MRI · reads
Functional MRI at 7 tesla is magnetic resonance imaging of the human brain that infers neuronal activity from blood oxygenation. In gradient-echo BOLD imaging the signal falls as local deoxyhemoglobin rises; after neuronal activity increases, blood flow rises more than oxygen use, so deoxyhemoglobin drops near active sites and in downstream draining veins. What it reads is BOLD image intensity, an indirect measure of neuronal activity, and the venous contribution limits spatial specificity. At 7 T in humans, spin-echo BOLD gave narrower estimated point-spread functions than gradient-echo BOLD.
Human functional MRI infers local changes in neuronal activity from task-invoked metabolic and hemodynamic responses, so its signal is an indirect measure of neuronal activity. In gradient-echo blood oxygenation level dependent (BOLD) imaging, the signal is inversely proportional to local deoxyhemoglobin content. After neuronal activity rises, local arterial cerebral blood flow increases more than oxygen consumption, which lowers deoxyhemoglobin in capillaries. This change propagates from capillaries at active sites downstream to draining veins, so large veins distant from those sites contribute functional mapping signals. That venous contribution impairs spatial specificity and adds to any intrinsic imprecision in coupling between blood flow and neuronal activity. Intermediary non-neuronal events limit specificity in space and how that specificity changes over time. A spatio-temporal point-spread function in human gray matter describes these limits. At 7 T in humans, estimated point-spread functions were narrower for spin-echo BOLD than for gradient-echo BOLD. These estimates reflect realistically achievable specificity at that field strength but are only upper bounds on the true widths. In human resting-state 7 T fMRI, functional connectivity between regional signals may reflect both monosynaptic and polysynaptic connections. Higher-resolution acquisition, minimal smoothing and nuisance regression using signals from individual ventricles can mitigate low signal-to-noise, partial volume effects and physiological aliasing.
Download the dataset (JSON, described by its JSON Schema).
Loading…