Predicting the macrovascular contribution to resting-state fMRI functional connectivity at 3 Tesla: A model-informed approach

Abstract

Macrovascular biases have been a long-standing challenge for fMRI, limiting its ability to detect spatially specific neural activity. Recent experimental studies, including our own (Huck et al., 2023; Zhong et al., 2023), found substantial resting-state macrovascular BOLD fMRI contributions from large veins and arteries, extending into the perivascular tissue at 3 T and 7 T. The objective of this study is to demonstrate the feasibility of predicting, using a biophysical model, the experimental resting-state BOLD fluctuation amplitude (RSFA) and associated functional connectivity (FC) values at 3 Tesla. We investigated the feasibility of both 2D and 3D infinite-cylinder models as well as macrovascular anatomical networks (mVANs) derived from angiograms. Our results demonstrate that: 1) with the availability of mVANs, it is feasible to model macrovascular BOLD FC using both the mVAN-based model and 3D infinite-cylinder models, though the former performed better; 2) biophysical modelling can accurately predict the BOLD pairwise correlation near to large veins (with R2 ranging from 0.53 to 0.93 across different subjects), but not near to large arteries; 3) compared with FC, biophysical modelling provided less accurate predictions for RSFA; 4) modelling of perivascular BOLD connectivity was feasible at close distances from veins (with R2 ranging from 0.08 to 0.57), but not arteries, with performance deteriorating with increasing distance. While our current study demonstrates the feasibility of simulating macrovascular BOLD in the resting state, our methodology may also apply to understanding task-based BOLD. Furthermore, these results suggest the possibility of correcting for macrovascular bias in resting-state fMRI and other types of fMRI using biophysical modelling based on vascular anatomy.

Macrovascular contributions to resting-state fMRI signals: A comparison between EPI and bSSFP at 9.4 Tesla

“…Z. Zhong et al, 2024). In this work, we were able to reproduce the observations on the rs-fMRI signal originating from the large draining veins through analysis of the cortical orientation relative to − → B 0 in all cortical depths.…”

The influence of basal CO2on neurofluid dynamics measured using resting-state BOLD fMRI

“…It is also noteworthy that the macrovascular ROIs used in this study are based on regions with high BOLD power, which may differ from real macrovascular structures based on recent biophysical models (X. Zhong et al, 2024).…”

Widespread Autonomic Physiological Coupling Across the Brain-Body Axis

“…basilar artery, internal carotid and posterior cerebral artery) overlap with areas of high CSF density (Figure 3D), suggesting decreases in arterial blood volume (and increases of CSF volume) may result in positive fMRI signal in these areas. Alternatively, opposing signals in the cerebral arteries may arise from partial-volume effects between arterial blood and surrounding tissue, such that blood volume fluctuations change the relative size of the intravascular component contribution to the signal (versus the extravascular component), resulting in changes in phase homogeneity within a voxel (Zhong et al, 2024).…”