**5. Stroke MRI with hyperpolarized 129Xe**

Because there is no background signal from xenon in biological tissue, and because the inhaled xenon is delivered to the brain by the blood flow, we would expect a perfusion deficit, such as could be seen in stroke, to reduce xenon concentration in the region of the deficit. Thermal polarization yields negligible xenon signal relative to hyperpolarized xenon; therefore, hyperpolarized xenon can be used as a tracer of cerebral blood flow (CBF). This subsection will describe that hyperpolarized 129Xe MRI is able to detect, *in vivo* hypoperfused area of focal cerebral ischemia—i.e., the ischemic core area of stroke, by using a rat permanent right middle cerebral artery occlusion (MCAO) model (Zhou et al., 2011a).

Stroke is the single most common reason for permanent disability and is the third leading cause of death in developed countries. During acute ischemic stroke, a core of brain cells at the center of the affected region dies quickly, and the damage subsequently spreads to surrounding tissue over the next few hours. Because they allow for the delineation of areas of ischemic neuronal injury and hypoperfusion within minutes after the induction of cerebral ischemia, conventional proton MRI, especially diffusion-weighted imaging (DWI) and perfusion weighted imaging (PWI), have been particularly useful in the diagnosis of acute ischemic stroke. The target of acute stroke therapy is the portion of the ischemic region

Hyperpolarized Xenon Brain MRI 135

the ADC lesion and TTC lesion areas, and the green area shows the nonischemic region. ROI analyses were performed to further characterize the Xe CSI tissue signals within the different observed tissue compartments defined by their respective ADC and TTC

Xenon signals from each ROI in the contralesional (left) hemisphere were set as a reference (100%), and xenon signals from each ROI in the ipsilesional (right) hemisphere were normalized to these signals. The xenon signal in the ischemic core (ROI1) dropped to 8.4±0.4% of the contralesional side signal, and the xenon signal in normal tissue (ROI2) remained the same. Moreover, the xenon signal in ROI1 was reduced significantly relative to the corresponding con-tralesional ROIs in the MCAO group, as well as the corresponding ipsilateral ROIs in control animals. Within the control group, no significant differences in the xenon signal were observed between the corresponding ROIs of both hemispheres.

Fig. 3. (a) Representative proton apparent diffusion coefficient (ADC) map image obtained 90 min after right middle cerebral artery occlusion (MCAO). The ischemic core is indicated

ipsilesional (right) hemisphere. The defined regions of interest (ROIs) are labeled as follows: ROI1, core; ROI2, normal tissue. The xenon signal intensity is given in arbitrary units. (c) Corresponding 2,3,5-triphenyltetrazolium chloride (TTC)-stained brain section of the same animal as in (a) and (b). (d) Tricolor map based on the ADC and TTC images shown in

Using a rat permanent right middle cerebral artery occlusion model, it has demonstrated that hyperpolarized 129Xe MRI is able to detect, *in vivo*, the hypoperfused area of focal cerebral ischemia, that is the ischemic core area of stroke. To the best of our knowledge, this is the first time that hyperpolarized 129Xe MRI has been used to explore normal and abnormal cerebral perfusion. The study shows a novel application of hyperpolarized 129Xe

by ADC values below 5.3 × 10-4 mm2/s (circled by a blue line). (b) Corresponding hyperpolarized 129Xe chemical shift image (CSI). There is a large signal void in the

(a) and (c). Green, red and blue represent nonischemic tissue, core and penumbra,

signatures.

respectively.

which is still potentially salvageable, that is the ischemic penumbra. MRI operationally defines the ischemic penumbra by the mismatch area of PWI–DWI. DWI detects changes in the apparent diffusion coefficient (ADC) of water molecules associated with early cytotoxic edema in ischemic stroke. Arterial spin labeling (ASL)-based PWI methods provide excellent anatomical information for the measurement of tissue perfusion. The ASL technique shows numerous advantages, such as noninvasive measurements of cerebral blood flow (CBF) quantifiable in standard units of mL/g/min, and is able to image multi-slices and multiregions of the brain. However, in some situations, PWI methods require the injection of gadolinium containing contrast agents to map relative CBF in order to identify the hypoperfused tissue. In addition to the conventional DWI and PWI techniques, van Zijl and coworkers have developed pH-weighted MRI to study stroke and ischemic penumbra. However, proton imaging has a large background signal in biological tissue, and contrast injection is an invasive approach. Moreover, contrast-associated nephrogenic systemic fibrosis has been reported after the use of gadolinium-based agents, and many patients with impaired renal function are not eligible to receive contrast media. In contrast, hyperpolarized 129Xe MRI shows great potential and advantages for the identification of hypoperfused brain tissue. Xenon is highly lipid soluble and lacks an intrinsic background signal in biological tissue (Albert et al., 1994). Duhamel and co-workers have studied CBF using intra-arterial injection of hyperpolarized 129Xe dissolved in a lipid emulsion (Duhamel et al., 2000). Alternatively, hyperpolarized 129Xe can be administered noninvasively by inhalation; following inhalation, 129Xe is absorbed into the bloodstream and delivered to the brain through the circulation. Because the spin-exchange optical pumping technique can enhance the 129Xe MR signal 10,000–100,000 times over thermal polarization, the dissolvedphase hyperpolarized 129Xe signal in the brain can be detected even at low concentrations. Because the xenon signal is proportional to CBF (Zhou et al., 2008), a decrease in the signal is expected to occur in areas of decreased CBF after the inhalation of hyperpolarized xenon gas. Hyperpolarized xenon imaging currently can not achieve a slice as thin as that obtained by ASL. In addition, ASL can be performed with substantially higher spatial resolution than hyperpolarized xenon imaging in brain tissue. However, the ASL technique requires two experiments (arterial spin labeled and controlled) to obtain CBF information. In this subsection, we report, for the first time, that hyperpolarized 129Xe MRI is able to detect areas of decreased CBF following middle cerebral artery occlusion (MCAO) in a single scan. These findings show the great potential and utility of hyperpolarized 129Xe MRI for stroke imaging, and further demonstrate that hyperpolarized 129Xe is a safe and noninvasive signal source for imaging diseases and function of the brain.

Figure 3a shows a representative proton ADC map obtained 90 min following MCAO. There is a large ischemic core within the ipsilesional (right) MCA territory, as indicated by ADC values below the critical threshold of 5.3 × 10-4 mm2/s for infarction (Meng et al., 2004). [The normal ADC value of rat brain tissue in the contralesional (left) hemisphere is (7.5±1.8)×10-4 mm2/s.] Figure 3b depicts the corresponding hyperpolarized 129Xe CSI, indicating signal reduction in large parts of the right hemisphere, consistent with the area typically experiencing decreased CBF following right MCAO in the model used (Duhamel et al., 2002). Figure 3c shows the TTC-stained brain section of the same animal as illustrated in Fig. 3a, b; the black line in this figure delineates the infracted brain tissue. Xenon CSI, shown in Fig. 3b, demonstrates reduced perfusion in brain tissue, ultimately leading to infarction, as shown by TTC staining in Fig. 3c. In Fig. 3d, the blue area represents the difference between

which is still potentially salvageable, that is the ischemic penumbra. MRI operationally defines the ischemic penumbra by the mismatch area of PWI–DWI. DWI detects changes in the apparent diffusion coefficient (ADC) of water molecules associated with early cytotoxic edema in ischemic stroke. Arterial spin labeling (ASL)-based PWI methods provide excellent anatomical information for the measurement of tissue perfusion. The ASL technique shows numerous advantages, such as noninvasive measurements of cerebral blood flow (CBF) quantifiable in standard units of mL/g/min, and is able to image multi-slices and multiregions of the brain. However, in some situations, PWI methods require the injection of gadolinium containing contrast agents to map relative CBF in order to identify the hypoperfused tissue. In addition to the conventional DWI and PWI techniques, van Zijl and coworkers have developed pH-weighted MRI to study stroke and ischemic penumbra. However, proton imaging has a large background signal in biological tissue, and contrast injection is an invasive approach. Moreover, contrast-associated nephrogenic systemic fibrosis has been reported after the use of gadolinium-based agents, and many patients with impaired renal function are not eligible to receive contrast media. In contrast, hyperpolarized 129Xe MRI shows great potential and advantages for the identification of hypoperfused brain tissue. Xenon is highly lipid soluble and lacks an intrinsic background signal in biological tissue (Albert et al., 1994). Duhamel and co-workers have studied CBF using intra-arterial injection of hyperpolarized 129Xe dissolved in a lipid emulsion (Duhamel et al., 2000). Alternatively, hyperpolarized 129Xe can be administered noninvasively by inhalation; following inhalation, 129Xe is absorbed into the bloodstream and delivered to the brain through the circulation. Because the spin-exchange optical pumping technique can enhance the 129Xe MR signal 10,000–100,000 times over thermal polarization, the dissolvedphase hyperpolarized 129Xe signal in the brain can be detected even at low concentrations. Because the xenon signal is proportional to CBF (Zhou et al., 2008), a decrease in the signal is expected to occur in areas of decreased CBF after the inhalation of hyperpolarized xenon gas. Hyperpolarized xenon imaging currently can not achieve a slice as thin as that obtained by ASL. In addition, ASL can be performed with substantially higher spatial resolution than hyperpolarized xenon imaging in brain tissue. However, the ASL technique requires two experiments (arterial spin labeled and controlled) to obtain CBF information. In this subsection, we report, for the first time, that hyperpolarized 129Xe MRI is able to detect areas of decreased CBF following middle cerebral artery occlusion (MCAO) in a single scan. These findings show the great potential and utility of hyperpolarized 129Xe MRI for stroke imaging, and further demonstrate that hyperpolarized 129Xe is a safe and noninvasive signal

source for imaging diseases and function of the brain.

Figure 3a shows a representative proton ADC map obtained 90 min following MCAO. There is a large ischemic core within the ipsilesional (right) MCA territory, as indicated by ADC values below the critical threshold of 5.3 × 10-4 mm2/s for infarction (Meng et al., 2004). [The normal ADC value of rat brain tissue in the contralesional (left) hemisphere is (7.5±1.8)×10-4 mm2/s.] Figure 3b depicts the corresponding hyperpolarized 129Xe CSI, indicating signal reduction in large parts of the right hemisphere, consistent with the area typically experiencing decreased CBF following right MCAO in the model used (Duhamel et al., 2002). Figure 3c shows the TTC-stained brain section of the same animal as illustrated in Fig. 3a, b; the black line in this figure delineates the infracted brain tissue. Xenon CSI, shown in Fig. 3b, demonstrates reduced perfusion in brain tissue, ultimately leading to infarction, as shown by TTC staining in Fig. 3c. In Fig. 3d, the blue area represents the difference between the ADC lesion and TTC lesion areas, and the green area shows the nonischemic region. ROI analyses were performed to further characterize the Xe CSI tissue signals within the different observed tissue compartments defined by their respective ADC and TTC signatures.

Xenon signals from each ROI in the contralesional (left) hemisphere were set as a reference (100%), and xenon signals from each ROI in the ipsilesional (right) hemisphere were normalized to these signals. The xenon signal in the ischemic core (ROI1) dropped to 8.4±0.4% of the contralesional side signal, and the xenon signal in normal tissue (ROI2) remained the same. Moreover, the xenon signal in ROI1 was reduced significantly relative to the corresponding con-tralesional ROIs in the MCAO group, as well as the corresponding ipsilateral ROIs in control animals. Within the control group, no significant differences in the xenon signal were observed between the corresponding ROIs of both hemispheres.

Fig. 3. (a) Representative proton apparent diffusion coefficient (ADC) map image obtained 90 min after right middle cerebral artery occlusion (MCAO). The ischemic core is indicated by ADC values below 5.3 × 10-4 mm2/s (circled by a blue line). (b) Corresponding hyperpolarized 129Xe chemical shift image (CSI). There is a large signal void in the ipsilesional (right) hemisphere. The defined regions of interest (ROIs) are labeled as follows: ROI1, core; ROI2, normal tissue. The xenon signal intensity is given in arbitrary units. (c) Corresponding 2,3,5-triphenyltetrazolium chloride (TTC)-stained brain section of the same animal as in (a) and (b). (d) Tricolor map based on the ADC and TTC images shown in (a) and (c). Green, red and blue represent nonischemic tissue, core and penumbra, respectively.

Using a rat permanent right middle cerebral artery occlusion model, it has demonstrated that hyperpolarized 129Xe MRI is able to detect, *in vivo*, the hypoperfused area of focal cerebral ischemia, that is the ischemic core area of stroke. To the best of our knowledge, this is the first time that hyperpolarized 129Xe MRI has been used to explore normal and abnormal cerebral perfusion. The study shows a novel application of hyperpolarized 129Xe

Hyperpolarized Xenon Brain MRI 137

High resolution proton images were taken of the rat head to provide an anatomical reference for hyperpolarized 129Xe images. In order to evaluate the distribution of hyperpolarized 129Xe in brain following an external sensory stimulus, we acquired MRS images before and after a pain producing stimulus that has a well-defined functional response that can be measured using traditional fMRI techniques. A baseline hyperpolarized 129Xe spectroscopic image was acquired from a coronal slice centered at the

The maximal, steady-state 129Xe brain signal occurred within 15 seconds after starting the ventilation with hyperpolarized 129Xe. Once verification of the xenon signal in the brain, a baseline 129Xe chemical shift image (CSI) was acquired that was centred in the plane corresponding to the proton reference image. Low flip angle used for CSI acquisition insured minimal loss of hyperpolarized 129Xe signal due to RF destruction, and the relatively long TR allowed continuous delivery of hyperpolarized 129Xe to the tissue, a steady–state concentration of hyperpolarized 129Xe was maintained in the brain thereby insuring constant signal intensity across the k-space. In a subset of animals (n=3) the animal's left forepaw was injected with a vehicle solution during baseline. Following acquisition of the baseline CSI image, the animal was ventilated for 10 minutes with O2 (isoflurane to allow for complete clearance of 129Xe magnetization from the brain). After that, the chemical irritant capsaicin (20 ul of 3 mg/ml) was injected into the animal's right forepaw (n= 6), and a second CSI was

A robust hyperpolarized 129Xe spectroscopic signal with one primary peak at 194.7 ppm developed within 15 seconds after starting the ventilation with hyperpolarized 129Xe. In order to determine the extent of hyperpolarized 129Xe distribution throughout the rat brain, a magnetic resonance spectroscopic image was acquired of the primary peak during the administration of hyperpolarized 129Xe (Figure 4). The four smaller resonances did not have sufficient SNR to produce spectroscopic images. Figure 4a shows an hyperpolarized 129Xe image taken in the axial plane. Addition of a color look-up table (Figure 4b) aided in visually delineating areas of low and high SNR. Figure 4c show a 1 mm proton slice in which the olfactory bulbs and cerebellum are visible. Overlay of the hyperpolarized 129Xe spectroscopic image onto the proton reference image (Figure 4d) revealed that the steadystate hyperpolarized 129Xe signal originated from within the brain tissue and further demonstrated a pattern of hyperpolarized 129Xe distribution throughout the brain with

Three of the six animals studied received a vehicle injection (saline) to the left forepaw immediately prior to the acquisition of the baseline image. 10 minutes after acquisition of the baseline hyperpolarized 129Xe MRS, the animal's right forepaw was injected with the chemical irritant capsaicin (20 ml of 3 mg/ml), and a second hyperpolarized 129Xe spectroscopic image was acquired. Responses from three individual animals are shown in Figure 5. Whereas baseline images showed some hyperpolarized 129Xe signal intensity in cortical and sub-cortical brain regions (Figure 5, left panel), images acquired following administration of capsaicin showed both higher hyperpolarized 129Xe signal intensity and an increased area of distribution within the brain (Figure 5, right panel). Superimposition of a rat brain atlas (Figure 5a) revealed that areas of hyperpolarized 129Xe signal increase occurred both bilaterally and

level of the anatomical reference slice.

**6.2 Experiment results and discussion** 

varying signal intensity in different brain regions.

acquired.

MRI for imaging stroke, and further demonstrates its capacity to serve as a complementary tool to proton MRI for the study of the pathophysiology during brain hypoperfusion. More importantly, these results indicate the possibility of the use of *in vivo* MRI to diagnose brain disease employing inhaled hyperpolarized gas, eliminating the potential adverse effects to the patient resulting from the injection of gadolinium-based contrast agents.
