Development of White Matter Fiber Covariance Networks Supports Executive Function in Youth

Abstract

The white matter architecture of the human brain undergoes substantial development throughout childhood and adolescence, allowing for more efficient signaling between brain regions that support executive function. Increasingly, the field understands grey matter development as a spatially and temporally coordinated mechanism that follows hierarchically organized gradients of change. While white matter development also appears asynchronous, previous studies have largely relied on anatomical atlases to characterize white matter tracts, precluding a direct assessment of how white matter structure is spatially and temporally coordinated. Here, we leveraged advances in diffusion modeling and unsupervised machine learning to delineate white matter fiber covariance networks comprised of structurally similar areas of white matter in a cross-sectional sample of 939 youth aged 8-22 years. We then evaluated associations between fiber covariance network structural properties with both age and executive function using generalized additive models. The identified fiber covariance networks aligned with the known architecture of white matter while simultaneously capturing novel spatial patterns of coordinated maturation. Fiber covariance networks showed heterochronous increases in fiber density and cross section that generally followed hierarchically organized temporal patterns of cortical development, with the greatest increases in unimodal sensorimotor networks and the most prolonged increases in superior and anterior transmodal networks. Notably, we found that executive function was associated with structural features of limbic and association networks. Taken together, this study delineates data-driven patterns of white matter network development that support cognition and align with major axes of brain maturation.

Dynamics of White Matter Architecture in Lexical Production among Middle-Aged Adults

“…From a methodological standpoint, our results advocate for a network-level approach to complement anatomically defined bundle atlases (Bullock et al, 2019; Bullock et al, 2022; Catani & Thiebaut De Schotten, 2008; Mori et al, 2008; Rojkova et al, 2016; Wasserthal et al, 2018), leading to novel associations between brain pathways free from explicit spatial constraints (Bagautdinova et al, 2023). Ultimately, this emphasizes the need to apply advanced data-driven tools (i.e., dimensionality reduction via NMF; Sotiras et al, 2015) to capture the complexity of WM changes.…”

“…That is, each network comprised of a unique set of structurally covarying voxels, indicating that the microstructural integrity changes (TW-FA) along these voxels covary across participants (i.e., either strongly correlated or anti-correlated). Given its recent application in investigating age-related (Bagautdinova et al, 2023; Lynch et al, 2023; Sotiras et al, 2015) and biological processes captured by microstructural changes (Patel et al, 2023; Robert et al, 2022), and given that TW-FA values are non-negative, we considered NMF as an appropriate approach for our research question.…”

“…NMF decomposes the matrix containing all TW-FA images ( A ) into two matrices: voxel matrix W ( voxels × k ) and subject matrix H ( k × participants ). The optimal number of networks k was determined using the procedure described by Bagautdinova et al (2023): We performed NMF 10 times, incrementing the number of networks at each iteration (i) from k=2 two k=20 by steps of 2. The reconstruction error ( RE ), calculated as the Frobenius norm of the difference between the input and reconstructed matrix (Eq.2), guides the selection of the optimal k .…”

“…NMF decomposes the matrix containing all TW-FA images (A) into two matrices: voxel matrix W (voxels × k) and subject matrix H (k × participants). The optimal number of networks k was determined using the procedure described by Bagautdinova et al (2023): We performed NMF 10 times, incrementing the number of networks at each iteration (i) from k=2 two k=20 by steps of 2. The reconstruction error (RE), calculated as the Frobenius norm of the difference between the input and reconstructed matrix (Eq.2), guides the selection of the optimal k. The gradient of this error (GradientRE), representing the difference between reconstruction errors of two successive iterations (Eq.3), helped identify the point after which decomposing A into more networks yielded diminishing returns on model fit (Figure A.4…”

Lateral frontoparietal functional connectivity based on individual sulcal morphology

“…In particular, the present findings beg the question of which LPFC and LPC pTS, if any, are directly connected via long-range white matter tracts. For example, the superior longitudinal fasciculus (SLF) is among the later-developing white matter bundles (Liang et al, 2022; Bagautdinova et al, 2023; Di Carlo et al, 2023). In particular, the middle branch of the SLF, connecting regions of the intraparietal sulcus to regions of the superior and middle frontal gyrus – i.e.…”