Multivariate prediction of white matter BOLD activity and its relationship to marginal connectivity
- 1Vanderbilt University Institute of Imaging Science, Vanderbilt University Medical Center, Nashville, TN, USA; Department of Radiology and Radiological Sciences, Vanderbilt University Medical Center, Nashville, TN, USA; Corresponding author at: Vanderbilt University Institute of Imaging Science, 1161 21st Ave. S, Medical Center North, AA-1105, Nashville, TN, 37232-2310, USA.
- 2Vanderbilt University Institute of Imaging Science, Vanderbilt University Medical Center, Nashville, TN, USA; Department of Electrical and Computer Engineering, Vanderbilt University, Nashville, TN, USA; Department of Biomedical Engineering, Vanderbilt University, Nashville, TN, USA; Department of Computer Science, Vanderbilt University, Nashville, TN, United States
Abstract
Functional MRI studies have traditionally focused on gray matter, whereas white matter BOLD signals have often been treated as weak or artifactual. However, white matter BOLD fluctuations are robustly detectable and contain reproducible functional information, yet most gray-to-white matter analyses rely on only marginal functional connectivity, which cannot separate pairwise coupling from shared variance among distributed cortical systems. Here, we used a multivariate cortical prediction framework to test whether spontaneous white matter BOLD activity can be predicted from distributed cortical gray matter BOLD activity, and how this predictive structure relates to marginal connectivity. Resting-state fMRI data from 81 Human Connectome Project young adult participants were analyzed using a strict white matter mask with no overlap with cortical predictors. For each white matter voxel, time series from 400 cortical parcels were used to predict white matter BOLD signals with nested leave-one-run-out ridge regression. Cortical BOLD activity modestly but reliably predicted white matter BOLD dynamics, demonstrating consistent cross-validated prediction accuracy across a broad spatial extent of the white matter. Ridge β coefficients strongly recapitulated marginal functional connectivity fingerprints, indicating a shared functional backbone, but their first gradients differed reproducibly. The difference between these gradients defined a β-minus-FC G1 divergence axis that remained stable after broadly controlling for anatomical, signal-related, and spatial factors. At the low end of the axis, enriched in callosal and capsular structures, pairwise gray-to-white matter connectivity was relatively strong, but white matter BOLD fluctuations were predicted less well by distributed cortical BOLD activity than that connectivity would suggest. At the high end, enriched in posterior projection and association pathways, pairwise connectivity was relatively weak, but distributed cortical BOLD activity predicted white matter BOLD signals better than expected. These findings identify a reproducible, tract-organized contrast between marginal connectivity and its covariance-adjusted predictive representation.
Related
- Multivariate connectivity deviations reveal task-evoked and clinical network reorganization
- Migraine-related differences in connectivity patterns during emotional audiovisual processing revealed by functional connectivity multivariate pattern analysis.
- Physically aligned forest fire risk prediction: A deep learning framework coupling fuel and climate multivariate factors
- Total phenolic content is associated with antioxidant activity independently of preliminary toxicity in medicinal plants from Sinaloa, Mexico: a multivariate approach
- 4-Diketopyrrole-2-carboxamide derivatives affecting HBV capsid assembly: structure-activity relationship around the GLP-26 scaffold.
BibTeX
@article{li2026multivariate,
title = {{Multivariate prediction of white matter BOLD activity and its relationship to marginal connectivity}},
author = {Muwei Li and Zhaohua Ding and John C Gore},
journal = {NeuroImage},
year = {2026},
month = oct,
doi = {10.1016/j.neuroimage.2026.122297},
url = {https://doi.org/10.1016/j.neuroimage.2026.122297},
note = {CC BY},
}
RIS
TY - JOUR
TI - Multivariate prediction of white matter BOLD activity and its relationship to marginal connectivity
AU - Muwei Li
AU - Zhaohua Ding
AU - John C Gore
PY - 2026
DA - 2026/10/30
JO - NeuroImage
DO - 10.1016/j.neuroimage.2026.122297
UR - https://doi.org/10.1016/j.neuroimage.2026.122297
AB - Functional MRI studies have traditionally focused on gray matter, whereas white matter BOLD signals have often been treated as weak or artifactual. However, white matter BOLD fluctuations are robustly detectable and contain reproducible functional information, yet most gray-to-white matter analyses rely on only marginal functional connectivity, which cannot separate pairwise coupling from shared variance among distributed cortical systems. Here, we used a multivariate cortical prediction framework to test whether spontaneous white matter BOLD activity can be predicted from distributed cortical gray matter BOLD activity, and how this predictive structure relates to marginal connectivity. Resting-state fMRI data from 81 Human Connectome Project young adult participants were analyzed using a strict white matter mask with no overlap with cortical predictors. For each white matter voxel, time series from 400 cortical parcels were used to predict white matter BOLD signals with nested leave-one-run-out ridge regression. Cortical BOLD activity modestly but reliably predicted white matter BOLD dynamics, demonstrating consistent cross-validated prediction accuracy across a broad spatial extent of the white matter. Ridge β coefficients strongly recapitulated marginal functional connectivity fingerprints, indicating a shared functional backbone, but their first gradients differed reproducibly. The difference between these gradients defined a β-minus-FC G1 divergence axis that remained stable after broadly controlling for anatomical, signal-related, and spatial factors. At the low end of the axis, enriched in callosal and capsular structures, pairwise gray-to-white matter connectivity was relatively strong, but white matter BOLD fluctuations were predicted less well by distributed cortical BOLD activity than that connectivity would suggest. At the high end, enriched in posterior projection and association pathways, pairwise connectivity was relatively weak, but distributed cortical BOLD activity predicted white matter BOLD signals better than expected. These findings identify a reproducible, tract-organized contrast between marginal connectivity and its covariance-adjusted predictive representation.
SN - 1095-9572
SN - 1053-8119
ER -