The nature and interpretation of BOLD signals in white matter - A review.

This review summarizes selected recent findings demonstrating the dependence of blood oxygenation level dependent (BOLD) signals in white matter (WM) on tissue microstructure, composition, vascular properties and metabolism, as well as their relationships to fMRI signals within gray matter (GM) networks. BOLD signals in WM are robustly detectable after a stimulus, and at rest their temporal variations reveal synchronized networks and correlated neural activities involving both WM and GM. However, to date, most analyses of brain fMRI data have ignored WM signals, and often have removed them as nuisance regressors. However, emerging evidence clearly demonstrates that WM BOLD signals represent potentially important and heretofore overlooked indicators of neural activities that are intimately related to how cortical regions communicate, and so should be incorporated into more complete models of brain functional organization. Here we review recent work that contributes to our understanding of their interpretation and significance. The factors that affect the magnitude and other characteristics of BOLD responses in WM are becoming more clear, and recent studies have demonstrated and quantified the relationships between BOLD signals and vascular and microstructural properties of WM tracts. These relationships depend on the degree of myelination and neurite and mitochondrial densities, but they also are qualitatively different when comparing different fiber types, notably association versus projection fibers. Some fully myelinated fibers appear to not show detectable BOLD effects. The relationships between WM and GM BOLD signals, the contributions of GM resting state correlations and signals to WM BOLD signals, and the engagement of WM in GM networks, are also becoming more clear. These findings supplement the growing literature demonstrating practical, clinical applications of BOLD in WM. The goal of this review is to highlight recent research that demonstrates how WM and GM activities are related, and to stimulate further investigations that may produce a more complete model of brain organization.