Page 1 of 28 Articles in PresS. J Appl Physiol (December 7, 2006). doi:10.1152/japplphysiol.00756.2006
Microsphere maps of regional blood flow and regional
ventilation
H. Thomas Robertson
1
Michael P. Hlastala
From: 1Departments of Medicine and of Physiology and Biophysics, University of
Washington, Seattle, WA 98195-6522
Address for correspondence: H. Thomas Robertson, Pulmonary and Critical Care
Medicine, Box 356522, University Hospital, Seattle, WA 98195-6522, Phone : 206-5433166, FAX 206-685-8673, email [email protected]
Running title: Microsphere maps of blood flow and ventilation
Copyright © 2006 by the American Physiological Society.
Downloaded from http://jap.physiology.org/ by 10.220.33.5 on June 14, 2017
1
Page 2 of 28
1
Abstract
Systematically mapped samples cut from lungs previously
labeled with intravascular and aerosol microspheres can
be
used
to
perfusion
create
and
high-resolution
regional
maps
ventilation.
of
regional
With
multiple
radioactive or fluorescent microsphere labels available,
this methodology can compare regional flow responses to
different interventions without partial volume effects or
registration errors that complicate interpretation of in
revealed
that
regional
flow
heterogeneity
increases
progressively down to an acinar level of scale.
pattern
of
characteristic
This
scale-dependent
heterogeneity
of a fractal distribution network,
is
and
suggests that the anatomic configuration of the pulmonary
vascular tree is the primary determinant of highresolution
regional
resolution,
the
flow
heterogeneity.
large-scale
gravitational
At
~2-cm3
gradients
of
blood flow per unit weight of alveolar tissue account for
less
than
5%
of
the
overall
flow
heterogeneity.
Furthermore, regional blood flow per gram of alveolar
tissue remains relatively constant with different body
positions, gravitational stresses and exercise. Regional
alveolar ventilation is accurately represented by the
deposition of inhaled 1.0-ዊm FMS aerosols, at least down
to
the
~2-cm3
level of scale.
Analysis of these
ventilation maps has revealed the same scale-dependent
property of regional ventilation heterogeneity, with a
strong correlation between ventilation and blood flow
maintained at all levels of scale.
obtained
from
microsphere
flow
The VA/Q distributions
maps of normal animals
Downloaded from http://jap.physiology.org/ by 10.220.33.5 on June 14, 2017
vivo imaging measurements.
Microsphere blood flow maps
examined at different levels of spatial resolution have
Page 3 of 28
2
agree
with
simultaneously
elimination
underestimate
technique
gas
exchange
acquired
VA/Q
multiple
inert
distributions,
impairment
in
diffuse
gas
but
lung
injury.
Key words: ventilation-perfusion ratio, radioactive microspheres, fluorescent
microspheres, aerosol
Downloaded from http://jap.physiology.org/ by 10.220.33.5 on June 14, 2017
Page 4 of 28
3
In-vivo
pulmonary
images
of
intravenously
injected
radionuclide-labeled
microspheres and radionuclide-labeled aerosol have been part of the clinical
diagnostic armamentarium for over 40 years. The resolution of the old in-vivo planar
gamma camera images was relatively weak, and lung flow mapping techniques
using experimental animals were subsequently developed to gain enhanced spatial
resolution. The microsphere mapping approach involves in-vivo administration of
intravascularly injected and/or aerosol labels, followed by fixation of the lung,
systematic cutting and mapping of the cut pieces, and measurement of label
reconstructed from this data have been employed in a wide range of experimental
studies. This review will address the advantages and limitations of this approach to
map regional pulmonary perfusion and regional ventilation, and outline the important
concepts that have arisen from studies utilizing the technique.
Advantages of microsphere maps
In
comparison
to
in-vivo imaging techniques utilizing radionuclide-labeled
microspheres (RMS) to estimate regional blood flow, destructive studies of lungs
previously labeled with intravenously injected RMS have three important
advantages. First, RMS regional blood flow measurements made on cut pieces of
lung are not influenced by partial volume artifact. Hence, provided there is adequate
signal from the label, the level of spatial resolution attainable by the destructive
approach is simply a function of the cut piece size. Second, for studies of responses
to different interventions, with each condition marked by a different injected
microsphere label, there is no ambiguity concerning the exact spatial region sampled
at the time of each measurement.
Finally, as all non-microsphere imaging
modalities to estimate regional blood flow require a mathematical model for
interpretation, the intravascular microsphere deposition method sidesteps the
assumptions of those models.
Downloaded from http://jap.physiology.org/ by 10.220.33.5 on June 14, 2017
concentrations in each piece. The spatial perfusion and/or ventilation maps that are
Page 5 of 28
4
Fluorescent microspheres (FMS) can be employed as flow markers for maps, as the
small tissue pieces sampled by this destructive methodology permit complete
extraction of fluorescence from either inhaled and deposited aerosol or intravenously
injected microspheres.
The availability of 13 different fluorescent labels has
substantially expanded investigational opportunities for simultaneous measurements
of both regional perfusion and regional ventilation with the small tissue samples (51).
The correlation for simultaneously administered different FMS colors uniformly
exceeds 0.99 for the standard ~2-cm3 lung piece size, and these measurements
correlate equally well with simultaneous measurements made with RMS (18).
Relative to RMS, FMS labels show only a modest signal overlap with combinations
inversion program (51). Properly stored FMS maintain their fluorescent intensity
over time and also remain stable over extended periods of time when injected into
experimental animals (23). Finally, FMS markers are particularly convenient to use
in studies of the lung, as the fluorescence can be completely extracted from dried
lung pieces by soaking in a solvent, without requiring tissue digestion, as is
necessary for FMS measurements in other organs (18).
Limitations of microsphere maps
An initial concern for any microsphere study of regional blood flow was whether
there might be a systematic bias, as 15-µm microspheres are large, unyielding
particles relative to the characteristics of erythrocytes. Bassingthwaighte et al (6)
utilized
mapped
measurements
of
an
atrially-injected
radionuclide-labeled
“molecular microsphere” that was fully extracted by myocardial capillary endothelium
and compared those measurements with data obtained by injection of standard 15µm RMS.
They found excellent correlation between the two regional flow
measurements, with only a modest tendency for microspheres to under-estimate the
very lowest flows and over-estimate the very highest flows.
Using an isolated
perfused lung, Beck (7) compared regional microsphere deposition with flow of
radionuclide-labeled erythrocytes, and found no significant difference between
measurements. Melsom et al (43) made a comparison of lung RMS blood flow maps
Downloaded from http://jap.physiology.org/ by 10.220.33.5 on June 14, 2017
of different labels, and as with RMS, this overlap is readily corrected by a matrix
Page 6 of 28
5
with pulmonary parenchymal extraction of a radionuclide-labeled “molecular
microsphere” and again demonstrated no significant difference between the two
measurements of regional flow.
A concern for interpretation of microsphere maps of pulmonary flow is that the lung
can only be represented by one volume, ordinarily the total lung capacity (TLC). The
in vivo apex to base density gradient in an upright lung or the in vivo ventral-todorsal density gradient in a supine lung are not appropriately represented when the
lung is extracted and air-dried inflated to TLC.
In the prone position, lung
parenchymal density is relatively uniform (33), and the most appropriate
made using in-vivo measurements acquired in the prone position.
Multiple lung samples are required to construct microsphere maps, and each of the
two generally used cutting options has pros and cons. If the entire organ is sampled
by cutting along a rectilinear grid, then the entire cardiac output and alveolar
ventilation during labeling are represented, but with the disadvantage that the
rectilinear sectioning produces a large fraction of irregularly sized pieces from the
periphery of the lung.
Without adjustment, the lower signal obtained from the
smaller edge pieces creates the appearance of a radial gradient of flow.
The
correction employed in most studies is to normalize the microsphere signal by the
weight of the air-dried piece that has been flushed clear of blood, the so-called “flow
per alveolus” correction. A problem with this convention is that it underestimates the
regional flow for lung pieces that include the relatively heavy large airways, or
regions with inflammation or scarring. Another issue with the “flow per alveolus”
correction is that for animals such as sheep and goats, with lungs heavily invested
with connective tissue, alveolar tissue content is less-well represented by tissue
weight. An alternative to weight normalization is to systematically sample a subset
of fixed-volume pieces, avoiding the most peripheral parts of lung, large conducting
airways and large vasculature. This approach has the advantage of appropriately
characterizing large-scale spatial gradients with fewer samples, and the uniform
Downloaded from http://jap.physiology.org/ by 10.220.33.5 on June 14, 2017
comparisons between microsphere maps and in-vivo measurements should be
Page 7 of 28
6
piece size provides an estimate of regional flow heterogeneity without the bias
introduced by giving identical statistical weight to tissue samples of different
volumes.
The disadvantage is that small-scale heterogeneity is less well
characterized and influences of the most peripheral and most central parts of the
lung are ignored.
Methods to create microsphere maps of flow
The initial studies utilizing measurements of intravenously injected RMS to
reconstruct microsphere flow maps of the lung employed different approaches to
lung fixation, sectioning, flow measurement and data reduction. Regardless of the
initially attributed to problems with the methods themselves.
Hogg et al (34)
compared distribution of RMS injected intravenously into dogs in different body
positions, followed by rapid freezing of the post-mortem animals.
Subsequent
transverse band saw cuts of the frozen thorax were mapped and selected cores
counted. Following intravenous microsphere injection, Greenleaf et al (28) inflated
postmortem dog lungs to TLC and air-dried the specimen before cutting it into 1-cm
transverse slices for gamma camera imaging of each slice. Beck and Rehder (8)
also studied microsphere-injected TLC-dried lungs, but systematically sampled and
mapped small cores from transverse slices of the dried specimen. Nicolaysen et al
(45) used circumferentially cut sections from partially inflated and frozen RMSlabeled lungs to refute the existence of a significant radial gradient of flow in the
lung.
After intravenous injection of a flow label, Glenny and Robertson (25)
employed in-situ supine tissue fixation and acquired gamma camera images of
transverse slices of the fixed lung.
Regardless of the approach to fixation,
sectioning and counting, the regional flow heterogeneity attained by these studies
revealed coefficients of variation ranging between 20% and 45%. These relatively
large flow heterogeneity measurements in normal animals were regarded with some
suspicion, as the lung was traditionally considered to have only modest large-scale
anatomic gradients of flow attributable to gravity, based on studies utilizing in-vivo
radionuclide images (56). In retrospect, the 0.3-cm3 to 2.5-cm3 resolution attained
Downloaded from http://jap.physiology.org/ by 10.220.33.5 on June 14, 2017
method, these studies revealed an unexpected extent of flow heterogeneity that was
Page 8 of 28
7
with the above-cited RMS studies were better than could be attained with in-vivo
radionuclide images from an earlier era, but the importance of resolution to
characterize physiologically relevant regional heterogeneity was not initially
appreciated.
Scale-dependent heterogeneity of lung blood flow
As additional high-resolution studies accumulated over time (8), (45), it became clear
that some factor unrelated to the gravitational gradient was responsible for a major
component of the spatial heterogeneity of regional blood flow.
Inspired by a
demonstration of scale-dependent heterogeneity of cardiac blood flow (37), (5) a
heterogeneity of regional blood flow increased in a similar predictable fashion as
resolution was enhanced (25). This scale-dependence of measured regional blood
flow heterogeneity can be understood as a consequence of the progressive
branching of the pulmonary arterial tree, where flows will be heterogeneous but
locally correlated.
The relationship between spatial resolution and the spatial
heterogeneity of pulmonary blood flow can be characterized by a fractal dimension
(27). In a subsequent study using an imaging cryomicrotome to obtain the spatial
coordinates of over 75,000 15-µm FMS injected into a rat lung, Glenny et al (20)
were able to characterize regional flow heterogeneity down to an acinar level of
scale. Figure 1 from that study illustrates how the logarithm of the coefficient of
variation of regional flow increases in a predictable fashion as the logarithm of
sampled volume of lung becomes progressively smaller.
The slope of the
relationship describes how well the flow in one region is correlated with the flow of its
neighboring region (27).
The steeper curve in Figure 1 demonstrates how this
relationship would differ if the distribution of microspheres in the lung were random
rather than locally correlated. In agreement with descriptions of a fractal dimension
of pulmonary blood flow in several different species (2, 12, 14, 46, 52), this study
demonstrated that the extent of measured flow heterogeneity continued to increase
as resolution improved, again following a fractal dimension that was not different
from the fractal dimension calculated from larger lung volume samples from the
Downloaded from http://jap.physiology.org/ by 10.220.33.5 on June 14, 2017
high-resolution study of lung perfusion distribution confirmed that the spatial
Page 9 of 28
8
same data set. This analysis utilizing microsphere perfusion maps established an
important concept relevant to any study of whole-organ perfusion, the property of
scale-dependent heterogeneity of flow.
Microsphere studies of large-scale flow gradients in the lung
High-resolution microsphere studies have provided important new information
concerning the regional changes in pulmonary blood flow attributable to changes in
position. Beck and Rehder (7) were the first investigators to specifically delineate
high-flow and low-flow regions of the lung whose perfusion was only minimally
altered by changes in body position. Glenny et al (22) compared regional blood flow
in “per alveolus” blood flow, in contrast to the complete reversal of flow gradients
that would be predicted if gravitational force were the only determinant of regional
lung blood flow distribution. As the lungs of larger animals would be expected to
show the largest gravitationally determined differences in blood flow, Hlastala et al
(30) conducted a study of pulmonary perfusion in awake horses. For four of the five
standing horses, higher regional blood flows were noted in the dorsal caudal part of
the lung, opposite of what would be predicted if gravity were the only determinant of
large-scale gradients. A similar predominance of posture-independent distribution of
blood flow was documented in anesthetized baboons, an animal that is ordinarily
upright and has an anatomic structure most similar to human lungs (19).
Investigating gravitational flow gradients by comparison of flows in prone and supine
postures is complicated by the compression of dorsal-caudal lung regions in
anesthetized animals in the supine posture (33).
The influence of gravity on
pulmonary blood flow distribution is best studied in a single position, with different Gforces as the independent variable. Riding in the NASA KC-135 Glenny et al (21)
administered intravascular FMS to anesthetized pigs experiencing G-forces from
parabolic flight pathways ranging between microgravity and 1.8G.
The airborne
laboratory provided repeated 22-second periods of microgravity, a sufficient time
interval to allow the FMS flow labels to be injected and lodge in the microvasculature
Downloaded from http://jap.physiology.org/ by 10.220.33.5 on June 14, 2017
in prone and supine positions with TLC-dried lungs, demonstrating only minimal shift
Page 10 of 28
9
(36). Figure 2A illustrates a normal extent of regional flow heterogeneity plotted
against height up the lung in a prone animal from that study at 1G. Figure 2B
demonstrates the shift in vertical flow distribution for the same animal during
microgravity. (For each run, the piece flows were normalized by the mean piece
flow for that gravitational condition. Overall cardiac output changed very little among
the gravitational interventions.) The large-scale gravitational gradient, indicated by
the slopes of the two lines, changes in the expected direction, but the effect is quite
small relative to the heterogeneity of flow noted within any isogravitational slice of
lung. The study compared prone and supine pulmonary blood flow distribution in
microgravity, 1G and 1.8G conditions, documenting large-scale gravitationally
the mean slope differences for each gravitational condition were small. A more
comprehensive statistical model incorporating the fixed flow effects of vascular
anatomy, posture, gravitational force, and error attributed 85% of the total variability
to structure and 2 to 3% of the observed variability in flow to the difference between
microgravity and 1G. Steeper and statistically significant large-scale gravitationally
directed gradients were demonstrable in animal centrifuge studies where larger G
forces could be generated (31). Of interest, however, was that even with centrifugeinduced forces of 3 to 6 G along the z-axis, the addition of a pressure suit to
maintain a more normal chest and abdominal wall configuration at high G-z forces
minimized the high G-z effect on the slope of the gravitational gradient within the
lung (13).
The microsphere maps demonstrating both high spatial blood flow heterogeneity and
only modest changes in large-scale flow gradients in the lung in response to different
postures or gravitational conditions suggested the hypothesis that the anatomy of
the pulmonary vasculature is the primary determinant of blood flow heterogeneity
and regional blood flow distribution, with gravitational forces contributing a relatively
small influence. This hypothesis received additional support from a modeling study
by Burrowes et al (11) that utilized high-resolution measurements of the complete
human pulmonary artery tree, predicting regional flow distribution based only on
Downloaded from http://jap.physiology.org/ by 10.220.33.5 on June 14, 2017
directed gradients of flow that increased between microgravity and 1.8G, although
Page 11 of 28
10
calculations of regional flow resistance calculated from detailed measurements of
the pulmonary artery tree. Predicted regional flow distributions were compared with
and without the force of gravity. The conclusion from this human pulmonary regional
flow model was that the influence of gravity was detectable but quite small relative to
the influence of the vascular anatomy.
Both the model study of Burrowes and
Tawhai and the microsphere maps describe the effect of gravity on “per-alveolus”
blood flow, and do not include the in-vivo effects of gravitationally mediated
parenchymal compression. An important difference between microsphere maps and
in-vivo flow images is that for supine and upright postures, the in-vivo maps show
higher flows per unit volume in dependent regions primarily because of parenchymal
density difference between non-dependent and dependent parenchyma can differ by
a factor of three (33). The important insight provided by the microsphere studies is
that the large-scale gravitational and postural changes in in-vivo blood flow
distribution measurements are primarily determined by shifts in parenchymal
density, supplemented by a small gravitationally-directed contribution to “per
alveolus” flow in the dependent lung regions.
Effects of changes in cardiac output on regional blood flow
One of the original regional blood flow findings in pump-perfused lung preparations
was that flow in the highest part of the lung would increase disproportionately with
increases in cardiac output, hence making overall flow more homogeneous (57).
Caruthers and Harris (12) provided evidence for this effect with microsphere
injections made into pump-perfused sheep lungs over a range of relatively low flows.
In intact animals with higher baseline perfusion, however, the change in spatial flow
distribution was not apparent. Parker and Hamm (46) injected microspheres into
dogs exercising at a trotting pace, and showed no difference in overall blood flow
heterogeneity between rest and exercise.
Bernard et al (9) used high aerobic
capacity thoroughbred horses to study the influence of maximal increases in cardiac
output on blood flow heterogeneity.
They noted no change in overall flow
heterogeneity between standing posture and full gallop. Large-scale gradients in the
Downloaded from http://jap.physiology.org/ by 10.220.33.5 on June 14, 2017
compression (17). In supine anesthetized animals at functional residual capacity, the
Page 12 of 28
11
exercising horses remained unchanged, with the highest blood flow consistently
delivered to the dorsal caudal portion of the lung during exercise. Melsom et al (41)
measured blood flow distribution in resting and exercising sheep and demonstrated
a modest redistribution of blood flow to the dorsal portion of the lung with exercise.
Studies of pharmacologic manipulation of cardiac output in anesthetized baboons
demonstrated no significant differences in flow heterogeneity or large-scale flow
gradients (26). Thus in intact animals, increasing the cardiac output above normal
resting levels has little to no influence on spatial pulmonary blood flow distribution.
These findings again support the hypothesis that in intact animals with normal or
increased cardiac outputs, the anatomic configuration of the pulmonary vasculature
Temporal changes in pulmonary blood flow distribution
Repeated measurements of flow distribution with injected microspheres have shown
good reproducibility, but a component of variability within pieces develops over time,
with the largest proportional changes most likely to occur in the low-flow pieces (24).
Measurement error also contributes to between-run differences in flow, but the
extent of this artifact can be estimated by simultaneous administration of two
different RMS or FMS labels (10), with the residual variability attributable to true
temporal shifts.
For studies of anesthetized dogs and pigs, the contribution of
temporal shifts to overall flow heterogeneity has generally ranged between 7 to 17%
of the summed heterogeneity (2, 23, 24). In response to the question of whether any
structure could be discerned within these temporal shifts, Glenny et al (24) applied
the statistical technique of cluster analysis to regional pulmonary blood flow
measurements in anesthetized dogs.
They demonstrated that these small
spontaneous temporal shifts in regional flow were not random, but rather took place
among lung pieces adjacent to each other. As might be expected from a branching
distribution system, when one region gained in relative flow, its immediate neighbor
tended to lose flow. The size of the units undergoing these spontaneous shifts was
larger than the ~2-cm3 piece size, generally in the 20-cm3 range. Some fraction of
the temporal variability observed in anesthetized animals may have been attributable
Downloaded from http://jap.physiology.org/ by 10.220.33.5 on June 14, 2017
is the primary determinant of regional flow distribution.
Page 13 of 28
12
to effects anesthesia or mechanical ventilation, as sequential FMS measurements of
regional pulmonary blood flow in awake, standing dogs over one week showed a
temporal coefficient of variation of only 7% (23). In this study of awake animals, the
lung pieces with the greatest variability still tended to be the low-flow pieces, but the
awake animals generally demonstrated less temporal variability than that previously
described with anesthetized animals.
Pulmonary vascular response to hypoxia
While the pulmonary vascular hemodynamic response to hypoxia has been studied
for decades, the issue of whether there might be regional differences in response
Melsom et al first
demonstrated consistent region-dependent changes in blood flow in response to
hypoxia in awake sheep (41). Hlastala et al (32, 39, 53) investigated the regional
blood flow response to different levels of hypoxia in both anesthetized dogs and
pigs.
Applying the cluster analysis approach previously utilized in the temporal
change study cited above, Hlastala et al were able to demonstrate relatively
stereotypical regional changes in flow distribution among all animals, with specified
regions receiving more blood flow during hypoxia and other regions receiving less
flow. These animal findings were relevant to the interpretation of the human highaltitude pulmonary edema, where a heterogeneous pulmonary vascular response to
hypoxemia has been postulated to explain the patchy nature of the pulmonary
edema observed in that condition.
Microsphere aerosols to map regional ventilation
Zeltner et al (58) administered a dilute aerosol of 0.9-µm FMS to guinea pigs, and
documented with histology that the microsphere deposition was almost exclusively
alveolar. This suggested that an FMS aerosol in that size range might be used to
produce maps of regional ventilation, provided an adequate fluorescence signal
could be obtained from post-mortem cut lung pieces. Both Melsom et al (43) and
Robertson et al (49) found that a five-minute administration of an inhaled 1.0-µm
FMS aerosol to goats or pigs yielded an adequate fluorescent signal from ~2cm3
Downloaded from http://jap.physiology.org/ by 10.220.33.5 on June 14, 2017
had not been apparent from in-vivo imaging studies.
Page 14 of 28
13
pieces of air-dried lung.
Measurements utilizing simultaneous administration of
different FMS aerosol labels demonstrated reproducibility that was comparable to
the earlier regional blood flow measurements made with the 15-µm injected FMS.
The aerosol deposition on large conducting airways was minimal relative to the total
fluorescence signal in the lung, so the proportional deposition of fluorescent label
among the subsequently cut pieces appeared to be an appropriate estimate of
regional alveolar ventilation (47).
These measurements of regional ventilation
demonstrated that the heterogeneity of ventilation was comparable to that described
for regional blood flow, at least down to the level of the ~2-cm3 piece size chosen for
these studies.
An important consequence of development of the FMS aerosol
simultaneous injection of intravenous 15-µm FMS labels, high-resolution maps of
VA/Q distribution could be created.
The validity of aerosol deposition measurements to represent high-resolution maps
of ventilation remains to be firmly established.
Although Melsom et al (43)
demonstrated equivalent microsphere ventilation map measurements with the 1.0µm microspheres and the ultra-fine particles of Technegas, even the smallest
aerosol will distribute differently from gas molecules at an alveolar level of scale (29).
While the highest resolution where aerosol deposition can still be interpreted as a
valid measure of regional ventilation remains to be determined, the results from FMS
ventilation maps at the ~2-cm3 level of scale can be compared to other highresolution methods for regional ventilation measurement. Kreck et al (38) developed
a regional ventilation model that utilized sequential CT lung images acquired during
the washin of 65% Xenon to create a high-resolution a map of regional ventilation.
Robertson et al (50) compared regional FMS ventilation measurements in five
anesthetized sheep with measurements with the CT xenon model described by
Kreck et al, where the CT analysis volume was adjusted to the ~2-cm3 size of the cut
lung pieces. Since the CT images were acquired with sheep imaged in the supine
posture, there were minor lung shape discrepancies, but the overall extent of
heterogeneity described by the two methods was comparable, and large-scale
Downloaded from http://jap.physiology.org/ by 10.220.33.5 on June 14, 2017
method for measurement of regional ventilation was that when paired with
Page 15 of 28
14
gradients of ventilation were likewise comparable. Hence at least at the ~2-cm3
level of scale, the aerosol microsphere maps of regional ventilation agree
reasonably well with an independent technique, but confirmation of these findings at
a higher level of resolution is still needed.
The ventilation maps obtained using the FMS aerosol allowed exploration of the
small scale and large-scale properties of regional ventilation. Altemeier et al (1)
performed fractal analysis of ventilation maps of prone pigs and demonstrated a
consistent scale-dependent progression in spatial heterogeneity as the volume
analyzed became smaller. The fractal dimension calculated from this ventilation
confirming that the rate of increase in spatial heterogeneity with decreasing scale
was comparable for the two parameters.
One appreciable difference between
microsphere ventilation and perfusion maps has been noted in large-scale
differences in regional ventilation between prone and supine postures. Altemeier et
al (2) showed that there was a large-scale decrement in dorsal-caudal aerosol
deposition in the anesthetized supine pig that was not noted in the prone position.
This marked regional decrease in ventilation was associated with a lesser dorsalcaudal decrement in perfusion that accounted for the increased hypoxemia noted in
the supine posture.
This study described the disproportionate gas exchange
influence exerted by the supine compression of the dorsal-caudal portion of the lung
in anesthetized animals, an issue that had not been separately analyzed in earlier
prone-supine comparisons of regional blood flow (22, 40).
FMS studies of VA/Q distribution
Simultaneous FMS measurements of both regional ventilation and regional blood
flow have clarified how the high extent of heterogeneity noted in both parameters
could still yield normal gas exchange. For animals examined at a fixed level of
scale, there is appreciable variability among animals in the extent of spatial
heterogeneity of regional ventilation and regional perfusion, but for any single
animal, the extent of heterogeneity of the two parameters is roughly comparable.
Downloaded from http://jap.physiology.org/ by 10.220.33.5 on June 14, 2017
data was comparable to the fractal dimension calculated for regional blood flow,
Page 16 of 28
15
Within animals, the correlation between regional ventilation and regional blood flow
has ranged between 0.76 and 0.92 (1, 49) with the higher correlations noted in those
animals with the greatest overall extent of VA and Q heterogeneity.
Figure 3
illustrates the range of heterogeneity of both ventilation and blood flow and the tight
correlation between the two parameters in a prone anesthetized pig from the study
of Robertson et al (49). Melsom et al (43) demonstrated comparable heterogeneity
and correlations between VA and Q in microsphere measurements in awake goats.
As a VA/Q measurement is made for every piece of the cut lung, calculations utilizing
the VA/Q distribution, measurements of mixed venous blood gases, cardiac output
normal anesthetized pigs, making FMS measurements of VA and Q and also
acquiring gas exchange data using the multiple inert gas elimination technique
(MIGET) (55). The VA/Q distributions derived from FMS and MIGET techniques
showed excellent agreement, with both calculated distributions able to predict
arterial blood gas values. Furthermore, utilizing the FMS VA/Q distributions and the
mixed venous inert gas concentrations, Altemeier et al were able to accurately
predict the individual inert gas retention and excretion values as an additional
validation of the concurrence between the FMS VA/Q distributions and gas exchange
characteristics of the lungs (3). However the agreement between calculations based
on FMS VA/Q distributions and MIGET measurements was weaker in both a bead
embolization injury model (4) and an endotoxin injury model (16). For both of those
studies, the calculations based on FMS data underestimated the extent of gas
exchange abnormality, and the authors postulated that the experimental injuries
produced a spatially distributed injury that could not be adequately represented by
the FMS method because injury was heterogeneous within the ~2-cm3 lung piece
size.
The comparison between FMS VA/Q distributions and MIGET VA/Q distributions in
normal animals (3) used FMS data with ~2-cm3 pieces of lung.
While the
heterogeneity of perfusion is scale-dependent below the resolution of that study, the
Downloaded from http://jap.physiology.org/ by 10.220.33.5 on June 14, 2017
and ventilation can be used to predict gas exchange. Altemeier et al (3) studied
Page 17 of 28
16
agreement of results between techniques for normal lungs suggests that overall
VA/Q heterogeneity may not exhibit the same scale-dependent property shown by its
two components. It seems plausible that the heterogeneity of ventilation at smaller
scales may increase in parallel with the heterogeneity of perfusion but that the
correlation between the two parameters remains stable and scale-independent.
While ventilation heterogeneity has not been measured at a scale smaller than ~2cm3, Robertson et al (48) applied the fractal ventilation data set of Altemeier et al (1),
to examine a range of different volumes of regional ventilation and perfusion
measurements in the same animal. Although the heterogeneity of both VA and Q
increased as the lung was subdivided into smaller and smaller volumes, the overall
cm3 level of scale shows that VA/Q heterogeneity does not demonstrate the scaledependent properties of its two components, or that the correlation between VA and
Q in a normal lung remains relatively constant regardless of the resolution attained
for the maps of flow.
Studies of the posture-related large-scale shifts in regional VA/Q using microsphere
techniques should yield similar findings to in vivo VA/Q imaging methods because in
vivo tissue compression will affect regional measurements of both VA and Q equally.
Mure et al (44) utilized combined FMS measurements of ventilation and blood flow in
prone and supine positions to show that VA/Q distribution was more uniform in the
prone position.
Utilizing multiple measurements performed in prone and supine
positions in anesthetized pigs, Altemeier (2) showed that regional “per alveolus”
ventilation was appreciably reduced in the dorsal-caudal regions of lung in the
supine position. That study also established the respective quantitative contributions
of posture, time, and experimental error to the heterogeneity of both regional
ventilation and regional blood flow. As with earlier studies, the major contribution to
regional heterogeneity of both VA and Q was fixed or anatomic, comprising 74% and
63% of the total heterogeneity respectively, with posture contributing 16% and 24%
and temporal changes accounting for the balance.
Downloaded from http://jap.physiology.org/ by 10.220.33.5 on June 14, 2017
VA/Q distribution did not broaden. Hence at least analysis of data down to the ~2-
Page 18 of 28
17
Given the heterogeneity of regional VA and Q in a normal lung, a simple mechanism
to explain the strong match of the two parameters would be to propose that the
heterogeneity of ventilation is a fixed property established during growth, and that
active pulmonary vasoconstriction adjusts flow to those fixed characteristics. This
hypothesis was tested by Melsom et al (42), utilizing microsphere maps to
characterize VA/Q distributions in awake sheep before and after inhalation of nitric
oxide. They noted no significant change in VA/Q distribution after NO inhalation.
Glenny et al (26) studied the distribution of perfusion alone in normal anesthetized
baboons before and after the infusion of PGI2, a potent pulmonary vasodilator, and
showed no appreciable change in perfusion distribution or A-aO2 difference. While
to VA in a normal lung, it still remains possible that long-sustained pulmonary
vasoaction might trigger morphologic adjustments in vascular structure to account
for the normal degree of VA/Q homogeneity.
Directions for future investigation:
The high-resolution microsphere flow maps of normal lung have revealed an
unexpected extent of ventilation and blood flow heterogeneity in an organ that
appears relatively uniform at a microscopic level. This heterogeneity is spatially
organized so that ventilation and blood flow remain tightly matched, as required for
normal gas exchange. High-resolution in-vivo images of the pulmonary arterial tree
are being incorporated in flow models to suggest that the anatomy of the pulmonary
arterial tree may be the primary determinant of this blood flow heterogeneity.
However these anatomically-based modeling studies have yet to be linked to data
acquired with microsphere flow maps, and this step will be essential to strengthen
the hypothesis that regional blood flow heterogeneity has a primary anatomic
explanation.
Explaining the basis of ventilation heterogeneity is more complex than perfusion
heterogeneity because regional ventilation is influenced both by conducting airway
anatomy and local compliance. Studies with pleural capsules (15) and implanted
Downloaded from http://jap.physiology.org/ by 10.220.33.5 on June 14, 2017
these findings appear to exclude simple vasoconstriction as the mechanism tuning Q
Page 19 of 28
18
lung markers (35) have demonstrated substantial small-scale variability in local
compliance.
Although it seems likely that local airways conductance and local
compliance are correlated, this has not been directly demonstrated. Hence, model
studies of regional ventilation based on airway dimensions alone are not likely to
agree well with microsphere maps, but those maps together with airway model
predictions may provide some insight into the correlation between local compliance
and airways conductance.
Finally, the aerosol ventilation maps are of uncertain
validity at higher resolution, for at some yet-to-be determined level of scale, the
extent of 1.0-µm aerosol deposition will fail to represent local ventilation. Better
estimates may come from smaller aerosols, and FMS particles in the 0.04-µm range,
regional ventilation.
Microsphere VA/Q maps have been applied to predict gas exchange abnormalities in
pulmonary embolus and pulmonary edema models, but calculations based on the
maps have underestimated the extent of hypoxemia (4, 16, 54).
It appears that
higher resolution maps will be required to appropriately predict the gas exchange
consequences of diffuse lung injury. In addition, most current studies of disease
models utilize small animals, again requiring high-resolution data.
The
cryomicrotome system used to produce the acinar-level maps of regional blood flow
in rat lungs (20) can image deposited FMS aerosols, offering the potential for highresolution ventilation measurements. However the aerosol particles are far below
the resolving power of the cryomicrotome system, and each cryomicrotome aerosol
image is homogenized by signal from aerosol deeper within the specimen. To date,
no correction for this homogenizing effect has been developed to provide accurate
local measurements of aerosol deposition. Hence new approaches are needed for
analysis of microscopic aerosol images to provide valid high-resolution VA/Q maps
for small animal disease models and for studies of mechanisms of VA/Q matching
during growth.
Grant Support
Downloaded from http://jap.physiology.org/ by 10.220.33.5 on June 14, 2017
now commercially available, may provide additional insight into high-resolution
Page 20 of 28
19
Preparation of this review was supported by National Institutes of Health Grants HL24163 and HL-073598
Downloaded from http://jap.physiology.org/ by 10.220.33.5 on June 14, 2017
Page 21 of 28
20
Bibliography
Downloaded from http://jap.physiology.org/ by 10.220.33.5 on June 14, 2017
1.
Altemeier WA, McKinney S, and Glenny RW. Fractal nature of regional
ventilation distribution. J Appl Physiol 88: 1551-1557, 2000.
2.
Altemeier WA, McKinney S, Krueger M, and Glenny RW. Effect of posture
on regional gas exchange in pigs. J Appl Physiol 97: 2104-2111, 2004.
3.
Altemeier WA, Robertson HT, and Glenny RW. Pulmonary gas-exchange
analysis by using simultaneous deposition of aerosolized and injected
microspheres. J Appl Physiol 85: 2344-2351, 1998.
4.
Altemeier WA, Robertson HT, McKinney S, and Glenny RW. Pulmonary
embolization causes hypoxemia by redistributing regional blood flow without
changing ventilation. J Appl Physiol 85: 2337-2343, 1998.
5.
Bassingthwaighte JB, King RB, and Roger SA. Fractal nature of regional
myocardial blood flow heterogeneity. Circ Res 65: 578-590, 1989.
6.
Bassingthwaighte JB, Malone MA, Moffett TC, King RB, Little SE, Link JM,
and Krohn KA. Validity of microsphere depositions for regional myocardial flows.
Am J Physiol 253: H184-193, 1987.
7.
Beck KC. Regional trapping of microspheres in the lung compares well
with regional blood flow. J Appl Physiol 63: 883-889, 1987.
8.
Beck KC and Rehder K. DIfferences in regional vascular conductances in
isolated dog lungs. J Appl Physiol 61: 530-538, 1986.
9.
Bernard SL, Glenny RW, Erickson HH, Fedde MR, Polissar N, Basaraba
RJ, and Hlastala MP. Minimal redistribution of pulmonary blood flow with
exercise in racehorses. J Appl Physiol 81: 1062-1070, 1996.
10.
Buckberg GD, Luck JC, Payne BD, Hoffman JIE, Archie JP, and Fixler
DE. Some sources of error in measuring regional blood flow with radioactive
microspheres. J Appl Physiol 31: 598-604, 1971.
11.
Burrowes KS and Tawhai MH. Computational predictions of pulmonary
blood flow gradients: Gravity versus structure. Respir Physiol Neurobiol, 2005.
12.
Caruthers SD and Harris TR. Effects of pulmonary blood flow on the
fractal heterogeneity in sheep lungs. J Appl Physiol 77: 1474-1479, 1994.
13.
Chornuk MA, Bernard SL, Burns JW, Glenny RW, Sheriff DD, Sinclair SE,
Polissar NL, and Hlastala MP. Effects of inertial load and countermeasures on
the distribution of pulmonary blood flow. J Appl Physiol 89: 445-457, 2000.
14.
Deem S, Hedges RG, McKinney S, Polissar NL, Alberts MK, and
Swenson ER. Mechanisms of improvement in pulmonary gas exchange during
isovolemic hemodilution. J Appl Physiol 87: 132-141, 1999.
15.
Fredberg JJ, Keefe DH, Glass GM, Castile RG, and Frantz ID, 3rd.
Alveolar pressure nonhomogeneity during small-amplitude high-frequency
oscillation. J Appl Physiol 57: 788-800, 1984.
16.
Gerbino AJ, McKinney S, and Glenny RW. Correlation between ventilation
and perfusion determines VA/Q heterogeneity in endotoxemia. J Appl Physiol 88:
1933-1942, 2000.
Page 22 of 28
21
Downloaded from http://jap.physiology.org/ by 10.220.33.5 on June 14, 2017
17.
Glazier JB, Hughes JM, Maloney JE, Pain MC, and West JB. Decreasing
alveolar size from apex to base in the upright lung. Lancet 2: 203-204, 1966.
18.
Glenny RW, Bernard S, and Brinkley M. Validation of fluorescent-labeled
microspheres for measurement of regional organ perfusion. J Appl Physiol 74:
2585-2597, 1993.
19.
Glenny RW, Bernard S, Robertson HT, and Hlastala MP. Gravity is an
important but secondary determinant of regional pulmonary blood flow in upright
primates. J Appl Physiol 86: 623-632, 1999.
20.
Glenny RW, Bernard SL, and Robertson HT. Pulmonary blood flow
remains fractal down to the level of gas exchange. J Appl Physiol 89: 742-748,
2000.
21.
Glenny RW, Lamm WJ, Bernard SL, An D, Chornuk M, Pool SL, Wagner
WW, Jr., Hlastala MP, and Robertson HT. Selected contribution: redistribution of
pulmonary perfusion during weightlessness and increased gravity. J Appl Physiol
89: 1239-1248, 2000.
22.
Glenny RW, Lamm WJE, Albert RK, and Robertson HT. Gravity is a minor
determinant of pulmonary blood flow distribution. J Appl Physiol 71: 620-629,
1991.
23.
Glenny RW, McKinney S, and Robertson HT. Spatial pattern of pulmonary
blood flow distribution is stable over days. J Appl Physiol 82: 902-907, 1997.
24.
Glenny RW, Polissar NL, McKinney S, and Robertson HT. Temporal
heterogeneity of regional pulmonary perfusion is spatially clustered. J Appl
Physiol 79: 986-1001, 1995.
25.
Glenny RW and Robertson HT. Fractal properties of pulmonary blood
flow: characterization of spatial heterogeneity. J Appl Physiol 69: 532-545, 1990.
26.
Glenny RW, Robertson HT, and Hlastala MP. Vasomotor tone does not
affect perfusion heterogeneity and gas exchange in normal primate lungs during
normoxia. J Appl Physiol 89: 2263-2267, 2000.
27.
Glenny RW, Robertson HT, Yamashiro S, and Bassingthwaighte JB.
Applications of fractal analysis to physiology. J Appl Physiol 70: 2351-2367,
1991.
28.
Greenleaf JF, Ritman EL, Sass DJ, and Wood EH. Spatial distribution of
pulmonary blood flow in dogs in left decubitus position. Am J Physiol 227: 230244, 1974.
29.
Henry FS, Butler JP, and Tsuda A. Kinematically irreversible acinar flow: a
departure from classical dispersive aerosol transport theories. J Appl Physiol 92:
835-845, 2002.
30.
Hlastala MP, Bernard SL, Erickson HH, Fedde MR, Gaughan EM,
McMurphy R, Emery MJ, Polissar N, and Glenny RW. Pulmonary blood flow
distribution in standing horses is not dominated by gravity [see comments]. J
Appl Physiol 81: 1051-1061, 1996.
31.
Hlastala MP, Chornuk MA, Self DA, Kallas HJ, Burns JW, Bernard S,
Polissar NL, and Glenny RW. Pulmonary blood flow redistribution by increased
gravitational force. J Appl Physiol 84: 1278-1288, 1998.
Page 23 of 28
22
Downloaded from http://jap.physiology.org/ by 10.220.33.5 on June 14, 2017
32.
Hlastala MP, Lamm WJ, Karp A, Polissar NL, Starr IR, and Glenny RW.
Spatial distribution of hypoxic pulmonary vasoconstriction in the supine pig. J
Appl Physiol 96: 1589-1599, 2004.
33.
Hoffman EA. Effect of body orientation on regional lung expansion: a
computed tomographic approach. J Appl Physiol 59: 468-480, 1985.
34.
Hogg JC, Holst P, Corry P, Ruff F, Housley E, and Morris E. Effect of
regional lung expansion and body position on pulmonary perfusion in dogs. J
Appl Physiol 31: 97-101, 1971.
35.
Hubmayr RD, Walters BJ, Chevalier PA, Rodarte JR, and Olson LE.
Topographical distribution of regional lung volume in anesthetized dogs. J Appl
Physiol 54: 1048-1056, 1983.
36.
Jaryszak EM, Baumgartner WA, Jr., Peterson AJ, Presson RG, Jr.,
Glenny RW, and Wagner WW, Jr. Selected contribution: measuring the response
time of pulmonary capillary recruitment to sudden flow changes. J Appl Physiol
89: 1233-1238, 2000.
37.
King RB, Bassingthwaighte JB, Hales JR, and Rowell LB. Stability of
heterogeneity of myocardial blood flow in normal awake baboons. Circ Res 57:
285-295, 1985.
38.
Kreck TC, Krueger MA, Altemeier WA, Sinclair SE, Robertson HT, Shade
ED, Hildebrandt J, Lamm WJ, Frazer DA, Polissar NL, and Hlastala MP.
Determination of regional ventilation and perfusion in the lung using xenon and
computed tomography. J Appl Physiol 91: 1741-1749, 2001.
39.
Lamm WJ, Starr IR, Neradilek B, Polissar NL, Glenny RW, and Hlastala
MP. Hypoxic pulmonary vasoconstriction is heterogeneously distributed in the
prone dog. Respir Physiol Neurobiol 144: 281-294, 2004.
40.
Melsom MN, Flatebo T, Kramer-Johansen J, Aulie A, Sjaastad OV,
Iversen PO, and Nicolaysen G. Both gravity and non-gravity dependent factors
determine regional blood flow within the goat lung. Acta Physiol Scand 153: 343353, 1995.
41.
Melsom MN, Flatebo T, and Nicolaysen G. Hypoxia and hyperoxia both
transiently affect distribution of pulmonary perfusion but not ventilation in awake
sheep. Acta Physiol Scand 166: 151-158, 1999.
42.
Melsom MN, Flatebo T, and Nicolaysen G. No apparent effect of nitric
oxide on the local matching of pulmonary perfusion and ventilation in awake
sheep. Acta Physiol Scand 168: 361-370, 2000.
43.
Melsom MN, Kramer-Johansen J, Flatebo T, Muller C, and Nicolaysen G.
Distribution of pulmonary ventilation and perfusion measured simultaneously in
awake goats. Acta Physiol Scand 159: 199-208, 1997.
44.
Mure M, Domino KB, Lindahl SG, Hlastala MP, Altemeier WA, and Glenny
RW. Regional ventilation-perfusion distribution is more uniform in the prone
position. J Appl Physiol 88: 1076-1083, 2000.
45.
Nicolaysen G, Shepard J, Onizuka M, Tanita T, Hattner RS, and Staub
NC. No gravity-independent gradient of blood flow distribution in dog lung. J Appl
Physiol 63: 540-545, 1987.
Page 24 of 28
23
Downloaded from http://jap.physiology.org/ by 10.220.33.5 on June 14, 2017
46.
Parker JC, Ardell JL, Hamm CR, Barman SA, and Coker PJ. Regional
pulmonary blood flow during rest, tilt, and exercise in unanesthetized dogs. J
Appl Physiol 78: 838-846, 1995.
47.
Robertson HT. Measurement of regional ventilation by aerosol deposition.
In: Complexity in Structure and Function in the Lung, edited by Hlastala MP and
Robertson HT. New York: Marcel Dekker, Inc., 1998.
48.
Robertson HT, Altemeier WA, and Glenny RW. Physiological implications
of the fractal distribution of ventilation and perfusion in the lung. Ann Biomed Eng
28: 1028-1031, 2000.
49.
Robertson HT, Glenny RW, Stanford D, McInnes LM, Luchtel DL, and
Covert D. High-resolution maps of regional ventilation utilizing inhaled fluorescent
microspheres. J Appl Physiol 82: 943-953, 1997.
50.
Robertson HT, Kreck TC, and Krueger MA. The spatial and temporal
heterogeneity of regional ventilation: comparison of measurements by two highresolution methods. Respir Physiol Neurobiol 148: 85-95, 2005.
51.
Schimmel C, Frazer D, and Glenny RW. Extending fluorescent
microsphere methods for regional organ blood flow to 13 simultaneous colors.
Am J Physiol Heart Circ Physiol 280: H2496-2506, 2001.
52.
Sinclair SE, McKinney S, Glenny RW, Bernard SL, and Hlastala MP.
Exercise alters fractal dimension and spatial correlation of pulmonary blood flow
in the horse. J Appl Physiol 88: 2269-2278, 2000.
53.
Starr IR, Lamm WJ, Neradilek B, Polissar N, Glenny RW, and Hlastala
MP. Regional hypoxic pulmonary vasoconstriction in prone pigs. J Appl Physiol
99: 363-370, 2005.
54.
Tsang JY, Lamm WJ, Starr IR, and Hlastala MP. Spatial pattern of
ventilation-perfusion mismatch following acute pulmonary thromboembolism in
pigs. J Appl Physiol 98: 1862-1868, 2005.
55.
Wagner PD, Saltzman HA, and West JB. Measurement of continuous
distributions of ventilation-perfusion ratios: theory. J Appl Physiol 36: 588-599,
1974.
56.
West JB and Dollery CT. Distribution of blood flow and ventilationperfusion ratio in the lung, measured with radioactive carbon dioxide. J Appl
Physiol 15: 405-410, 1960.
57.
West JB, Dollery CT, and Naimark A. Distribution of blood flow in isolated
lung: relation to vascular and alveolar pressures. J Appl Physiol 55, 1964.
58.
Zeltner TB, Sweeney TD, Skornik WA, Feldman HA, and Brain JD.
Retention and clearance of 0.9-micron particles inhaled by hamsters during rest
or exercise. J Appl Physiol 70: 1137-1145, 1991.
Page 25 of 28
24
Figure Legends:
Figure 1: Plot of the coefficient of variation of regional blood flow in a rat lung as a
function of the unit size of lung volume used for the calculation. The slope of the line
on this log-log plot describes a fractal dimension that represents the extent of local
correlation of the flow measurements.
The steeper slope represents the same
calculation made after the regional flow to each volume unit was randomly shuffled,
so the flow to one volume unit bears no relation to the flow of its immediate spatial
neighbors. By this approach to analysis, the steeper slope with a fractal dimension
of 1.50 represents a random distribution of flow. (Figure from Glenny et al (20))
normal gravity (1G), plotted against distance of lung piece from the ventral surface of
the lung. Figure2B: Weight-normalized blood flow for the same lung pieces in the
prone pig during microgravity, plotted against distance from ventral surface of the
lung. (Data from Glenny et al (21))
Figure 3: Correlation between regional blood flow and regional ventilation in a prone
anesthetized pig (Figure from Robertson et al (49)).
Downloaded from http://jap.physiology.org/ by 10.220.33.5 on June 14, 2017
Figure 2A: Weight-normalized per-piece blood flow in prone anesthetized pig at
Page 26 of 28
Downloaded from http://jap.physiology.org/ by 10.220.33.5 on June 14, 2017
Plot of the coefficient of variation of regional blood flow in a rat lung as a function of the
unit size of lung volume used for the calculation. The slope of the line on this log-log plot
describes a fractal dimension that represents the extent of local correlation of the flow
measurements. The steeper slope represents the same calculation made after the
regional flow to each volume unit was randomly shuffled, so the flow to one volume unit
bears no relation to the flow of its immediate spatial neighbors. By this approach to
analysis, the steeper slope with a fractal dimension of 1.50 represents a random
distribution of flow. (Figure from Glenny et al (20))
Page 27 of 28
Downloaded from http://jap.physiology.org/ by 10.220.33.5 on June 14, 2017
Figure 2A: Weight-normalized per-piece blood flow in prone anesthetized pig at normal
gravity (1G), plotted against distance of lung piece from the ventral surface of the lung.
Figure2B: Weight-normalized blood flow for the same lung pieces in the prone pig during
microgravity, plotted against distance from ventral surface of the lung. (Data from Glenny
et al (21))
Page 28 of 28
Downloaded from http://jap.physiology.org/ by 10.220.33.5 on June 14, 2017
Correlation between regional blood flow and regional ventilation in a prone anesthetized
pig (Figure from Robertson et al (49)).