Whole-brain perfusion mapping in mice by dynamic BOLD MRI with transient hypoxia

Abstract

Non-invasive mapping of cerebral perfusion is critical for understanding neurovascular and neurodegenerative diseases. However, perfusion MRI methods cannot be easily implemented for whole-brain studies in mice because of their small size. To overcome this issue, a transient hypoxia stimulus was applied to induce a bolus of deoxyhemoglobins as an endogenous paramagnetic contrast in blood oxygenation level-dependent (BOLD) MRI. Based on stimulus-duration-dependent studies, 5 s anoxic stimulus was chosen, which induced a decrease in arterial oxygenation to 59%. Dynamic susceptibility changes were acquired with whole-brain BOLD MRI using both all-vessel-sensitive gradient-echo and microvascular-sensitive spin-echo readouts. Cerebral blood flow (CBF) and cerebral blood volume (CBV) were quantified by modeling BOLD dynamics using a partial-volume-corrected arterial input function. In the mouse under ketamine/xylazine anesthesia, total CBF and CBV were 112.0 ± 15.0 ml/100 g/min and 3.39 ± 0.59 ml/100 g (n = 15 mice), respectively, whereas microvascular CBF and CBV were 85.8 ± 6.9 ml/100 g/min and 2.23 ± 0.27 ml/100 g (n = 7 mice), respectively. Regional total vs. microvascular perfusion metrics were highly correlated but a slight mismatch was observed in the large-vessel areas and cortical depth profiles. Overall, this non-invasive, repeatable, simple hypoxia BOLD-MRI approach is viable for perfusion mapping of rodents.

Keywords

BOLD cerebral blood flow cerebral blood volume dynamic susceptibility contrast (DSC)transient hypoxia

Introduction

Proper cerebral perfusion is critical for maintaining brain homeostasis, delivering oxygen and nutrients, and removing neurometabolic wastes. Cerebral perfusion is thus a good biomarker of neurovascular and neurodegenerative diseases such as stroke, vascular dementia, Alzheimer’s disease, and tumor.¹ Therefore, it is critical to quantify perfusion metrics, including cerebral blood flow (CBF) and cerebral blood volume (CBV), with high sensitivity and repeatability for the follow-up progression and treatment of diseases. Numerous mouse models that recapitulate neurological disease phenotypes or progression also manifest with brain perfusion changes. This makes it imperative to non-invasively measure perfusion metrics in mice.

Magnetic resonance imaging (MRI) is widely used to obtain CBF and/or CBV using endogenous arterial blood or exogenous contrast agents as tracer.² Arterial spin labeling (ASL) techniques rely on magnetically labeled arterial water as an endogenous tracer and permit quantitative CBF measurements that can be repeated during the same experiment.^3–5 However, the sensitivity of ASL depends on the spin labeling time and T₁ of arterial blood. In mice, a high spatial resolution is required because of their small brain volume of ∼0.5 cm³ (vs. 2 cm³ in rats). Consequently, CBF in mice is often obtained in a few selected slices with extensive averaging.^6–13 Alternatively, dynamic susceptibility contrast (DSC) MRI uses a bolus injection of a paramagnetic contrast agent whose concentration changes are dynamically observed via MRI during the first passage of the contrast agent.^14,15 Because contrast agents change both T₁ and T₂/T₂*, a proper sampling scheme to separate these two relaxation parameters is required for modeling of quantification. Leakage of contrast agents through the blood–brain barrier can be significant in pathological conditions such as a tumor, which can complicate quantification.^16–18 Nonetheless, the DSC-MRI technique is highly sensitive, straightforward, and widely used in humans. Application of DSC MRI to mice is technically difficult because it requires an extremely short bolus injection of contrast agents into blood and rapid imaging. Therefore, there are currently limited studies in the literature on mouse cerebral perfusion.^19–21

Similar to blood oxygenation level-dependent (BOLD) fMRI,²² paramagnetic deoxyhemoglobin has been recently used as an endogenous contrast agent for DSC MRI in humans.^23,24 A change in deoxyhemoglobin concentration modulates T₂/T₂* but not T₁.²⁵ Recently, a method was introduced to obtain perfusion measurements using BOLD and hyperoxia-modulated changes in the concentration of deoxygenated hemoglobin.²⁶As hyperoxia is known to cause an increase in the blood oxygenation level primarily localized to the venous vasculature, this contrast mechanism has been proposed as a means to measure venous blood volume.²⁷ Accordingly, a better approach is to modulate the arterial and venous oxygenation saturation level by applying hypoxia. With the adoption of DSC modeling to dynamic BOLD MRI with ∼25 s hypoxic stimulation, CBF and CBV were successfully obtained in humans.²⁴

In this study, we developed a BOLD-DSC MRI protocol with a transient hypoxia gas stimulus for mouse perfusion imaging and measured whole-brain quantitative CBF and CBV at 9.4 T. We examined the stimulus duration-dependent BOLD-MRI response, arterial oxygen saturation level changes, and associated baseline physiological changes. To quantify CBF accurately, the arterial input function was obtained by correcting the partial-volume effect of the arterial vessels. Whole-brain CBF and CBV of mice were mapped using all-sized-vessel-sensitive gradient-echo and microvessel-sensitive spin-echo data collection.

Materials and methods

Theoretical background

Deoxyhemoglobin is used as an endogenous blood-pool contrast agent; its dynamic change in response to transient hypoxia²⁸ is used for DSC perfusion modeling, similar to that used for the exogenous contrast agent.

Quantitative perfusion measurements from dynamic ΔR₂* changes due to hypoxia

The change in the $T_{2}^{*}$ -weighted MRI signal intensity due to hypoxia²⁹ can be described as follows:

\frac{S (t)}{S_{0}} = \frac{f {\cdot e}^{- TE \cdot (R_{2, blood}^{*} + Δ R_{2, blood}^{*} (t))} + (1 - f) \cdot e^{- TE \cdot (R_{2}^{*} + Δ R_{2}^{*} (t))}}{f {\cdot e}^{- TE \cdot R_{2, blood}^{*}} + (1 - f) \cdot e^{- TE \cdot R_{2}^{*}}} ≅ e^{- TE \cdot Δ R_{2}^{*} (t)}

(1)

where S(t) is the signal intensity at time t,

S_{0}

is the baseline signal intensity, f is the volume fraction of blood, TE is the echo time,

R_{2, blood}^{*} and R_{2}^{*}

are the baseline transverse relaxation rates of blood and tissue, and

Δ R_{2, blood}^{*} (t)

and

Δ R_{2}^{*} (t)

are the changes in transverse relaxation rates of blood and extravascular tissue due to hypoxia, respectively. Since blood R₂/R₂* is much larger than tissue R₂/R₂* at ultrahigh fields^30,31 and f is on the order of a few percent, it is assumed that the intravascular contribution is negligible here. Extravascular

Δ R_{2}^{*}

induced by hypoxia is linearly dependent on the concentration change of deoxyhemoglobin within the voxel.^22,25,32,33 To determine cerebral hemodynamics, we assumed that transient hypoxia does not change CBV, CBF, and oxygen consumption, producing the same magnitude of change in oxygen saturation levels in all sizes and types of vessels.

To obtain quantitative perfusion indices from transient hypoxia data, we adopted theories in the widely used DSC MRI, which measures first-pass MRI signal change following the bolus injection of exogenous contrast agent.^{14,15,34–36} The exogenous plasma-based contrast agent concentration in the conventional DSC-MRI theory was replaced with the endogenous deoxyhemoglobin concentration. The change in deoxyhemoglobin concentration $C (t)$ is directly related to ${Δ R}_{2}^{*} (t)$ as follows:

C (t) = k {Δ R}_{2}^{*} (t) = \frac{ρ}{H_{f}} \cdot CBF \cdot (C_{a} (t) ⨂ R (t))

(2)

where k is a proportional constant related to the field strength and pulse sequence;

H_{f}

is the hematocrit level difference between systemic arterial blood and capillary (Hct_art/Hct_capillary); ρ is the brain tissue density of 1.04 g tissue/ml volume;³⁷ C_a(t) is the arterial input function (AIF), that is, the concentration of contrast entering the region of interest (ROI) at time t, which is related to

Δ R_{2}^{*} (t)

of arterial blood; R(t) is the tissue residue function, that is, the fraction of the injected tracer in the vasculature at time t with R(0) = 1 and R(∞) = 0; and

⨂

indicates the convolution operation. Since capillary hematocrit level values have been reported to be 74% by Cremer et al.³⁸ and 81% by Levin et al.³⁹ of systemic hematocrit levels,

H_{f}

is assumed to be 1.3.

Then, CBV (unit of ml blood/g tissue) can be quantified from the area under the curve (AUC) of $Δ R_{2}^{*} (t)$ and the integration of AIF as follows:

CBV = \frac{H_{f}}{ρ} \frac{\int C (t)}{\int C_{a} (t)} = \frac{H_{f}}{ρ} \frac{\int {Δ R}_{2}^{*} (t)}{\int {Δ R}_{2, a}^{*} (t)}

(3)

where

{Δ R}_{2, a}^{*} (t)

Δ R_{2}^{*} (t)

of arterial blood.

CBF (unit of ml blood/g tissue/min) can be obtained using equation (2) by deconvoluting the measured tissue concentration–time curve $Δ R_{2}^{*} (t)$ with AIF. Here, a singular value decomposition method was used for deconvolution of the concentration–time curve.¹⁵ Note that $Δ R_{2} (t)$ should be used instead of $Δ R_{2}^{*} (t)$ in equations (1) to (3) for microvessel-sensitive spin-echo BOLD measurements. Using the central volume theorem,^40,41 the mean transit time (MTT) (in seconds) can be calculated as MTT = CBV/CBF.

Determination of arterial input function

Accurate determination of AIF is important for the quantification of cerebral hemodynamics. However, extracting AIF of pure arterial intravascular blood signals is difficult because of the partial-volume effect with a smaller-than-voxel arterial vessel diameter.^42–46 Thus, a large blood-vessel voxel with the largest value in the area under the ${Δ R}_{2}^{*} (t)$ curve in the brain region was selected, assuming that ${Δ R}_{2, LV}^{*} (t)$ originates only from the blood pool. Because the time integrals of the AIF and a large blood vessel should be identical, the time integral of the arterial voxel $Δ R_{2}^{*} (t)$ was scaled to match that of the large-vessel voxel, such that:

C_{a} (t) = k {Δ R}_{2, a}^{*} (t) = k {Δ R}_{2, art}^{*} (t) \frac{\int {Δ R}_{2, LV}^{*} (t)}{\int {Δ R}_{2, art}^{*} (t)}

(4)

where

{Δ R}_{2, art}^{*} (t)

and

{Δ R}_{2, LV}^{*} (t)

represent the hypoxia-induced

{Δ R}_{2}^{*}

curve of the arterial and large blood-vessel voxels, respectively.

Determination of relative CBV with iron oxide contrast agent

To examine the assumptions that i) CBV change is minimal during transient hypoxia and ii) large vessel voxels contain only intravascular signals, and to compare baseline CBV values determined by hypoxia vs. contrast agents, exogeneous susceptibility contrast monocrystalline iron oxide nanoparticle (MION) (Feraheme, AMAG Pharmaceuticals, USA; diluted with saline for an accurate injection dose) was injected into blood. Transient hypoxia experiments were performed without and with various doses of MION. The transverse relaxation rate change by MION under hypoxia is related to absolute CBV change due to hypoxia as:⁴⁷

Δ R_{2, MION + Hyp}^{*} = α (CBV + Δ CBV {(t)) χ}_{MION} = R_{2, MION}^{*} + Δ R_{2, MION}^{*} (t)

(5)

where α is a constant related to imaging and vascular parameters, and

χ_{MION}

is the mean magnetic susceptibility of MION.

R_{2, MION}^{*}

is the change in transverse relaxation rates by MION at baseline, which corresponds to relative CBV index, while

Δ R_{2, MION}^{*} (t)

is the MION-induced relaxation change under hypoxia without the hemoglobin-based effect. Then, the relative blood volume change by hypoxia can be calculated from baseline

R_{2, MION}^{*}

and evoked

Δ R_{2, MION}^{*} (t)

Animal preparation and physiological monitoring

All animal experiments were approved by the Institutional Animal Care and Use Committee of Sungkyunkwan University, and were conducted in compliance with the Animal Welfare Act (AWA), National Institutes of Health Guide for the Care and Use of Laboratory Animals, and ARRIVE guidelines 2.0 (Animal Research: Reporting in Vivo Experiments).⁴⁸ Twenty-eight male mice (C57BL/6, 25–30 g, ages 2–5 months old, Orient Bio, Korea) were used. Mice were initially sedated using an intraperitoneal (IP) injection of a mixture of ketamine (100 mg/kg) and xylazine (10 mg/kg), and then maintained with a subsequent IP bolus injection of 25 mg/kg of ketamine and 1.25 mg/kg of xylazine at approximately every 45 min as needed.^49,50

During the experiment, the heart rate and peripheral oxygen saturation (S_pO₂) values were monitored and recorded using an MR-compatible optical pulse oximeter (SA Instruments, USA) attached to the tail. A rectal temperature probe and respiration pad were used to record the temperature and respiration rate, respectively. End-tidal CO₂ was monitored using a multi-parameter physiological monitor (LifeWindow LW9x, Digicare Biomed Tech, USA). Physiological data were monitored using a physiological monitoring system (Model 1030, Small Animal Instruments Inc., USA) and recorded using a data acquisition system (Biopac Systems, Inc., USA). The temperature of the mice was maintained at 37 ± 0.5°C using a circulating warm water blanket.

Transient hypoxia paradigm for perfusion measurements

Figure 1(a) shows a schematic of the experimental setup for the respiratory stimulation MRI. A three-channel programmable gas mixer (GSM-3 Gas Mixer, CWE, Inc. Ardmore. PA 19003 U.S.A) synchronized with the MRI scanner was used to precisely control the composition of output gas mixture from 100% N₂ and 100% O₂ input gases with in-house MATLAB-based software. The delivery time from the gas mixer to the mouse is dependent on the length and diameter of the tubing and the gas flow rate (1 L/min). The gas transfer delay from the gas mixer to the nose cone of the cradle was 5.51 ± 0.24 s, as measured by the CO₂ sensor, and was corrected for data analysis. Transient hypoxia was induced by reducing the fraction of inspired oxygen (FiO₂) from 44% to 0%, typically for 5 s. In our all experiments, a 44% FiO₂ condition was set as the baseline.

Figure 1.

Experimental setup and anoxia duration-dependent responses. (a) Schematic of experimental setup. 100% N₂ and 100% O₂ gases were mixed by a computer-controlled gas mixer and delivered at a flow rate of 1 L/min to a mouse’s nose cone, which was attached to the home-built cradle. (b) The experimental procedure of anoxia stimulus with injections of 1 mg/kg MION for determining relative CBV changes. (c,d) BOLD-MRI responses to 3 s, 5 s, and 10 s anoxic stimulation in the whole cerebral cortex. Negative response was observed due to an increase in paramagnetic deoxyhemoglobin, and its peak response was −6.6 ± 0.1% (n = 5 mice; red), −8.2 ± 0.3% (n = 8; green) and -10.9 ± 0.8% (n = 3; blue) for 3 s, 5 s and 10 s stimulus, respectively. (e,f) Relative CBV changes of the entire cortex responding to 3 s, 5 s, and 10 s anoxic stimulation, computed from the difference between relaxation rate changes due to anoxia with different doses of MION (see Supplementary Figure 1 for details). Three combinations were used for checking consistency (e: no vs. 1 mg/kg MION, black; no vs. 2 mg/kg MION, yellow; 1 vs. 2 mg/kg MION, pink) and then averaged (f). The CBV increase was 9.4 ± 1.3% and 11.2 ± 1% during 3 s (n = 5; red) and 5 s anoxia (n = 8; green), respectively, whereas a larger and prolong CBV change was observed during 10 s anoxic stimulus (n = 3; blue). Gray shade or color bars represent the stimulation period. For clarity, only one-side error bars (standard deviation) are presented.

Arterial blood gas analysis

Arterial blood gas analysis (ABGA) was performed using various FiO₂ settings (n = 6 mice). The femoral artery was catheterized by a catheter made using a PE-10 tube. Mice were placed on a cradle with the same setting for MRI experiments with added ECG monitoring. 0.95 $μ l$ of arterial blood was withdrawn under three different steady-state respiratory conditions, including normoxia (FiO₂ 44%; 30% O₂/70% air) and mild hypoxia (FiO₂ 20% and 10%). Arterial blood gases (pO₂ and pCO₂), pH, and HCO₃ were measured using an i-STAT CG8+ portable clinical analyzer (Abbott Point of Care, USA). Arterial oxygen saturation (S_aO₂) levels were determined from the pO₂ values using the mouse oxyhemoglobin dissociation curve.^51,52

MRI acquisitions

All MRI experiments were performed on a Bruker Biospec 9.4 T/30 cm horizontal bore instrument with an actively shielded 12.0 cm diameter insert. The mouse brain was positioned as close as possible to the isocenter of the magnet. A quadrature birdcage coil (86 mm inner diameter) was used for excitation, and an actively decoupled Bruker planar surface coil (10 mm inner diameter) positioned on top of the mouse head was used for detection. T₂*-weighted single-shot gradient-echo echo-planar images (GE-EPI) were acquired for total vascular weighted perfusion mapping with the following parameters: repetition time (TR)/echo time (TE) = 1000/11 ms, flip angle = 50°, spatial resolution = 156 × 156 × 500 μm³, 20 continuous slices with interleaved order, and dummy scans of 10. Alternatively, T₂-weighted single-shot spin-echo echo-planar images (SE-EPI) were acquired for microvascular perfusion mapping with the same parameters, except TE = 22 ms and flip angles = 90/180°.

To characterize BOLD responses to hypoxic stimulation and to quantify perfusion values, three experimental designs were used: 1) duration of anoxic stimulation, 2) determination of arterial oxygen saturation change, and 3) quantification of regional perfusion values. The first two experiments were performed with GE-EPI, while the last experiments were performed with GE-EPI and SE-EPI.

Experiment #1. Dependence of hypoxia stimulus duration on vascular physiology (n=10 mice). Here, three stimulus durations of 3, 5, and 10 s were examined. On each animal, the following anoxia stimulus block design was applied: 20 s baseline – 3 s (5 s) stimulation – 40 s baseline – 5 s (10 s) stimulation – 40 s baseline. Two different stimulus lengths (3 s and 5 s, n = 7 mice; 5 s and 10 s, n = 3 mice) were used in each trial. To examine underlying assumptions of our model, relative CBV changes were determined. A dose of 1 mg Fe/kg body weight of MION solution was introduced through the tail vein three times (Figure 1(b)). Five to seven hypoxic stimulus runs were performed for each condition.

Experiment #2: Estimation of arterial oxygen saturation level (S_aO₂) during transient 5s anoxia. The BOLD contrast was used to determine oxygen saturation changes. For comparison with arterial blood gas analyses measured at a steady state (see the “Arterial blood gas analysis” section), 50 s stimulation of 10% and 20% FiO₂ hypoxia was used to obtain steady-state BOLD MRI in addition to a 5 s transient hypoxia (n = 5 mice used in Expt. #1). The experimental paradigm consisted of 20 s baseline – 5 s anoxia – 60 s baseline – 50 s of 10% FiO₂ – 50 s of 20% FiO₂ – 45 s baseline. Five to seven runs were conducted.

Experiment #3: Quantification of CBF and CBV. To obtain high-quality perfusion images, a 5 s anoxic stimulus was used. Five to ten runs of GE-EPI BOLD MRI were performed on 15 mice (including 10 mice used in Expt. #1) before and after 3–5 mg/kg MION for comparing baseline CBV values determined by DSC BOLD (equation (3)) vs. steady-state MION (equation (5)). In a separate group (n = 7 mice), 20 to 30 runs were repeated for SE-EPI collections without MION.

Data processing

Physiological data analysis

The recorded time-domain heartbeat and respiratory signals were transformed into the frequency domain through fast Fourier transformation. Then, the maximum value in the 120–360 bpm range was selected to obtain the heart rate (bpm, beats per min) from the pulse-oximetry pulse data and the respiration rate (bpm, breaths per min) from the motion-sensitive respiration signals. Dynamic pulse and respiration rates were averaged over repeated MRI trials in the same imaging session, resulting in averaged dynamic pulse and respiration rates for each subject.

Overall BOLD-MRI analysis

All BOLD data for each mouse were analyzed using MATLAB (MathWorks, USA), the Analysis of Functional Neuroimages (AFNI) package,⁵³ FMRIB Software Library (FSL),⁵⁴ and Advanced Normalization Tools (ANTs).⁵⁵

For individual EPI images, the following preprocessing steps were performed: slice timing correction, motion correction, linear detrending for signal drift removal, and trial averaging within the same imaging session. Spatial smoothing using a Gaussian kernel with a 156 μm full-width half-maximum (FWHM) was included to minimize potential misalignment and improve statistical power. All repeated BOLD-MRI runs in the same experimental condition were averaged in each animal. The preprocessed EPI images (156 × 156 × 500 μm³) were co-registered to the high-resolution anatomical T₂-weighted image (78 × 78 ×500 μm³). Then, BOLD images were linearly interpolated 2 times, leading to a nominal resolution of 78 × 78 μm². The anatomical T₂-weighted image was spatially co-registered with the mouse brain atlas from the Allen Mouse Brain Atlas⁵⁶ using both affine and nonlinear transformations. The Allen Mouse Brain Atlas was used to define thirty right- and left-hemisphere ROIs covering most of the brain regions, and cortical layer ROIs in the primary somatosensory cortex for the cortical depth-dependent perfusion values.

Determination of dynamic CBV and S_aO₂ change by transient hypoxia (Expt. #1 – #2)

In 10 mice (Expt. #1), two mice were excluded from the analysis due to incomplete MION studies: one died after the 2nd MION injection and the other did not receive the appropriate MION dose. Relative CBV time courses in response to transient hypoxia were calculated in the whole cortex ROI of each animal with equation (5).⁴⁷

To determine the S_aO₂ change from MRI data (Expt. #2), hypoxia-induced $Δ R_{2}^{*}$ of the entire cortex at a steady-state condition (i.e., 46-50 s after the onset of 50 s stimulus) was determined for 10% and 20% FiO₂ hypoxia relative to baseline 44% FiO₂, then correlated with corresponding S_aO₂ changes. Peak $Δ R_{2}^{*}$ of 5 s anoxia was determined from 3 data points around the peak, then $Δ$ S_aO₂ was determined from the relationship between $Δ$ S_aO₂ and $Δ R_{2}^{*}$ at a steady state.

Estimation of corrected arterial input function

Candidate vascular voxels were selected based on a two-group hierarchy clustering method,⁵⁷ which clustered the vascular and tissue voxels in the first iteration. i) The voxel with the largest AUC value among the blood-vessel candidate voxels was selected for ${Δ R}_{2, LV}^{*} (t) .$ The ${Δ R}_{2}^{*}$ curves before and after MION injection were compared to confirm the assumption that the intravascular contribution of the BOLD signal was predominant in the large-vessel voxel (12 out of 15 animals were used from Expt. #3: data from three mice were excluded because of incomplete MION studies). ii) One arterial-like voxel was chosen for ${Δ R}_{2, art}^{*} (t)$ with arterial response behaviors such as a single-peaked shape, the fastest arrival time, time to peak at the end of hypoxic stimulation, and the largest wash-in and wash-out rates. The location of arterial-like voxels was examined using a cerebral vascular atlas based on a combination of MRI and computed tomography.⁵⁸ The area under the $Δ R_{2}^{*}$ curve of the arterial-like voxel was adjusted using equation (4). The adjusted arterial $Δ R_{2}^{*}$ time course of each animal was fitted using a gamma-variate function to reduce signal fluctuations and defined as the corrected AIF ( ${Δ R}_{2, a}^{*} (t))$ for the perfusion analysis.

Determination of quantitative perfusion values (Expt. #3)

Dynamic BOLD data of 5 s anoxic stimulus without MION were used for quantification of CBF and CBV. GE-EPI BOLD-DSC MRI in 15 animals was used for the quantification of total vascular-weighted perfusion with subject-specific AIF, while SE-EPI BOLD-DSC MRI in 7 animals was used to quantify microvascular perfusion values (mv-CBV and mv-CBF). Since SE-EPI AIF could not be obtained from arterial vessels due to the signal loss by flow-induced dephasing, and intravascular ${Δ R}_{2}^{*}$ and $Δ R_{2}$ are expected to be similar in the motional narrowing regime,³¹ the average corrected AIFs determined from GE-EPI data were used as AIF. In each animal, perfusion values were determined using equations (2) and (3) on a ROI-by-ROI and voxel-by-voxel basis. Mean perfusion maps were obtained by averaging the voxel-wise values of all animals.

Subsequently, multiple comparisons were also performed. 1) To check the consistency of perfusion measurements, perfusion metrics of the right- and left-hemisphere ROIs were compared. Bland-Altman plots were generated as percent differences between right and left ROI values vs. means of right and left ROI values. 2) Regional perfusion values determined using GE-BOLD MRI and SE-BOLD MRI were compared. 3) Regional quantitative perfusion values obtained from GE-BOLD MRI (CBV and CBF) or microvascular SE-BOLD MRI (mv-CBV and mv-CBF) were correlated with experimental parameters (∫ ${Δ R}_{2}^{*}$ and max ${Δ R}_{2}^{*}$ or ∫ $Δ R_{2}$ and max $Δ R_{2}$ ). 4) In the 12 mice with successful MION studies, the total CBV values obtained from GE-BOLD MRI (without MION) were compared with the relative CBV indices of $R_{2, MION}^{*}$ obtained from the pre- and post-MION images (without the hypoxic challenge).

Statistical analysis

Standard normal distributions of perfusion values were confirmed by one-sample Kolmogorov-Smirnov test. Pearson’s correlation (r) was used to measure the degree of the relationship between two linearly related perfusion values. The significance was considered according to the p-value <0.05. All results are presented as the mean ± standard deviation (SD).

Results

Choice of stimulus duration

A proper stimulation duration was necessary to obtain sufficient sensitivity and AIF. Initially, 3 s, 5 s, and 10 s anoxic stimuli were used (Figure 1). During anoxic stimulation, the magnitude of decrease in BOLD signal in the cerebral cortex increased for longer stimulations (Figure 1(c)). The 3 s, 5 s and 10 s stimuli induced a negative peak change of −6.6 ± 0.1% (n = 5 mice; red), −8.2 ± 0.3% (n = 8; green), and −10.9 ± 0.8% (n = 3; blue), respectively (Figure 1(d)). For 10 s stimulation, a post-stimulus overshoot was observed.

To investigate the impact of brief anoxic stimulus on neurovascular physiology, relative CBV changes in the entire cortex were measured with different MION doses (Figure 1(e) and (f) and Supplementary Figure 1 for anoxia-evoked responses at three different MION levels). Since the contrast between the 2 mg/kg and 3 mg/kg MION doses was small (see Supplementary Figure 1), the 3 mg/kg MION data were not included in the CBV calculation. Relative CBV changes measured using different combinations are quite similar, thus averaged (Figure 1(e)). The 3 s and 5 s anoxic stimuli induced an average CBV increase of 9.4 ± 1.3% and 11.2 ± 1.0%, respectively (Figure 1(f)), and quickly returned to the pre-stimulus baseline level. The 10 s anoxia stimulus induced a prolonged increase in CBV. Although both 3 s and 5 s stimuli induced similar CBV changes, the latter was chosen for obtaining more data points of larger BOLD responses in further perfusion studies.

Systemic physiology changes responding to 5 s anoxic stimulus

Next, we determined the oxygen saturation level change induced by the 5 s anoxia. BOLD-DSC MRI signals (yellow) were acquired with S_pO₂ responses (blue) at different FiO₂ levels (green) of 10%, 20%, and 44% in addition to 5 s anoxia (Figure 2(a)). S_pO₂ followed BOLD-MRI responses but with a delay of a few seconds (Figure 2(b)) $.$ However, the low S_pO₂ value peripherally measured in the tail might reflect a change in the concentration of deoxyhemoglobin in arterial and venous blood, poor reliability, and their dependence on the sensor site.^59–61 Therefore, arterial blood was sampled under steady-state conditions of 10%, 20%, and 44% FiO₂ (Supplementary Table 1). The S_aO₂ values were 35.6 ± 5.9%, 74.3 ± 7.8%, and 97.7 ± 0.8% at FiO₂ of 10%, 20%, and 44%, respectively (Figure 2(c)).

Figure 2.

Physiological changes induced by 5 s anoxic stimulation. (a,b) BOLD-MRI signals (yellow) and peripheral oxygen saturation responses measured in tail (blue; S_pO₂) were acquired during different FiO₂ levels (green) of 10%, 20%, and 44% in addition to 5 s anoxia. The red rectangle section in Figure 2(a) has been expanded in Figure 2(b). The S_pO₂ response followed BOLD MRI responses but with a few seconds of delay. Upward red arrows represent arterial blood sampling periods. (c) Box plot of arterial blood oxygenation levels measured at steady-state condition of 10%, 20%, and 44% FiO₂ (left) and ${Δ R}_{2}^{*}$ vs. ΔS_aO₂ induced by FiO₂ of 10% and 20% from FiO₂ of 44% (right). Note that arterial blood gases were reported in Supplementary Table 1. The 5 s anoxia (FiO₂ of 0%) induced a peak ${Δ R}_{2}^{*}$ of 8.1 ± 0.5 s⁻¹ (n = 5), resulting in S_aO₂ of 59.0 ± 7.1% (dashed line). Red squares represent mean, while red dots represent median. (d) Heart rate (red) and respiratory rate change (yellow) responding to the 5 s anoxic stimulation. Gray shade represents the stimulation period. For clarity, only one-side error bars (standard deviation) are presented.

As BOLD-DSC MRI signals reflect changes in cerebral oxygen saturation,^22,25,32,33 ${Δ R}_{2}^{*}$ values due to several respiratory challenges of anoxia, apnea, and hypercapnia were used to determine oxygen saturation changes. ${Δ R}_{2}^{*}$ values induced by FiO₂ from 44% to 10% and FiO₂ from 44% to 20% were 12.6 ± 2.1 s⁻¹ and 5.3 ± 1.5 s⁻¹, respectively (Figure 2(c)), and 5 s anoxia induced a peak ${Δ R}_{2}^{*}$ of 8.1 ± 0.5 s⁻¹. Since ΔS_aO₂ to hypoxic challenge at a steady state is linearly correlated with ${Δ R}_{2}^{*}$ of BOLD MRI (see the linear equation ΔS_aO₂ = 5.27 ${Δ R}_{2}^{*}$ − 4.47), the 5 s anoxia induced a decrease in S_aO₂ to 59.0 ± 7.1%. Additionally, the 5 s anoxic stimulus increased both the heart rate and self-breathing respiratory rate (Figure 2(d)). The heart rate (red) increased from 210 ± 20 bpm prior to hypoxic stimulation to 244 ± 12 bpm during 5 s of hypoxia, and the breathing rate (yellow) also increased from an average of 181 ± 16 bpm to 215 ± 19 bpm.

Determination of corrected arterial input function

To quantify CBF and CBV accurately, it is critical to obtain an AIF. First, one large-vessel- and one arterial-like voxel were identified for each animal (blue and red dots in the images in Figure 3(a) and (c) and Supplementary Figure 2). The ${Δ R}_{2}^{*}$ curves of large vessels showed varying peak intensity, duration, time to peak, and wash-in and wash-out rates (Figure 3(a)). To evaluate the assumption that the large-vessel voxels only contain vessels, MION (3-5 mg/kg) was injected to suppress the intravascular signal contribution. The area under the large-vessel curves decreased by 88.7 ± 8.3% (n = 12 mice) after MION injection, and the mean signal change to 5 s anoxic stimulation was not statistically different from 0% (Figure 3(b)), indicating that our assumption is valid. Arterial-like curves from 15 mice had similar curve shapes but showed variance in peak intensities (Figure 3(c)), whereas corrected AIFs showed similar peak intensities and curve shapes among 15 animals (Figure 3(d)). This corrected AIF was used to quantify CBF and CBV.

Figure 3.

Determination of animal-wise arterial input function with the partial volume correction. The large-vessel and arterial-like voxels were found based on the area under the curve (AUC) and characteristics of dynamic ${Δ R}_{2}^{*}$ response to a 5 s anoxic stimulation (see Supplementary Figure 2 for 15 mice). (a,b) Animal-wise ${Δ R}_{2}^{*}$ responses of the large-vessel voxel without (a) and with 3–5 mg/kg monocrystalline iron oxide nanoparticle (MION) (b). After intravenous injection of 3–5 mg Fe/kg MION, the area under the mean ${Δ R}_{2}^{*}$ response curve decreased by 88.7 ± 8.3% (n = 12) and was not statistically different from 0% (inset box plot), indicating that the intravascular contribution is predominant in the large-vessel voxel. The red time course represents the average of all subject’s data, the blue voxel in inset EPI image represents large vessel, and the green bar represents a 5 s anoxic stimulus. (c,d) Animal-wise ${Δ R}_{2}^{*}$ response of the arterial-like voxel (c) and the gamma-variate fitted curves (corrected AIF) of partial volume-corrected arterial-like voxel responses (d). To compensate the partial-volume effect of the arterial-like ${Δ R}_{2}^{*}$ response curve, the area under the arterial-like curve was adjusted to the area under the large-vessel curve. Red voxel in inset EPI image represents arterial-like vessel.

Regional total vs. microvascular cerebral perfusion values of mice

To obtain total and microvascular perfusion values, hypoxia-induced GE-BOLD MRI in 15 animals (92 runs with 5-10 runs per animal) and SE-BOLD MRI in 7 animals (175 runs with 20-30 runs per animal) were used, respectively. Thirty bilateral ROIs were selected based on the Allen Mouse Brain Atlas (Figure 4(a)). As examples, regional GE-BOLD MRI and SE-BOLD MRI responses to 5 s anoxic stimulation are shown (Figure 4(b) for the somatosensory area; see Supplementary Figure 3 for representative regions). CBF and CBV values were quantified by modeling BOLD-DSC responses with the corrected AIF (Supplementary Figure 4). Regional perfusion values between the left and right hemispheres were subsequently examined. Total perfusion metrics (CBV and CBF) showed more variation between the two hemispheres than microvascular perfusion metrics (mv-CBV and mv-CBF), likely due to the asymmetric large vasculatures between the two hemispheres (Figure 4(c) and (d) and Supplementary Figure 5 for Bland-Altman plots). These regional total and microvascular perfusion values were summarized in Table 1. Microvascular perfusion values were lower than the total perfusion values, but both total and microvascular perfusion values are highly correlated (Figure 4(e)). The magnitude of reduction using SE-BOLD relative to GE-BOLD MRI depends on the macrovascular contributions. Microvascular perfusion value was close to total perfusion in white matter corpus callosum (see also Supplementary Figure 3 for BOLD-DSC responses) but only to 40-50% of total perfusion rates in the dentate gyrus and medulla.

Figure 4.

ROI analysis of gradient-echo and spin-echo BOLD-MRI responses to 5 s anoxic stimulus. Dynamic gradient-echo (GE; n = 15 mice) and spin-echo (SE; n = 7 mice) BOLD MRI were performed for measuring total and microvascular perfusion values, respectively. For each animal, 5–10 runs were averaged for GE BOLD and 20–30 runs for SE BOLD. (a) Thirty regions of interest (ROI) were defined based on the Allen Mouse Brain Atlas with anterior-posterior coordinates relative to bregma. (b) Regional GE- and SE-BOLD MR signal responses were obtained from the somatosensory ROI, where black and red lines represent individual and mean signal curves, respectively (see Supplementary Figure 3 for other representative ROIs). Cerebral blood flow (CBF) and cerebral blood flow (CBV) were quantified by modeling BOLD-DSC responses with the corrected AIF. The perfusion ratio of GE- to SE-BOLD is 1.1–1.3 for the somatosensory area. The inset image shows the ROI, while the green bar represents 5 s anoxic stimulus. (c,d) Perfusion values between left and right hemispheres were examined. Bland-Altman plots of right and left ROI values are shown in Supplementary Fig. 5. Each data point was derived from the same-colored ROI in each subject. Total vascular perfusion values (CBV and CBF) have more variations between the two hemispheres than microvascular perfusion metrics (mv-CBV and mv-CBF), likely due to the unsymmetric large vasculatures between the two hemispheres. (e) Microvascular perfusion values are highly correlated with total vascular values. Each data point represents a mean perfusion value in the same-colored ROI, and its values can be found in Table 1. CC: corpus callosum; MB: midbrain; TH: thalamus; HY: hypothalamus; P: pons; MY: medulla; CBX: cerebellar cortex; CBN: cerebellar nuclei; STR: striatum; PAL: pallidum; OLF: olfactory areas; VISC: visceral area; VIS: visual areas; TEa: temporal association areas; SS: somatosensory areas; RSP: retrosplenial area; PTLp: posterior parietal association area; PL: prelimbic area; PERI: perirhinal area; ORB: orbital area; MO: somatomotor areas; ILA: infralimbic area; GU: gustatory areas; ECT: ectorhinal area; AUD: auditory areas; AI: agranular insular area; ACA: anterior cingulate area; SUB: subiculum; DG: dentate gyrus; CA: cornu ammonis.

Table 1.

Regional total and microvascular CBV and CBF of ketamine/xylazine-anesthetized mouse.

	CBV (ml/100 g)		CBF (ml/100 g/min)
	Total^a	Micro^b	Total^a	Micro^b
Whole brain	3.39 (0.59)	2.23 (0.27)	112.0 (15.0)	85.8 (6.9)
Cerebral cortex
Somatomotor (MO)	3.35 (0.64)	2.53 (0.35)	110.4 (16.3)	95.8 (10.0)
Somatosensory (SS)	3.20 (0.64)	2.53 (0.37)	107.2 (11.9)	97.8 (10.2)
Gustatory (GU)	3.15 (0.60)	2.09 (0.32)	110.1 (16.7)	79.5 (8.0)
Visceral (VISC)	2.75 (0.58)	1.77 (0.25)	86.7 (18.2)	65.3 (7.9)
Auditory (AUD)	2.87 (0.54)	1.99 (0.18)	92.9 (10.5)	75.3 (4.7)
Visual (VIS)	2.71 (0.53)	2.20 (0.33)	87.4 (12.1)	84.3 (7.5)
Anterior cingulate (ACA)	4.51 (0.82)	2.46 (0.31)	154.1 (31.2)	92.9 (10.2)
Prelimbic (PL)	3.23 (0.70)	2.20 (0.40)	111.9 (29.3)	84.5 (11.5)
Infralimbic (ILA)	5.44 (0.83)	2.29 (0.42)	186.0 (33.1)	91.3 (14.6)
Orbital (ORB)	3.62 (0.65)	2.32 (0.34)	123.0 (22.1)	87.5 (8.4)
Agranular insular (AI)	3.34 (0.62)	2.08 (0.29)	111.1 (17.2)	77.0 (8.9)
Retrosplenial (RSP)	3.84 (0.63)	2.86 (0.39)	128.1 (26.2)	110.5 (10.9)
Posterior parietal association (PTLp)	2.91 (0.51)	2.50 (0.41)	94.4 (15.5)	95.2 (10.7)
Temporal association (TEa)	2.55 (0.49)	1.60 (0.20)	83.1 (19.7)	60.6 (6.5)
Perirhinal (PERI)	2.79 (0.69)	1.63 (0.34)	92.4 (19.6)	59.9 (11.7)
Ectorhinal (ECT)	2.60 (0.52)	1.46 (0.15)	84.1 (16.1)	55.3 (7.1)
Olfactory areas (OLF)	3.44 (0.58)	1.87 (0.36)	118.9 (19.3)	68.8 (10.5)
Hippocampal formation
Ammon's horn (CA)	2.24 (0.38)	1.66 (0.19)	73.2 (15.3)	62.7 (4.7)
Dentate gyrus (DG)	5.12 (0.90)	2.30 (0.30)	171.9 (37.4)	88.6 (10.1)
Subiculum (SUB)	3.63 (0.75)	2.07 (0.25)	126.7 (33.1)	79.5 (9.2)
Cerebral nuclei
Pallidum (PAL)	2.98 (0.59)	1.51 (0.21)	99.7 (18.5)	59.5 (7.0)
Striatum (STR)	2.81 (0.47)	1.86 (0.21)	93.8 (14.1)	69.2 (5.6)
Thalamus (TH)	3.66 (0.65)	2.42 (0.35)	122.5 (18.0)	97.1 (11.1)
Hypothalamus (HY)	3.14 (0.61)	1.69 (0.24)	100.9 (22.4)	65.5 (9.3)
Brain stem
Medulla (MY)	2.84 (0.67)	1.11 (0.55)	90.5 (20.9)	45.2 (21.8)
Pons (P)	2.75 (0.77)	1.74 (0.21)	88.4 (15.8)	67.3 (8.1)
Midbrain (MB)	4.88 (1.04)	2.70 (0.34)	157.5 (22.1)	106.5 (8.8)
Cerebellar cortex (CBX)	3.66 (1.52)	2.31 (0.27)	117.2 (33.9)	89.5 (8.6)
Cerebellar nuclei (CBN)	3.34 (0.51)	1.76 (0.68)	108.5 (32.8)	66.0 (21.8)
Corpus callosum (CC)	1.76 (0.31)	1.46 (0.17)	55.6 (9.6)	53.7 (4.6)

^aTotal-vascular CBV and CBF were obtained from dynamic gradient-echo BOLD MRI responding to 5 s anoxic stimulation (n = 15 mice). Perfusion values were quantified by modeling BOLD DSC with subject-specific AIF, and reported as means (standard deviations in parentheses). The ROIs were determined based on the Allen Atlas.

^bMicrovascular CBV and CBF were obtained from dynamic SE-BOLD MRI responding to 5 s anoxic stimulation with an averaged AIF from GE-BOLD MRI (n = 7 mice).

In the case where the AIF is not properly determined, the ${Δ R}_{2}^{*}$ or $Δ R_{2}$ response curve to hypoxia could be directly used for determining relative total and microvascular perfusion parameters, since AUC (∫ ${Δ R}_{2}^{*}$ or ∫ $Δ R_{2}$ ) and peak intensity (max ${Δ R}_{2}^{*}$ or max $Δ R_{2}$ ) were well correlated with quantitative CBV or mv-CBV and CBF or mv-CBF values, respectively (see Supplementary Figure 6).

Cerebral perfusion maps obtained from GE- and SE-BOLD MRI

The mean total perfusion maps of the 15 mice were generated from GE-BOLD MRI (Figure 5(a) and (c)), whereas the mean microvascular perfusion maps of the 7 mice were generated from SE-BOLD MRI (Figure 5(b) and (d)). CBV and CBF maps showed similar contrast in most brain regions for both total- and micro-vasculature, but high perfusion values obtained from GE-BOLD MRI (red arrows) were considerably reduced by the use of SE-BOLD MRI. The differences in perfusion values in the cortical, white matter (green arrowhead), and hippocampal regions were clearly distinguished. The ventricle area (black arrowhead) had no perfusion as expected.

Figure 5.

Whole-brain perfusion maps of mouse under ketamine and xylazine anesthesia. Total (a,c) and microvascular (b,d) perfusion maps were obtained from dynamic gradient-echo (GE; n = 15 mice) and spin-echo (SE; n = 7 mice) BOLD-MRI responses to a 5 s anoxic stimulation, respectively. For each animal, 5–10 runs were averaged for GE BOLD and 20–30 runs for SE BOLD. Cerebral blood volume (CBV) (a,b) and cerebral blood flow (CBF) maps (c,d) have similar contrasts for both total- and micro-vasculature. Black, green, and red arrows indicate cerebrospinal fluid, white matter, and vascular areas, respectively. In the brainstem medulla area (blue arrows), perfusion values were not computed due to low signal intensities (<half of the standard deviation).

For 12 mice with 3-5 mg/kg MION, quantitative total CBV values obtained from hypoxia-induced GE-BOLD MRI (without MION) were highly correlated with relative CBV indices obtained from steady-state MRI data before and after MION injection (see Supplementary Figure 7), indicating that hypoxia BOLD-DSC MRI measures CBV reliably. Voxel-wise correlation coefficient was higher with a higher dose of MION (r = 0.70 for 5 mg/kg vs. 0.63 for 3 mg/kg).

Cortical depth-dependent perfusion

Cortical depth-dependent perfusion values were examined in the somatosensory area for determining the feasibility of high-resolution perfusion studies. Cortical layers and sub-regions in the primary somatosensory cortex were defined using the Allen Atlas (Figure 6(a)). Total CBV and CBF peaked at layer 2/3 (L2/3), while microvascular CBV and CBF peaked at layer 4 (L4) for all sub-regions (Figure 6(b)). An average perfusion ratio of combined L2/3 and L4 to L6 is 1.53 and 1.39 for GE- and SE-based measurements, respectively. This indicates that larger-diameter vessels contribute more to the upper cortical layers.

Figure 6.

Depth-dependent perfusion values of the mouse somatosensory cortex. (a) Total and microvascular CBV and CBF maps (extracted from Figure 5) were compared in the somatosensory region (see ROI described in top row and left column), which was extensively investigated by somatosensory fMRI. One hemisphere brain atlas covering the somatosensory cortex is shown in a coronal view with anterior-posterior coordinates relative to bregma. Color perfusion maps in the T₂-weighted image domain are presented with layer ROIs in the primary somatosensory cortex. Cortical layers and color-coded sub-regions in the primary somatosensory cortex were defined by Allen Atlas: lower limb, upper limb, trunk, nose, mouth, barrel field, and unassigned primary somatosensory area. (b) Total vascular CBV and CBF peaked at layer 2/3 (L2/3), while microvascular CBV and CBF peaked at layer 4 (L4). Note that the nearby sub-regions were combined to enhance the sensitivity for layer-dependent perfusion profiles. Data are represented as mean with one-side error bars (standard deviation).

Discussion

Herein, we presented a non-invasive perfusion MRI technique with dynamic BOLD contrast and short transient anoxia (5 s), allowing us to obtain quantitative CBF and CBV mapping in the entire mouse brain. Since the hypoxic approach utilizes paramagnetic deoxyhemoglobin as an endogenous tracer, it can be easily implemented in small animal studies with a gas switching system. Compared to existing exogenous contrast agent perfusion methods, the hypoxia approach has a more economical, reproducible, and repeatable tracer delivery with a much faster wash-out time. Consequently, whole-brain perfusion measurements can be repeated almost every minute. Since the newly developed perfusion MRI approach does not require any sophisticated pulse sequence, hardware, catheterization or blood gas analysis, it can be readily used to measure quantitative perfusion values and changes induced by drug treatments and vascular reactivity in health and diseases.

Anoxic stimulus and its effects to vascular physiology

The strength and duration of the hypoxic stimulus are important factors for maximizing sensitivity while minimizing physiological side effects. Our strategy was to adopt a strong and short stimulus that is sufficient to generate BOLD contrasts and maintain normal cerebral physiology. An increase in heart rate and respiration rate in response to the 5 s anoxic stimulus is likely due to the activation of peripheral chemoreceptors, which detect changes in arterial blood O₂/CO₂ and initiate reflexes in cardiorespiratory systems (respiratory rate and heart rate) for maintaining oxygen homeostasis.^62–65

Hypoxia dilates the cortical vessels, thereby enhancing the delivery of oxygen to the brain.^66,67 Anoxic stimulus duration of 10 s (or longer) violated the assumption that CBV does not change significantly, thus is not recommended. The 5 s anoxic stimulation induced a peak CBV increase of 11% in the mouse cerebral cortex, which can cause an error in absolute perfusion values up to 11%. To reduce hypoxia-induced physiology changes, a mild hypoxic stimulus can be considered. If arterial vessels used for the AIF curve are dilated similarly to cortical vessels, then overestimation would be cancelled out. Therefore, we did not attempt an error correction in our quantification. Nonetheless, the relative CBV and CBF values are well-preserved.

Errors of BOLD-DSC perfusion measurement and its translation to humans

An important issue is accuracy and precision of perfusion values measured by DSC MRI. In the Gd-DSC perfusion community, error sources have been extensively discussed.^68–70 Error sources of the proposed BOLD-DSC approach are very similar to those in Gd-DSC (e.g., SNR, temporal resolution, AIF, deconvolution, and additionally hypoxia-induced physiology changes), as both change the magnetic susceptibility in the blood vessels. One simplifying factor in our approach is that the intravascular MRI signal is minimal at ultrahigh fields.

The macrovascular contribution to GE-BOLD-based measurements leads overestimation of perfusion values. As the extravascular dephasing effect in GE BOLD is extensively described in the fMRI field,^29,31,71 it is closely dependent on the diameter and orientation of vessels and can extend to voxels proximal to the large vessels. This dephasing effect around large vessels can be refocused by the 180° radiofrequency pulse in SE-BOLD MRI. By combining GE and SE measurements, a weighted vessel size index can be determined, which is highly useful for angiogenesis research.⁷²

Translation of the BOLD-DSC method to humans should be cautious, even if it has been used in humans with ∼25 s anoxia stimulus at 3 T (inducing −7.9% in the gray matter).²⁴ The underlying assumption of our BOLD model is that the intravascular contribution is minimal, which is satisfied at ultrahigh fields. When the BOLD-DSC approach is applied at low fields, the dynamic BOLD model should include the intravascular BOLD contribution.⁷³ In human BOLD-DSC applications, the AIF is delayed and dispersed, unlike a bolus injection of contrast agent, causing additional errors of quantification. Thus, quantification of perfusion in humans is likely complicate.

Regional CBV and CBF values of mouse under ketamine and xylazine anesthesia

Cerebral hemodynamics are sensitive to the mouse strain, age, sex, and anesthetics used. Thus, comparisons with previously reported values should be performed cautiously. Here, we chose quantitative perfusion values obtained from C57BL/6 mice under ketamine/xylazine anesthesia. Quantitative CBF measured with gold-standard C¹⁴-iodoantipyrine autoradiography⁷⁴ is consistent with our data: 100 ml/100 g/min in the somatosensory cortex (vs. 107 ml/100 g/min in our GE-BOLD data), 121 ml/100 g/min in the thalamus (vs. 123), and 78 ml/100 g/min in the corpus callosum (vs. 56). When CBF was measured with ASL MRI, Zuloaga et al.⁹ reported ∼120 ml/100 g/min in the cortex and ∼130 ml/100 g/min in the hippocampus.

Quantitative total CBV and microvascular CBV with <100 μm diameter vessels were measured using micro-computed tomography (CT) by filling the vasculature with a radio-opaque silicon rubber.⁷⁵ Average difference between total CBV and microvascular CBV is approximately 30%, but it is as much as 70% in the cortex. In our study, average difference between GE-based total CBV and SE-based microvascular CBV was 20–60% in the cortex, hippocampus, thalamus, and corpus callosum. Total CBV measured using micro-CT was 7.9% (vs. 3.6 ml/100 g in our study) in the cerebral cortex, and 3.7% (vs. 3.7 ml/100 g) in the thalamus, and 3.1% (vs. 1.8 ml/100 g) in the corpus callosum. Since large vessels near the surface of the cortex are not easily visible in GE-EPI images (see Figures 5 and 6), our GE-based CBV in the cortex is apparently similar to microvascular CBV of 4.6% in micro-CT studies. In general, CBV values measured using BOLD DSC with hypoxia agree well with those measured using micro-CT⁷⁵ and steady-state MION MRI studies (see Supplementary Figure 7).

Depth-dependent perfusion characteristics in the cortical somatosensory area

In depth-dependent perfusion mapping, the highest microvascular CBF and CBV peaks occurred at L4, while the highest total CBF and CBV peaks occurred at L2/3. This can be explained by the vessel-size-dependent laminar distribution;⁷⁶ the total vascular volume showed a peak at the cortical surface due to large pial vessels and decreased with an increase in cortical depth, while the microvascular volume density (diameter <6 µm) was the highest at L4 in the primary sensory area.⁷⁶ In the mouse barrel cortex, penetrating arterioles and venules had median diameters of 11 µm and 9 µm, respectively.⁷⁷

Limitation and alternatives of BOLD DSC with transient anoxia stimulus

Although the BOLD-DSC approach can be easily implemented, a computer-controlled gas switching system is needed. In our setup, three gases can be mixed as needed, thus CO₂ gas can be added for vascular reactivity studies. When volatile anesthesia is used, it is critical to ensure stable or no anesthesia during hypoxic stimulation. Since the short duration without anesthesia may not change perfusion much, 5 s anoxia without anesthesia may be used as a stimulus. In our study, GE- and SE-based perfusion maps were separately obtained. Thus, future studies can use a combination of spin- and gradient-echo EPI (SAGE EPI) proposed by Schmiedeskamp et al.^78,79 for obtaining both total and microvascular perfusion metrics simultaneously.

Supplemental Material

sj-pdf-1-jcb-10.1177_0271678X221117008 - Supplemental material for Whole-brain perfusion mapping in mice by dynamic BOLD MRI with transient hypoxia

Supplemental material, sj-pdf-1-jcb-10.1177_0271678X221117008 for Whole-brain perfusion mapping in mice by dynamic BOLD MRI with transient hypoxia by DongKyu Lee, Thuy Thi Le, Geun Ho Im and Seong-Gi Kim in Journal of Cerebral Blood Flow & Metabolism

Footnotes

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was funded by the Institute for Basic Science in Korea (IBS-R015-D1).

Acknowledgements

The authors acknowledge Mr. Chanhee Lee for his assistance with maintaining the 9.4 T MRI system, and Profs. Kamil Uludag, Fernando Calamante, and Ashley Stokes for helpful discussions.

Authors’ contributions

S-GK, DKL and TTL conceptualized and designed the study. DKL, TTL and GHI performed the data acquisition. DKL and TTL analyzed data. DKL, TTL and S-GK drafted the manuscript. DKL, TTL, GHI and S-GK contributed to the data interpretation and approved the final draft.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

ORCID iD

Seong-Gi Kim

Supplemental material

Supplemental material for this article is available online.

References

Edvinsson L, MacKenzie ET and McCulloch J. Cerebral blood flow and metabolism. New York: Raven Press, 1993, vol. 1185, pp. 1–680.

Calamante

Thomas

Pell

, et al. Measuring cerebral blood flow using magnetic resonance imaging techniques. J Cereb Blood Flow Metab 1999; 19: 701–735.

Detre

Leigh

Williams

, et al. Perfusion imaging. Magn Reson Med 1992; 23: 37–45.

Kim

SG.

Quantification of relative cerebral blood flow change by flow‐sensitive alternating inversion recovery (FAIR) technique: application to functional mapping. Magn Reson Med 1995; 34: 293–301.

Alsop

Detre

JA.

Reduced transit-time sensitivity in noninvasive magnetic resonance imaging of human cerebral blood flow. J Cereb Blood Flow Metab 1996; 16: 1236–1249.

Duhamel

Callot

Tachrount

, et al. Pseudo‐continuous arterial spin labeling at very high magnetic field (11.75 T) for high‐resolution mouse brain perfusion imaging. Magn Reson Med 2012; 67: 1225–1236.

Gao

Goodnough

Erokwu

, et al. Arterial spin labeling‐fast imaging with steady‐state free precession (ASL‐FISP): a rapid and quantitative perfusion technique for high‐field MRI. NMR Biomed 2014; 27: 996–1004.

Qin

, et al. Steady pulsed imaging and labeling scheme for noninvasive perfusion imaging. Magn Reson Med 2016; 75: 238–248.

Zuloaga

Johnson

Roese

, et al. High fat diet-induced diabetes in mice exacerbates cognitive deficit due to chronic hypoperfusion. J Cereb Blood Flow Metab 2016; 36: 1257–1270.

10.

Hirschler

Munting

Khmelinskii

, et al. Transit time mapping in the mouse brain using time‐encoded pCASL. NMR Biomed 2018; 31: e3855.

11.

Munting

Derieppe

MPP

Suidgeest

, et al. Influence of different isoflurane anesthesia protocols on murine cerebral hemodynamics measured with pseudo‐continuous arterial spin labeling. NMR Biomed 2019; 32: e4105.

12.

Zhang

Liu

, et al. Increased cerebral vascularization and decreased water exchange across the blood-brain barrier in aquaporin-4 knockout mice. PLOS One 2019; 14: e0218415.

13.

Gottschalk

Look‐locker FAIR TrueFISP for arterial spin labelling on mouse at 9.4 T. NMR Biomed 2020; 33: e4191.

14.

Villringer

Rosen

Belliveau

, et al. Dynamic imaging with lanthanide chelates in normal brain: contrast due to magnetic susceptibility effects. Magn Reson Med 1988; 6: 164–174.

15.

Østergaard

Weisskoff

Chesler

, et al. High resolution measurement of cerebral blood flow using intravascular tracer bolus passages. Part I: mathematical approach and statistical analysis. Magn Reson Med 1996; 36: 715–725.

16.

Aronen

Gazit

Louis

, et al. Cerebral blood volume maps of gliomas: comparison with tumor grade and histologic findings. Radiology 1994; 191: 41–51.

17.

Boxerman

Schmainda

Weisskoff

RM.

Relative cerebral blood volume maps corrected for contrast agent extravasation significantly correlate with glioma tumor grade, whereas uncorrected maps do not. AJNR Am J Neuroradiol 2006; 27: 859–867.

18.

Paulson

Schmainda

KM.

Comparison of dynamic susceptibility-weighted contrast-enhanced MR methods: recommendations for measuring relative cerebral blood volume in brain tumors. Radiology 2008; 249: 601–613.

19.

Pike

Stoops

Langford

, et al. High‐resolution longitudinal assessment of flow and permeability in mouse glioma vasculature: sequential small molecule and SPIO dynamic contrast agent MRI. Magn Reson Med 2009; 61: 615–625.

20.

Ielacqua

Schlegel

Füchtemeier

, et al. Magnetic resonance Q mapping reveals a decrease in microvessel density in the arcaβ mouse model of cerebral amyloidosis. Front Aging Neurosci 2015; 7: 241.

21.

Montagne

Nikolakopoulou

Huuskonen

, et al. APOE4 accelerates advanced-stage vascular and neurodegenerative disorder in old Alzheimer’s mice via cyclophilin a independently of amyloid-β. Nat Aging 2021; 1: 506–520.

22.

Ogawa

Lee

Kay

, et al. Brain magnetic resonance imaging with contrast dependent on blood oxygenation. Proc Natl Acad Sci U S A 1990; 87: 9868–9872.

23.

Poublanc

Sobczyk

Shafi

, et al. Perfusion MRI using endogenous deoxyhemoglobin as a contrast agent: preliminary data. Magn Reson Med 2021; 86: 3012–3021.

24.

Chai

Coloigner

, et al. Quantitative perfusion mapping with induced transient hypoxia using BOLD MRI. Magn Reson Med 2021; 85: 168–181.

25.

Thulborn

Waterton

Matthews

, et al. Oxygenation dependence of the transverse relaxation time of water protons in whole blood at high field. Biochim Biophys Acta 1982; 714: 265–270.

26.

MacDonald

Berman

AJL

Mazerolle

, et al. Modeling hyperoxia-induced BOLD signal dynamics to estimate cerebral blood flow, volume and mean transit time. NeuroImage 2018; 178: 461–474.

27.

Blockley

Griffeth

Germuska

, et al. An analysis of the use of hyperoxia for measuring venous cerebral blood volume: comparison of the existing method with a new analysis approach. NeuroImage 2013; 72: 33–40.

28.

Turner

Le Bihan

Moonen

, et al. Echo‐planar time course MRI of cat brain oxygenation changes. Magn Reson Med 1991; 22: 159–166.

29.

Kim

Ogawa

Biophysical and physiological origins of blood oxygenation level-dependent fMRI signals. J Cereb Blood Flow Metab 2012; 32: 1188–1206.

30.

Lee

Silva

Ugurbil

, et al. Diffusion‐weighted spin‐echo fMRI at 9.4 T: microvascular/tissue contribution to BOLD signal changes. Magn Reson Med 1999; 42: 919–928.

31.

Uludağ

Müller-Bierl

Uğurbil

An integrative model for neuronal activity-induced signal changes for gradient and spin echo functional imaging. NeuroImage 2009; 48: 150–165.

32.

Jezzard

Heineman

Taylor

, et al. Comparison of EPI gradient‐echo contrast changes in cat brain caused by respiratory challenges with direct simultaneous evaluation of cerebral oxygenation via a cranial window. NMR Biomed 1994; 7: 35–44.

33.

Kida

Yamamoto

Tamura

Interpretation of BOLD MRI signals in rat brain using simultaneously measured near‐infrared spectrophotometric information. NMR Biomed 1996; 9: 333–338.

34.

Rosen

Belliveau

Vevea

, et al. Perfusion imaging with NMR contrast agents. Magn Reson Med 1990; 14: 249–265.

35.

Simonsen

Østergaard

Vestergaard-Poulsen

, et al. CBF and CBV measurements by USPIO bolus tracking: reproducibility and comparison with Gd‐based values. J Magn Reson Imaging 1999; 9: 342–347.

36.

Østergaard

Principles of cerebral perfusion imaging by bolus tracking. J Magn Reson Imaging 2005; 22: 710–717.

37.

Bothe

Bodsch

Hossmann

KA.

Relationship between specific gravity, water content, and serum protein extravasation in various types of vasogenic brain edema. Acta Neuropathol 1984; 64: 37–42.

38.

Cremer

Seville

MP.

Regional brain blood flow, blood volume, and haematocrit values in the adult rat. J Cereb Blood Flow Metab 1983; 3: 254–256.

39.

Levin

Ausman

JI.

Relationship of peripheral venous hematocrit to brain hematocrit. J Appl Physiol 1969; 26: 433–437.

40.

Stewart

GN.

Researches on the circulation time in organs and on the influences which affect it: parts I–III. J Physiol 1893; 15: 1–89.

41.

Meier

Zierler

KL.

On the theory of the indicator-dilution method for measurement of blood flow and volume. J Appl Physiol 1954; 6: 731–744.

42.

Lin

Celik

Derdeyn

, et al. Quantitative measurements of cerebral blood flow in patients with unilateral carotid artery occlusion: a PET and MR study. J Magn Reson Imaging 2001; 14: 659–667.

43.

Chen

Smith

Frayne

The impact of partial‐volume effects in dynamic susceptibility contrast magnetic resonance perfusion imaging. J Magn Reson Imaging 2005; 22: 390–399.

44.

van Osch

van der Grond

Bakker

CJ.

Partial volume effects on arterial input functions: shape and amplitude distortions and their correction. J Magn Reson Imaging 2005; 22: 704–709.

45.

Van der Schaaf

Vonken

Waaijer

, et al. Influence of partial volume on venous output and arterial input function. AJNR Am J Neuroradiol 2006; 27: 46–50.

46.

Knutsson

Börjesson

Larsson

, et al. Absolute quantification of cerebral blood flow in normal volunteers: correlation between Xe‐133 SPECT and dynamic susceptibility contrast MRI. J Magn Reson Imaging 2007; 26: 913–920.

47.

Kim

Harel

Jin

, et al. Cerebral blood volume MRI with intravascular superparamagnetic iron oxide nanoparticles. NMR Biomed 2013; 26: 949–962.

48.

Percie Du Sert

Hurst

Ahluwalia

, et al. ARRIVE guidelines 2.0: updated guidelines for reporting animal research. J Cereb Blood Flow Metab 2020; 40: 1769–1777.

49.

Shim

Jung

Schlegel

, et al. Mouse fMRI under ketamine and xylazine anesthesia: robust contralateral somatosensory cortex activation in response to forepaw stimulation. NeuroImage 2018; 177: 30–44.

50.

You

Kim

SG.

Characterization of brain-wide somatosensory BOLD fMRI in mice under dexmedetomidine/isoflurane and ketamine/xylazine. Sci Rep 2021; 11: 13110.

51.

Gray

Steadman

JM.

Determination of the oxyhaemoglobin dissociation curves for mouse and rat blood. J Physiol 1964; 175: 161–171.

52.

Shim

Lee

Kim

SG.

BOLD fMRI and hemodynamic responses to somatosensory stimulation in anesthetized mice: spontaneous breathing vs. mechanical ventilation. NMR Biomed 2020; 33: e4311.

53.

Cox

RW.

AFNI: software for analysis and visualization of functional magnetic resonance neuroimages. Comput Biomed Res 1996; 29: 162–173.

54.

Smith

SM, MJ

Woolrich

Beckmann

, et al. Advances in functional and structural MR image analysis and implementation as FSL. NeuroImage 2004; 23: S208–S219.

55.

Avants

Tustison

Song

Advanced normalization tools (ANTS). Insight j 2009; 2: 1–35.

56.

Lein

Hawrylycz

, et al. Genome-wide atlas of gene expression in the adult mouse brain. Nature 2007; 445: 168–176.

57.

Peruzzo

Bertoldo

Zanderigo

, et al. Automatic selection of arterial input function on dynamic contrast-enhanced MR images. Comput Methods Programs Biomed 2011; 104: e148–e157.

58.

Dorr

Sled

Kabani

Three-dimensional cerebral vasculature of the CBA mouse brain: a magnetic resonance imaging and micro computed tomography study. Neuroimage 2007; 35: 1409–1423.

59.

Jubran

Pulse oximetry. Crit Care 1999; 3: R11–7.

60.

Yamaya

Bogaard

Wagner

, et al. Validity of pulse oximetry during maximal exercise in normoxia, hypoxia, and hyperoxia. J Appl Physiol (1985) 2002; 92: 162–168.

61.

Salyer

JW.

Neonatal and pediatric pulse oximetry. Respir Care 2003; 48: 386–396. discussion 397.

62.

Narkiewicz

van de Borne

Pesek

, et al. Selective potentiation of peripheral chemoreflex sensitivity in obstructive sleep apnea. Circulation 1999; 99: 1183–1189.

63.

Halliwill

Morgan

Charkoudian

Peripheral chemoreflex and baroreflex interactions in cardiovascular regulation in humans. J Physiol 2003; 552: 295–302.

64.

Kara

Narkiewicz

Somers

VK.

Chemoreflexes – physiology and clinical implications. Acta Physiol Scand 2003; 177: 377–384.

65.

Prabhakar

Peng

YJ.

Peripheral chemoreceptors in health and disease. J Appl Physiol (1985) 2004; 96: 359–366.

66.

Kontos

Wei

Raper

, et al. Role of tissue hypoxia in local regulation of cerebral microcirculation. Am J Physiol 1978; 234: H582–H591.

67.

Ulatowski

Bucci

Razynska

, et al. Cerebral blood flow during hypoxic hypoxia with plasma-based hemoglobin at reduced hematocrit. Am J Physiol 1998; 274: H1933–H1942.

68.

Willats

Calamante

The 39 steps: evading error and deciphering the secrets for accurate dynamic susceptibility contrast MRI. NMR Biomed 2013; 26: 913–931.

69.

Knutsson

Ståhlberg

Wirestam

Absolute quantification of perfusion using dynamic susceptibility contrast MRI: pitfalls and possibilities.

MAGMA 2010; 23: 1–21.

70.

Calamante

Arterial input function in perfusion MRI: a comprehensive review. Prog Nucl Magn Reson Spectrosc 2013; 74: 1–32.

71.

Kim

Ugurbil

High-resolution functional magnetic resonance imaging of the animal brain. Methods 2003; 30: 28–24.

72.

Valable

Lemasson

Farion

, et al. Assessment of blood volume, vessel size, and the expression of angiogenic factors in two rat glioma models: a longitudinal in vivo and ex vivo study. NMR Biomed 2008; 21: 1043–1056.

73.

Schulman

, et al. Deoxyhemoglobin versus gadolinium as contrast in dynamic susceptibility perfusion imaging: simulations and scan validations. Proc Intl Soc Mag Reson Med 2022; 30: 337.

74.

LaRue

Hogg

Sagare

, et al. Method for measurement of the blood–brain barrier permeability in the perfused mouse brain: application to amyloid-β peptide in wild type and Alzheimer’s Tg2576 mice. J Neurosci Methods 2004; 138: 233–242.

75.

Chugh

Lerch

, et al. Measurement of cerebral blood volume in mouse brain regions using micro-computed tomography. Neuroimage 2009; 47: 1312–1318.

76.

Tsai

Kaufhold

Blinder

, et al. Correlations of neuronal and microvascular densities in murine cortex revealed by direct counting and colocalization of nuclei and vessels. J Neurosci 2009; 29: 14553–14570.

77.

Blinder

Tsai

Kaufhold

, et al. The cortical angiome: an interconnected vascular network with noncolumnar patterns of blood flow. Nat Neurosci 2013; 16: 889–897.

78.

Schmiedeskamp

Straka

Newbould

, et al. Combined spin‐ and gradient‐echo perfusion‐weighted imaging. Magn Reson Med 2012; 68: 30–40.

79.

Schmiedeskamp

Andre

Straka

, et al. Simultaneous perfusion and permeability measurements using combined spin- and gradient-echo MRI. J Cereb Blood Flow Metab 2013; 33: 732–743.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

2.68 MB

Whole-brain perfusion mapping in mice by dynamic BOLD MRI with transient hypoxia

Abstract

Keywords

Introduction

Materials and methods

Theoretical background

Quantitative perfusion measurements from dynamic ΔR2* changes due to hypoxia

Determination of arterial input function

Determination of relative CBV with iron oxide contrast agent

Animal preparation and physiological monitoring

Transient hypoxia paradigm for perfusion measurements

Arterial blood gas analysis

MRI acquisitions

Data processing

Physiological data analysis

Overall BOLD-MRI analysis

Determination of dynamic CBV and SaO2 change by transient hypoxia (Expt. #1 – #2)

Estimation of corrected arterial input function

Determination of quantitative perfusion values (Expt. #3)

Statistical analysis

Results

Choice of stimulus duration

Systemic physiology changes responding to 5 s anoxic stimulus

Determination of corrected arterial input function

Regional total vs. microvascular cerebral perfusion values of mice

Cerebral perfusion maps obtained from GE- and SE-BOLD MRI

Cortical depth-dependent perfusion

Discussion

Anoxic stimulus and its effects to vascular physiology

Errors of BOLD-DSC perfusion measurement and its translation to humans

Regional CBV and CBF values of mouse under ketamine and xylazine anesthesia

Depth-dependent perfusion characteristics in the cortical somatosensory area

Limitation and alternatives of BOLD DSC with transient anoxia stimulus

Supplemental Material

sj-pdf-1-jcb-10.1177_0271678X221117008 - Supplemental material for Whole-brain perfusion mapping in mice by dynamic BOLD MRI with transient hypoxia

Footnotes

Funding

Acknowledgements

Authors’ contributions

Declaration of conflicting interests

ORCID iD

Supplemental material

References

Supplementary Material

Quantitative perfusion measurements from dynamic ΔR₂* changes due to hypoxia

Determination of dynamic CBV and S_aO₂ change by transient hypoxia (Expt. #1 – #2)