Evidence for a vascular contribution to diffusion FMRI at high b value

Abstract

Recent work has suggested that diffusion-weighted functional magnetic resonance imaging (FMRI) with strong diffusion weighting (high b value) detects neuronal swelling that is directly related to neuronal firing. This would constitute a much more direct measure of brain activity than current methods and represent a major advance in neuroimaging. However, it has not been firmly established that the observed signal changes do not reflect residual vascular effects, which are known to exist at low b value. This study measures the vascular component of diffusion FMRI directly by using hypercapnia, which induces blood flow changes in the absence of a change in neuronal firing. Hypercapnia elicits a similar diffusion FMRI response to a visual stimulus including a rise in percent signal change with increasing b value, which was reported for visual activation. Analysis of the response timing found no evidence for an early response at high b value, which has been reported as evidence for a nonhemodynamic response. These results suggest that a large component of the diffusion FMRI signal at high b value is vascular rather than neuronal.brain activation ͉ diffusion MRI ͉ functional MRI ͉ neuronal swelling F unctional neuroimaging has enabled major advances in the study of normal and pathological brain function. However, the methods that provide the greatest coverage and spatial resolution, including positron emission tomography (1, 2) and magnetic resonance imaging (MRI) (3, 4), are indirect measures of neuronal activity based on metabolically driven changes in blood flow. These hemodynamic measures suffer from spatial and temporal confounds (5, 6), are nonlinearly related to neuronal firing (7,8), and depend on baseline hemodynamics that are uncoupled from the activity of interest (9, 10). An imaging method that detects neuronal activity more directly while achieving whole-brain coverage would therefore represent a significant advance for neuroscience.One alternative method is diffusion-weighted functional MRI (DFMRI), which attenuates the MRI signal in a manner that depends on the amount of motion (diffusion and flow) in the underlying tissue. In general, this attenuation is described by a factor exp(ϪbD), where D is the apparent diffusion coefficient of the local tissue, and b is an acquisition parameter that describes the strength of diffusion contrast. At low b value, true diffusive motion is more difficult to detect, and the ''apparent diffusion'' is dominated by local blood flow (11)(12)(13)(14). In general, diffusion weighting is directional, so that attenuation depends on the diffusion and flow parameters along the applied direction. Different signal sources have been proposed in DFMRI depending on the b value. Early work used low b value, which is thought to reflect tissue perfusion (11-16). Recent work suggested that DFMRI at high b value detects cellular swelling that is a direct consequence of neural firing (17, 18), which would constitute a more direct measure of neuronal activity.A recent study at high b valu...

Assessing the sensitivity of diffusion MRI to detect neuronal activity directly

“…The debate about whether diffusion MRI can detect neuronal activity more directly and accurately than conventional hemodynamicbased fMRI methods has never stopped since it was proposed (10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20). Our hybrid test bed, which allows for simultaneous monitoring of neuronal activity via intracellular calcium fluorescence imaging during MR acquisition on the organotypic cortical cultures, provides a unique opportunity for direct, time-dependent comparisons of the diffusion MR signals with neuronal activity without any hemodynamic or other physiologic confounds.…”

“…Two major reasons for this may be a dearth of experiments that convincingly establish its neurophysiological basis and the poor reproducibility of the originally reported changes in diffusion MRI signals by different laboratories. The inability to detect the predicted changes using fDMRI and the possible confounds of hemodynamic contributions in fDMRI measurements in vivo do not argue for a robust connection between changes in diffusion MRI and underlying neuronal activity (17)(18)(19)(20). Thus, "ground-truth" experiments, potentially establishing a connection between the changes in diffusion MRI and underlying neuronal activity, are needed, particularly to shed light on the possible biophysical basis of the fDMRI signal.…”

Water-Diffusion Slowdown in the Human Visual Cortex on Visual Stimulation Precedes Vascular Responses

“…By contrast, the indirect T2* t component is expected to share the temporal profile of the deoxyhemoglobin vascular content (BOLD) time course (Boxerman et al, 1995;van Zijl et al, 1998). This indirect vascular effect on T2* t could also easily explain the increase in the diffusion-sensitized signal, observed during vascular challenges, such as hypercapnia (Miller et al, 2007), which is known to decrease the vascular deoxy-Hb content. Miller et al have reported an apparent increase of the signal response to hypercapnia with the b value.…”

“…This hypothetical mechanism, if confirmed, would be more closely associated with neuronal activation, and would mark a significant departure from the blood flow-based MRI approach, thereby potentially offering improved spatial and temporal resolution. However, the assumption of a nonvascular origin of the water-diffusion signal is a subject of controversy (Jin and Kim, 2008;Miller et al, 2007;Yacoub et al, 2008). Furthermore, vascular responses that are faster than BOLD signals have also been observed through measurements of the regional cerebral blood volume (rCBV) and the total local hemoglobin content (Huppert et al, 2006;Lu et al, 2003;Malonek et al, 1997).…”

Functional localization of the human color center by decreased water displacement using diffusion‐weighted fMRI

“…Diffusion‐weighted imaging is inherently sensitive to BOLD signal changes, and the extent to which BOLD sources contribute to the overall DfMRI signal continues to be the subject of controversy (Miller et al. 2007; Autio et al. 2011; Kuroiwa et al.…”