Content uploaded by George Allan Johnson
Author content
All content in this area was uploaded by George Allan Johnson on Jul 20, 2015
Content may be subject to copyright.
Mechanical Ventilation for Imaging the Small Animal Lung
Laurence W. Hedlund and G. Allan Johnson
Abstract
This review emphasizes some of the challenges and benefits
of in vivo imaging of the small animal lung. Because me-
chanical ventilation plays a key role in high-quality, high-
resolution imaging of the small animal lung, the article
focuses particularly on the problems of ventilation support,
control of breathing motion and lung volume, and imaging
during different phases of the breathing cycle. Solutions for
these problems are discussed primarily in relation to mag-
netic resonance imaging, both conventional proton imaging
and the newer, hyperpolarized helium imaging of pulmo-
nary airways. Examples of applications of these imaging
solutions to normal and diseased lung are illustrated in the
rat and guinea pig. Although difficult to perform, pulmo-
nary imaging in the small animal can be a valuable source
of information not only for the normal lung, but also for the
lung challenged by disease.
Key Words: lung; magnetic resonance imaging; mechani-
cal ventilation; pulmonary; rodent; thorax; ventilator
Introduction
The small animal is an important subject for models of
pulmonary disease (Crapo et al. 1992; Gardner et al.
1993), and in vivo imaging of the lung is an important
tool for many of these studies (Cutillo 1996). However, the
lung is a difficult organ to image in vivo, especially in the
small animal. There are two major challenges to imaging the
lungs: they are in almost constant motion and the tissue
density is low, thus there is little substance for x-ray attenu-
ation or proton signal for magnetic resonance (MR
1
) imag-
ing. Lung imaging is also complicated by the incessant
motion of the heart. Because of the importance of the small
animal in pulmonary research, we have devoted consider-
able effort over the past 15 yr to developing noninvasive,
nondestructive methods for in vivo lung imaging in the
small animal. For this work, we have used two types of MR
studies: the more conventional method using proton (
1
H)
imaging and the more recently developed method using hy-
perpolarized
3
He imaging for pulmonary gas spaces. Al-
though the emphasis in this review is on ventilation support
and control for MR imaging of the lung, many of the same
issues and solutions apply to other noninvasive imaging
modalities such as conventional x-ray, microcomputed to-
mography, single photon emission computed tomography,
and positron emission tomography.
Mechanical ventilation is important for in vivo studies
in small animals for several reasons. Ventilation provides a
method for maintaining proper gas exchange, especially
when respiratory depressant anesthetics are used. It provides
a convenient way to administer easily controllable gaseous
anesthetics. It also provides a way to synchronize imaging
to the breathing cycle to significantly diminish the delete-
rious effects of breathing motion on resolution of images of
both the thorax and the upper abdomen. Finally, using syn-
chronous ventilation, imaging data can be captured from the
lung during specific phases of the breathing cycle, which is
especially important in studies using special gases such as
hyperpolarized helium and xenon. This brief review focuses
on the following: (1) use of mechanical ventilation for mo-
tion control for lung imaging, (2) synchronizing image data
acquisition to selected phases of the breathing cycle, and (3)
examples of application of these techniques to magnetic
resonance imaging of pulmonary models in small animals.
Control of Breathing Motion for
Lung Imaging
Breathing motion typically produces partial volume blurring
and artifacts when samples are imaged over a significant
part of the breath cycle or multiple cycles. Severity of mo-
tion degrading effects are dependent on rate and amplitude
of breathing motion, imaging frequency, and, of course,
spatial resolution. The deleterious effects of breathing
motion are typically minimized, especially in the clinical
setting, by having subjects hold their breath after a deep
inspiration. A variation on this strategy can be used with
the anesthetized small animal on mechanical ventilation.
In Figure 1 (left), image artifacts that occur when MR
acquisition occurs during spontaneous breathing of an anes-
thetized rat are shown. The weaker secondary images (ghost
artifacts) reflect the primary image and are related to phase
differences in the raw data space (Fourier) inasmuch as
the structures occupy different positions during the breath
cycle (Wood and Henkelman 1985). These artifacts, com-
bined with image blurring, almost completely obscure ana-
tomic detail. For example, also in Figure 1 (right) is an
Laurence W. Hedlund, Ph.D., and G. Allan Johnson, Ph.D., are Professors
in the Department of Radiology and in the Center for In Vivo Microscopy
at Duke University Medical Center, Durham, North Carolina.
1
Abbreviations used in this article: ADC, apparent diffusion coefficient;
HF, high frequency; HP, hyperpolarized; MR, magnetic resonance.
Volume 43, Number 3 2002 159
Figure 1 MR images of the upper abdomen of an anesthetized rat acquired during spontaneous breathing (left) and synchronous with
breathing during end-expiration (right) reveal the clearly defined body wall, spinal cord, and abdominal detail. DA, descending aorta; AD,
left adrenal gland, with stomach below, and inferior vena cava (IVC) as it courses through the liver with hepatic vasculature clearly showing.
Reprinted with permission from Hedlund LW, Johnson GA, Mills GI. 1986b. Magnetic resonance microscopy of the rat thorax and
abdomen. Invest Radiol 21:843–846.
Figure 2 Spin echo images from an anesthetized dog obtained during ventilation at 16 breaths per min, about 300 cc per breath (left) and
during high-frequency ventilation at 1620 breaths per min (27 Hz), about 3 cc per breath (right). Reprinted with permission from Hedlund
LW, Deitz J, Nassar R, Herfkens R, Vock P, Dahlke J, Kubek R, Effmann E, Putman C. 1986a. A ventilator for magnetic resonance imaging.
Invest Radiol 21:18-23.
160 ILAR Journal
image obtained while using the same imaging sequence ex-
cept the image data was acquired only when the lungs were
at end expiratory volume, at a time when there is little lung
motion and when the lungs are at their highest tissue den-
sity. The improvement in image resolution is clear because
many of the anatomic details of the upper abdomen are
easily identified.
There are several ways to synchronize imaging to the
breathing cycle. In the example described above (Figure 1),
the ventilator generated a pulse on each breath, which in
turn triggered imaging. Some commercial small animal ven-
tilators can generate an output signal to trigger external
devices. Alternatively, a signal can be detected from the
chest, such as from a chest bellows, and this signal can be
used to generate a signal for imaging (Ehman et al. 1984).
This method is useful for studies with spontaneous breath-
ing animals.
Another approach to minimizing the image-degrading
effects of breathing motion is to lessen the amplitude of
motion by using very small tidal volumes. This approach
can be used with high-frequency (HF
1
) jet ventilation,
which utilizes tidal volumes that are a fraction of conven-
tional volumes delivered at high rates (up to 25 Hz) (Froese
and Bryan 1981; Rehder et al. 1983). Very small tidal vol-
umes minimize the partial volume blurring and, at high
frequencies, normal gas exchange is maintained. We ex-
plored this approach early in our attempts to control breath-
ing motion blurring (Hedlund et al. 1986a). Figure 2 is a
comparison between ghosting artifacts during normal ven-
tilation (left, 16 breaths/min, about 300 cc/breath) and those
occurring during HF ventilation (right, 1620 breaths/min,
about 3 cc/breath). Although ghosting artifacts are less se-
vere with HF ventilation, there is sufficient image degreda-
tion to make this unacceptable. This comparison of the
impact of two types of breathing motion indicates that
ghosting artifacts in MR are not just dependent on the mag-
nitude of motion but also on its frequency in relation to
imaging frequency. Although this method is not appropriate
Figure 3 MR-compatible ventilator for small animals. Blocks or components outside the shaded area are located outside the strong fringe
field of the imaging magnet and include the Macintosh computer, gas sources, and pneumatic computer interface controller. In the shaded
area is the breathing valve, which is in the imaging magnet attached directly to the animal’s endotracheal tube. The other components shown
in this shaded box are within1mofthemagnet. There are additional components shown here that are used not for normal ventilation and
imaging but instead, for hyperpolarized gas imaging (see text). This ventilator operates as an open system (see references for details).
Reprinted with permission from Hedlund LW, Mo¨ller HE, Chen XJ, Chawla MS, Cofer GP, Johnson GA. 2000a. Mixing oxygen with
hyperpolarized
3
He for small-animal lung studies. NMR Biomed 13:202-206.
Volume 43, Number 3 2002 161
for MR imaging of the lung, it may be useful for other
imaging modalities that have lower spatial resolution and a
lower susceptibility to motion artifacts. As seen in Figure 2
(right), HF ventilation does minimize motional blurring of
other thoracic structures such as descending aorta, azygous
vein, pulmonary vessels, vena cava, and chest wall.
For small animal lung imaging in our laboratory, we
have used a form of scan synchronous ventilation for both
the normal and disease models. We use a custom-built ven-
tilator that is MR compatible. Some commercially available
small animal ventilators offer external triggering for syn-
chronizing ventilation with external events such as imaging,
and one, which we know of, is MR compatible (CWE, Inc,
Ardmore, Pennsylvania). These commercial systems do not
offer flexibility of breathing cycle control required for many
types of lung studies needed for MR imaging and spectros-
copy. To use the typical commercial ventilators with MR, it
is necessary to place them at a safe distance from the mag-
net and therefore some distance from the animal. This is
possible with long hoses (several meters) between the
breathing gas sources and the animal. However, long hoses
can create troublesome dead space problems and significant
transit times delays for inspiration and expiration. To avoid
these problems, we designed a MR-compatible ventilator
that can be used with small animals (e.g., mice to large
guinea pigs). A key component of this ventilator is a pneu-
matically controlled plastic breathing valve that attaches
directly to the endotracheal tube, thus eliminating the prob-
lem of large gas volumes in ventilator hoses between inspi-
ratory and expiratory valves and thereby decreasing dead
space. The nature of the design and control of this valve
allows us to generate a wide variety of breathing patterns
that can be accurately synchronized to image acquisition.
The current ventilator is based on one described earlier
(Hedlund et al. 1986a) and has been subsequently modified
for the small animal (Hedlund et al. 1996; Shattuck et al.
1997). Our current ventilator (Hedlund et al. 2000a,b) is
shown schematically in Figure 3. The ventilator is con-
trolled by a custom LabVIEW (National Instruments, Aus-
tin, Texas) application, operates in a Macintosh computer
(Apple Computer, Cupertino, California), and uses a digital
output board (National Instruments). This output then con-
trols electropneumatic valves that pneumatically control a
plastic breathing valve (Figure 4). The breathing valve at-
taches directly to the endotracheal tube of the animal to
minimize ventilator dead space. This ventilator is an open
system without any rebreathing of gases.
Because of the small size and internal volume of these
breathing valves (Figure 4), the dead space of the valve is
Figure 4 (Left) The upper breathing valve (scale bar 1 cm) is a four-port valve for hyperpolarized gas delivery and the lower two-port valve
system is for normal air-isoflurane delivery. Hedlund LW, Cofer GP, Owen SJ, Johnson GA. 2000a. MR-compatible ventilator for small
animals: Computer-controlled ventilator for proton and noble gas imaging. Magn Res Imaging 18:753-759. (Right) The four-port breathing
valve is attached to the endotracheal tube of an anesthetized rat. Note the electrocardiogram leads on the front paws. Reprinted with
permission from Hedlund LW, Mo¨ller HE, Chen XJ, Chawla MS, Cofer GP, Johnson GA. 2000b. Mixing oxygen with hyperpolarized
3
He
for small-animal lung studies. NMR Biomed 13:202-206.
Figure 5 Breathing patterns that are possible with the computer
controlled ventilator. Filled wave forms reveal times of gas flow
and changes in lung volumes during a single breath. The vertical
hatched bar denotes the beginning and duration of the image data
acquisition. Each imaging segment is initiated by a trigger from the
ventilator.
162 ILAR Journal
Figure 6 (Right) Breathing valve (also in Figure 4) illustrates how the inspiration valve port (IN) opens to allow for inspiration, then closes
momentarily; and how the expiration valve port (EX) opens for exhalation. The wave form plots on the left indicate the relation between
the digital computer signal and valve ports that operate to control inspiratory, expiratory gas flow, and generation of imaging triggers.
Reprinted with permission from Hedlund LW, Cofer GP, Owen SJ, Johnson GA. 2000a. MR-compatible ventilator for small animals:
Computer-controlled ventilator for proton and noble gas imaging. Magn Res Imaging 18:753-759.
Figure 7 Macintosh display of our physiological monitor reveals the following: (a) electrocardiogram; (b) airway pressure wave form; (c)
exhaled CO
2
; (d) DC output pulse to trigger scanner; and cumulative records of (e) body temperature, (f) exhaled CO
2
, and (g) heart rate.
The ventilator control panel (h) is also evident. Ventilator and monitor applications can also run on PC computers. Reprinted with permission
from Johnson GA, Turnbull DH, Fitzsimons EG, eds. 1999. In vivo microscopy: Technologies and applications. NIH Workshop for Small
Animal Imaging, Gaithersburg, MD.
Volume 43, Number 3 2002 163
approximately 1.5 to 3% of the tidal volume, which is usu-
ally 3 to 6 cc for rats or guinea pigs. The length of hose
needed to provide breathing gases from their sources is not
critical in this design because the primary breathing control
component is attached directly to the animal’s endotracheal
tube. This feature is a potential advantage for any applica-
tion where mechanical ventilation is needed but for which
access to the animal is restricted by distance or space re-
strictions. This particular ventilator design also meets the
much more stringent requirements for lung imaging with
hyperpolarized gases, as discussed below.
Some of the breathing patterns that can be generated
with this ventilator are shown in Figure 5. Each of these
patterns has advantages for certain types of studies. The
trigger generated from the ventilator computer (vertical
hatched bar) is positioned at different points in the breathing
cycle. Thus by using scan synchronous ventilation, imaging
can be restricted to occur during end-expiration (1st and 2nd
rows), when the lungs are at end-expiratory volume (func-
tional residual capacity or FRC), or during long inspiratory
gas flow (3rd line), during a short breath hold (4th line), or
during longer breath holds equal to the duration of several
breaths after a single full inspiration (last line). This last
breath pattern of extended breath hold is particularly useful
for MR spectroscopy experiments, when longer data collec-
tion times are needed (Mo¨ller et al. 2001).
These breathing patterns are computer generated by ad-
justing the timing of opening and closing of the inspiration
and expiration valve ports on the breathing valve, as shown
in Figure 6. In Figure 6 (right), the breathing valve and the
operation of the individual gas ports can be seen—inspi-
ration (IN) and expiration (EX). Also illustrated in Figure 6
(left) is how timing of opening and closing valve ports
produces these breathing patterns. For example, for the nor-
mal pattern, the inspiration port opens for a preset duration,
then closes shortly before the expiration valve port to allow
exhalation. The various breathing patterns are created by
changing the durations of inspiration and delays to the ex-
piration port opening. The pair of valve ports for hyperpo-
larized (HP
1
) gas and air input are used for hyperpolarized
gas imaging, as discussed below.
Physiological Monitoring
for In Vivo Imaging
In vivo imaging studies with MR require anesthesia for
which, to ensure safety and humane treatment, animals must
be physiologically monitored and supported. In our labora-
tory, we maintain appropriate levels of anesthesia and
support normal body temperatures with physiological moni-
toring. All of our basic procedures for imaging are nonin-
vasive and thus allow us to perform survival studies. First,
we anesthetize the animal with a very short-acting barbitu-
rate, methohexital (45 mg/kg intraperitoneally), intubate
perorally with a Quick Cath (Baxter, Deerfield, Illinois)
cut to appropriate length, insert a rectal thermistor, and
tape pediatric electrocardiogram electrodes to the paws.
Next, anesthesia is maintained with isoflurane (1.5-3.5%)
delivered with the ventilator. Electrocardiogram, airway
pressure, exhaled CO
2
, body temperature, and heart rate are
displayed on a computer monitor (Apple Computer, Cuper-
tino, California) (Figure 7) using signal processors (Coul-
bourn Instruments, Lehigh Valley, Pennsylvania), an A/D
computer board (National Instruments, Austin, Texas), and
a LabVIEW (National Instruments) application. The moni-
tor application also incorporates a feedback control loop for
automatically maintaining body temperature (Qiu et al.
1997) by using warm air circulated through the bore of the
imaging magnet. Also shown in Figure 7 is the front control
panel for the ventilator, which operates in the same com-
puter. It is essential to keep the small animal physiologically
stable during the course of imaging if in vivo imaging is to
be successful. During imaging sessions that may vary from
a fraction of an hour to several hours, the physiological
stability of the animal is maintained primarily in terms of
anesthesia level, heart rate, and body temperature. The im-
aging examples described below utilize these physiological
monitoring and control systems, as well as the ventilation
methods previously described.
Figure 8 Axial view of an anesthetized rat imaged with a projec-
tion encoding sequence and scan synchronous ventilation and elec-
trocardiogram gating. 1. Lung parenchyma; 2. lobar blood vessels;
3. pulmonary artery above number; 4. pulmonary vein above num-
ber; 5. vena cava left of number; 6. descending aorta to the right;
7. axygous vein (above); 8. left ventricle; 9. right ventricle; 10.
intrathoracic vein (lateral), intrathoracic artery (medial). Pixel size
is 195 × 195 m. Reprinted with permission from Hedlund LW,
Gewalt SL, Cofer GP, Johnson GA. 1996. MR microscopy of the
lung. In: Cutillo A, ed. Application of Magnetic Resonance to the
Study of the Lung: Armonk NY: Futura Press. p 401-415.
164 ILAR Journal
Examples of Scan Synchronous
Ventilation for MR Imaging
One example of an MR image of an anesthetized, 300-g rat
obtained by scan synchronous ventilation is shown in Figure
8. This image was acquired during end-expiration lung vol-
ume using cardiac gating and a very short echo delay time
(345 sec) projection sequence (Gewalt et al. 1993) that
captures the very weak, short-lived signal from the lungs.
The very rapid decay of the MR signal that occurs in the
lungs, and not in other organs, is due to the lung’s high
susceptibility resulting from the extensive network of gas-
tissue interfaces. There is also a lack of signal from the
airways, which is to be expected for structures devoid of
protons. The lack of motion artifacts (ghosting) from the
heart and lungs is due to the combination of projection
imaging sequence, cardiac gating, and scan synchronous
ventilation.
The importance of combined cardiac gating and scan
synchronous ventilation is also seen in Figure 9, in which
the heart of an anesthetized rat at six phases of the cardiac
cycle is shown, based on use of a protocol similar to that
used for Figure 8. The series of images were obtained at
20-msec intervals through the cardiac cycle starting from 1
msec after the QRS spike. In this example, the changes in
left ventricular wall thickness and lumen diameter can be
seen clearly, in addition to the alterations in diameters of
various blood vessels (coronary, aorta, pulmonary) as they
change over the course of the cardiac cycle.
An application of scan synchronous ventilation and car-
diac gating in a model of lung disease is shown in Figure 10.
Here we have used conventional spin echo imaging of four
rats that were exposed to 85% oxygen at one atmosphere for
1 to 14 days. Within 1 day (A), there was little observable
change from normal except increased peribronchial signal
intensity; however, by days 4 (B) and 5 (C), there were
significant pulmonary edema and pleural effusion that was
completely reversed by day 7 (D). This example clearly
Figure 9 Projection acquisition images from an anesthetized rat at six intervals of the cardiac cycle starting from the QRS spike and
progressing in 20-msec intervals (left to right) to the end of systole (middle bottom row) and then to diastole (far right bottom) 40 msec
later. Delay time was 250 micro sec. Pixel size is 195 × 195 m. Reprinted with permission from Hedlund LW, Gewalt SL, Cofer GP,
Johnson GA. 1996. MR microscopy of the lung. In: Cutillo A, ed. Application of Magnetic Resonance to the Study of the Lung: Armonk
NY: Futura Press. p 401-415.
Volume 43, Number 3 2002 165
demonstrates how the sensitivity of MR imaging to protons
of water can be used to advantage in this type of model of
lung injury.
In the next example (Figure 11), we instilled paraquat
into the left lung of anesthestized rats and imaged them 1, 7,
and 14 days later using a spin echo sequence with cardiac
gating and scan synchronous ventilation. Paraquat or methyl
viologen is known to produce an acute pulmonary edema
and chronic fibrosis. Within 1 day (Figure 11A), there was
evidence of edema by the presence of bright signal in the
left lung, and within 7 days (B), partial reabsorption is re-
vealed by reduction of the high signal area. Based on his-
tological findings, fibrosis is present by 14 days. Edema was
confirmed by wet/dry weight measurements. The unilateral
injury also results in compensatory hyperexpansion of the
contralateral normal right lung.
In all of the examples above (Figures 8-11), images
were obtained using scan synchronous ventilation with the
lung at end-expiratory volume. This is the lowest gas vol-
ume of the lung, which maximizes the proton signal and
makes images maximally sensitive to detecting changes in
water content of the lungs. However, in some instances,
performing comparison imaging of the lung at two volumes,
end-expiratory and end-inspiratory, can yield valuable pul-
monary information, as described below, when performing
hyperpolarized helium imaging of the lung.
Hyperpolarized Gas Imaging of
Pulmonary Airways
A limitation of most conventional, noninvasive in vivo im-
aging methods for the lung is that they depend on either the
Figure 10 Axial spin echo images from four anesthetized rats that had been exposed to 85% oxygen atmosphere for 1 (A), 4 (B), 5 (C),
and 7 (D) days. The increased signal intensity in the lung parenchyma in B and C is evidence of pulmonary edema, and the homogeneous
white band surrounding dorsal and lateral periphery of the lung is fluid of pleural effusion. Pixel size is 195 × 195 m. This effusion is
completely resorbed by 7 days (D). Reprinted with permission from Hedlund LW, Gewalt SL, Cofer GP, Johnson GA. 1996. MR microscopy of
the lung. In: Cutillo A, ed. Application of Magnetic Resonance to the Study of the Lung: Armonk NY: Futura Press. p 401-415.
166 ILAR Journal
presence of tissue water or hydrogen (protons) or x-ray
attenuating material. Most of the volume of the lungs, how-
ever, is composed of gases N
2
,O
2
, and CO
2
and water
vapor, which provide poor substrates for imaging. Thus,
most of the lung cannot be imaged directly by conventional
methods, inasmuch as gas spaces of the lungs and extra
pulmonary airways are seen only as voids surrounded by the
relatively low density of the lung parenchyma. However,
MR imaging of the lung changed profoundly in 1994 when
HP
3
He became available. When introduced into the lungs,
this gas provides a rich MR signal source for directly im-
aging the lung’s gas spaces. A group of physicists at Prince-
ton University, headed by William Happer, had previously
developed a laser method for polarizing
3
He and
129
Xe so
that these gases could be used as signal sources for MR
imaging (Albert et al. 1994; Happer et al. 1984; Johnson et
al. 1998). Our laboratory was fortunate to work with this
group, and together we produced the first live animal lung
image with HP
3
He (Black et al. 1996). Hyperpolarized
3
He
has the advantage of being a MR signal source that is about
10 times greater than an equivalent number of protons typi-
cally involved in conventional MR imaging, and He is not
readily absorbed by body tissues (Middleton et al. 1995).
The gas is polarized by a laser process and not by a mag-
netic field. Thus, as soon as the HP
3
He polarization is used
for imaging, the gas is no longer available as a signal
source, unlike protons in conventional MR, which are re-
polarized by the magnet. An HP
3
He imaging session must
use a fresh supply of polarized
3
He, and imaging is therefore
limited by the total amount of HP
3
He available. In our
laboratory, this amount is usually about 1.5 l and requires
about 10 hr to polarize to ∼30%. The total number of images
that can be generated in an imaging session is dependent on
the specific types of images, which can range from single
slices to 3D data sets of the entire lung.
Given the limited supply of HP
3
He, it must be delivered
very carefully to the animal. Several critical issues are in-
volved in this process. Because the gas undergoes a natural
depolarization in the reservoir before imaging use, it must
be used within several hours. Contact with metals or oxygen
hasten HP
3
He depolarization (Saam et al. 1995). We have
incorporated features in the ventilator (Figure 3) to minimize
unnecessary depolarization from oxygen and metals. For in-
stance, HP
3
He is mixed with oxygen at the last possible mo-
ment at the start of inspiratory gas flow. In addition, the HP
3
He reservoir and tubing supplying gas to the breathing valve
and breathing valve itself contain no metal parts (Hedlund et
al. 2000a). These and many other issues of hyperpolarized gas
imaging have been summarized recently (Kauczor 2000).
Examples of HP
3
He imaging are shown in Figure 12.
On the left is a 5-mm-thick proton projection acquisition of
the thorax of an anesthetized rat obtained during short
breath holds at full inspiration. The lungs are present here as
empty space devoid of signal. This example is the typical
image of the chest showing the thoracic wall and vascula-
ture. On the right is a 5-mm-thick slice from the same rat
obtained with short breath holds at full inspiration of HP
3
He. Note the appearance of only the lungs and extrapul-
monary airways, that is, only the spaces occupied by the
gas. The signal void areas around the bronchi and within the
lung are spaces occupied by major and minor blood vessels,
which are seen in the proton image (left). The lungs are seen
without superimposed soft tissue structures or vasculature,
in part because
3
He is not absorbed into the tissue. We see
Figure 11 Spin echo images from three anesthetized rats 1 (A), 7 (B), and 14 (C) days after instillation of paraquat (0.5 mg/kg) into the
left lung (right). Within 1 day, there is evidence of edema in the left lung (bright signal area); by day 7, there is partical resorption as seen
by reduction in bright signal area; and by day 14, there is fibrosis based on histological findings. Pixel size is 195 × 195 m. Reprinted
with permission from Hedlund LW, Gewalt SL, Cofer GP, Johnson GA. 1996. MR microscopy of the lung. In: Cutillo A, ed. Application
of Magnetic Resonance to the Study of the Lung: Armonk NY: Futura Press. p 401-415.
Volume 43, Number 3 2002 167
here a direct image of the intrathoracic gas spaces from the
trachea to the very distal spaces consisting of terminal bron-
chioles and alveoli. Imaging, in this case, was synchronized
by the ventilator computer to occur after the completion of
a short inspiration of HP
3
He during multiple short breath
holds (Figure 6B). This example represents the most effi-
cient way to use the very limited supply of HP
3
He. It is
possible to obtain registered images without moving the
animal by using a dual frequency imaging coil-proton (83
MHz) and
3
He (64 MHz).
With proper synchronization, it is possible to acquire
images at different phases of the breathing cycle, as seen in
Figure 13 (Chen et al. 1998). The vertical hatching in the
small icons reveals when in the breath cycle the image was
obtained. When image acquisition occurs during early in-
spiration, we are able to capture images of gas inflow before
the lungs are completely filled with gas and, as a result, we
are able to see the structure of the conductive airways more
clearly. The exquisite detail of the airway tree are easily
seen extending to approximately the 5th generation of
branching. With the image at full inspiration (right), we can
readily see the extent of complete filling of the lungs.
From early results with HP
3
He, it was clear that
3
He
imaging of the lung could be a sensitive indicator of re-
gional pulmonary ventilation. Application to the clinical
realm warrants some caution because of the great flow rate
and diffusion differences between helium and air; however,
early results revealing ventilation defects in human disease
are promising (de Lange et al. 1999; Kauczor et al. 1997).
In a study with adult guinea pigs, we tested the sensitivity of
HP
3
He to detect intraluminal localized bronchial obstruc-
tions in the lung (Hedlund et al. 1997). Figure 14 reveals
how we placed a catheter in the right or left mainstem
bronchus for injection of a small amount of a fast-drying
surgical cement to create a localized airway obstruction. On
the right is an HP
3
He gas image of the guinea pig airways
obtained during a short period of early inspiration. Clearly
shown in the right lung (image left) are the cranial, middle,
caudal, and accessory lobe bronchi; and similarly in the left
lung, we see the cranial, middle, and caudal lobe bronchi.
Figure 15 comprises two examples of placement of the
bronchial obstruction. In the first (top) example, images
were obtained in early inspiration to reveal the fine detail of
the bronchial tree, and the left caudal lobe bronchus was
blocked, as can be seen from the image on the right com-
pared with the preblockage image on the left. In the second
(bottom) example, imaging was performed during several
breath holds after full inspiration to detect any defects in
filling of the lungs. The blockage was placed in the right
caudal lobe bronchus. The postblockage image on the right
reveals that gas flow to both the accessory and the right
caudal lobes was obstructed.
It is clear that imaging the small animal lung with HP
3
He is a very effective method for determining abnormali-
ties of the gas flow and distribution in the lungs. We antic-
ipate that this method will also be valuable in studies relating
to airway constriction and dilatation in such models as asthma.
In another example of a model of lung disease, we used
HP
3
He to examine an elastase model of emphysema in the
rat (Chen et al. 2000). Elastase instilled into the trachea
Figure 12 Anesthetized rat imaged with a dual frequency coil for proton (left) and hyperpolarized
3
He (right) The proton image reveals
very little of the lung because of its low tissue density, whereas the hyperpolarized image reveals most of the lung because it is filled with
3
He. The images are 5-mm-thick slices at the same level.
168 ILAR Journal
Figure 13 Hyperpolarized
3
He images of an anesthetized guinea pig revealing airways during early inspiration (left) and full extent of gas
filling (right). Reprinted with permission from Chen XJ, Chawla MS, Hedlund LW, Mo¨ller HE, MacFall JR, Johnson GA. 1998. MR
microscopy of lung airways with hyperpolarized
3
He. Magn Res Med 39:P79-84.
Figure 14 (Left) Method of inserting a catheter through the side of the endotracheal tube and then into the right or left mainstem bronchus
to the level of caudal lobe. (Right) Airway structure from an anesthetized guinea pig.
Volume 43, Number 3 2002 169
results in the breakdown of the elastic components in the
tissue adjacent to the airspaces and an overall enlargement
of alveolar volume. Our study was based on the phenom-
enon that changes in the diffusion of
3
He can be used to
detect changes in the microstructure of the lungs. By using
diffusion-sensitive MR imaging, it is possible to measure
the apparent diffusion coefficient (ADC
1
)of
3
He. The term
apparent is used because the mobility of the gas is really a
function of the microstructure that restricts the true diffu-
sion of the gas. This measurement can be used in turn to
determine the volume available for
3
He diffusion. For in-
stance, ADC of
3
He in the trachea, where its diffusion is
Figure 15 (Top) Hyperpolarized
3
He images obtained during early inspiration, and (bottom) during multiple short breath holds after full
inspiration. In each example, the left image reveals the lungs before bronchial obstruction and the right, after obstruction is placed. In the
top pair, the result of placing an obstruction in the left caudal lobe is evident. In the bottom pair, the effect of blockage in the right caudal
lobe bronchus is shown. In this case, gas flow to both the right caudal and accessory lobes is blocked. Reprinted with permission from
Hedlund LW, Chen XJ, Chawla MS, Cofer GP, Cates G, Happer W, Wheeler CT, Johnson GA. 1997. Pulmonary airway obstruction in an
animal model: MRI detection using hyperpolarized
3
He. ISMRM 5th Scientific Meeting, p. 183.
170 ILAR Journal
relatively unrestricted, is 2.4 cm
2
/sec; whereas in the gas
exchange region of the normal lung in the alveolar spaces
where
3
He diffusion is severely restricted, the ADC is 0.16
cm
2
/sec (Chen et al. 1999b). Based on these measurements,
we reasoned that microstructural changes in the alveolar
spaces of the lung due to an emphysema-like condition
would be reflected in a change in
3
He ADC. We expected
that as these spaces were enlarged by treatment with elas-
tase to mimic panacinar emphysema, the
3
He ADC would
be increased. To test this hypothesis, we treated rats with
elastase to reduce the elastic components of the lungs and
then imaged the animals 4 wk later. We compared the ADC
of
3
He in the lungs of normal and elastase treated animals at
end-expiratory volume and during a breath hold after full
inspiration. The normal lung exhibited a significant reduc-
tion in ADC from full-inspiratory to end-expiratory volume,
as would be expected for a normal lung. In other words, the
apparent diffusion of
3
He at full inspiration, when alveoli
are fully expanded, is much greater than the ADC at end-
expiratory volume, when alveoli are at their smallest vol-
ume. However, in the elastase-treated animals that suffered
a loss in elastic properties, there was little difference be-
tween the full-inspiration and end-expiration ADC, indicat-
ing that the alveoli at end-expiratory volume remained
nearly fully expanded because of the damage to the elastic
components of the lungs. Conventional histological exami-
nation of the lung in each case confirmed the extent of
damage in the treated lungs and the normal condition of the
control animals. In this case we used HP
3
He as a tool for
quantitatively assessing changes in the lung by measuring
ADC and these changes were detected independent of vi-
sually detectable changes in the lung images. It would not
have been possible to perform these studies without the use
of HP
3
He and the ability to control image data acquisition
at two different lung volumes.
Using computer-controlled ventilation (Hedlund et al.
2000b), we are able to control the delivery of HP
3
He to the
animal carefully and to synchronize image data capture at
any phase of the breathing cycle. In Figure 16 are shown HP
3
He images from an anesthetized guinea pig obtained by
capturing imaging data in a series of 100-msec intervals
starting at the beginning of inspiration (Viallon et al. 1999).
MR imaging utilized a special radial acquisition in a CINE
mode, which collects image data dynamically and allows
observation of the inflow of HP
3
He into the lungs. Figure
16a reveals the summation of the images over the entire
800-msec acquisition. In the eight images that follow and
represent 100-msec intervals from the beginning of inspi-
ration, it is possible to visualize the HP
3
He at the beginning
of inspiration (b), as it moves from the extrapulmonary
airways in to the individual lobar bronchi (c), and finally to
the most distal gas exchange regions (d-g) at the end of
inspiration. The distribution of gas during the short breath
hold can be seen in 16h, and the beginning of exhalation can
be seen in 16i. Similarly, it is possible to observe inspiratory
gas flow in the axial view in 50-msec intervals of early
inspiration (Figure 17). With this kind of imaging and gas
delivery control, it will be possible to perform dynamic
pulmonary function analysis in the small animal, measuring
regional gas flow velocity and volume and evaluating how
these may be changed by drug treatments in models of lung
Figure 17 Hyperpolarized
3
He axial views acquired at 50-msec
intervals from the beginning of inspiration. Reprinted with per-
mission from Viallon M, Cofer GP, Suddarth SA, Mo¨ller H, Chen
XJ, Chawla MS, Hedlund LW, Cre´millieux Y, Johnson GA. 1999.
Functional MR microscopy of the lung with hyperpolarized
3
He.
Mag Res Med 41:787-792.
Figure 16 Coronal projection views acquired dynamically with
hyperpolarized
3
He from an anesthetized guinea pig. (a) Com-
posite of all eight images; (b-g) acquired at 100-msec intervals
from the start of inspiration; (h) during 100 msec of a breath hold
at full inspiration; (i) during the early phase of expiration. Re-
printed with permission from Viallon M, Cofer GP, Suddarth SA,
Mo¨ller H, Chen XJ, Chawla MS, Hedlund LW, Cre´millieux Y,
Johnson GA. 1999. Functional MR microscopy of the lung with
hyperpolarized
3
He. Mag Res Med 41:787-792.
Volume 43, Number 3 2002 171
disease. Such regionally specific information is not cur-
rently available with conventional pulmonary function tests
that provide only global assessments.
The limits of spatial resolution for proton and HP
3
He
imaging are not clearly known; however, by using both
together, we may have sufficient spatial and contrast reso-
lution to image the smallest structural units of the lung—
alveoli or a small collection of them. Early thoughts on the
limits of spatial resolution of HP
3
He imaging suggest that
the gas images would have poor resolution because of the
high diffusion rate of this
3
He in free space. However, as
previously documented (Chen et al. 1999a), the apparent
diffusion coefficient of
3
He in the structurally restricted
spaces of the lung is much lower than in unrestricted spaces.
This knowledge has led to a reassessment of the lower limits
of resolution with
3
He and, as can be seen in Figure 18,
exceptional spatial detail in the anesthetized rat lung using
3
He. Registered proton images from this same study can be
seen in Figure 19. We believe these images are the highest
resolution lung images with HP
3
He yet obtained (Johnson
et al. 2001): voxel size is 117 × 117 × 468 m. These
images clearly reveal helium within alveoli, that is, at the
very edge of the lungs. However, at this time, we cannot
claim to have resolved the individual alveolus, which is on
the order of 100 m in diameter in the rat. The ability to see
registered HP
3
He and proton images at these high resolu-
tions will no doubt be useful in future work in critically
examining the morphological changes occurring in vivo in
small animal models of pulmonary disease.
Conclusions
In this brief review we have confirmed that it is possible,
with a combination of techniques, to perform high-quality
in vivo MR imaging of the small animal lung, both in nor-
mal and disease models. The problem of lung motion can be
Figure 18 Six selected coronal slices from a three-dimensional data set from an anesthetized rat lung using HP
3
He. The most intense signal
structures here are conductive airways. Areas devoid of signal within the lung are spaces occupied by blood vessels. Acquisition time for
this set was approximately 20 min. Proton images at the same coronal levels are shown in Figure 19. Reprinted with permission from
Johnson GA, Cofer GP, Hedlund LW, Maronpot RR, Suddarth SA. 2001. Registered
1
H and
3
He magnetic resonance microscopy images
of the lung. Magn Res Med 45:365-370.
172 ILAR Journal
resolved by synchronizing image acquisition to the breath-
ing cycle. Furthermore, the effect of low tissue density of
the lung can be minimized by imaging during the time of the
highest lung density, at end-expiratory volume. The struc-
tural characteristic of the tissue-gas interfaces of lung re-
sulting in MR high susceptibility and weakening the already
poor proton signal can be resolved by using a projection
sequence with a very short echo time. Finally, the gas spaces
of the lung can now be imaged directly using hyperpolarized
helium in the breathing mixture. Thus, with the proper tools,
the small animal lung can be studied dynamically in vivo in
survival studies. This capability creates many possibilities for
basic studies of the lung and the development of better meth-
ods for small animal pulmonary research.
Acknowledgments
We thank the many researchers who have contributed to the
work described here (see cited references) and our funding
agencies (National Institutes of Health, National Center for
Research Resources [P4105959], and National Heart, Lung,
and Blood Institute). We also convey special thanks for
many years of help from Ted Wheeler for animal support
and Elaine Fitzsimons for manuscript preparation.
References
Albert MS, Cates GD, Driehuys B, Happer W, Saam B, Springer CS Jr,
Wishnia A. 1994. Biological magnetic resonance imaging using laser-
polarized 129Xe. Nature 370:199-201.
Black RD, Middleton H, Cates GD, Cofer GP, Driehuys B, Happer W,
Hedlund LW, Johnson GA, Shattuck MD, Swartz J. 1996. In vivo He-3
MR images of guinea pig lungs. Radiology 199:867-870.
Chen XJ, Chawla MS, Hedlund LW, Mo¨ller HE, MacFall JR, Johnson GA.
1998. MR microscopy of lung airways with hyperpolarized
3
He. Magn
Res Med 39:P79-84.
Chen XJ, Hedlund LW, Mo¨ller HE, Chawla MS, Maronpot RR, Johnson
GA. 2000. Detection of emphysema in rat lungs using magnetic reso-
nance measurements of
3
He diffusion. Proc Natl Acad SciUSA
97:11478-11481.
Figure 19 Proton views at the same levels as shown in Figure 18. The low signal areas bounded by high signal structures correspond to
airways seen in Figure 18, and the high signal structures seen here are large blood vessels that correspond to low signal areas in Figure 18.
Reprinted with permission from Johnson GA, Cofer GP, Hedlund LW, Maronpot RR, Suddarth SA. 2001. Registered
1
H and
3
He magnetic
resonance microscopy images of the lung. Magn Res Med 45:365-370.
Volume 43, Number 3 2002 173
Chen XJ, Mo¨ller HE, Chawla MS, Cofer GP, Driehuys B, Hedlund LW,
Johnson GA. 1999a. Spatially resolved measurements of hyperpolar-
ized gas properties in the lung in vivo. Part I: Diffusion coefficient.
Magnetic Resonance in Medicine 42:721-728.
Chen XJ, Mo¨ller HE, Chawla MS, Cofer GP, Driehuys B, Hedlund LW,
MacFall JR, Johnson GA. 1999b. Spatially resolved measurements of
hyperpolarized gas properties in the lung in vivo. Part II: T2*. Mag-
netic Resonance in Medicine 42:729-737.
Crapo J, Miller FJ, Mossman B, Pryor WA, Kiley JP. 1992. Environmental
lung diseases. Am Rev Resp Dis 145:1506-1512.
Cutillo AG, ed. 1996. Application of Magnetic Resonance to the Study of
Lung: Armonk NY: Futura Publishing Company, Inc. p 607.
de Lange EE, Mugler JP III, Brookeman JR, Knight-Scott J, Truwit JD,
Teates CD, Daniel TM, Bogorad PL, Cates GD. 1999. Lung air spaces:
MR imaging evaluation with hyperpolarized 3He gas. Radiology 210:
851-857.
Ehman R, McNamara M, Pallack M, Hricak H, Higgins C. 1984. Magnetic
resonance imaging with respiratory gating: Techniques and advantages.
Am J Radiol 143:1175-1182.
Froese AB, Bryan AC. 1981. High frequency ventilation. Am Rev Resp
Dis 123:249-250.
Gardner DE, Crapo JD, McClellan RO, eds. 1993. Toxicology of the Lung:
Target Organ Toxicology Series: New York: Raven Press. p 688.
Gewalt SL, Glover GH, MacFall JR, Hedlund LW, Johnson GA. 1993. MR
microscopy of the rat lung using projection reconstruction. Magnetic
resonance in medicine 29:99-106.
Happer W, Miron E, Schaefer S, Schreiber D, Wijngaarden W, Zeng X.
1984. Polarization of the nuclear spins of noble-gas atoms by spin
exchange with optically pumped alkali-metal atoms. Phys Rev A 29:
3092-3110.
Hedlund LW, Chen XJ, Chawla MS, Cofer GP, Cates G, Happer W,
Wheeler CT, Johnson GA. 1997. Pulmonary airway obstruction in an
animal model: MRI detection using hyperpolarized
3
He. Proceedings of
the 5th Scientific Meeting of ISMRM [author: please write out] held
month, year in city, state. p 183.
Hedlund LW, Cofer GP, Owen SJ, Johnson GA. 2000a. MR-compatible
ventilator for small animals: Computer-controlled ventilator for proton
and noble gas imaging. Magn Res Imaging 18:753-759.
Hedlund LW, Deitz J, Nassar R, Herfkens R, Vock P, Dahlke J, Kubek R,
Effmann E, Putman C. 1986a. A ventilator for magnetic resonance
imaging. Invest Radiol 21:18-23.
Hedlund LW, Gewalt SL, Cofer GP, Johnson GA. 1996. MR microscopy
of the lung. In: Cutillo A, ed. Application of Magnetic Resonance to the
Study of the Lung: Armonk NY: Futura Press. p 401-415.
Hedlund LW, Johnson GA, Mills GI. 1986b. Magnetic resonance micros-
copy of the rat thorax and abdomen. Invest Radiol 21:843-846.
Hedlund LW, Mo¨ller HE, Chen XJ, Chawla MS, Cofer GP, Johnson GA.
2000b. Mixing oxygen with hyperpolarized
3
He for small-animal lung
studies. NMR Biomed 13:202-206.
Johnson GA, Cofer GP, Hedlund LW, Maronpot RR, Suddarth SA. 2001.
Registered
1
H and
3
He magnetic resonance microscopy images of the
lung. Magn Res Med 45:365-370.
Johnson GA, Hedlund LW, MacFall JR. 1998. A new window into the
lung. Phys World: 35-38.
Johnson GA, Turnbull DH, Fitzsimons EG, eds. 1999. In vivo microscopy:
Technologies and applications. NIH Workshop for Small Animal Im-
aging, Gaithersburg, MD.
Kauczor HU, ed. 2000. Current issues in hyperpolarized gases in MRI:
Biomedical investigations and clinical applications: NMR in Biomedi-
cine. Vol 13. Philadelphia: John Wiley & Sons, Ltd.
Kauczor HU, Ebert M, Kreitner KF, Nilgens H, Surkau R, Heil W, Hof-
mann D, Otten EW, Thelen M. 1997. Imaging of the lungs using 3He
MRI: Preliminary clinical experience in 18 patients with and without
lung disease. J MRI 7:538-543.
Middleton H, Black R, Saam B, Cates G, Cofer G, Guenther B, Happer W,
Hedlund L, Johnson G, Juvan K, Swartz J. 1995. MR imaging with
hyperpolarized He-3 gas. Magn Res Med 33:271−275.
Mo¨ller HE, Hedlund LW, Chen XJ, Carey MR, Chawla MS, Wheeler CT,
Johnson GA. 2001. Measurements of hyperpolarized gas properties in
the lung. Part III:
3
He T1. Magn Res Med 45:421-430.
Qiu H, Cofer GP, Hedlund LW, Johnson GA. 1997. Automated feedback
control of body temperature for small animal studies with MR micros-
copy. IEEE Trans Biomed 44:1107-1113.
Rehder K, Schmid E, Knopp T. 1983. Long-term high-frequency ventila-
tion in dogs. Am Rev Resp Dis 128.
Saam B, Happer W,Middleton H. 1995. Nuclear relaxation of
3
He in the
presence of O
2
. Phys Rev A 52:862-865.
Shattuck MD, Gewalt SL, Glover GH, Hedlund LW, Johnson GA. 1997.
MR microimaging of the lung using volume projection encoding. Magn
Res Med 38:938-942.
Viallon M, Cofer GP, Suddarth SA, Mo¨ller H, Chen XJ, Chawla MS,
Hedlund LW, Cre´millieux Y, Johnson GA. 1999. Functional MR mi-
croscopy of the lung with hyperpolarized
3
He. Mag Res Med 41:787-
792.
Wood M, Henkelman R. 1985. NMR image artifacts from periodic motion.
Med Phys 12:143-151.
174 ILAR Journal
