| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department of Anesthesia, Massachusetts General Hospital, and Harvard Medical School, Boston, Massachusetts 02114
ABSTRACT
INTRODUCTION
GENERAL THEORY
METHODS
RESULTS
DISCUSSION
APPENDIX
ACKNOWLEDGEMENTS
FOOTNOTES
REFERENCES
ABSTRACT 
Mijailovich, Srboljub, Steven Treppo, and José G. Venegas. Effects of lung motion and tracer kinetics corrections on
PET imaging of pulmonary function. J. Appl. Physiol. 82(4): 1154-1162, 1997.
A method to assess the three-dimensional
distribution of alveolar ventilation-perfusion ratio
(
A/
) by
imaging the lungs with positron emission tomography (PET) during a
constant-rate intravenous infusion of 13NN-labeled saline
solution was developed by C. G. Rhodes, S. O. Valind, L. H. Brudin, P. E. Wollmer, T. Jones, P. D. Buckingham, and J. M. B. Hughes (J. Appl. Physiol. 66: 1896-1904 and 1905-1913, 1989). We
have modified this methodology to obtain high-resolution, low-noise PET
images of local A/
where lung motion artifact was eliminated by respiratory
gating of image collection. In addition, we have refined and
implemented the methods to assess local alveolar ventilation by imaging
the washout of equilibrated 13NN and local perfusion by
imaging the distribution of an intravenous bolus of
13NN-labeled saline solution during apnea. This paper
experimentally evaluates the effect of the implemented modifications in
mechanically ventilated and anesthetized dogs. We found that the lack
of gating had no significant effect on the average recovered
A/, but the spatial
heterogeneity [pixel-by-pixel coefficient of variation squared
(CV2) = SD2/mean2] was
underestimated by 14%. The lack of gating during the washout underestimated the average specific ventilation by 11% and decreased the corresponding CV2 by 50%.
regional lung function; respiratory gating; motion artifacts; nitrogen-13; positron emission tomography; alveolar ventilation; perfusion; ventilation-perfusion ratio
INTRODUCTION LOCAL LUNG FUNCTION has been assessed by positron
emission tomography (PET), from the noninvasive quantification of the
spatial distribution of positron-emitting tracer isotopes.
Local alveolar gas volume (VA) and specific ventilation
(s Distribution of pulmonary blood flow has been measured by PET after
intravenous bolus injection of radioactive-labeled microspheres (4) and
intravenous injection of 15O-labeled water (7). Pulmonary
blood flow measured from microspheres has the advantage that it is
linearly related to measured activity over entire physiological range,
but the long half-life of the labeled compounds is a disadvantage for
making sequential complementary PET studies. 15O, in
contrast, has a short half-life (2.05 min), but tissue activity is
nonlinearly related to pulmonary blood flow and is strongly dependent
on local water tissue content (7). Furthermore, because the transit
time of the water through the lung is very short (<20 s),
short-duration PET images have to be acquired, limiting the signal-to-noise ratio and/or resolution of the resulting
images.
On the basis of the low solubility of 13NN in blood and
tissues, Hales et al. (3) measured regional perfusion using a planar positron camera by imaging the lungs after an intravenous bolus injection of 13NN-labeled saline during apnea. The validity
of this method rests on the assumption that, on arriving to the
pulmonary capillary vessels, most of the tracer gas migrates to
alveolar spaces where it remains during imaging. Although this
technique overcomes the problems of 15O and radioactively
labeled microspheres, the extent to which the alveolar tracer
concentration during imaging changes (because of mixing between
adjacent regions or readsorption of the 13NN by the
capillary blood) has not been quantified.
Rhodes and co-workers (1, 2, 5, 6, 11) pioneered the tomographic
measurements of alveolar ventilation perfusion ratio
( We have refined the methods of Venegas et al. (14, 15), Hales et al.
(3), and Rhodes et al. (1, 2, 5, 6, 11) to assess independently and at
high resolution local Glossary
A) have been assessed with PET from
images collected during equilibrated rebreathing of 13NN
tracer and during a subsequent washout of the tracer (13, 15). In those
studies, correction for radioactive tracer decay was done by assuming
constant tracer concentration during each image of the washout.
Although this assumption leads to a good approximation for
short-duration images, it can substantially underestimate
sA when long-duration images are
collected during the washout.
A/) during a constant
infusion (CI) of 13NN in saline solution by using PET.
Possible shortcomings of this method are as follows: the spatial
averaging created by breathing motion of the lungs, the level of noise
when imaging at high spatial resolution, the need to include the right
ventricle in the imaged cross section, and the inaccuracies caused to
partial volume effects.
A from the washout
of equilibrated 13NN, local perfusion
() from the distribution of an intravenous bolus of 13NN-labeled saline solution, and
A/ by using CI
technique, respectively. This paper describes in detail these
methodologies and evaluates experimentally these refinements in
mechanically ventilated dogs.
A(t)
Instantaneous tracer activity per unit mass at time t, in
µCi/g
A0
Tracer activity per unit mass at a reference time (e.g., start of
imaging), in µCi/g
ANN
Tracer activity per unit mass of 13NN at the beginning of
imaging, in µCi/g
ACO
Tracer activity per unit mass of 11CO at the beginning of
imaging, in µCi/g
Cpa(t), Cpv(t)
Instantaneous tracer mass concentration in arterial and venous blood,
respectively, in g/ml
CA(t)
Instantaneous tracer mass concentration in alveoli, in g/ml
CA0
Mass tracer concentration of the gas during lung equilibration, in g/ml
CCO
Mass tracer concentration of 11CO in blood, in g/ml
CCO1, CCO2
Mass tracer concentration of 11CO in blood at the beginning
and at the end of the blood volume imaging, in g/ml
f
Breathing frequency, breaths/min
fa
Fraction of arterial pulmonary blood in the blood volume image (assumed
to be equal to 0.42)
fCI
Decay correction factor of the CI image to the reference time at the
start of the image collection
fb
Decay correction factor of the blood volume image to the reference time
at the start of the image collection
fVA
Decay correction factor of the gas volume content image to the
reference time at the start of the image collection
fV
Decay correction factor of the gas volume content image to the
reference time at the start of the washout image collection
f
Factor to account for tracer removal by pulmonary blood
fgated
Correction factor that fills gaps during gated image collection
Kcam
Calibration factor of the camera, in
counts · µCi
1 · s1
nb(t)
Instantaneous voxel count rate registered by PET at time t, in
counts · voxel
1 · s1
np
The ideal plateau tracer voxel count rate of a perfusion image that
would have been reached in the absence of tracer kinetics, in counts/s
iFlow rate of the bolus of 13NN-labeled saline solution
infused into the right atrium, in ml/min
Regional perfusion, in ml · min
1 · cm3 thorax
s
A = A/VASpecific ventilation of the alveolar gas volume element, in l/min
S = 1S + 2S
Total no. of counts in a voxel of an ungated image, in counts/voxel
2S
Total no. of counts in a voxel of a gated image, in counts/voxel
Sb
Total no. of counts in a voxel of the blood volume image, in
counts/voxel
Sbn
Voxel counts originating from arterial blood that are normalized to the
CI image, in counts/voxel
SCI
Total number of counts in a voxel of the CI image, in counts/voxel
S
Total no. of counts in a voxel of the blood flow image, in counts/voxel
SR
Total no. of counts in a voxel of the blood flow image that includes
the tracer arrival period and the rise of count rate period, in
counts/voxel
SVA
Total number of counts in a voxel of the gas volume content image, in
counts/voxel
Sw
Total no. of counts in a voxel of the washout image, in counts/voxel
ti
Infusion time of a bolus, in s
Tarr
Arrival time of the tracer, measured from initiation of bolus infusion
till the tracer arrival in a voxel, in s
Tb
Collection time of blood volume image, in s
TCI
Collection time of CI image, in s
T
Collection time of the perfusion image, in s
TR
Collection time of the initial bolus infusion image that includes
Tarr and time of rising concentration, in s
TVA
Collection time of the gas volume content scan, in s
Tw
Collection time of the washout image scan, in s
Tw 0
Time from the beginning of collection of gas content image to the start
of breathing of the tracer-free gas, in s
VA
Gas volume content, in ml/cm3 thorax
Va, Vv
Arterial and venous blood volume, respectively, in ml/cm3
thorax
AAlveolar ventilation, defined as tidal respiratory ventilation
(VT · f ) minus dead-space
ventilation (Vds · f ) (5), in ml · min
1 · cm3 thorax
tBreathing period, in s
tgatedCollection time of a gated image within a breathing period, in s
Blood-to-gas partition coefficient, in ml gas/ml blood
dTracer decay time constant, in s
eEffective washout time constant, in s
e bEffective ventilation time constant of the 11CO tracer from
the pulmonary bloodstream, in s
KEffective time constant of the tracer kinetics, in s
Time constant of the tracer mass kinetics, in s
wWashout time constant, in s
bVentilation time constant of the 11CO tracer from the
pulmonary bloodstream, in s
GENERAL THEORY
PET enables accurate, noninvasive measurement of the regional distribution of a positron-emitting isotope tracer within the lung in well-defined volume elements (voxels). The number of radioactive decay events detected per unit time (count rate) is proportional to the amount of the total radioactivity of the tracer in that voxel. In a typical voxel within the lung field, the tracer may be distributed among lung tissue and alveolar gas and pulmonary blood compartments.
Following the approach by Rhodes and co-workers (5), a lumped-parameter
model of tracer kinetics in a voxel is derived under the following
assumptions: 1) distributions of
A, , and
A/ are assumed uniform
within a voxel; 2) blood flow and ventilation are assumed to be
continuous and invariant during the measurement period; 3) when
the tracer is intravenously infused, it is assumed to be mixed with
blood and thus distributed along the pulmonary arterial tree in
proportion to ; 4) tracer concentrations of
end-capillary blood and alveolar gas are in equilibrium and distributed in accordance with the partition coefficient .
A differential equation describing radioactive tracer kinetics within a
voxel can be formulated from a tracer mass balance: the rate of change
of tracer mass is equal to the rate of tracer influx of the tracer by
the arterial perfusion (Cpa), minus the rate of
tracer removal by ventilation
(ACA) and by pulmonary venous
outflow (CA)
|
(1) |
The specific activity of the tracer in a voxel can be expressed as the product of tracer mass concentration [C(t), in g/ml] and the tracer activity per unit mass [A(t), in µCi/g]. Because the tracer activity per unit mass decays exponentially with time, instantaneous tracer activity A(t), is
|
where A0 is tracer activity per unit mass at a reference
time (e.g., start of imaging), t is time elapsed from that
reference, and d is the tracer decay time constant.
Neglecting the tracer activity within the tissue compartment, the
number of counts per voxel S in an PET image is equal to the
integral over collection time T of the total count rates from the
alveolar and arterial and venous blood compartments
![<IT>S</IT> = <IT>K</IT><SUB>cam</SUB> <LIM><OP>∫</OP><LL><IT>t</IT><SUB>0</SUB></LL><UL><IT>t</IT><SUB>0</SUB>+T</UL></LIM> [V<SC>a</SC>C<SC>a</SC>(<IT>t</IT>) + VabCpa(<IT>t</IT>) + VvCv(<IT>t</IT>)]](/content/vol82/issue4/fulltext/1154/img003.gif)
|
|
(2) |
where Cv(t), Vv, Cpa(t), and Va are the
regional tracer mass concentration and volume of pulmonary venous and
arterial blood, respectively; and Kcam is the
calibration factor of the camera correlating total activity per cubic
centimeter thorax, with a voxel rate of coincidence counts
registered by the PET camera. Equations 1 and 2 are applied to different experimental conditions to assess
A, , VA,
and A/ with PET.
Respiratory Gating
To minimize motion artifacts during breathing, image collection is gated in synchrony with the mechanically controlled breathing. A ventilatory frequency f is fixed at 12 breaths/min, with inspiratory time equal to 30% of the breathing period (t). Triggered at
the start of inhalation by the ventilator, the PET camera acquires two
images of equal duration (2.48 s) and waits for a new trigger signal to
repeat the cycle, averaging the data over a predetermined total
collection time. The first of these images covers inhalation and the
early part of exhalation, whereas the second image covers the last part
of exhalation when the lung motion is small and its volume is
essentially at resting functional residual capacity (FRC). The second
image is referred to as "gated," with a number of counts per
voxel (2S). Images created by adding the two
images to cover most of the entire breathing cycle (4.96 of 5.0 s) are
referred to as "ungated" images, with a number of counts per
voxel S = 1S + 2S.
Assessment of VA Content and
sA
1d, radioactive decay slope;
, ventilation decay slope; 1e, effective decay slope. Tw0 is time from the beginning of collection
of gas-content image to start of breathing of tracer-free gas.
VA content. Because of the low solubility of 13NN in blood and tissue, during steady state the voxel activity of rebreathing gas is proportional to the local VA content. Thus neglecting the effect of the adsorption of tracer by the blood (
CA term in Eq. 1), the
average VA per voxel can be estimated as
|
(3) |
A.
Washout images reflect a simultaneous exponential decay in mass
concentration of tracer (represented by a time constant
w) and an exponential decay in radioactivity of the
tracer (represented by decay time constant d) (Fig. 1).
It can be shown that the total counts per voxel collected during a
washout (Sw) are a function of an effective
exponential time constant (e) that combines the time
constants of these two exponential decays: e = (1/w + 1/d)1. The value
of e can be calculated from Eq. A5a (derived in
Appendix) as
|
(4) |
Tw 0/
d
is a decay correction factor for the VA content image to a
reference time at the initiation of the washout.
The sA, defined as the inverse of
w, becomes
|
(5) |
= 1 + (/A) is a factor to
account for tracer removal by pulmonary blood assumed to be 1 for
normal lungs.
Assessment of
i for infusion time ti
(Fig. 2). A sample of the infusate is then
collected to assess its specific activity
(A0Ci) in a cross-calibrated well counter.
Simultaneously with the start of infusion, a series of sequential
images is collected while the subject remains in apnea.
1 and
S2 are total nos.
of counts per voxel of 3 perfusion images.
SR1, SR2, and
S (hatched area) are
corrections for tracer kinetics. TR is collection time of
the image that covers rising time of alveolar concentration and must be
larger than sum of arrival time Tarr and bolus infusion ti. T is total
collection time of perfusion image.
Because of the low-solubility 13NN in blood and tissues,
most of the tracer diffuses into alveolar gas at first pass, and the alveolar distribution of 13NN approximates
(3). An idealized time course of alveolar tracer
concentration in a voxel involves an arrival time Tarr, a
period of linear increase in concentration during infusion
ti, and a period where CA remains at a
constant plateau value (Fig. 2). In reality, the tracer may diffuse to
neighboring regions and/or may be readsorbed by the pulmonary
circulation. To assess and correct for these possible sources of error,
a sequence of image collection was designed, consisting of a first
image including the infusion and increasing concentration phases (with
counts SR and duration
TR 
ti + Tarr) and
two additional images during the plateau phase (with voxel counts
S
and
S
and duration
T
/2 each) (Fig. 2).
Assuming that during the plateau phase the alveolar tracer activity
changes exponentially, its equivalent time constant (K), which incorporates radioactive tracer decay d and the
tracer removal , can be derived as a function
of S1 and
S2
|
(6) |
After correction for tracer kinetics and tracer radioactive decay, the
voxel values for (in ml/cm3) can then be
calculated as
|
(7) |
|
is the ideal plateau tracer count rate that would have been reached in the absence of local tracer loss and radioactive decay.
Assessment of
A/
A/ ratio
|
(8) |
Our methodology to obtain the values of Cpa and CA with PET is as follows. During steady-state breathing with tracer-free gas, infusion of 13NN-labeled saline solution, with specific activity CCIA0, is initiated at a constant flow rate of 15 ml/min. With the start of infusion, a series of four sequential gated images are collected. Images acquired after a steady-state concentration is reached are summed into an image with total collection time TCI and voxel counts SCI. During the steady state, a sample of pulmonary arterial blood is collected to assess its specific activity decay corrected to the initiation of image collection (CpaA0). Alveolar tracer concentration CA is assessed from SCI, after subtracting the counts contributed by radioactivity of the local pulmonary arterial blood (Sbn).
To estimate Sbn, an additional blood volume scan is performed. Red blood cells are labeled by inhalation of 11C-labeled CO (11CO; half-life of 20 min) until the activity has reached a suitable steady-state count rate. After allowing enough time (10 min) for the tracer to wash away from the alveolar gas space, a gated image is collected, with voxel counts of 2Sb and collection time Tb. Blood samples are taken at the beginning and at the end of collection to asses the initial and final tracer specific activity. Sbn is obtained from the counts recorded in the gated blood volume image 2Sb (see Appendix, Eq. A12 for details), assuming that 40% of regional blood volume was arterial tree and thus labeled with 13NN during the CI protocol (6). CA is thus estimated as
|
(9) |
where fCI is the decay correction factor of CI image to the reference time at the start of the image collection.
An ungated A/ image is
derived from the ungated SCI and Sb
images and by setting fgated = 1 in Eqs. 9 and A12.
METHODS
Animal Studies
Detailed experimental protocols, animal preparation, and image processing are described in an accompanying paper (9). Five mongrel dogs were studied in the prone position and five in the supine position. The animals were positioned within the PET camera to obtain an image of the transverse slice of the lungs with the greatest cross-sectional area.Image Analysis
Images of VA, sA,
, and
A/ were generated for
gated and ungated conditions. These images were characterized in terms of mean voxel value and spatial heterogeneity defined as the
pixel-by-pixel coefficient of variation (CV = SD/mean) of the voxel
values within the lung field. The procedure to define the lung field is
described in detail in the accompanying paper (9).
To exclude areas outside the lung field, the heart, and large blood vessels, we used the same mask for all analyzed images. We developed a master mask by using the following three-step iterative procedure. Voxels outside the lung field were eliminated by thresholding the CI image and refined until no extraneous points remained (mean threshold 33 ± 6% of the peak voxel value). A similar procedure was used to isolate the heart and large vessels from the blood volume (Vb) image (mean threshold 55 ± 5% of the peak voxel value) and subtracted from the lung field mask. Finally, this mask was further refined to exclude areas that heart and lungs share during imaging period because of heart motion (at most 2-3 voxels wide, where voxel size was 0.1925 × 0.1925 × 0.5 cm2).
Statistical Analysis
Values are means ± SE obtained from five measurements (5 prone or 5 supine dogs). Also, measured mean values and CV2 values were compared by using Student's paired t-test. Statistical significance was taken at P < 0.05 level.RESULTS
Image Gating
PET images of regional VA, sA, and
A/ are all acquired
during breathing. Because the imaging device remains stationary in
space, these PET images can be expected to be affected by breathing in
two distinct ways. First, there may be an underestimation of regional
heterogeneity because of spatial averaging; and, second, the expansion
of the lung parenchyma during breathing could result in a greater mean
voxel gas content compared with that at FRC.
As expected, VA without gating was higher on average (10.5 and 5.7% for prone and supine positions, respectively) compared with VA measured during the last part of expiration. We have chosen to present results as CV2 rather than as the CV, because the contributions of different factors to heterogeneity such as nose or vertical gradients are additive only when expressed in terms of CV2. Reflecting the spatial averaging created by lung motion, the regional heterogeneity (CV2) of VA ungated was significantly lower compared with VA gated (16.9 and 12.2%, respectively) for prone and supine positions.
Gating created no significant differences in mean
A/ values,
either prone or supine (Table 1). In
contrast, CV2 of
A/ was significantly
lower when measured without gating, compared with gating (16.4 and
12.3% decrease for prone and supine positions, respectively).
|
|||||||||||||||||||||||||||||||||||||||
Mean values of gated sA were
significantly higher (9.1% increase in prone and 12.5% in supine
positions) than those for ungated sA, and
the CV2 values for gated sA
were also significantly higher than these for ungated
sA (44.9% increase in prone and 55.4%
in supine animals) (Table 1).
Tracer Decay Correction Method for Washout Images
Correction for radioactive tracer decay in PET is generally carried out for each image by assuming that the tracer concentration is constant during the image-collection washout maneuver. The error incurred by such an assumption during a washout maneuver depends on the ratio of tracer washout time constantw to the radioactive decay
time constant d and the image collection time. This
error is greatest for rapid washouts and increases dramatically with
increasing collection times. In our 3-min washout images with typical
w of 20 s, this error would have been ~8%.
Blood Contribution to Local Activity Distribution During CI
The estimated contribution of tracer activity in pulmonary arterial blood to total activity per voxel during CI of 13NN saline after masking heart and large blood vessels (9) was 15.9 ± 2.1% for prone and 12.8 ± 2.2% for supine positions. Correction for the contribution of tracer in pulmonary blood did not significantly change the CV2 (i.e., increase of 5.7 ± 5.3% in prone and decrease of 3.3 ± 4.3% in supine positions).Tracer Kinetics During Imaging
image
for time variations in local tracer concentration during the imaging increased the mean values of by 13.7 ± 4.0% in
supine and by 7.9 ± 4.0% in prone positions. The CV2 of
was decreased by 13.5 ± 6.5% in supine and by 5.5 ± 4.2% in prone positions by this correction.
DISCUSSION
We implemented methodologies for assessing the local distribution of
VA, , A,
and A/ with PET
including the following enhancements: 1) elimination of lung
motion artifact by respiratory gating; 2) implementation of a
proper correction for radioactive tracer decay during washouts;
3) implementation of a robust method of correction for the
contribution of activity in pulmonary arterial blood to the CI images;
and 4) correction for the effect of temporal changes in
tracer concentration during collection of images. These corrections affected the mean value and/or the
CV2 of the VA, ,
A, and
A/ images in different
degrees.
The theoretical model used to interpret and analyze the PET images
(from mechanically ventilated dogs) during CI (5) is based on several
assumptions and simplifications. Among them are that the distributions
of A, , and
A/ are uniform within a
pixel; that the tracer is well mixed with the blood during infusion; and that the lungs are ventilated by continuous rather than a tidal
ventilation. Rhodes and co-workers (5) have discussed in detail the
shortcomings and validity of the model. We are limiting this discussion
to areas affected by our modifications of the technique.
Effect of Gating
We found that all images acquired with gating had higher spatial heterogeneity (CV2) than those acquired without gating (Table 1). This finding was consistent with the expected spatial averaging effect caused by lung motion during breathing, which should be reduced by imaging the lungs during end exhalation only. However, the average increase in CV2 of the VA andA/ images by
gating was not greater than 17%. This suggests that, in normal animals
breathing with a total volume as large as 20 ml/kg, interregional
differences are relatively small at the length scale of the breathing
motion. In our study, we used low respiratory frequency (~12
breaths/min) and short inspiratory times to ensure that VA
was close to FRC during the second (gated) image collection. Thus, to
maintain normocapnea, total volume had to be relatively large (22 ml/kg). Given that our total volume was much higher than that of
spontaneous breathing in humans, the fact that the effect of gating
on A/ was so small
suggests that motion artifacts must have been even smaller than the
data reported by Rhodes et al. (6). A much greater (
50%) increase
in the CV2 of sA uncovered by
gated imaging suggests that a significant component of heterogeneity in
ventilation must occur at the length scales comparable to lung motion
during breathing.
Aside from spatial averaging of breathing motion, other mechanisms
could have also contributed to the lower CV2 seen without
gating. Gating reduced by one-half the total imaging time and reduced
the total number of counts per voxel, thus increasing the contribution
of statistical noise to the heterogeneity of the gated images (12). The
primary images of CI and VA, used to generate
A/, had very high
counts (>5,000 counts/resolution element) in which noise was
estimated to contribute <2% to the corresponding CV2.
Thus, we are confident that noise was not responsible for the increased
heterogeneity of the A/
images seen with gating. In contrast, the average contribution of noise
to the CV2 of sA with gating
was as much as 20%. Subtraction of the contribution of noise from both
gated and ungated sA images decreased the corresponding CV2 but it did not appreciably change the
percent increase in CV2 uncovered by gating. Therefore, the
increased noise in sA images between
gated and ungated cannot explain the increase in CV2.
Analysis of Washouts
In the past (12-14), we have assessed sA with the 13NN-washout
technique by using the exponential approximation. Analysis included correction for radioactive tracer that assumed a constant tracer concentration during each image collected. The error induced by this
assumption is small when several short images are collected, and each
of these images is individually corrected for radioactive decay before
summation for the calculation of sA.
However, the error induced may become significant when, as in our
study, long (60-s) washout images are collected. This source of error
was eliminated by measuring an equivalent time constant that combined radioactive decay and tracer washout. Because the radioactive decay
time constant of 13NN is known, the tracer washout time
constant can be readily solved by using Eq. 5.
Contribution of Activity in Arterial Blood to CI Images
In addition to the activity originating from alveolar spaces, images acquired during the CI protocol include a contribution from the activity in pulmonary blood. To correct for this contribution, Rhodes et al. (6) used a method that relied on a measurement of pulmonary arterial blood specific activity from an area of interest defined in the right ventricle. Because of partial volume effects, this measurement becomes quite unreliable in small animals. In our study, we obtained samples of pulmonary arterial blood during CI and of systemic blood during the blood volume scan and we measured their specific activity in a well counter previously cross calibrated with the PET camera to eliminate this source of error. We found that neglecting to include the effect of pulmonary blood activity resulted in an average underestimation ofA/ of ~15% in normal lungs. This value is in good agreement with the
number reported for humans by Rhodes et al. (6). Correction for
pulmonary arterial blood activity left the heterogeneity of A/ relatively
unchanged.
Tracer Kinetics During Imaging
One could expect that interregional mixing by diffusion or cardiogenic
oscillations could have decreased the heterogeneity in
recovered by our technique. However, if that was the
case, one would have expected that the tracer kinetics correction
(which follows and corrects for dynamic changes in local concentration) would have increased the heterogeneity of the image.
The fact that tracer kinetics correction in normal lungs was so small
strongly supports the concept that interregional mixing during
measurement period, if present, occurred at length scales lower than
the spatial resolution length of our images. Although the overall
correction was rather small in the normal dogs, our preliminary studies
showed that it could be extremely important in atelectatic lungs (8).
In summary, we have refined and successfully implemented imaging
methodologies to assess VA,
sA, , and
A/ at high spatial
resolution. The effects of these refinements are small in normal
animals but could be much greater in diseased lungs.
ACKNOWLEDGEMENTS
This work was supported by the National Heart, Lung, and Blood Institute Grant HL-38267.
FOOTNOTES
Address for reprint requests: J. Venegas, Dept. of Anesthesia, Room 255 (CLN-2), Massachusetts General Hospital, Fruit St., Boston, MA 02114.
Received 22 March, 1995; accepted in final form 12 November 1996.
APPENDIX
Gas Volume Image
After a sufficient time for equilibration, the concentration of the 13NN-labeled tracer gas in a closed breathing circuit is approximately constant. Because there is no perfusion of tracer (Cpa = 0), and neglecting adsorption of the tracer by the blood ( 0), the number of counts per voxel in
SVA can be calculated by integrating
Eq. 2
|
(A1) |
|
Gating Correction Factor for VA
If one integrates instantaneous tracer activity over acquisition timetg for each breath, the total number of counts
2SVA is equal to sum of
these integrals during the collection time TVA.
Because net collection time per breath is reduced to a fraction of a
breathing period, 2SVA
should be multiplied by a correction factor fgated to account for the reduction in collection time
|
(A2) |
t is breathing period; tg is the
collection time of the gated image, and then VA can be
calculated from Eq. 3.
In the case where the breathing period is much shorter than the time constant of radioactive decay of the tracer, the correction factor is approximately equal to
|
(A3) |
Alveolar sA
|
(A4) |
A = A/VA is the specific
ventilation; and f = 1 + /A is
a blood adsorption correction factor. The number of
counts in the washout image according to Eq. 2 is
|
(A5) |
e is an effective time constant such that
|
|
Tw 0 /d
is tracer activity per unit mass at the beginning of collection of
washout image. A ratio of Eqs. A1 and A5 yields to an
implicit function of e that can be solved iteratively
|
(A5a) |
Tw 0/d,
is a decay correction factor of SVA
image to the beginning of breathing tracer-free gas.
Alveolar Tracer Content During Steady-State CI
During the CI protocol, the influx of tracer into the alveolar space is constant and equal to · Cpa, and the
lungs are being ventilated with tracer-free gas. The solution of the
differential Eq. 1 under these conditions becomes
|
(A6) |
is defined in Eq. A4.
At steady-state
(sA ft 
1),
Eq. A6 simplifies to
|
(A7) |
|
(A8) |
d)/(1 
eTCI/d)
is the decay correction factor.
Assessment of Local Blood Volume
During the collection of the blood volume scan, some of the 11CO absorbed by the red blood cells is released and washed out at a slow rate. To account for this effect, two arterial blood samples are withdrawn, one at the beginning of the image collection and one at the end. Assuming a monoexponential 11CO washout, its time constant
b can be
calculated as
|
(A9) |
The voxel count rate nb at time t is a product of the blood volume in the voxel Vb, concentration of the tracer, and a term that accounts for the radioactive decay of the tracer
|
(A10) |
The total number of counts per voxel in the collected image Sb is obtained by integrating Eq. A10 over the collection time Tb.
|
(A11) |
e b)/(1 eTb/eb)
is a combined radioactive decay and tracer washout decay correction
factor; e b = 
d/( + d) is effective time constant; Va is arterial blood volume; and
CCOACO is specific activity of the blood
corrected to the beginning of recording time.
Contribution of Activity in Arterial Blood Volume During the CI Protocol
The estimated counts originated from pulmonary arterial blood, values of Sb are normalized by the ratio of blood-specific activities and collection times during CI and blood volume and by assuming that a fraction fa of total lung blood volume is pulmonary arterial blood
|
(A12) |
REFERENCES
| 1. |
Brudin, L. H.,
C. G. Rhodes,
S. O. Valind,
T. Jones,
B. Johnson,
and
J. M. B. Hughes.
Interrelationships between regional blood flow, blood volume, and ventilation in supine humans.
J. Appl. Physiol.
76:
1205-1210,
1994.
|
| 2. |
Brudin, L. H.,
C. G. Rhodes,
S. O. Valind,
T. Jones,
B. Johnson,
and
J. M. B. Hughes.
Relationships between regional ventilation and vascular and extravascular volume in supine humans.
J. Appl. Physiol.
76:
1195-1204,
1994.
|
| 3. |
Hales, C. A.,
B. Ahluwalia,
and
H. Kazemi.
Strength of pulmonary vascular response to regional alveolar hypoxia.
J. Appl. Physiol.
38:
1083-1087,
1975.
|
| 4. |
Mintun, M. A.,
M. M. Ter-Pogossian,
M. A. Green,
L. L. Lych,
and
D. P. Schuster.
Quantitative measurement of regional pulmonary blood flow with positron emission tomography.
J. Appl. Physiol.
60:
317-324,
1986.
|
| 5. |
Rhodes, C. G.,
S. O. Valind,
L. H. Brudin,
P. E. Wollmer,
T. Jones,
and
J. M. B. Hughes.
Quantification of regional / ratios in humans by use of PET. I. Theory.
J. Appl. Physiol.
66:
1896-1904,
1989.
|
| 6. |
Rhodes, C. G.,
S. O. Valind,
L. H. Brudin,
P. E. Wollmer,
T. Jones,
P. D. Buckingham,
and
J. M. B. Hughes.
Quantification of regional / ratios in humans by use of PET. II. Procedure and normal values.
J. Appl. Physiol.
66:
1905-1913,
1989.
|
| 7. | Schuster, D. P. Positron emission tomography: theory and its application to the study of lung disease. Am. Rev. Respir. Dis. 139: 818-840, 1989. [Medline] . |
| 8. | Seow, D., C. Andrie, A. Head, and J. Venegas. Assessment of
intrapulmonary perfusion () and shunt
(s) with positron emission tomography
(Abstract). Am. J. Respir. Crit. Care Med. 151, Suppl.
4, pt. 2: A520, 1995.
|
| 9. |
Treppo, S.,
S. M. Mijailovich,
and
J. G. Venegas.
Contributions of pulmonary perfusion and ventilation to heterogeneity in A/ measured by PET.
J. Appl. Physiol.
82:
1163-1176,
1997.
|
| 10. |
Tsuzaki, K.,
S. C. A. Hales,
D. J. Strieder,
and
J. G. Venegas.
Regional lung mechanics and gas transport in lungs with inhomogenous compliance.
J. Appl. Physiol.
75:
206-216,
1993.
|
| 11. | Valind, S. O., P. E. Wollmer, and C. G. Rhodes. Application of positron emission tomography in the lung. In: Positron Emission Tomography, edited by M. Reivich, and A. Alavi. New York: Liss, 1995, p. 387-412. . |
| 12. | Venegas, J. G., S. Treppo, and S. M. Mijailovich. Contributions of statistical noise to spatial heterogeneity of PET images of pulmonary function. In: Medical Imaging 1994: Physiology and Function from Multidimensional Images (Proceedings), edited by E. A. Hofman, and R. S. Acharya. Newport Beach, CA: SPIE-T Int. Soc. for Optical Engineering, 1994, vol. 2168, p. 361-368. . |
| 13. |
Venegas, J. G.,
Y. Yamada,
C. Burnham,
and
C. A. Hales.
Local gas transport in eucapnic ventilation: effects of gravity and breathing frequency.
J. Appl. Physiol.
68:
2287-2295,
1990.
|
| 14. |
Venegas, J. G.,
Y. Yamada,
J. Custer,
and
C. A. Hales.
Effect of respiratory variables on regional gas transport during high-frequency ventilation.
J. Appl. Physiol.
64:
2108-2118,
1988.
|
| 15. |
Yamada, Y.,
C. Burnham,
C. A. Hales,
and
J. G. Venegas.
Regional mapping of gas transport during high-frequency and conventional ventilation.
J. Appl. Physiol.
66:
1209-1218,
1989.
|
This article has been cited by other articles:
|
|
 |
|
  G. Musch, G. Bellani, M. F. Vidal Melo, R. S. Harris, T. Winkler, T. Schroeder, and J. G. Venegas Relation between Shunt, Aeration, and Perfusion in Experimental Acute Lung Injury Am. J. Respir. Crit. Care Med., February 1, 2008; 177(3): 292 - 300. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Schroeder, M. F. Vidal Melo, G. Musch, R. S. Harris, T. Winkler, and J. G. Venegas PET Imaging of Regional 18F-FDG Upta |
