Assessment of lung microstructure with magnetic resonance imaging of hyperpolarized Helium-3

https://doi.org/10.1016/j.resp.2005.05.002Get rights and content

Abstract

Magnetic resonance imaging of the apparent diffusion coefficient (ADC) of hyperpolarized Helium-3 is a new technique for probing pulmonary microstructure in vivo. The aim of this study was the assessment of potential sources of systematic errors of the ADC measurement. The influence of macroscopic motion was determined by measurements at two different delays after initiating the breath-hold, and before and after cardiac arrest. An intercentre comparison was performed in two age- and lung function-matched groups of lung-healthy volunteers at two research sites. Moreover, measurements of diffusion anisotropy were performed. We found no dependency of the ADC as a function of the delay after stop of inspiration. The influence of cardiac motion was less than 10%. In the intercentre comparison study, an excellent agreement between the two sites was found. First measurements of the diffusion tensor of intrapulmonary Helium-3 are shown.

Introduction

The microstructure of the lung is complex. This complex structure containing airways and alveolar sacs is a prerequisite for optimal gas exchange. Since lung diseases like emphysema lead to a destruction of the microstructure, there is growing interest in early detection and assessment of lung disease using non-invasive imaging methods. In general, pulmonary function tests and computed tomography are used for this purpose.

Recently, new methods based on magnetic resonance imaging (MRI) have been developed for diagnostic imaging of the lung. They can be used to measure lung microstructure by an MRI examination obtained during the inspiration of a special gas. A short review about principles and applications of this method will be given in the following paragraphs.

Pulmonary air spaces of various sizes and geometries are separated by solid tissue. Vital tissue and gases have extremely different specific densities. One cubic centimetre ( = 1 g) of water contains 6.7 × 1022 protons, which are the source of the MRI signal. Therefore, one gram of (solid) lung tissue will contain on the order of 1023 protons. In gas space on the other hand, occurrence of protons is extremely low. Air, which is saturated with water at 37 °C has a vapor pressure of 47 mmHg, i.e. the proton density is 3.3 × 1018 cm−3 or four orders of magnitude smaller than that of water (or in tissue). This extremely low density of protons needs to be compensated for in order to enable MRI of intrapulmonary gas. Moreover, the MRI signal from protons in the gas phase would not be detectable because it would superimpose with the signal from protons in solid lung tissue. In consequence, air space gives no observable signal in conventional clinical MRI.

Recently, specially prepared exogenous gases have been used successfully for MRI of lung gas space: hyperpolarized Helium-3 and hyperpolarized Xenon-129.

Hyperpolarization (Bouchiat et al., 1960, Colegrove et al., 1963) is a technique where the polarization of nuclei, another determinant of the MRI signal intensity, can be boosted by up to a factor of 105. In clinical MRI, the nuclear polarization (on the order of 10−6) is generated by the magnetic field of the MRI scanner. During the hyperpolarization procedure momentum is transferred by optical pumping from laser light to the nuclei of noble gases (polarization 50–80%). The hyperpolarized gas can be stored for up to 200 h in special glass cells (W. Heil, unpublished).

Nuclear spins of hyperpolarized gas are in a non-equilibrium state and will decay towards their thermal equilibrium polarization (10−6) because of several mechanisms: (i) relaxation by paramagnetic substances like oxygen molecules in lung gas space (Saam et al., 1999) or paramagnetic centers in the walls of the storage vessel; (ii) radiofrequency irradiation, e.g. by the radiofrequency pulses inherent to the MRI pulse sequence, and (iii) by magnetic gradient fields.

Albert et al. (1994) were the first to demonstrate that hyperpolarized Xenon-129 provides sufficient MRI signal so that its intrapulmonary distribution can be visualized. Today, most clinical applications use hyperpolarized Helium-3 (MacFall et al., 1996, Kauczor et al., 1996) because it can be polarized to a higher degree and because it yields higher signal intensity than hyperpolarized Xenon-129. MRI using hyperpolarized Helium-3 has developed to a promising technique for both visualization of ventilation defects (c.f., Fig. 1) and for functional MRI of the lung. Rapid MRI during and after inspiration of a short Helium-3 bolus gives visual (Saam et al., 1999, Schreiber et al., 2000, Wild et al., 2003) and quantitative (e.g., in mL Helium-3/mm2, Lehmann et al., 2004a) information about the temporal and spatial kinetics of the intrapulmonary gas distribution (c.f., Fig. 2). From the relaxation properties of Helium-3 in the presence of oxygen, the regional oxygen partial pressure as well as its rate of decrease during a breath-hold can be determined on a region-of-interest basis (Deninger et al., 1999, Eberle et al., 1999), after voxel binning (Deninger et al., 2002, Fischer et al., 2004) and with high spatial resolution (Lehmann et al., 2004b, Eberle and Schreiber, 2005, c.f., Fig. 3).

Helium-3 MRI permits an almost noninvasive determination of lung function. The only invasive procedure in the context of Helium-3 MRI is the administration of the Helium-3 gas. An overview of methodological and clinical studies using hyperpolarized Helium-3 and other gases can be found in review articles by Kauczor et al. (2001), Mills et al. (2003), and van Beek et al. (2004).

The high diffusion coefficient of gases can be used to probe the lung's microstructure. At 760 mmHg and room temperature (20 °C), the calculated rate of self-diffusion of pure Helium-3 is D = 1.8 ± 0.2 cm2/s (Bock, 1997), while the rate for dilute Helium-3 in N2 is D = 0.798 ± 0.018 cm2/s at 324 K and 1 atm (Liner and Weissman, 1972). Brownian motion results in a root-mean-square (rms) atomic displacement Δx along a single axis during an observation time Δobs of Δx = (2obs)1/2. For an in vivo Helium-3 diffusion measurement, 300–500 mL of hyperpolarized gas will normally be administered to a patient with approximately 6 L total lung capacity. Therefore, Helium-3 can be considered dilute and during a typical observation time of Δ = 4.6 ms, the rms displacement will be 0.86 mm. For airways which are significantly larger than this dimension Helium-3 diffusion will be unrestricted and the free diffusion coefficient D will be measured (Fig. 4a/left).

Alveoli are significantly smaller (0.086 mm radius in humans (Klimment, 1973). Thus, walls of alveoli, as well as those of bronchioles, alveolar ducts, sacs, and other branches of the airway tree, serve as obstacles to the path of diffusing Helium-3 atoms. They restrict the maximum Helium-3 displacement. Therefore, a diffusion coefficient smaller than D is measured. This is usually referred to as the apparent diffusion coefficient ADC. Only at very short observation times Δobs  (2 × 0.086 mm)2/2D  0.18 ms Helium-3 diffusion in the alveoli can be considered as free and the ADC would be equal to D. In general, such short observation times cannot be achieved on clinical MRI systems.

The ADC can be measured by the Stejskal (1965) pulsed field gradient experiment. A gradient field is a magnetic field whose amplitude G changes in a linear fashion along an arbitrary direction. The Stejskal–Tanner method comprises an MRI pulse sequence in which a special bipolar magnetic field gradient (Fig. 4b) imposes a phase shift, which is proportional to the displacement of the observed nuclei during the individual random walk of the particle. The net phase shifts of all particles within an MRI voxel lead to a decrease of the signal intensity according toS/S0=exp(bADC)where S denotes the MRI signal intensity in an image which was obtained with an MRI pulse sequence with a diffusion gradient (i.e., diffusion weighted image), and S0 the signal intensity in an image obtained with identical imaging parameters but without a diffusion gradient (i.e., reference image). b denotes the amount of diffusion weighting which is generated by the diffusion gradient. Since the bipolar field gradient can vary along any selected orientation, diffusion can be measured in any arbitrary direction. The amount of diffusion sensitization b of this trapezoidal G(t) can be calculated from its temporal variation (Basser et al., 1994; Yablonskiy et al., 2002):b=γ20t0tG(t)dt2dt=(γGDiff)2δ2Δδ3+τδ22Δδ+Δτ76δτ+815τ2where γ denotes the gyromagnetic ratio of the observed nuclei (γHelium-3 = 20.389 × 107 rad T−1 s−1), τ the ramp-up and ramp-down time of the field gradients, δ the duration of one lobe, and Δ the delay of the negative lobe with respect to the start of the positive lobe (Fig. 5). Δ is related to the previously mentioned observation time by Δobs = Δ + δ. The b-value for τ = 300 μs, δ = Δ = 2300 μs, GDiff = 12 mT/m μs b = 3.89 s/cm2. With these parameters, the signal intensity of freely diffusing Helium-3 is reduced to 4.0% of that in the reference image (c.f., Eq. (1), Fig. 5).

Chen et al. (1999) and Saam et al. (2000) were the first to report restricted diffusion in the parenchyma of healthy lungs. They reported an average ADC of 0.19 ± 0.06 cm2/s (Chen et al., 1999) and a distribution between 0.17 and 0.25 cm2/s, respectively. Differences between the ADC values arise from different species (guinea pigs versus humans) and from the selection of different parameters for the diffusion gradient. With other diffusion gradient parameters, a recent study (Morbach et al., 2005) reported average ADC values of 0.173 ± 0.045 and 0.276 ± 0.075 cm2/s (mean value ± single standard deviation) in healthy volunteers and patients with pulmonary emphysema, respectively. In the latter study the reproducibility of the ADC measurement was measured to be 2% if a global histogram evaluation was performed. In a region-of-interest (ROI) evaluation with user-selected ROIs, the reproducibility was 5.1 and 6.1% in healthy volunteers and patients with emphysema, respectively.

The ADC measurement appears to be sensitive to variations of the alveolar size due to gravitational effects (Schreiber, 2002, Fichele et al., 2004). Both studies report that the ADC is significantly higher in the non-dependent uppermost regions of the lung compared to the dependent lowermost regions of the lung. Fichele also examined posture dependent gravitational effects. The difference between non-dependent and dependent regions was 0.012 cm2/s on average, i.e. approximately 9% of the normal ADC value in a healthy lung. In that study, gravitational effects were maximal in the left-lateral decubitus position.

Most MRI voxels contain different sizes of airways. Therefore, in those voxels where a significant volume is filled by a larger airway, the diffusion should be unrestricted along the airway and may be restricted perpendicular to the orientation of the airway, i.e. an anisotropic distribution of the ADC would be expected in this case (Fig. 7). Chen et al. (1999) reported ADC measurements along three orientations (cranio–caudal, left–right, anterior–posterior) in guinea pigs but did not detect any anisotropy.

With the airways modeled as long cylinders and accounting for diffusion along (DL) and perpendicular (DT) to the airway orientation, Yablonskiy et al. (2002) demonstrated that anisotropic diffusion in humans exists on an acinar level (c.f., Fig. 6). Moreover, they were able to measure the characteristic outer diameter R of the airway. In healthy volunteers, they measured DT to be almost eight times smaller (0.09–0.13 cm2/s) than the free Helium-3 diffusion coefficient, and R (0.38–0.36 mm) was in excellent agreement with the in vitro measurements (0.35 mm) of Haefeli-Bleuer and Weibel (1988). In patients with severe emphysema nearly all DT (0.18–0.39 cm2/s) and R (0.38–0.73 mm) values were elevated when compared with those in healthy volunteers.

Peces-Barba et al. (2003) have shown in an elastase-induced panacinar emphysema model in rats that a higher ADC value is in fact related to larger geometric dimensions of the intrapulmonary gas space. They compared Helium-3 diffusion images obtained ex vivo with the alveolar internal area, a microscopic measurement of airway size. A satisfactory correlation of r = 0.71 was obtained. In patients with lung emphysema, indirect validation of the Helium-3 diffusion measurement with high-resolution computed tomography demonstrated a good (r = 0.8) correlation between the ADC value and the emphysema index derived from computed tomography (Ley et al., 2004). While correlation of the ADC with forced expiratory volume in 1 s, a parameter determined by spirometry, was moderate (r = −0.7) in that study and good in another study (r = −0.8) by Salerno et al. (2002), the latter authors found an excellent negative correlation (r = −0.93) between the ADC and the forced vital capacity.

The ADC measurement is a measurement of motion on a microscopic scale. However, in living species the lung experiences macroscopic motion during the measurement due to the beating heart. Moreover, remaining gas flow and diffusion of gases may be present at the beginning of the breath-hold period during which the MRI measurement is performed. Therefore, our first aim was to measure the Helium-3 ADC in a pig model before and after cardiac arrest, and at different times after initiating a breath-hold.

A second aim was the comparison of ADC values obtained with identical measurement techniques at two different research centers. In theory, results should be equal. However, this hypothesis has not been tested before.

A third aim was to measure if anisotropic Helium-3 diffusion occurs in healthy volunteers. The cylindrical shape of the airways (c.f., Fig. 6) imposes a non-spherical geometry on Helium-3 diffusion. Diffusion is less restricted along the airway, and severely restricted by the airway's walls perpendicular to that direction. If this type of anisotropic diffusion influences the ADC measurement on a macroscopic (i.e., voxel) level, it may result in systematic errors if data from different research sites are compared, and if the orientation of the diffusion gradient is not well controlled, or if anisotropy changes because of lung disease. Moreover, in a regional analysis of lung microstructure, knowledge of potential regional variations of diffusion anisotropy is essential to enable comparison of measured ADC values with the reference ADC value for normal, healthy tissue.

Section snippets

General experimental setup

MR measurements were performed on a Siemens Magnetom Vision whole-body MRI system (1.49 T) equipped with broadband capabilities for MRI at the Helium-3-frequency (48.44 MHz). For transmission and reception at the Helium-3 frequency a dedicated custom-built chest coil (Fraunhofer Institute, St. Ingbert, Germany) was used. The coil is a dual ring construction with a sensitive volume of 450 mm × 365 mm × 340 mm (L × W × H). It was manually tuned to the Helium-3 Larmor frequency for each subject.

For imaging, a

Influence of macroscopic motion on Helium-3 ADC

Fig. 8 shows exemplary images from these measurements. The trachea as well as the two mainstem bronchi are well visible before and after cardiac arrest, they show the expected high ADC values 0.70 cm2/s, while lung parenchyma is in the normal range of ADC values (<0.20 cm2/s).

An analysis of the pooled data of all animals showed that the median ADC measured with a 5 s delay after inspiration was 1.3% lower than the ADC of the baseline measurement performed with a 1 s delay (Fig. 9). For the two

Discussion

Hyperpolarized gas MRI is an innovative new technology for the assessment of lung function. Moreover, using the rapid diffusion of the gas atoms allows for probing pulmonary microstructure, i.e. emphysema. In this study we examined potential sources for systematic errors in the Helium-3 ADC measurement: macroscopic motion, intercentre agreement and anisotropy.

The ADC measurement is a sensitive measurement of the microscopic behavior of the gas atoms. However, sensitivity of the measurement on

Acknowledgements

This work was supported in part by the German Research Council (DFG grant # FOR 474), by the European Union (“PHIL”) and by MAIFOR. Thanks go to Florian Meise for assitance in the data analysis, to the members of PHIL and to the collaborators of the Mainz and Heidelberg pulmonary MRI research teams.

References (50)

  • M. Bock

    Simultaneous T2* and diffusion measurements with 3He

    Magn. Reson. Med.

    (1997)
  • M.A. Bouchiat et al.

    Nuclear polarization in 3He gas induced by optical pumping and dipolar exchange

    Phys. Rev. Lett.

    (1960)
  • X.J. Chen et al.

    Spatially resolved measurements of hyperpolarized gas properties in the lung in vivo. Part I: diffusion coefficient

    Magn. Reson. Med.

    (1999)
  • F.D. Colegrove et al.

    Polarization of 3He gas by optical pumping

    Phys. Rev.

    (1963)
  • T.E. Conturo et al.

    Tracking neuronal fiber pathways in the living human brain

    Proc. Natl. Acad. Sci. U.S.A.

    (1999)
  • A.J. Deninger et al.

    Assessment of a single-acquisition imaging sequence for oxygen-sensitive 3He-MRI

    Magn. Reson. Med.

    (2002)
  • B. Driehuys et al.

    High-volume production of laser-polarized Xe-129

    Appl. Phys. Lett.

    (1996)
  • Eberle, B., Ebert, M., Großmann, T., Heil, W., Kauczor, H.U., Lauer, L., Markstaller, K., Otten, E.W., Surkau, R.,...
  • B. Eberle et al.

    Analysis of intrapulmonary O2 concentration by MR imaging of inhaled hyperpolarized helium-3

    J. Appl. Physiol.

    (1999)
  • Eberle, B., Schreiber, W.G., 2005. Regional Oxygen Balance. In: Lipson D.A., van Beek, E.J.R., (Eds.). Functional Lung...
  • S. Fichele et al.

    MRI of helium-3 gas in healthy lungs: posture related variations of alveolar size

    JMRI

    (2004)
  • M.C. Fischer et al.

    Single-acquisition sequence for the measurement of oxygen partial pressure by hyperpolarized gas MRI

    Magn. Reson. Med.

    (2004)
  • B. Haefeli-Bleuer et al.

    Morphometry of the human pulmonary acinus

    Anat. Rec.

    (1988)
  • H.U. Kauczor et al.

    Pulmonary ventilation imaged by magnetic resonance: at the doorstep of clinical application

    Eur. Respir. J.

    (2001)
  • H.U. Kauczor et al.

    Normal and abnormal pulmonary ventilation: visualization at hyperpolarized He-3 MR imaging

    Radiology

    (1996)
  • Cited by (48)

    • Low-temperature heat transfer mediums for cryogenic applications

      2023, Journal of the Taiwan Institute of Chemical Engineers
    • MRI Acquisition Techniques

      2017, Hyperpolarized and Inert Gas MRI: From Technology to Application in Research and Medicine
    • Hyperpolarized <sup>3</sup>HE Gas MRI Studies of Pulmonary Disease

      2017, Hyperpolarized and Inert Gas MRI: From Technology to Application in Research and Medicine
    • Multinuclear magnetic resonance evaluation of tumor tissue

      2015, Journal of Applied Biomedicine
      Citation Excerpt :

      129Xe MR has potential as a tool for studying the brain, especially for the assessment of cerebral blood flow (CBF) related to the brain function, evaluation of lung ventilation in lung cancer. MR imaging of hyperpolarized 3He gas in the lungs have included the characterization of age related microstructure (Schreiber et al., 2005) and cystic fibrosis (Mentore et al., 2005). Recent advances in hyperpolarization have shown that hyperpolarized gases have great potential in the clinical MR (Lilburn et al., 2013).

    • Random walk simulation of the MRI apparent diffusion coefficient in a geometrical model of the acinar tree

      2009, Biophysical Journal
      Citation Excerpt :

      Following this reasoning, ADC should be expected to depend on the degree of lung inflation, since, as the lung inflates to take up more gas, alveolar sacs could be expected also to inflate (especially taking into account that it is in those structures where most of the gas resides; ∼90%). Furthermore, although the influence of experimental setup and parameters on ADC measurements has been studied (9–13), there is little understanding of how the complex combination of experimental parameters and lung tissue properties end up determining ADC values. Some of the factors which could influence ADC values include degree of lung inflation, subject posture, and gas mixing, as well as random and systematic errors related to the behavior of gradient systems and other MRI hardware, ventilating devices, data processing algorithms, etc.

    View all citing articles on Scopus

    This paper includes material from the dissertation thesis of Andreas Morbach.

    View full text