Hyperpolarized Gas MRI: Technique and Applications

Synopsis

Functional imaging today offers a rich world of information that is more sensitive to changes in lung structure and function than traditionally obtained pulmonary function tests. Hyperpolarized helium (³He) and xenon (¹²⁹Xe) MR imaging of the lungs provided new sensitive contrast mechanisms to probe changes in pulmonary ventilation, microstructure and gas exchange. With the recent scarcity in the supply of ³He the field of hyperpolarized gas imaging shifted to the use of cheaper and naturally available ¹²⁹Xe. Xenon is well tolerated and recent technical advances have ensured that the ¹²⁹Xe image quality is on par with that of ³He. The added advantage of ¹²⁹Xe is its solubility in pulmonary tissue, which allows exploring specific lung function characteristics involved in gas exchange and alveolar oxygenation.

With a plethora of contrast mechanisms, hyperpolarized gases and ¹²⁹Xe in particular, stands to be an excellent probe of pulmonary structure and function, and provide sensitive and non-invasive biomarkers for a wide variety of pulmonary diseases.

1. Introduction

Lung diseases affect the conducting airways, the gas exchange parenchyma or both. As a result, they have a high impact on patients’ morbidity, and quality of life. Chronic obstructive pulmonary disease (COPD), for example, is an umbrella term for progressive and irreversible obstructive diseases that affect the airways as well as the parenchyma. After cardiovascular diseases and cancer, COPD is predicted to be the third leading cause of death in the US ¹. Similarly, airway diseases such as asthma have a high prevalence and a large economic impact with $56 billion in associated healthcare costs². Furthermore, interstitial lung diseases (ILD), while less prevalent than COPD or asthma, also have a high morbidity. This is partially attributable to a lack of viable therapeutic options to treat the underlying diseases ³. However, equally problematic is the lack of sensitive biomarkers that can be used to diagnose these diseases earlier, better monitor progression and show early therapeutic response.

The current gold standard for diagnosis and monitoring treatment of pulmonary diseases is spirometric pulmonary function testing (PFTs). However, by assessing the lung only on a global basis, PFT metrics generally lack the ability to detect functional changes associated with the small airways and gas exchange regions. This insensitivity of PFTs to the functioning of the distal lung (Fig 1) has led these regions to be referred to as the ‘silent zone’ ⁴. Additionally, these metrics largely rely on subject effort, causing significant measurement uncertainty and variability. Hence, current therapy is guided largely by patients’ symptoms and survival ^5–7. Given the large burden on our healthcare system and growing prevalence of pulmonary disease, there exists a need for improved diagnostic tools and quantitative metrics to better diagnose and quantify pulmonary disease progression and help accurately measure response to therapy.

Figure 1.

Anatomical relationship between the purely air conducting zone and the transitional/respiratory zones. Pulmonary function test (PFTs) lack the ability to detect early changes at the level of the small airways, the so-called ‘silent zone’ for PFTs.

Adapted from Weibel E.R. Morphometry of the Human Lung. Heidelberg Springer, 1963; 111, with permission.

To overcome the lungs’ many compensatory mechanisms and improve sensitivity to early disease, it is necessary to exploit regional information obtained by imaging. To this end, hyperpolarized (HP) gas magnetic resonance (MR) imaging provides a non-invasive, ionizing-radiation-free method to image pulmonary structure and function ^8–13. While the early days of HP gas MRI saw extensive development and progress using the noble gas ³He, its recent scarcity and concomitant increase in price (~2000$/liter) ¹⁴ has caused a transition to the cheaper and naturally available ¹²⁹Xe (~30$/liter, natural abundance). An added advantage of ¹²⁹Xe is obtained by virtue of its solubility in pulmonary tissues, which leads to two signal sources, distinct from the xenon in the airspaces. ¹²⁹Xe can be separately detected in the pulmonary barrier tissues and the red blood cells in the capillary network, compartments in which it exhibits distinct resonant signal frequencies. These three ¹²⁹Xe resonances can provide quantitative regional information on the fundamental function of the lung – gas-exchange¹⁵.

This article will briefly touch upon the history of HP gas MR imaging, describe the key technical properties, and discuss current and future applications and opportunities for pulmonary imaging.

2. History and Safety

In 1994, the first HP gas MR imaging studies were carried out ex-vivo using the noble gas isotope ¹²⁹ Xenon (¹²⁹ Xe)¹⁶. In 1997, Mugler et al. used ¹²⁹Xe to conduct the first ground breaking studies in humans ¹⁷. While this represented rapid advancement from mouse to clinical translation, these studies were limited by relatively low ¹²⁹ Xe polarization (1–2%), which resulted in low signal intensities. This caused research interest to transition quickly to the other stable inert gas isotope. ³He, which has a larger gyromagnetic ratio than ¹²⁹Xe (32.4 MHz/T Vs. 11.77 MHz/T), offered a simpler and more mature polarization technology (~30%), and corresponding larger signal intensities. Additionally, unlike ¹²⁹Xe which in large enough alveolar concentrations (>70%)¹⁸ is known to exhibit anesthetic properties, ³He does not have any physiological side effects, and was considered to be a better starting point for clinical imaging. Indeed the known anesthetic properties of xenon led to it being regulated as a drug, and further increased the barriers to its use in research. Interestingly, it would later be recognized that the tissue solubility of ¹²⁹Xe that contribute to its anesthetic properties, actually provided new and exciting opportunities for imaging of the lung beyond what is possible with ³He.

³He MR imaging entered clinical research in 1996^19,20 and soon expanded to multi-center clinical studies ²¹. The results of the ventilation studies showed significant correlation to conventional PFTs in patients with COPD, asthma, and cystic fibrosis (CF). Diffusion weighted imaging yielded the apparent diffusion coefficient, which is a sensitive marker of alveolar enlargement, and this marker was significantly increased in subjects with emphysema ^8,9,22–24.

The problem with ³He HP MR imaging is twofold. First, the only source of ³He comes from the decay of tritium, which is exclusively derived from the past production of nuclear warheads in the US. The supply from the current stockpile is becoming progressively depleted and access increasingly limited. Secondly, a large portion of the current stockpile has been allocated for homeland security applications to detect emitted neutrons from smuggled plutonium. These reasons have driven up costs significantly to approximately $800–2000 per liter depending on academic versus commercial use ¹⁴. With these higher costs and lower availability, ³He HP MR imaging while having contributed greatly to the creation of this field, is not economically sustainable.

Recent progress in ¹²⁹ Xe polarization technology led Patz et al. to reintroduce ¹²⁹ Xe MR imaging in humans ²⁵. Xenon has a long history of safe use as a contrast agent in computed tomography (CT) lung imaging studies ²⁶, which was confirmed by Driehuys et al, who rigorously tested the safety and tolerability of inhaling multiple, undiluted 1-liter volumes of hyperpolarized ¹²⁹Xe ²⁷. No major adverse effects were reported in total of 44 study subjects including healthy volunteers and COPD patients. Among the reported symptoms were mild dizziness, par-/hypoesthesia and euphoria, which were transient for approximately 1–2min. No subject showed changes in laboratory tests or ECG. With the advent of more efficient polarizers, resulting in improved ¹²⁹Xe polarization ²⁸, one can expect better image quality with a lower volume of xenon, and the described symptoms are likely to be diminished further. In fact, a second safety study by Shukla and co-workers showed that inhalation of only 0.5 liter volumes caused subjects to experience few or no symptoms²⁹.

3. Technique

Traditional MR imaging of the lungs is fraught with a number of difficulties. Conventional MR scanners are primarily tuned to excite hydrogen protons (¹ H) that are present in abundance in water molecules. However the lungs have only very low ¹H density (~20%) compared to other anatomical structures. Combined with a long relaxation time, the signal available for imaging is minimal. Further complicating the image acquisition is the inhomogeneous magnetic environment of the lung, which introduces significant susceptibility artifacts that further challenge conventional MR acquisitions. However, these problems of MR-based lung imaging are not faced by the external gaseous contrast media (³He or ¹²⁹ Xe), which instead image the airways and airspaces within the lungs rather than the surrounding tissues (Fig 2). This greatly reduces the problems of unfavorable longitudinal and transverse relaxation times faced by ¹H MRI in the lung. On the other hand, MR imaging of a gas is challenging because it’s density is typically about 4 orders of magnitude less than that of protons (¹H density of water ~100 Mol/l versus ¹²⁹ Xe gas density ~0.01 Mol/l). This lower density commensurately reduces the available signal intensity and would appear to make gas MRI impossible. To circumvent this limitation, a physical process called hyperpolarization is used to increase the magnetization of these particular gases by about 5 orders of magnitude. Hyperpolarization more than overcomes their low density and makes MR-based imaging of these inhaled gases feasible within a single breath hold.

Figure 2.

Coronal mid lung proton MR image without (a) and with (b) hyperpolarized ¹²⁹ Xe Gas contrast (green overlay). It is obvious that proton MR imaging and the lungs are not really close friends: Conventional MR imaging excites and detects hydrogen nuclei (protons) in water. The lungs have a very low proton density and those that are present are difficult to image given their unfavorable relaxation characteristics. However, beyond the challenge of imaging parenchymal structure by MRI, there is an enormous need to image its function. This led to the development of contrast techniques using noble gases such as ³He or ¹²⁹ Xe, which nicely demonstrate the lack of ventilation in a large right upper lobe bullae (arrow) in this patient.

3.1 Gas Polarization

3.1.1. Physics

The physics around polarizing gases dates back to the atomic physics publications in the 1960’s^30,31. This technology remained of purely intellectual interest until the introduction of high power lasers afforded the ability to polarize large volumes of gases on reasonable time scales. This paved the way for the use of hyperpolarized gases in medical imaging. The schematic shown in Figure 3 describes the sequence of events involved in increasing the nuclear spin polarization of either ³He or ¹²⁹Xe. This involves two processes – 1) optical pumping, and 2) spin exchange^30,31.

Figure 3.

Schematics explain sequence involved in gas polarization. Under normal conditions, half of the nuclear spins within the gas volume are pointed up, along the magnetic field direction, and half are pointed down. This leads to zero polarization. If we put this sample in a large magnetic field (1.5 – 3.0 Tesla), it is slightly more favorable for spins to be up, but this leads to polarization of only a few parts per million. In hyperpolarization, we seek to have nearly all nuclei spins polarized in one direction. As shown in panel (a), from a physics perspective, all that is needed to transform this unpolarized sample of 3 down and 3 up spins, is to add 3 quanta of angular momentum (a) to flip the down spins to up, and polarize the sample. (b) We cannot directly flip nuclear, but we can flip electron spins in an alkali metal atom rubidium, which by absorbing angular momentum from laser photons, allows its outer-shell valence electron to become spin polarized. This process is known as optical pumping (Fig 3b). (c) Mother Nature takes care of the rest when ¹²⁹ Xe or ³He nuclei collide with Rb and transfer polarization from its valence electron to the nuclear spin of the noble gas atom. This process is known as spin exchange.

As shown in Fig 3a depicting 6 nuclear spins, under normal conditions, roughly half of the nuclear spins within the gas volume are aligned parallel to the direction of the main magnetic field and half against it. This situation leads to zero signal since the MR signal is determined by the difference in the number of spins aligned in either direction - polarization. Hence, the most ideal scenario that would give us the most signal would have all the nuclear spins pointed along one direction, or 100% polarized. If we were to put this ensemble of spins into a large magnetic field like that of a 1.5T or 3.0T scanner, it would cause slightly more spins to align with rather than against the field. However, this difference while sufficient for imaging with ubiquitous sources like water, is not nearly sufficient for imaging dilute gases like ³He or ¹²⁹Xe. An alternative way of viewing our hypothetical system is to note that we need only to add 3 quanta of angular momentum to turn every downward pointed nuclear spin to one that is up and thereby align every nuclear spin in the same direction. Thus, hyperpolarization simply involves adding angular momentum to the system.

To achieve this, circularly polarized laser light is used as a carrier of angular momentum. However, nuclei cannot directly absorb laser light, so we use an intermediary to absorb the light and its angular momentum. Specifically, laser light is tuned to the principal resonance of an alkali metal atom such as rubidium, which causes its single outer-shell valence electron to become aligned. This process is known as optical pumping (Fig 3b) ^32,33. Only alkali metal atoms with electron spins that are down can absorb the light, so simply illuminating the alkali vapor with circularly polarized resonant light will convert the entire sample to the spin up direction. Once a valence electron spin has been flipped up, it remains in this state until collisions cause it to depolarize. But then it simply absorbs another photon and returns to the aligned state.

Subsequently, through collisions between polarized electron spins of the alkali metal rubidium and the ¹²⁹ Xe or ³He nuclei, the alignment from rubidium valence electron is transferred to the noble-gas nucleus. This process is referred as spin exchange (Fig 3c)^30–33. After the spin-exchange collision, the rubidium electrons become aligned again by absorbing additional laser light and continue to build polarization within the available noble-gas nuclei. Currently available optical pumping and spin exchange techniques deliver polarizations of ~40–80% for ³He and approximately 10 – 40% for ¹²⁹ Xe. Recently, very high peak ¹²⁹Xe polarization levels have been demonstrated in diluted mixtures ³⁴

3.1.2. Gas polarization Technique

A simplified schematic for xenon polarization is shown in the Figure 4.

Figure 4.

Schematic of the device used to hyperpolarize ¹²⁹ Xe. The optical pumping and spin-exchange process occurs in the optical cell in a flowing mixture of 1% hyperpolarized ¹²⁹ Xe, 89% ⁴He and 10% N₂. Once the mixture flows out of the optical cell, ¹²⁹ Xe can be separated from the ⁴He N₂ buffer gases by exploiting the fact that xenon freezes readily at the 77 K temperature of liquid nitrogen, while the other gases remain gaseous. Once a sufficient quantity of HP ¹²⁹ Xe has been frozen and accumulated, it is thawed and evacuated into a perfluoropolymer bag for delivery to the patient and subsequent imaging. In these bags, the hyperpolarized state has about a 1 hour half life, allowing ample time for delivery.

Optical pumping, is carried out using Rubidium (Rb) which is contained within a glass optical cell. This optical cell is housed in an oven, surrounded by two Helmholtz coils that generate a small, but homogeneous 20 G magnetic field. The Rb in the optical cell is heated to ~150° C to produce a Rb vapor pressure equal to about 1ppm of the total gas density in the cell. During this time, 10–100W of circularly polarized laser light tuned to the D1 transition of rubidium illuminates the optical cell. The Rb vapor absorbs a significant fraction (>50%) this incident laser light, which polarizes the valence electron spins on the Rb atoms in the vapor.

Spin exchange is initiated when a mixture of 1% ¹²⁹Xe, 89% ⁴He and 10% N₂ is directed to flow through the optical cell and interact with the optically pumped Rb atoms. The buffer gases (helium and nitrogen) serve to pressure broaden the Rb absorption cross section to enable a large fraction of laser light to be productively absorbed. Rb-¹²⁹Xe spin exchange transfers electron spin polarization to the ¹²⁹Xe nuclei through a combination of binary collisions and the formation of transient Van der Waals complexes. The gas flow rate through the system is regulated such that ¹²⁹Xe emerges from the cell highly polarized. To now separate the ¹²⁹Xe from the ⁴He and N₂ buffer gases, it is cryogenically accumulated in a cold finger immersed in liquid nitrogen. As xenon has a higher freezing point than both these gases (~165K), it is selectively frozen out and separated from the helium and nitrogen, which are vented out of the system. Once sufficient xenon has been accumulated, the frozen xenon is thawed, and the gaseous, polarized xenon, is dispensed into a perfluoropolymer bag. The xenon polarization is then measured using a low-field NMR-based system and the delivered to the patient. Commercially available systems can produce liter quantities xenon polarized to 10–15% in 1 hour. New discoveries in polarization physics are certain to enhance both the production rate and ¹²⁹Xe polarization in the near future.

This dose of xenon is then administered to the subject in the MRI scanner. Prior to inhalation, the subject is coached to inhale to total lung capacity and exhale to functional residual capacity, twice. The contents of the bag are then inhaled through a 6-mm ID Tygon tube which is connected to a mouth piece. Typical scans use between 200-ml and 1000-ml of ¹²⁹Xe and may be mixed with a buffer gas like helium or nitrogen to create a full inhalation volume.

3.2 MR Scanner Configuration

A conventional MR scanner is typically configured to work primarily on the proton frequency. However, through minor commercially available additions to the MR hardware and software, any scanner can be used to image hyperpolarized gases. As xenon has a gyromagnetic ratio of 11.77 MHz/T which is almost a factor of ~4 lower than that of protons (42.5 MHz/T), at typical field strengths of 1.5 T, it has a much lower resonance frequency of 17.66 MHz. Because most scanners typically use a narrow-banded RF amplifier optimized only for exciting at the proton frequency, non-proton nuclei such as xenon require a separate broad-banded RF amplifier to be installed. Similarly, the receive-side of the scanner is typically also tuned to operate just at the proton frequency and thus must also be broad-banded. Fortunately, such upgrades are available from all major scanner vendors and known as a “multinuclear package”. As with most MR applications, a dedicated transmit-receive coil is also necessary, and must of course be designed to operate at the xenon frequency. It is preferred that such coils are “proton-blocked” to enable an anatomical image to be acquired using the scanners body coil without moving the patient. These RF coils are commercially available (Clinical MR Solutions, WI; Rapid Biomedical, Rimpar Germany) and at our institution we use a quadrature vest coil that delivers optimal SNR (Clinical MR Solutions, WI).

Proton MRI has the advantage of a renewable thermal signal, which affords it the utility of a multitude of MR pulse sequences. However, the non-thermal signal of hyperpolarized gases enforces a number of constraints in the choice of pulse sequences ³⁵. Firstly, once inhaled, the hyperpolarized signal has a finite lifetime in the lung (T₁) which is typically around 20 s. Note that T₁ for hyperpolarized gases refers to timescale with which the polarized state decays down to negligibly small thermal equilibrium values, rather than the conventional beneficial recovery of longitudinal magnetization. The T₁ of xenon in lung is primarily affected by dipolar interactions with paramagnetic, molecular oxygen. Fortunately, T1 effects are rendered somewhat negligible by acquiring images as rapidly as the scanner will allow. A second significant consideration is that the RF pulses used to excite the xenon signal also deplete its magnetization by the cosine of the flip-angle applied. Hence, a traditional spin echo pulse sequence using 90° and 180° RF pulses cannot generally be used for HP gas imaging. Secondly, being a gaseous contrast agent, xenon also has high diffusivity that can depend on the local gas composition (0.14 cm²/s at infinite dilution) ^23,36. This means that the gradients applied just to acquire the image can actually induce diffusion-based attenuation of the signal if not properly considered. Lastly, clinical imaging using hyperpolarized gases require that the acquisition be completed in a short breath-hold (~15 s). These factors confine hyperpolarized gas imaging to fast pulse sequences with low flip angles that can adequately sample all of k-space while minimizing image blurring. This means that most HP gas imaging is done using fast gradient echo based sequences.

4. Clinical Applications

4.1. Ventilation Imaging

The technical robustness and performance of HP ³He gas MR imaging of the lungs was confirmed in multiple clinical studies since its introduction in 1997 ^19,20. HP ³He MRI was predominately applied to produce spin density images of the gas distribution in order to quantify regions that lack any ventilation – ventilation defects. These defects may be caused by obstruction of airways as seen in diseases like asthma, or it may be due to alveolar tissue destruction as seen in emphysema. Given the wide dynamic range of signal intensities in the ventilation images, these intensities can be grouped into 4 clusters to analyze the signal distribution. Clusters of low or absent MR signal within the lungs corresponded well to ventilation defects and allowed to detect and quantify functional ventilation impairment like in asthma, COPD or cystic fibrosis (CF) patients (Fig 5). Since data acquisition is completed during a single breath-hold, this renders static ventilation information from early in the wash-in period. Other dynamic ventilation properties, such as gas flow characteristics including delayed gas filling, are not as readily attainable although significant progress has been made recently in this arena ³⁷.

Figure 5.

Quantified ventilation defects using a 4 color ventilation mask with color red representing the most impaired ventilation in a asthma patient Pre (a) and Post (b) bronchodilator therapy. From these images, the ventilation defect percentage (VDP) can be calculated. In this patient with FEV1=31%, the pre-bronchodilator VDP = 30%; post-bronchodilator, the FEV₁ remained at 31%, while VDP reduced significantly to 19%. Based on ATS criteria patient would represent a non-responder to bronchodilatation by PFT criteria, while quantified HP gas ventilation confirms a significant and positive effect to bronchodilator therapy.

Traditionally, because of the significantly larger polarization available using HP ³He compared to ¹²⁹Xe, the ³He image quality was superior. More recently, significant advances in the polarization technology and careful tuning of the MR acquisition have yielded ¹²⁹Xe ventilation images that rival those obtained with ³He. And as for ventilation defects, despite its lower signal to noise ratio, ¹²⁹Xe with its higher density and lower diffusivity than ³He seems to be more sensitive to ventilation defects ³⁸. Currently, using a somewhat larger volume of ¹²⁹Xe (up to 1 liter per scan) counterbalances the decreased SNR of ¹²⁹Xe compared to ³He (normally 0.1–0.3 liters per scan) (Fig 6).

Figure 6.

Hyperpolarized ³He ¹²⁹ Xe images of healthy volunteer, old but otherwise healthy individual, smoker without COPD, COPD, asthma and IPF patients. Hyperpolarized ¹²⁹ Xe images accurately depict different ventilation patterns related to underlying pulmonary disease.

4.2 Diffusion-weighted Imaging

Diffusion-weighted MRI has been extensively validated, and widely used with hyperpolarized gases to calculate the apparent diffusion coefficient (ADC) of the gas ^8,9,22–24. The ADC image is obtained by acquiring gas images with and without applying diffusion sensitizing gradients. The value of this contrast derives from the fact that the diffusion of the gases is highly constrained by the architecture of the normal lung. However, in diseases such as emphysema, where airspaces become significantly enlarged, the gases are free to diffuse (Fig 7). Thus, diffusion weighting can differentiate normal from enlarged airspaces by the degree of signal attenuation observed. The signal intensities in the weighted and non-weighted ventilation images are then used to calculate the ADC on a voxel-by-voxel basis. Thus, ADC maps exhibit low values in healthy lung, whereas in emphysematous lungs elevated ADC values are often observed. In fact, in addition to revealing emphysematous changes, ³He or ¹²⁹Xe ADC values have been shown to be sensitive to early microstructural changes in asymptomatic smokers ⁸ and in second hand smokers ³⁹. Even within healthy individuals, ADC MRI has been shown to be sensitive to age–related changes in alveolar size ⁴⁰. Clinical comparisons to CT densitometry ⁴¹ showed that ADC strongly correlates with DL_CO and that ADC measurements may reveal subclinical emphysematous changes before seen on High Resolution CT (HRCT) ⁸. While the majority of ADC imaging to date has involved ³He MRI, it has recently been shown that ADC imaging is also feasible with ¹²⁹Xe. Fig 8 shows examples of ¹²⁹Xe ADC images acquired in patients with COPD. It illustrates the highly elevated ADC value seen in emphysematous bullae, but also a more subtle distribution in patients where disease is less severe ²³.

Figure 7.

Diffusion-weighted imaging of hyperpolarized ¹²⁹ Xe can reveal airspace enlargement. The schematics compares diffusion of gas atoms in normal lung airspaces (a) versus enlarged airspaces (b). In normal airspaces the diffusion of ¹²⁹Xe is constrained by the normal alveolar architecture. However, when alveolar spaces become enlarged such as occurs in emphysema, ¹²⁹Xe diffusion is no longer constrained, and measured apparent diffusion coefficients (ADC) become higher.

Figure 8.

Apparent diffusion coefficient (ADC) MRI using HP ¹²⁹ Xe gas. The color scale shows escalating enlargement of airspaces from normal lung tissue (blue) to severe bullous emphysematous parenchymal destruction (red). Note parenchymal destruction depicted by CT paralleling the change of ADC values.

5. Future applications

5.1. ¹²⁹Xe Dissolving imaging

Compared to helium, xenon has a lower gyromagnetic ratio, and images have moderately lower SNR. But, the useful and unique property of ¹²⁹ Xe that sets it apart from ³He, comes by virtue of its moderate solubility in pulmonary tissues ^18,42 (Fig 9). As a result, xenon diffuses into the alveolar capillary membrane, and when it does, it experiences a distinct frequency shift from the gas-phase of about 3.5 kHz (on a 1.5T scanner) or 198 ppm in relative terms. Xenon diffuses further into the capillary blood stream, where it transiently binds with the hemoglobin in the red blood cells (RBCs), and when it does, it experiences a larger chemical shift of 217 ppm. These two resonances of xenon are known as the ‘dissolved-phase’, and as xenon follows the same gas transfer pathway as oxygen, these resonances to first order, give us information about gas-exchange in the lung. Hence, while the gas-phase resonance of xenon can be used to provide information relating to ventilatory distribution and microstructure (ADC), the dissolved-phase resonances can be used to probe diffusive gas-exchange.

Figure 9.

Hyperpolarized ¹²⁹Xe can be separately detected in airspaces, interstitial tissues, and red blood cells (RBCs). A small fraction of the inhaled ¹²⁹Xe dissolves in pulmonary tissues and blood plasma (referred to as the barrier tissues), and changes its MR frequency dramatically compared to ¹²⁹Xe left in the airspaces. When ¹²⁹Xe diffuses further into the RBCs it changes its detection frequency again. Because ¹²⁹Xe follows essentially the same pathway as oxygen, these spectroscopic properties of ¹²⁹Xe present an enormously powerful means to directly assess pulmonary gas exchange.

Imaging this dissolved-phase comes with its own set of challenges. Firstly, the magnetization or signal intensity in the dissolved-phase is only ~2% of that in the alveolar spaces, or the gas-phase. Secondly, in addition to the lower signal intensity, the T₂* of the dissolved-phase is extremely rapid, at ~2 ms ⁴³. Lastly, as the dissolved-phase resonances are ~200 ppm from the gas-phase, which on a 1.5T scanner is about 3.8 kHz, the RF excitation pulses must be made sufficiently frequency-selective so as to not excite the larger gas-phase magnetization pool.

Because of these limitations, the early efforts in imaging the dissolved-phase employed indirect methods such as Xenon polarization Transfer Contrast (XTC) ⁴⁴. As the xenon resonances are in dynamic exchange with one another, RF pulses applied to the dissolved-phase caused slight attenuation of the gas-phase signal. This concept was used to indirectly map out the dissolved-phase distribution. However, with the onset of increased polarization and rapid pulse sequences, this method soon gave way to approaches that imaged the dissolved ¹²⁹Xe directly. By using frequency-selective RF pulses and a 3D radial pulse sequence, the first dissolved phase images in humans were acquired in 2010 ⁴⁵. These images were of a lower resolution due to the small signal intensity of dissolved phase ¹²⁹Xe but already showed intriguing aspects of lung function. Soon after the introduction of this direct dissolved-phase imaging approach, Mugler et al., showed the merit of acquiring the ‘source’ gas-phase distribution in the same breath ⁴⁶. This ability allowed for the quantification of the dissolved-phase distribution. This was soon extended to using a radial acquisition strategy, which afforded the ability to quantify the impact of posture on the gas-transfer distribution ¹⁵

However, because the pathway xenon follows to reach the RBCs is identical to that of oxygen, it is the ability to separately detect ¹²⁹Xe transfer to RBCs that is of fundamental importance. The value of doing so, even on a whole-lung basis was recently demonstrated. Spectra of ¹²⁹Xe in the dissolved phase were acquired in subjects with idiopathic pulmonary fibrosis and exhibited greatly reduced ¹²⁹Xe transfer to the RBCs compared to that observed in healthy volunteers ¹⁵. This work thus showed that separating the dissolved ¹²⁹Xe resonances was critical to enable probing diffusion limitation caused by interstitial thickening in the lung. ¹²⁹Xe measurements correlated strongly with DLCO but also showed that the frequency of the ¹²⁹Xe RBC resonance may be a sensitive probe of blood oxygenation at the level of the capillary bed. While this work provided a global marker for gas-exchange impairment, it also highlighted the need to separately detect xenon uptake in the barrier tissues and RBCs by imaging.

Separate imaging of ¹²⁹Xe in barrier and RBCs is a problem that shares similarities to separating fat and water in ¹H MRI. Because the two resonances are similarly spaced, one can borrow from the vast library of fat-water separation algorithms. One such approach was employed by Qing et al., who used the Hierarchical IDEAL algorithm to image all three resonances of xenon in a single breath ⁴⁷. Alternatively, the 1-point Dixon strategy has also shown promise, and may be more robust against the short T₂* of the dissolved-phase ¹²⁹Xe signal. This technique was also recently shown to allow imaging all three resonances of xenon in a single breath ⁴⁸. One such example of images of xenon in all three compartments is shown in figure 10. Note that in the patient with IPF, numerous focal defects are visible in the Xe-RBC transfer images. These defects correspond well to fibrotic changes seen on CT.

Figure 10.

Dissolved Hyperpolarized ¹²⁹Xe MR imaging in healthy individual and patient with idiopathic pulmonary fibrosis (IPF).

(a) Recent improvements in 3D radially acquired, breath-hold images of inhaled ¹²⁹Xe now enable simultaneous depiction of ¹²⁹Xe in airspaces as well as ¹²⁹Xe transfered to the barrier and RBC compartments. (b) Note that in the patient with IPF, numerous focal defects are visible in the Xe-RBC transfer images. These defects nicely correspond to fibrotic changes seen on CT (yellow arrows).

6. Conclusions

While the field of pulmonary medicine has relied for over 100 years on the pulmonary function tests, increasing evidence is mounting that this is not sufficient. Instead, functional imaging now offers a rich world of information that is more sensitive to changes in lung structure and function than PFTs. While MRI may not historically have been the first choice to image the lungs, hyperpolarized ³He and ¹²⁹Xe have begun to change this view. These gases have provided new sensitive contrast mechanisms to probe changes in pulmonary ventilation, microstructure and gas exchange. Given the recent scarcity in the supply of helium and the associated increase in price, the field has adopted the cheaper and naturally available xenon. Xenon has been shown to be well tolerated in healthy volunteers and subjects with disease, and recent technical advances have ensured that the xenon image quality in on par with that of helium. The added advantage of xenon is that it exhibits two distinct new resonances that permit tracking its diffusion into the barrier tissues and the RBCs. Careful investigation of the amplitude, width and the frequency of these resonances carries additional important information about alveolar oxygenation that is only now beginning to be understood. Perhaps, most importantly, the past several years have now demonstrated that the day has arrived where ¹²⁹Xe can be imaged 3-dimensionally in all three compartments of the lung (airspace, barrier, RBC) to provide a fundamental view of pulmonary gas-exchange without requiring ionizing radiation. With a plethora of contrast mechanisms, hyperpolarized gases and ¹²⁹Xe in particular, stands to be an excellent probe of pulmonary structure and function, and provide sensitive and non-invasive biomarkers for a wide variety of pulmonary diseases. Interestingly, combined with recent advancements in structural imaging of the lung using ¹H MRI ⁴⁹, and 3D perfusion measurements, the day may finally be dawning that MR becomes the modality of choice for evaluating cardiopulmonary function comprehensively and non-invasively.

Key points.

• Hyperpolarized helium (³He) and xenon (¹²⁹Xe) MR imaging of the lungs provided sensitive contrast mechanisms to probe changes in pulmonary ventilation, microstructure and gas exchange.

• Recent scarcity in the supply of ³He shifted the field of hyperpolarized gas imaging to the use of cheaper and naturally available ¹²⁹Xe.

• Xenon (¹²⁹Xe) has been shown to be well tolerated in healthy volunteers and patients with various pulmonary diseases.

• Current technology allows ¹²⁹Xe to be imaged 3-dimensionally in all three compartments of the lung (airspace, barrier, red blood cells) and provides a fundamental new view of pulmonary gas-exchange.