The determination of both spatial and temporal characteristics of the HDR to focal brain activity is a topic of great relevance as it dictates the accuracy of functional neuroimaging techniques in mapping activation regions, establishes the ultimately achievable spatial and temporal resolution, and influences the interpretation of the data. In this aim, we are working on optimizing stimulus parameters and measuring, in space and in time, the ensuing HDR, with the long-term goal of determining the ultimate spatial domain of CBF control and its associated temporal evolution. We believe such work will require extremely brief stimuli, delivered under well controlled conditions, to elicit minute, yet measurable vascular events, which can presumably serve as the building blocks of the integrative CBF response to more complex stimuli. Our previous work on the characterization of BOLD and CBF changes to increased neural activity has shown that these two hemodynamic signals have distinct spatial and temporal characteristics across the cortical depth. Specifically, on the spatial domain, we observed that BOLD activation maps were consistently larger than the CBF maps and skewed towards the pial surface, due to contamination of the BOLD signal by large draining veins produced. The CBF maps, on the other hand, were better confined to the anatomical borders of S1, with the deeper layers of the cortex presenting the strongest resting CBF and CBF response to functional stimulation in the. On the temporal domain, with the development of an ultra-fast method to measure CBF changes, we were able to demonstrate that the onset time of the CBF response in S1 occurred significantly earlier than changes in deoxyhemoglobin concentration, and that CBF had a significant cortical heterogeneity of fMRI onset times, with deep regions of the cortex responding to stimulus significantly earlier than the pial surface. Faster CBF dynamics than BOLD were also found in humans. In coming to NIH and working at a higher spatial resolution using BOLD contrast at 11.7T, we observed that the BOLD response also presented a consistent heterogeneity of fMRI onset times and amplitudes across the cortex, with the earliest onset times corresponding anatomically to layer IV, and with superficial and deeper layers starting significantly later. The amplitude of BOLD signal changes varied with the cortical depth from the pial surface. Changes in the supragranular layers were 44% bigger than changes in the intermediate layers, located only approximately 700 m below, and 144% larger than the bottom layer, located approximately 1.4 mm below the pial surface. The main implications of these findings are that (i) BOLD (and also CBF) have distinct amplitude and temporal characteristics, which vary spatially across cortical layers and may be used to study laminar communication;and (ii) in most human fMRI, data is acquired at course spatial resolutions, insufficient to resolve subcortical depths, implying that the functional signal changes detected could be dominated by large surface vessels. In fact, we also observed significant contamination of BOLD by draining veins, even when a spin-echo EPI sequence is used at 11.7T, indicating that draining veins affect the spatial localization of BOLD fMRI. Our previous work comparing BOLD and CBF fMRI shows that draining veins do not affect CBF quantification, and similar results were obtained when using iron-oxide to measure CBV. Furthermore, the CBF temporal features where tighter (shorter onset, faster post-stimulus decay) than BOLD. Taken together, the above results raise the possibility that the temporal and spatial characteristics of BOLD fMRI are significantly affected and compromised by dispersive effects from draining veins that are difficult to eliminate. Based on the above, and to better understand the spatial and temporal features of BOLD, we hypothesized that the temporal resolution of BOLD fMRI is limited by vascular events with different time scales, and in particular by transit times through the vasculature. To test this hypothesis, we set out to measure the hemodynamic impulse response (HIR) function for BOLD and CBV in the rat brain. Working in collaboration with Dr. Duyn, we used an m-sequence approach to present short stimuli in a pseudo-random order, in a way that can be easily unraveled to reveal the HIR. The m-sequence constitutes an efficient and high SNR approach to measuring the HIR to brief stimuli. The spatial extent of the BOLD response was significantly larger than the one of the CBV response, indicating substantial contributions from large draining veins in the peripheral regions. In both the BOLD and CBV maps, the regions with the largest t-scores were located in the middle layers of the cortex, in agreement with our previous studies as well as with other animal-based fMRI studies at high field strength. The CBV response begins by the first point at 0.5 s after stimulus onset, suggesting that the CBV changes are dominated by vasodilatation of the arterial vessels (capillaries, arterioles, and arteries), rather than an increase in venous blood volume, in agreement with previous measurements of CBV changes by fMRI and optical imaging. On the other hand, the BOLD onset is significantly delayed with respect to the CBV onset. The mean difference in onset times between CBV and BOLD responses was 0.440.24s, indicating that the BOLD contrast is limited by the transit of oxygenated blood across the capillaries to the venous side of the vasculature. Second, the CBV IR shows rapid dynamics on both the rise and fall of the curve, which suggests fast-acting mechanisms of vasodilatation at the onset of the response and of vasoconstriction at the end of the response. Because of this, the CBV curve shows significantly narrower dispersion (full-width-at-half-maximum, FWHM = 1.370.11s) and time-to-peak (TTP = 1.650.15s) than the BOLD curve (FWHM = 1.920.22s and TTP = 2.180.14s). Third, the CBV curve shows a lengthy and slowly decaying tail that remains elevated for several seconds beyond the duration of the task. This slow recovery behavior is consistent with previous investigations of the CBV response to long (30s) stimuli, and has also been observed in optical imaging estimates of the HIR in rats and CBV fMRI in macaques. We need to pursue further experiments to understand the origins of the slow return of CBV to baseline. It could be that the observed biphasic CBV dynamics results from the interactions of two different vascular entities with distinct vasodilatory mechanisms, e.g., arteries/arterioles associated with the fast, active increase and decrease of CBV, while venules/veins explain the slow, passive dynamics. Alternatively, multiple dilatory mechanisms could be at play within the same vascular compartment, e.g., local vs. systemic, or control by smooth muscle cells vs. pericytes. Preliminary analysis shows that both fast and slow components of the CBV HIR have a similar spatial distribution: they do not involve pial vasculature, and every voxel in the cortex activated by the CBV HIR shows both components.
Showing the most recent 10 out of 44 publications
