As high-resolution functional magnetic resonance imaging (fMRI) and fMRI of cortical layers become more widely used, the question how well high-resolution fMRI signals reflect the underlying neural processing, and how to interpret laminar fMRI data becomes more and more relevant. High-resolution fMRI has shown laminar differences in cerebral blood flow (CBF), volume (CBV), and neurovascular coupling. Features and processes that were previously lumped into a single voxel become spatially distinct at high resolution. These features can be vascular compartments such as veins, arteries, and capillaries, or cortical layers and columns, which can have differences in metabolism. Mesoscopic models of the blood oxygenation level dependent (BOLD) response therefore need to be expanded, for instance, to incorporate laminar differences in the coupling between neural activity, metabolism and the hemodynamic response. Here we discuss biological and methodological factors that affect the modeling and interpretation of high-resolution fMRI data. We also illustrate with examples from neuropharmacology and the negative BOLD response how combining BOLD with CBF- and CBV-based fMRI methods can provide additional information about neurovascular coupling, and can aid modeling and interpretation of high-resolution fMRI.
fMRI at High Spatial Resolution: Implications for BOLD-Models
Different types of negative BOLD responses in early visual cortex of anesthetized macaques. (A–C): negative BOLD adjacent to positive BOLD in V1 operculum (A) corresponds to an increase in CBV (B) and a decrease in CBF (C). The contrast-agent based CBV-response is inverted and shows a signal decrease (violet) for an increase in CBV. The stimulus was a gapped rotating black-and-white checkerboard. Adapted from Goense et al. (2012), with permission. (D–F): negative BOLD in peripheral V1 (“pV1”) and extrastriate areas such as visual area V2, in response to a full-field rotating black-and-white checkerboard. The negative BOLD signal (D) is associated with a decrease in CBV (E) and CBF (F). Note that the negative CBF-signal does not show many significant voxels, despite having a sizable negative response, however this is the result of the high spatial resolution and the reduced sensitivity in the center of the brain due to its large distance from the receiver coils (Goense et al., 2010). (G–I): negative BOLD response in V1 to a full-field rotating checkerboard stimulus (G). The negative BOLD is associated with an increase in CBV (H) and CBF (I) as also seen in rodents (Schridde et al., 2008). Interestingly, the foveal area (“f”) showed a positive BOLD response. Acquisition parameters: (A–C, G–I) 4.7T, (D–F) 7T; for BOLD, 8-segment GE-EPI, resolution LRxAP 500 × 375 μm, TE 20 ms, TR 750 ms; for CBV, 8-segment GE-EPI, resolution 500 × 375 μm, TE 12–20 ms, TR 750 ms, 8 mg/kg MION (B and H) or Feraheme (E); for CBF, FAIR-based ASL, single shot EPI, resolution 500 × 500 μm (C), 500 × 312 μm (F) and 744 × 500 μm (H), TE 9.5–12 ms, TI 925–1300 ms, TR 3000–4500 ms.
Max Planck Institute for Biological Cybernetics
