is related to the fine-tuning of the brain network. However, the developmental trajectory of the large-scale white matter (WM) network is not fully understood. Here, using graph theory, we examine developmental changes in the organization of WM networks in 180 typically-developing participants. WM networks were constructed using whole brain tractography and 78 cortical regions of interest were extracted from each participant. The subjects were first divided into 5 equal sample size (n = 36) groups (early childhood: 6.0-9.7 years; late childhood: 9.8-12.7 years; adolescence: 12.9-17.5 years; young adult: 17.6-21.8 years; adult: 21.9-29.6 years). Most prominent changes in the topological properties of developing brain networks occur at late childhood and adolescence. During late childhood period, the structural brain network showed significant increase in the global efficiency but decrease in modularity, suggesting a shift of topological organization toward a more randomized configuration. However, while preserving most topological features, there was a significant increase in the local efficiency at adolescence, suggesting the dynamic process of rewiring and rebalancing brain connections at different growth stages. In addition, several pivotal hubs were identified that are vital for the global coordination of information flow over the whole brain network across all age groups. Significant increases of nodal efficiency were present in several regions such as precuneus at late childhood. Finally, a stable and functionally/anatomically related modular organization was identified throughout the development of the WM network. This study used network analysis to elucidate the topological changes in brain maturation, paving the way for developing novel methods for analyzing disrupted brain connectivity in neurodevelopmental disorders.

No MeSH data available.


Related in: MedlinePlus