Single-vessel cerebral blood flow fMRI to map blood velocity by phase-contrast imaging

Current approaches to high-field fMRI provide two means to map hemodynamics at the level of single vessels in the brain. One is through changes in deoxyhemoglobin in venules, i.e., blood oxygenation level-dependent (BOLD) fMRI, while the second is through changes in arteriole diameter, i.e., cerebral blood volume (CBV) fMRI. Here we introduce cerebral blood flow (CBF)-fMRI, which uses high-resolution phase-contrast MRI to form velocity measurements of flow and demonstrate CBF-fMRI in single penetrating microvessels across rat parietal cortex. In contrast to the venule-dominated BOLD and arteriole-dominated CBV fMRI signal, the phase-contrast -based CBF signal changes are highly comparable from both arterioles and venules. Thus, we have developed a single-vessel fMRI platform to map the BOLD, CBV, and CBF from penetrating microvessels throughout the cortex. This high-resolution single-vessel fMRI mapping scheme not only enables the vessel-specific hemodynamic mapping in diseased animal models but also presents a translational potential to map vascular dementia in diseased or injured human brains with ultra-high field fMRI. Summary We established a high-resolution PC-based single-vessel velocity mapping method using the high field MRI. This PC-based micro-vessel velocity measurement enables the development of the single-vessel CBF-fMRI method. In particular, in contrast to the arteriole-dominated CBV and venule-dominated BOLD responses, the CBF-fMRI shows similar velocity changes in penetrating arterioles and venules in activated brain regions. Thus, we have built a noninvasive single-vessel fMRI mapping scheme for BOLD, CBV, and CBF hemodynamic parameter measurements in animals.

Abstract 33 Current approaches to high-field fMRI provide two means to map hemodynamics at the 34 level of single vessels in the brain. One is through changes in deoxyhemoglobin in 35 venules, i.e., blood oxygenation level-dependent (BOLD) fMRI, while the second is 36 through changes in arteriole diameter, i.e., cerebral blood volume (CBV) fMRI. Here we 37 introduce cerebral blood flow (CBF)-fMRI, which uses high-resolution phase-contrast MRI 38 to form velocity measurements of flow and demonstrate CBF-fMRI in single penetrating 39 microvessels across rat parietal cortex. In contrast to the venule-dominated BOLD and 40 arteriole-dominated CBV fMRI signal, the phase-contrast -based CBF signal changes are 41 highly comparable from both arterioles and venules. Thus, we have developed a single- 42 vessel fMRI platform to map the BOLD, CBV, and CBF from penetrating microvessels 43 throughout the cortex. This high-resolution single-vessel fMRI mapping scheme not only 44 enables the vessel-specific hemodynamic mapping in diseased animal models but also 45 presents a translational potential to map vascular dementia in diseased or injured human 46 brains with ultra-high field fMRI. Summary 55 We established a high-resolution PC-based single-vessel velocity mapping method using 56 the high field MRI. This PC-based micro-vessel velocity measurement enables the 57 development of the single-vessel CBF-fMRI method. In particular, in contrast to the 58 arteriole-dominated CBV and venule-dominated BOLD responses, the CBF-fMRI shows 59 similar velocity changes in penetrating arterioles and venules in activated brain regions. 60 Thus, we have built a noninvasive single-vessel fMRI mapping scheme for BOLD, CBV, 61 and CBF hemodynamic parameter measurements in animals.

67
Cerebral blood flow (CBF) is a key readout of neuronal processing and viability in normal 68 and diseased brain states 1 . Changes in CBF may be monitored directly within individual 69 blood vessels through the use of optical-based particle tracking techniques 2 . A variety of 70 imaging methods have been developed to measure CBF across multiple spatial scales 71 from capillary beds to the vascular network in animal brains, including multi-photon 72 microscopy 3 , near-infrared spectroscopy (NIRS) 4 , optical coherence tomography 5 , 73 optoacoustic imaging 6 , or laser doppler and speckle imaging 7, 8 . In particular, the doppler-74 based functional ultrasound imaging method provides a unique advantage to detect the 75 CBF in the brain with a high spatiotemporal resolution, which can be readily applied for 76 awake animal imaging 9-11 . However, these methods share a common barrier that the 77 spectrum-specific signal transmission cannot effectively pass the skull of animals without 78 significant loss of the signal-to-noise ratio (SNR). Typically, a craniotomy or procedure to 79 thin the skull is needed to detect the hemodynamic signal 2 . While current techniques 80 support transcranial imaging into the superficial layers of the cortex, only functional MRI 81 (fMRI) provides a noninvasive approach for measuring hemodynamic signals throughout 82 the brain. 83 Changes in CBF may be detected by fMRI based on arterial spin labeling (ASL), 84 in which water protons in a major upstream vessel are spin-polarized with an external 85 field 12-14 . Two other fMRI-based techniques provide indirect information about changes in 86 CBF. Blood oxygenation level-dependent (BOLD) fMRI is used to determine changes in 87 the ratio of deoxy-to oxyhemoglobin in the blood and is a measure of changes in brain 88 metabolism 12, 15, 16 . Cerebral blood volume (CBV) fMRI is used to measure changes in 89 blood volume, i.e., essentially changes the diameter of arterioles, based on the use of 90 exogenous or endogenous contrast agents to differentiate blood from brain tissue 12, 17 .

91
Phase-contrast (PC) MRI relies on gradient-oriented dephasing of magnetized 92 protons to map the velocity, i.e., direction and speed, of blood flow 18,19 . The ASL-based 93 CBF fMRI technique detects local changes in the flow of blood through brain tissue but 94 does not show orientation-specific information related to the alignment of vessels 20 . Past 95 works with 7 T MR scanning showed that PC-MRI can be used to measure flow in the 96 perforating arteries through the while matter or the lenticulostriate arteries in the basal 97 ganglia of human brains 21-24 . However, the SNR was insufficient in these prior studies to 98 map changes in flow, and thus changes in CBF. 99 Here, we report on a PC-MRI method to detect the vessel-specific changes in 100 blood velocity in single trials. Compared with past implementations of PC-MRI 21, 25-28 , we 101 have implemented a small surface radio frequency (RF) coil with the high field MRI, i.e., 102 14.1 T for improved SNR. This further allows us to map the BOLD-and CBV-fMRI from 103 individual penetrating venules and arterioles, which span 20 to 70 µm diameter, with high 104 spatial resolution 29-31 .

106
Phantom validation of high-resolution PC-based flow velocity measurement 107 For calibration, we constructed an in vitro capillary tubing circulatory system to mimic 108 penetrating vessels, with flow rates from 1 to 10 mm/s ( Figure 1A). A 2D PC-MRI slice is 109 aligned perpendicular to the capillary tubing ( Figure 1A, B) and provides a voxel-specific 110 measurement of the flow velocity through two tubes with the upward flow (positive sign, 111 bright dots in Figure 1B) and two tubes with the downward flow (negative sign, dark dots 112 in Figure 1B), as well as a control tube. We observe a monotonic and near-linear relation 113 between the velocity measured by PC-MRI and the true velocity: Vmeas = 114 (0.67 ± 0.01) Vpump + (0.02 ± 0.11) mm/s at echo time (TE) = 5.0 ms ( Figure 1C). The 115 small offset could be caused by eddy current effects and other gradient-related scaling 116 errors of the PC-MRI sequence 32-34 . We further observe that the measured velocities are 117 relatively insensitive to the value of TE ( Figure 1C). 118 We implemented the high-resolution PC-MRI for in vivo measurement of blood flow  Figure 1). This was essential for the high-124 resolution mapping with a fast sampling rate of the single-vessel flow velocity over a 125 complete hemisphere of the rat brain (Figure 2A and Supplemental Figure 1E).

126
In vivo PC-based flow velocity mapping of penetrating microvessels 127 We first acquired the single-vessel A-V map by aligning a 500 µm thick 2D MRI slice 128 perpendicular to the penetrating vessels through layer V of one hemisphere (Figure 2A Figure 3Av). In contrast, the CBF-fMRI signal is 158 observed in both penetrating arterioles and venules (Figure 3Aiv). 159 The stimulus-evoked responses of all three fMRI signals were studied with an 160 on/off block design (Figures 3B-E). Group analysis shows that the positive BOLD signal 161 from venule voxels is significantly higher than the arteriole-specific BOLD signal 162 ( Figure 3C). In contrast, the arteriole dilation leads to an earlier CBV-weighted negative 163 fMRI signal, which is much stronger and faster than the signal from passive venule-  the arterial blood gas, administrate drugs, and constantly measure the blood pressure. 251 After catheterization, rats were secured in a stereotaxic apparatus, a custom-made RF 252 coil was fixed above the skull with cyanoacrylate glue (454, Loctite). After surgery, 253 isoflurane was switched off and a bolus of α-chloralose (80 mg/kg, Sigma-Aldrich) was 254 intravenously injected. A mixture of α-chloralose (26.5 mg/kg/h) and the muscle relaxant 255 (pancuronium bromide, 2 mg/kg/h) was continuously infused to maintain the anesthesia 256 and minimize the motion artifacts. Throughout the whole experiment, the rectal 257 temperature of rats was maintained at 37°C by using a feedback heating system. All  arteriole (right). Note that the venule is a bright voxel and arteriole is a dark voxel in the flow map, which is opposite to the A-V map and also the slightly different time course due to the altered spatial resolution.