Postacquisition suppression of large‐vessel BOLD signals in high‐resolution fMRI

Abstract

Large-vessel BOLD contamination is a serious impediment to localization of neural activity in high-resolution fMRI studies. A new method is presented which estimates and removes the fraction of BOLD signal that arises from oriented vessels, such as cerebral and pial veins in a voxel, by measuring their influence on the phase angle of the complex valued fMRI time series. A maximum likelihood estimator based on a linear leastsquares fit of the BOLD signal phase to the BOLD signal magnitude in a voxel is shown to efficiently suppress the BOLD effect from these larger veins, whose activation is not well colocalized with the neural response. In high-resolution in vivo fMRI data at 4 T, it is estimated that the method is sensitive to the phase changes in the cerebral, larger intracortical, and pial veins. The technique requires no special pulse sequence modifications or acquisition strategies, and is computationally fast and intrinsically robust. At field strengths of 0.5-7 Tesla (1-5), the intensity changes in MRI observed during functional activation using gradient-echo EPI are dominated by the intravascular BOLD changes in the cortical cerebral veins and larger pial veins (macrovascular BOLD). This increases the effective point spread function of the fMRI technique in high-resolution imaging studies (6,7), since these larger veins can extend from several to tens of millimeters from the site of neuronal activity (8). Even at lower resolutions, larger veins can give rise to uncertainty in localization. For example, the presence of a draining vein in the central sulcus (2,8) makes distinguishing pre-from postcentral gyrus activity in the cortex extremely difficult, with obvious implications for other applications, such as presurgical planning in the motor cortex. While many clever techniques, such as spin-echo EPI, spiral, diffusion-weighted EPI, and perfusion-weighted methodologies, have been proposed to suppress the macrovascular BOLD effect, the use of gradient-echo EPI has been virtually ubiquitous in BOLD applications over the past decade because of its speed, multislice capability, ease of acquisition, and robust signal.The objection to the use of the macrovascular BOLD signal in certain brain mapping applications is based on the inhomogeneous distribution of veins in the brain (Ref. 8 and references therein). Cortical cerebral veins can range from 0.5-2.5 mm in radius and are typically found every 5-30 mm along the cortical surface. Because of the folded nature of the cortex, some of these veins can be found in sulci while yet others cross sulci on their way to venous sinuses. The cerebral veins collect deoxygenated blood received from the pial vein network, a dense, branching collection of 25-250-m-radius vessels located every 0.2-3 mm along the cortical surface. The smallest pial veins are formed from the right-angle emergence of an intracortical vein (10 -60 m radius) that penetrates the cortex tangential to the surface. These tangential penetrations occur every 0.25-1.5 mm on the cortical surface, and with ...

Functional quantitative susceptibility mapping (fQSM)

“…Only a few studies have investigated BOLD contrast by also using the phase information in time-series (Arja et al, 2010;Bianciardi et al, 2013;Chen et al, 2013;Hagberg et al, 2008;Hagberg et al, 2012;Hahn et al, 2009;Menon, 2002;Petridou et al, 2009;Rowe, 2005;Rowe and Logan, 2004;Rowe et al, 2007;Tomasi and Caparelli, 2007). At the spatial resolution allowed by conventional magnetic fields (e.g.…”

“…1.5 T and 3 T), the BOLD phase effect, despite being stronger than the magnitude effect at microscopic level, is averaged out due to the orientation dependence of microscopic field perturbation effects, hence substantial phase contrast can only be found near a few large veins of diameter comparable to the voxel dimensions. At high spatial resolution, the phase of the fMRI time-series has been used to identify the dominant non-local BOLD effects due to large veins and to remove their contribution from the conventional magnitude BOLD statistical maps (Menon, 2002). Recently, a novel biophysical model for phase changes in BOLD fMRI based on the Lorentz-sphere approach was proposed (Zhao et al, 2007).…”

Macrovascular Contribution in Activation Patterns of Working Memory

“…The macrovascular contribution is a serious concern for blood oxygenation level dependent (BOLD)-functional magnetic resonance imaging (fMRI) studies because this not well-localized contribution dominates the BOLD-fMRI signal change (Menon, 2002). Thus, BOLD signal changes can reflect neuronal activity at more remote locations (from several to tens of millimeters) from the activation site (Duvernoy, 1999), reducing spatial localization in gradient-echo planar imaging (GE-EPI) based BOLD-fMRI studies (Engel et al, 1997;Menon and Goodyear, 1999).…”

“…While the extravascular BOLD-fMRI signal results from brain tissue other than blood, the intravascular BOLD-fMRI signal results from blood capillaries (2 to 4 mm radius), pial veins (10 to 250 mm), and cortical veins (0.5 to 2.5 mm) (Menon, 2002). The intra-, and extravascular components of the BOLD signal reflect local magnetic field variations produced by changes in the fractional oxygen saturation of hemoglobin during the fMRI tasks (Boxerman et al, 1995b;Menon et al, 1995;Ugurbil et al, 1999); however, the intravascular pool accounts for the majority of the BOLD-fMRI dependent signal (Boxerman et al, 1995a).…”

“…In contrast, the fMRI-stimulus modulates the phase component of the signal coming from the macrovascular compartment, since the more homogeneous magnetic field partially preserves the coherence of the magnetization inside the voxel. Thus, the macrovascular contribution in activated networks can be determined by using phase information embedded in the BOLD-fMRI signal and could be removed from activation patterns (Menon, 2002).…”