Format

Send to

Choose Destination
Neuroimage. 2018 Nov 24;186:637-646. doi: 10.1016/j.neuroimage.2018.11.043. [Epub ahead of print]

Detrended connectometry analysis to assess white matter correlates of performance in childhood.

Author information

1
Department of Psychology, University of Cincinnati, Cincinnati, OH, USA; Pediatric Neuroimaging Research Consortium (PNRC), Cincinnati Children's Hospital Medical Center, Cincinnati, OH, USA.
2
Pediatric Neuroimaging Research Consortium (PNRC), Cincinnati Children's Hospital Medical Center, Cincinnati, OH, USA; Division of Biostatistics and Epidemiology, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, USA; College of Medicine, Department of Pediatrics, University of Cincinnati, Cincinnati, OH, USA.
3
Pediatric Neuroimaging Research Consortium (PNRC), Cincinnati Children's Hospital Medical Center, Cincinnati, OH, USA; College of Medicine, Department of Pediatrics, University of Cincinnati, Cincinnati, OH, USA; Division of Neurology, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, USA. Electronic address: darren.kadis@cchmc.org.

Abstract

The white matter of the brain develops in a robust, regionally-variant, nonlinear manner during childhood. To relate white matter connectivity to performance, these regional nonlinear effects of age must be accounted for. Here, we identify white matter correlates of gross intellectual functioning using cutting-edge diffusion analyses inside a data-driven two-step regression framework. A total of 98 participants, ages 3-18 years, were included in the analyses. First, white matter connectivity was modeled as a function of age for each fiber direction at each voxel, extracted from the spin distribution function, using a 6th-order B-spline. The smoothing parameter for each direction was chosen by minimizing generalized cross-validation (GCV), which prevents overfitting while remaining sensitive to potentially nonlinear effects of age. In the second step, the resulting Gaussian residuals were modeled as a function of either full-scale IQ (FSIQ), or of verbal IQ (VIQ) and performance IQ (PIQ), using a linear regression framework (connectometry). Graph theoretical analyses were also performed to assess how each predictor relates to global topological changes, including average clustering coefficient, characteristic path length, global efficiency, average local efficiency, and small worldness. Analyses revealed widespread positive associations between white matter connectivity and FSIQ, including regions of the corpus callosum, fornix, and corticothalamic tracts (FDRq < .05). A separate regression model revealed a selective positive relationship between VIQ and white matter connectivity in predominately frontal tracts (e.g., anterior corticothalamic radiations, fornix, anterior corpus callosum, frontopontine tracts); in contrast, PIQ predicted white matter connectivity in the posterior brain (e.g. parietopontine tracts, posterior corticothalamic radiations, posterior corticostriatal projections), (FDRq < .05). No negative correlations were observed. Graph analyses revealed FSIQ, VIQ while controlling for PIQ, and PIQ while controlling for VIQ increase clustering coefficient, global efficiency, local efficiency, and small worldness, and decrease characteristic path length of the network. Results indicate regional white matter changes related to cognitive skills in childhood, independent of age.

Supplemental Content

Full text links

Icon for Elsevier Science
Loading ...
Support Center