ORIGINAL ARTICLE
Optimization and Standardization of Lung Densitometry in
the Assessment of Pulmonary Emphysema
Berend C. Stoel, PhD* and Jan Stolk MD, PhD†
Abstract: Currently, lung densitometry for the assessment of pulmonary emphysema has been fully validated against pathology,
pulmonary function, and health status, and it is therefore being
applied in pharmacotherapeutic trials. Nevertheless, its application
for the early detection of emphysema has not yet been introduced in
daily clinical practice. The main reason for this is the fact that it is
not yet regarded a fully optimized and standardized technique. In
this work, an overview is given on the current status of different
standardization aspects that play an important role in this, ie, image
acquisition, choice of densitometric parameter and image processing. To address these issues, solutions have been sought from the
literature and from original data from previous studies. Standardization and optimization of lung densitometry has reached a more
advanced stage than has been reported so far. If normal values will
become available, this technique will be feasible for clinical practice. As a result, standardization for the detection and assessment of
other density-related lung diseases can be achieved in a shorter
period of time.
Key Words: emphysema, computed tomography, quantitative,
lung densitometry
(Invest Radiol 2004;39: 681– 688)
L
ung densitometry owes its existence to the linear relationship between the physical density of soft tissue and its
x-ray attenuation. Because pulmonary emphysema is characterized by a decrease in density—attributable to a combination of loss of lung tissue, decreased pulmonary blood flow,
and increased areas of air trapping—25 years ago the link was
made to measure the extent of emphysema with computed
tomography (CT) by making use of this relationship.1– 4 In
Received January 30, 2004 and accepted for publication, after revision, May
14, 2004.
From the *Division of Image Processing, Department of Radiology, and
†Department of Pulmonology, Leiden University Medical Center, Leiden, The Netherlands.
Reprints: Berend C. Stoel, PhD, Division of Image Processing, Department
of Radiology, Leiden University Medical Center, Albinusdreef 2, 2333
ZA Leiden, The Netherlands. E-mail: [email protected]
Copyright © 2004 by Lippincott Williams & Wilkins
ISSN: 0020-9996/04/3911-0681
the subsequent years, a diversity of parameters has been
developed to measure lung density. Different investigators
have validated their parameters against pathology,5–10 against
pulmonary function,11–18 and against health status and exercise tests.19 –22 It has been applied in studies to describe the
natural history of the disease, for patient selection, planning
and validation of Lung Volume Reduction Surgery (LVRS),
and it is now being used in drug evaluation trials. From a
former study on the efficacy of alpha1-antitrypsin augmentation therapy,23 it has been concluded that CT densitometry is
more sensitive to assess the progression of emphysema than
lung function tests. Its application for other lung diseases also
has been explored, such as sarcoidosis,24 –25 asthma,26 –29
pulmonary edema,30 –33 acute respiratory distress syndrome,34 and cystic fibrosis.35
Despite these advances, lung densitometry has not yet
been applied for early detection of emphysema (or any other
lung disease) in daily clinical practice as was foreseen in the
early publications on this subject. The reason for this lies in
the fact that CT has primarily been optimized in the radiologic clinic for visual interpretation and has not been standardized for quantitative measurements.36,37 In contrast to
CT, equipment for pulmonary function testing could be
standardized to a higher level because they are exclusively
designed to measure lung function.
Therefore, consensus is needed on how to standardize
and optimize the CT acquisition protocol but also on which
densitometric parameter to use and how to automatically
compute these parameters from the resulting images in a
standardized and reproducible way. In this review, we will
address these issues, based on results from the literature
(from the multidisciplinary view of pulmonologists, radiologists, physicists, and computer scientists) and by additional
data from previous studies of our research group.
CHOICE OF DENSITOMETRIC PARAMETER
Numerous densitometric parameters have been defined
and validated during the past 2 decades, the most basic
parameter of which is the mean lung density (MLD) calculated by averaging the densities over all pixels in the image
that represent the lungs. Although correlation has been re-
Investigative Radiology • Volume 39, Number 11, November 2004
681
Stoel and Stolk
Investigative Radiology • Volume 39, Number 11, November 2004
ported with pulmonary function tests,15–17 MLD is found to
be less reproducible and therefore less sensitive to density
changes than other densitometric parameters,38 because of its
inherent sensitivity to noise.39
Coxson and coworkers40 developed a measure of the
expansion of the lung, expressed as the volume of gas per
gram of tissue. They defined this parameter as the inverse of
the MLD (ie, specific volume of the entire lung) minus a
constant of 1/1.065 mL/g, which is assumed to be the specific
volume of lung tissue. A similar measure has been developed
by Gattinoni et al,41 assuming a specific volume of tissue of
1 mL/g. However, as these measures are directly related to
the MLD, they have the same drawbacks of limited reproducibility.
The relative area (RA) of low attenuation has been
defined as the percentage of pixels within the lungs with a
density lower than a predefined threshold. RA is most commonly used because of its implementation on CT consoles,
called “density mask,” which made this analysis procedure
available to a large group of investigators. Different thresholds have been explored in the literature, ranging from ⫺770
HU to ⫺960 HU. The RA measurement has been validated
with different thresholds by comparison with measures of
static and dynamic volume measurements,6,11,42– 44 and quantitative pathologic morphology.5,6,9 –10
The nth percentile point has not been used in research
as frequently as the relative area measurement, since it
requires somewhat more specific software. The nth percentile
point, Perc(n), is defined as the threshold value for which n
percent of all pixels have a lower density. Also the percentile
point has been validated with pulmonary function tests14 and
pathology.7 In the assessment of emphysema, it has been
shown that the nth percentile point is equally sensitive between the 10th and 20th percentile, with approximately an
optimum at the 12th percentile.45 In practice, the 15th percentile point is being used.
The relative area and nth percentile point are closely
related, since they are basically each other’s inverse func-
tions. This can be best appreciated when the definitions of RA
and Perc(n) are illustrated using the cumulative histogram of
the densities (see Fig. 1). The cumulative histogram can then
be used as a look-up table: the RA-950 is the y-value at which
x ⫽ ⫺950 HU and the 15th percentile point is found by
determining the x-value for which y ⫽ 15%.
Because the histogram usually has a uni-modal shape,
the cumulative histogram is an increasing function with
generally an S-shape, featuring 2 plateaus; one at y ⫽ 0% and
one at y ⫽ 100%. As a result of these plateaus, a whole range
of threshold values for the RA-parameter will yield exactly
the same percentage, losing its discriminating power and its
sensitivity is highly dependent on the threshold used. This
effect translates into a skewed distribution of RA values in a
patient population due to this cut-off effect at 0%, as presented by Dowson et al.46
The nth percentile point, however, does not have this
drawback: each percentile corresponds to a distinct density
value and its sensitivity is therefore less dependent on the
choice of percentile. This difference in sensitivity can be
demonstrated by plotting the RA parameter against the 15th
percentile point (see Fig. 2). This relation is calculated from
results of a study by Stolk et al.21 The 15th percentile point
can detect changes in the range of ⫺900 to ⫺800 HU,
whereas the RA-950 parameter cannot discriminate any differences, because of the cut-off at 0%. For this reason, the
early onset of emphysema cannot be detected by RA-950, and
consequently the 15th percentile point is preferred.
Other densitometric measures have been developed that
are more specific to a certain aspect of a lung disease or
application. For example, diffuse air trapping has been measured by calculating the difference in density between inspiratory and expiratory scans.40,47–50 In other applications,
densitometric parameters have been defined for patient selection for lung volume reduction surgery, measuring the heterogeneity of the distribution of emphysema.51–55 More sophisticated, texture-based parameters have been developed to
classify different lung diseases.56 However, at this point in
FIGURE 1. Definition of relative area
below ⫺950 HU (RA-950) and 15th
percentile point (Perc15). A, Typical
histogram of densities in the lung. B,
Cumulative histogram by which
Perc15 and RA-950 have been defined.
682
© 2004 Lippincott Williams & Wilkins
Investigative Radiology • Volume 39, Number 11, November 2004
Optimization of Lung Densitometry
FIGURE 2. Relation between RA-950
and Perc15, revealing the lack of
sensitivity of the RA-950 parameter
to detect early signs of emphysema
(Perc15 greater than ⫺900 HU).
time these new techniques have not been validated to the
same extent as the relative area or percentile point.
OPTIMIZATION AND STANDARDIZATION OF
THE CT ACQUISITION PROTOCOL
Generally, manufacturers of CT scanners and radiologists are mainly focused on the optimization of the CT
acquisition protocol for visual interpretation. This optimization does not imply, however, that the resulting standard
protocols are also adequate for lung densitometry.
Different aspects of the CT acquisition protocol have
been examined in the past, in an attempt to optimize specifically for lung densitometry. Different criteria have been used
for this optimization. Mishima et al proposed an mA-setting
that is optimized for the accuracy of the density measurements, in the sense that the measured density values should
reproduce the actual densities in a phantom.57 This may not
be relevant, however, if normal values are available or if
progression is to be measured: in those cases precision is of
much more importance than accuracy. The same authors proposed an optimal setting of the slice thickness based on correlation with pulmonary function tests (PFTs) and subjectively
rated image quality.58 By doing so, the density measurement is
considered as a surrogate measure for these PFTs—with its
inherent limited specificity—and not as an autonomous parameter, that may not perfectly correlate with PFTs.
Kemerink and coworkers, on the other hand, have
developed a more specific measure of the ability of a CT
scanner to discriminate between different densities of foamlike material, called the density resolution.59 In an experiment, which remained relatively unknown, Kemerink used an
anthropomorphic thorax phantom in which 2 pieces of foam
were placed with 2 distinct densities. The histogram of the
© 2004 Lippincott Williams & Wilkins
resulting CT image should therefore ideally reveal 2 distinct
peaks. The ability to discriminate these peaks was defined by
the density resolution and was compared between different
spatial resolutions in the z-direction (set by the collimation
and slice thickness) and in the image plane (determined by
reconstruction filter). It was concluded that the highest density resolution could be obtained with a low spatial resolution
(ie, 5-mm slices and smooth reconstruction filters), which is
considered to result in poor image quality. Thus, it has been
demonstrated that there are conflicting requirements of the
CT protocol for visual and quantitative assessment.
The results of the work of Kemerink have been used to
develop a standardized protocol that could give comparable
densitometric results from different CT scanners.60 In a
research project supported by the European Commission
(SPREAD, QLG1-2000-01752), a standard protocol has been
defined for 5 different CT scanners.39 Typically, such a low
dose acquisition protocol comprises, depending on the manufacturer, 140 kVp, 40 mAs, pitch factor 1.5, 5-mm collimation (for multi-slice CT scanners with more than 4 detector
rows, a smaller collimation is used with the same reconstruction parameters), reconstructed with a slice thickness of 5
mm, 2.5-mm increment, and a smooth reconstruction filter. In
a phantom study, it has been found that this standardization
could still be improved, since small differences in sensitivity
remained between multislice CT scanners. Whether this is
truly caused by differences in the acquisition protocol or by
intrinsic differences in the CT design remains to be determined.39
As demonstrated by Stolk et al, a high level of repeatability can be achieved even with a markedly reduced x-ray
dosage of approximately 0.7 mSv.61 Apart from the settings
of the CT scanner itself, also patient-related issues concern-
683
Stoel and Stolk
Investigative Radiology • Volume 39, Number 11, November 2004
ing the acquisition protocol should be optimized and standardized; especially the optimal inspiration level and the
development of methods to keep the lung volume constant
during follow-up or correct for any changes, if needed.
Inspiration Level
Different studies have been performed to determine the
optimal inspiration level with which CT scans should be
acquired. The requirements for this optimum are highly
dependent on the lung disease involved. For example, for
visual rating of chronic obstructive pulmonary disease
(COPD) and, more specifically, of focal air trapping, expiratory scans yield better correlations with pulmonary function
tests than inspiratory scans.50,62 If on the other hand objective
densitometric measurements are made, 10% vital capacity
(VC) and 90% VC scans have equal discriminating power
between COPD and healthy individuals.47 In discriminating
between idiopathic pulmonary fibrosis, COPD and healthy
subjects by densitometry, scans taken at 50% VC are found to
be more accurate than at 80% VC.38 Because mostly highresolution CT (HRCT) was used in the aforementioned studies, the single slices acquired at a certain level of inspiration
and expiration do not exactly represent the same anatomic
information, which may account for the differences in discriminating power.
For follow-up measurements, it is much more important to obtain a highly reproducible densitometric measurement. In that respect, it has been found that 90%VC is more
reproducible than 10% VC by an order of 3.63 Additionally,
inspiration scans have the obvious benefit that the chances for
the patient of holding their breath are favorable.
FIGURE 3. Relation between difference in inspiration level
between repeated CT scans and difference in lung density
measured with Perc15.
Kalender et al64 therefore suggested to use spirometrically gated CT. This procedure has been evaluated by Gierada et al,65 who concluded that the repeatability could not
Volume Control or Correction for Volume
Differences?
One of the main confounders in lung densitometry is
the fact that density is highly influenced by the inspiration
level during the CT scan: as lung volume increases the lung
density will decrease. In Figure 3, data are shown from the
repeatability study of Stolk et al, in which patients were
scanned twice within a 2-week period by 2 different CT
scanners from the same manufacturer.61 On the x-axis the
difference in lung volume between the 2 (inspiratory) scans is
plotted against the resulting difference in 15th percentile
point, showing a significant correlation. The outlier in this
graph originates from a patient with a bimodal histogram
corresponding with a highly affected region in the basal area
and a compressed upper region (Fig. 4); this results in a
disturbed relation between volume and density.
Progression of emphysema generally is characterized
by hyperinflation and consequently by a gradual increase in
total lung capacity and the resulting decrease in lung density
could entirely be explained by this, thereby overestimating its
progression in terms of lung density.
684
FIGURE 4. 3D visualization of the localization of emphysema in
a patient with a bimodal histogram. The purple areas indicate
areas of low density, predominantly located in the lower
regions. Red: left lung; blue: right lung; green: trachea.
© 2004 Lippincott Williams & Wilkins
Investigative Radiology • Volume 39, Number 11, November 2004
be improved by spirometric control. In the so-called DutchDanish study24 the CT scans were spirometrically gated in an
attempt to keep the lung volume constant during follow-up,
based on a total lung capacity measurement before CT scanning. This procedure can simply be evaluated by calculating
the difference in CT volume from the baseline volume (see
Fig. 5). These results indicate that despite of the spirometric
control the lung volume could differ by more than 30%.
As a consequence, in longitudinal studies the correction
for volume differences should be performed afterward using
statistical procedures, for example by regarding the (log) lung
volume as a covariate.45 If one CT scan is acquired per
follow-up moment, these corrections can only be made under
the assumption that no systematic changes occur in the lung
volume over time. If this assumption does not hold true,
corrections can only be done when at least 2 density measurements are made per visit. As a consequence of the
spirometric control being unsuccessful, the most reproducible
measurement can only be obtained when volume CT is
performed covering the entire lungs, instead of single slice at
predefined anatomic levels.
IMAGE PROCESSING
In the literature, limited attention has been given to the
standardization of image processing, whereas its reproducibility plays a major role in the overall reproducibility of the
densitometric measurement. If the analysis is conducted automatically, the interobserver variability can be reduced by a
factor of five.66 At our division of image processing we have
developed a software program PulmoCMS in corporation
with a software company MEDIS medical systems (Leiden,
the Netherlands) with special focus on the reproducibility to
measure progression of pulmonary emphysema. To achieve
this, the analysis consists of 3 parts: (1) calibration for air and
FIGURE 5. Inability of spirometric control to keep the lung
volume constant during a period of 3 to 5 years.
© 2004 Lippincott Williams & Wilkins
Optimization of Lung Densitometry
blood; (2) (semi-) automatic selection of the lungs in the
three-dimensional image data; and (3) calculation and subsequent analysis of the density distributions.
Calibration
Factors such as changing spectra of the x-ray tube
caused by ageing, software updates that may include changes
in the reconstruction algorithms, and errors in the calibration
procedures in the CT scanner may negatively affect the
constancy of the density measurement.67 To correct for these
influences, the density data from the CT images are calibrated
a posteriori in the software program. In this way CT densitometry is less dependent on the type of CT scanner or its
state of maintenance.
The calibration procedure is based on 2 measurements
in the images: 1 measurement for the density of air, sampled
outside the patient; and 1 measurement of the density of
blood, taken from samples in the descending aorta. The
reference values for these measurements are fixed at ⫺1000
HU and 50 HU, respectively. This calibration procedure also
accounts for possible changes in the actual blood density:
since blood is one of the main contents that is being measured
in the lungs, changes in hematocrit and hemoglobin levels
(for example, from emphysema) would affect the lung density measurement. Therefore, if blood is taken as a reference
for calibration, these influences are accounted for, as well.
The measurement of blood density is carried out semiautomatically by indicating 2 points within the descending
aorta: one at the lowest slice where the lungs appear and one
at a level where the 2 lung halves circumvent the aorta. Next,
samples are taken automatically with the smallest standard
error of the mean.
The calibration for air is performed fully automatically.
In each slice, an area is defined outside the patient above the
sternum, in which a circular region is found with the smallest
standard error of the mean. As a result only samples are taken
from air, excluding any clothing (see Fig. 6). If necessary, the
user can manually or semiautomatically adjust the blood and
air samples.
FIGURE 6. Semi-automatic recalibration of the image data
based on samples from the aorta (A) and outside the patient
(B).
685
Stoel and Stolk
Investigative Radiology • Volume 39, Number 11, November 2004
Segmentation
The lungs are detected automatically by a so-called
region growing algorithm (Fig. 7A). A seed point is automatically placed in the trachea and this initial location expands
until it has reached the borders of the lungs. For this region
growing technique, a (calibrated) threshold value of ⫺380
HU is used that has been validated by Zagers et al by
comparison with manually drawn contours by an experienced
radiologist.66
A well-known drawback of region growing is leakage
into regions outside the intended object. Therefore, the outer
contours of the patient are detected beforehand, and this
region is taken as a mask in which the region may growing to
prevent leakage.
To reach a high level of reproducibility, the main stem
carina is used as a distinct anatomic landmark to define the
lungs (Fig. 7B and C). The entire trachea is detected and
excluded from the segmentation down to the level of the
carina, resulting in a definition of the lungs that is identical
throughout the follow-up period.
The left and right lung cannot be separated by the region
growing technique, since the septum between the lung halves is
too thin at some sites in the chest. Therefore, the septa are
detected in a separate step (Fig. 7D.). After completion, the
FIGURE 7. Segmentation of the lungs. Initial result (A), tracheal segmentation (B), Carina recognition (C), and detection
of the septa (D).
686
trachea and septa are subtracted from the initial lung segment
(Fig. 7A) and the left and right lung are identified.
Histogram Analysis
The resulting histogram is calibrated based on the
measurements in the aorta and outside the patient. Subsequently, 4 different parameters are calculated from the histogram: the MLD, the lung weight, the nth percentile point, and
the relative area below an indicated threshold value. Lung
volume is calculated by adding the volumes of pixels within
the lung, accounting for possible overlap between slices.
Quality Control
Special features are added in the analysis software to
make it possible to review each analysis session by a socalled audit trail, including time stamps. To prevent human
errors, the contours that have been produced by the user are
automatically checked for inconsistencies, such as overlapping contours between left and right lungs and inclusion of
areas above the predefined threshold.
CONCLUSIONS
Since lung densitometry has been validated against
pathology, lung function tests and health status, it is today
considered an important surrogate outcome parameter for clinical drug evaluation trials. As a result of this technique being
applied in these trials, standard procedures for image acquisition
and image analysis have already become available.
From the literature overview and additional data presented in this review, it can be concluded that, with the
current knowledge, the 15th percentile point is the most
sensitive densitometric parameter to use for the assessment of
(progression of) pulmonary emphysema.
Spirometric control cannot be employed successfully.
Therefore, density measurements should be corrected for
lung volume afterward and consequently volume CT scans
are required to determine the lung volume. Although CT
scans taken at inspiration level has proven to be most reproducible, more elaborate methods of correcting for lung volume may further improve the reproducibility of the measurement. This would require further investigation.
Standardization of the image acquisition protocol has
been implemented using a low radiation dose, but some small
improvements can still be made even with the most modern
CT scanners. The technical abilities of modern CT scanners
evolve at such a rapid pace, that standardization of their
acquisition protocols will be a continuous task.
Standardization of the image analysis procedures has
been underexposed in the literature. A proposal has been
made in this paper as to how this can be accomplished. To
assess the influence of inter- and intraobserver and interscan
variability on the overall variability, additional research is
however needed.
© 2004 Lippincott Williams & Wilkins
Investigative Radiology • Volume 39, Number 11, November 2004
To make lung densitometry available for clinical practice, normal values are required that originate from the
standardized CT protocol and image analysis procedures. If a
more precise densitometric measurement is required for follow-up, a standardized way is to be developed to statistically
correct for inspiration differences over time.
Despite the fact that some issues still need to be
resolved, standardization and optimization of lung densitometry has reached a higher standard than thus far has been
appreciated. Based on experience obtained during the development of optimized and standardized procedures for assessing pulmonary emphysema, we predict that the development
of these procedures will accelerate its application for other
lung diseases such as sarcoidosis, asthma, pulmonary edema,
acute respiratory distress syndrome, and cystic fibrosis.
REFERENCES
1. Wegener OH, Koeppe P, Oeser H. Measurement of lung density by
computed tomography. J Comput Assist Tomogr. 1978;2:263–273.
2. Robinson PJ, Kreel L. Pulmonary tissue attenuation with computed
tomography: comparison of inspiration and expiration scans. J Comput
Assist Tomogr. 1979;3:740 –748.
3. Rosenblum LJ, Mauceri RA, Wellenstein DE, et al. Density patterns in
the normal lung as determined by computed tomography. Radiology.
1980;137:409 – 416.
4. Goddard PR, Nicholson EM, Laszlo G, et al. Computed tomography in
pulmonary emphysema. Clin Radiol. 1982;33:379 –387.
5. Gevenois PA, deMaertelaer V, DeVuyst P, et al. Comparison of computed density and macroscopic morphometry in pulmonary emphysema.
Am J Respir Crit Care Med. 1995;152:653– 657.
6. Gevenois PA, DeVuyst P, deMaertelaer V, et al. Comparison of computed density and microscopic morphometry in pulmonary emphysema.
Am J Respir Crit Care Med. 1996;154:187–192.
7. Gould GA, MacNee W, McLean A, et al. CT measurements of lung
density in life can quantitate distal airspace enlargement–an essential
defining feature of human emphysema. Am Rev Respir Dis. 1988;137:
380 –392.
8. Hayhurst MD, MacNee W, Flenley DC, et al. Diagnosis of pulmonary
emphysema by computerised tomography. Lancet. 1984;2:320 –322.
9. Muller NL, Staples CA, Miller RR, et al. ‘Density mask’. An objective
method to quantitate emphysema using computed tomography. Chest.
1988;94:782–787.
10. Bankier AA, de Maertelaer V, Keyzer C, et al. Pulmonary emphysema:
subjective visual grading versus objective quantification with macroscopic morphometry and thin-section CT densitometry. Radiology.
1999;211:851– 858.
11. Baldi S, Miniati M, Bellina CR, et al. Relationship between extent of
pulmonary emphysema by high-resolution computed tomography and
lung elastic recoil in patients with chronic obstructive pulmonary disease. Am J Respir Crit Care Med. 2001;164:585–589.
12. Beinert T, Brand P, Behr J, et al. Peripheral airspace dimensions in
patients with COPD. Chest. 1995;108:998 –1003.
13. Biernacki W, MacNee W. Ocena postepu rozedmy pluc za pomoca
tomografii komputerowej u chorych na przewlekla obturacyjna chorobe
pluc. 关Evaluation of emphysema progression as measured by computed
tomography in patients with chronic obstructive pulmonary disease兴.
Pneumonol Alergol Pol. 1996;64:620 – 626.
14. Gould GA, Redpath AT, Ryan M, et al. Lung CT density correlates with
measurements of airflow limitation and the diffusing capacity. Eur
Respir J. 1991;4:141–146.
15. Guenard H, Diallo MH, Laurent F, et al. Lung density and lung mass in
emphysema. Chest. 1992;102:198 –203.
16. Haraguchi M, Shimura S, Hida W, et al. Pulmonary function and
regional distribution of emphysema as determined by high-resolution
© 2004 Lippincott Williams & Wilkins
Optimization of Lung Densitometry
computed tomography. Respiration. 1998;65:125–129.
17. Heremans A, Verschakelen JA, Van fraeyenhoven L, et al. Measurement
of lung density by means of quantitative CT scanning. A study of
correlations with pulmonary function tests. Chest. 1992;102:805– 811.
18. Zompatori M, Gavelli G, Fabbri M, et al. Bilancio della gravita
dell’enfisema polmonare mediante tomografia computerizzata. Confronto fra tomografia computerizzata ad alta risoluzione, valutazione
obiettiva con maschera di densita e test funzionali. 关The assessment of
the severity of pulmonary edema by computed tomography. A comparison between high-resolution computed tomography, objective evaluation with density masks and functional tests兴. Radiol Med (Torino).
1993;86:54 – 61.
19. Dowson LJ, Newall C, Guest PJ, et al. Exercise capacity predicts health
status in alpha(1)-antitrypsin deficiency. Am J Respir Crit Care Med.
2001;163:936 –941.
20. Dowson LJ, Guest PJ, Stockley RA. Longitudinal changes in physiological, radiological, and health status measurements in alpha(1)-antitrypsin deficiency and factors associated with decline. Am J Respir Crit
Care Med. 2001;164:1805–1809.
21. Stolk J, Ng WH, Bakker ME, et al. Correlation between annual change in
health status and computer tomography derived lung density in subjects
with alpha(1)-antitrypsin deficiency. Thorax. 2003;58:1027–1030.
22. Schwaiblmair M, Beinert T, Seemann M, et al. Relations between
cardiopulmonary exercise testing and quantitative high-resolution computed tomography associated in patients with alpha-1-antitrypsin deficiency. Eur J Med Res. 1998;3:527–532.
23. Dirksen A, Dijkman JH, Madsen F, et al. A randomized clinical trial of
alpha(1)-antitrypsin augmentation therapy. Am J Respir Crit Care Med.
1999;160:1468 –1472.
24. Gilman MJ, Laurens R-GJ, Somogyi JW, et al. CT attenuation values of
lung density in sarcoidosis. J Comput Assist Tomogr. 1983;7:407– 410.
25. Uppaluri R, Hoffman EA, Sonka M, et al. Interstitial lung disease - A
quantitative study using the adaptive multiple feature method. Am J
Respir Crit Care Med. 1999;159:519 –525.
26. Biernacki W, Redpath AT, Best JJ, et al. Measurement of CT lung
density in patients with chronic asthma. Eur Respir J. 1997;10:2455–
2459.
27. Mitsunobu F, Ashida K, Hosaki Y, et al. Decreased computed tomographic lung density during exacerbation of asthma. Eur Respir J.
2003;22:106 –112.
28. Newman KB, Lynch DA, Newman LS, et al. Quantitative computed
tomography detects air trapping due to asthma. Chest. 1994;106:105–109.
29. Pifferi M, Caramella D, Ragazzo V, et al. Low-density areas on highresolution computed tomograms in chronic pediatric asthma. J Pediatr.
2002;141:104 –108.
30. Gattinoni L, Pelosi P, Vitale G, et al. Body position changes redistribute
lung computed-tomographic density in patients with acute respiratory
failure. Anesthesiology. 1991;74:15–23.
31. Hedlund LW, Vock P, Effmann EL, et al. Hydrostatic pulmonary edema.
An analysis of lung density changes by computed tomography. Invest
Radiol. 1984;19:254 –262.
32. Kato S, Nakamoto T, Iizuka M. Early diagnosis and estimation of pulmonary congestion and edema in patients with left-sided heart diseases from
histogram of pulmonary CT number. Chest. 1996;109:1439 –1445.
33. Wollmer P, Rhodes CG. Lung density and pulmonary edema. Am
Heart J. 1985;110:1323.
34. Puybasset L, Cluzel P, Gusman P, et al. Regional distribution of gas and
tissue in acute respiratory distress syndrome. III. Consequences for the
effects of positive end-expiratory pressure. Intens Care Med. 2000;26:
1215–1227.
35. Goris ML, Zhu HJ, Blankenberg F, et al. An automated approach to
quantitative air trapping measurements in mild cystic fibrosis. Chest.
2003;123:1655–1663.
36. Madani A, Keyzer C, Gevenois PA. Quantitative computed tomography
assessment of lung structure and function in pulmonary emphysema. Eur
Respir J. 2001;18:720 –730.
37. Hoffman EA, Reinhardt JM, Sonka M, et al. Characterization of the
interstitial lung diseases via density-based and texture-based analysis of
computed tomography images of lung structure and function. Acad
Radiol. 2003;10:1104 –1118.
687
Stoel and Stolk
Investigative Radiology • Volume 39, Number 11, November 2004
38. Beinert T, Behr J, Mehnert F, et al. Quantitative Computertomographie
der Lunge–Respiratorisch kontrollierte Diagnostik diffuser Lungenerkrankungen. 关Quantitative computerized tomography of the lung–
respiration controlled diagnosis of diffuse lung diseases兴. Pneumologie.
1995;49:678 – 683.
39. Stoel BC, Bakker ME, Stolk J, et al. Comparison of the sensitivities of
5 different computed tomography scanners for the assessment of the
progression of pulmonary emphysema: a phantom study. Invest Radiol.
2004;39:1:1–7.
40. Coxson HO, Mayo JR, Behzad H, et al. Measurement of lung expansion
with computed tomography and comparison with quantitative histology.
J Appl Physiol. 1995;79:1525–1530.
41. Gattinoni L, Pesenti A, Bombino M, et al. Relationships between lung
computed tomographic density, gas exchange, and PEEP in acute respiratory failure. Anesthesiology. 1988;69:824 – 832.
42. Arakawa A, Yamashita Y, Nakayama Y, et al. Assessment of lung
volumes in pulmonary emphysema using multidetector helical CT:
comparison with pulmonary function tests. Comput Med Imaging
Graph. 2001;25:399 – 404.
43. Kinsella M, Muller NL, Abboud RT, et al. Quantitation of emphysema
by computed tomography using a ’density mask’ program and correlation with pulmonary function tests. Chest. 1990;97:315–321.
44. Park KJ, Bergin CJ, Clausen JL. Quantitation of emphysema with
three-dimensional CT densitometry: comparison with two-dimensional
analysis, visual emphysema scores, and pulmonary function test results.
Radiology. 1999;211:541–547.
45. Dirksen A, Friis M, Olesen KP, et al. Progress of emphysema in severe
alpha 1-antitrypsin deficiency as assessed by annual CT. Acta Radiol.
1997;38:826 – 832.
46. Dowson LJ, Guest PJ, Hill SL, et al. High-resolution computed tomography scanning in alpha1-antitrypsin deficiency: relationship to lung
function and health status. Eur Respir J. 2001;17:1097–1104.
47. Lamers RJ, Thelissen GR, Kessels AG, et al. Chronic obstructive
pulmonary disease: evaluation with spirometrically controlled CT lung
densitometry. Radiology. 1994;193:109 –113.
48. Lee KW, Chung SY, Yang I, et al. Correlation of aging and smoking
with air trapping at thin-section CT of the lung in asymptomatic subjects.
Radiology. 2000;214:831– 836.
49. Vikgren J, Bake B, Ekberg-Jansson A, et al. Value of air trapping in
detection of small airways disease in smokers. Acta Radiol. 2003;44:
517–524.
50. Kauczor HU, Hast J, Heussel CP, et al. Focal airtrapping at expiratory
high-resolution CT: comparison with pulmonary function tests. Eur
Radiol. 2000;10:1539 –1546.
51. Cederlund K, Tylen U, Jorfeldt L, et al. Classification of emphysema in
candidates for lung volume reduction surgery—A new objective and
surgical oriented model for describing CT severity and heterogeneity.
Chest. 2002;122:590 –596.
52. Coxson HO, Whittall KP, Nakano Y, et al. Selection of patients for lung
688
53.
54.
55.
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
volume reduction surgery using a power law analysis of the computed
tomographic scan. Thorax. 2003;58:510 –514.
Flaherty KR, Kazerooni EA, Curtis JL, et al. Short-term and long-term
outcomes after bilateral lung volume reduction surgery—Prediction by
quantitative CT. Chest. 2001;119:1337–1346.
Gierada DS, Yusen RD, Villanueva IA, et al. Patient selection for lung
volume reduction surgery: an objective model based on prior clinical
decisions and quantitative CT analysis. Chest. 2000;117:991–998.
Nakano Y, Coxson HO, Bosan S, et al. Core to rind distribution of
severe emphysema predicts outcome of lung volume reduction surgery.
Am J Respir Crit Care Med. 2001;164:2195–2199.
Uppaluri R, Hoffman EA, Sonka M, et al. Computer recognition of
regional lung disease patterns. Am J Respir Crit Care Med. 1999;160:
648 – 654.
Mishima M, Hirai T, Jin Z, et al. Standardization of low attenuation area
versus total lung area in chest X-ray CT as an indicator of chronic
pulmonary emphysema. Front Med Biol Eng. 1997;8:79 – 86.
Mishima M, Itoh H, Sakai H, et al. Optimized scanning conditions of
high resolution CT in the follow-up of pulmonary emphysema. J Comput
Assist Tomogr. 1999;23:380 –384.
Kemerink GJ, Kruize HH, Lamers RJ, et al. Density resolution in
quantitative computed tomography of foam and lung. Med Phys. 1996;
23:1697–1708.
Kemerink GJ, Kruize HH, Lamers RJ, et al. CT lung densitometry:
dependence of CT number histograms on sample volume and consequences for scan protocol comparability. J Comput Assist Tomogr.
1997;21:948 –954.
Stolk J, Dirksen A, van der Lugt AA, et al. Repeatability of lung density
measurements with low-dose computed tomography in subjects with
alpha-1-antitrypsin deficiency-associated emphysema. Invest Radiol.
2001;36:648 – 651.
Spiropoulos K, Trakada G, Kalamboka D, et al. Can high resolution
computed tomography predict lung function in patients with chronic
obstructive pulmonary disease? Lung. 2003;181:169 –181.
Lamers RJ, Kemerink GJ, Drent M, et al. Reproducibility of spirometrically controlled CT lung densitometry in a clinical setting. Eur Respir J.
1998;11:942–945.
Kalender WA, Rienmuller R, Seissler W, et al. Measurement of pulmonary parenchymal attenuation: use of spirometric gating with quantitative CT. Radiology. 1990;175:265–268.
Gierada DS, Yusen RD, Pilgram TK, et al. Repeatability of quantitative
CT indexes of emphysema in patients evaluated for lung volume reduction surgery. Radiology. 2001;220:448 – 454.
Zagers R, Vrooman HA, Aarts NJ, et al. Quantitative analysis of
computed tomography scans of the lungs for the diagnosis of pulmonary
emphysema. A validation study of a semiautomated contour detection
technique. Invest Radiol. 1995;30:552–562.
Stoel BC, Vrooman HA, Stolk J, et al. Sources of error in lung
densitometry with CT. Invest Radiol. 1999;34:303–309.
© 2004 Lippincott Williams & Wilkins