|
 |
INTRODUCTION |
|---|
TOP
INTRODUCTION
TECHNICAL CONSIDERATIONS
NORMAL STRUCTURE AND FUNCTION
DESCRIPTIVE STUDIES IN ASTHMA
FUTURE DIRECTIONS FOR THE...
SUMMARY
REFERENCES
|
|---|
Technical advances in medical imaging have provided an opportunity to examine the airways of the lung more accurately than ever before. The aim of this article is to review the use of computed tomography (CT) and more particularly, High-resolution CT (HRCT) in the investigation of airway structure and function in health and disease. Although we review both qualitative and quantitative studies, the quantitative studies are emphasized because of the enormous potential provided by the digital data on which this imaging modality is based.
HRCT has allowed visualization of airways and parenchyma in much greater detail than conventional CT and plain radiography and has made possible the investigation of the site, magnitude, and distribution of airway narrowing in vivo. Technical improvements have increased the spatial resolution of HRCT, making it theoretically possible to examine small airways. Because technical factors and the methods of analysis significantly influence the quality of HRCT images as well as their quantitative interpretation, we start with a brief review of the recommended methods.
| |
TECHNICAL CONSIDERATIONS |
|---|
|
TOP
INTRODUCTION
TECHNICAL CONSIDERATIONS
NORMAL STRUCTURE AND FUNCTION
DESCRIPTIVE STUDIES IN ASTHMA
FUTURE DIRECTIONS FOR THE...
SUMMARY
REFERENCES
|
|---|
Units of Measurement
HRCT data, like all radiographic data, are based on the variable absorption of X-rays by tissues. This is measured by
Hounsfield units (HU) and is arbitrarily set so that water has a
value of 0 HU and the range of density values ranges from
1,000 HU (which is the value for air) to a maximum of
+1,024 HU (which is the value for dense bone). Tissue densities can be calculated from HU by the following formula: density = (HU + 1,000)/1,000 g/ml. HRCT images are displayed
by assigning the range of HU to a 16-level gray scale, which is
approximately the limit of distinction between shades of gray
that the human eye can resolve. The window level is the HU at
which the midpoint of this 16-level gray scale is set. The window width is the range (in HU) over which this gray scale
spans, any pixels with values above the upper limit of the window width appear white, and any pixels with values below the lower limit of the window width appear black. Figure 1 shows
the relationship between HU and window levels and widths.
In this example any pixels with values above 500 HU will appear white while any pixels with values below 500 will appear black. The "lung window" level used for viewing and
photographing CT studies for clinical examination of lung parenchyma and airways is ~ 700 and the window width is usually between 1,000 and 1,500 HU.
|
Voxel and Pixel Size
Collimation is synonymous with "slice thickness" which, for HRCT, is usually 1 to 1.5 mm. The field of view (FOV) can be varied to accommodate the anatomical region of interest and is generally 40 × 40 cm for human lung studies. Matrix size is a measure of the number of pixels used to store the image information and should be the maximal number possible for the scanner which is usually 512 × 512. A FOV of approximately 13 × 13 cm produces the maximal spatial resolution; smaller FOVs do not improve resolution further (1). Using such a FOV and 512 × 512 matrix, the pixel size is 0.25 × 0.25 mm (0.06 mm2). The collimation thickness is the third dimension of the voxel, which represents a volume of lung being scanned. Using 1 mm collimation, combined with maximal pixel resolution, sets the voxel size at 0.25 × 0.25 × 1.00 mm (0.06 mm3 or 0.00006 ml). To resolve structures smaller than the pixel size their interface must be perpendicular to the plane of section (Figure 2).
|
Accurate definition of the inner and outer airway walls is required when performing quantitative studies of airway lumen and wall dimensions. Specifically, it is necessary to accurately identify the interface between lumenal air and airway wall, and between airway wall and the surrounding lung parenchyma. Although technical factors that can affect apparent airway wall thickness and airway lumen diameter may not be significant in everyday clinical practice where qualitative interpretation suffices, they are clearly important when making quantitative measurements on airways. Thus, the results of studies should be interpreted with knowledge of the inherent accuracy of the scanning techniques and analysis algorithms. Collimation, window level, window width, FOV, reconstruction algorithm, and the CT image analysis algorithm may affect the quality of the CT image and the accuracy of airway wall dimension measurements.
Window Levels and Widths
The window level has been shown to be especially important
in quantitative studies of airway lumen and wall dimensions.
The results of studies in which the relationship between window level and airway measurements has been examined, have
shown that airway wall thickness measurements are greater
and lumen areas less, when images are photographed at lower
window levels (2, 3). In some studies, window levels and
widths have been adjusted to make airways easier to visualize
and then measurements have been made using these images.
For example, Senéterre and coworkers used a protocol in
which the window level was adjusted to find the lowest value
at which the airway wall appeared to be an unbroken ring and
then these settings were used to make measurements. They set
the window width at 2 HU and although they found that the measurements were reproducible using these settings, they
noted that during these adjustments, some airways' lumen disappeared completely (4). This suggests that the window levels
resulted in overestimation of airway wall thickness and thus
underestimation of airway lumen area. They used a different
window level between and within subjects and because they
did not make any comparison with a known standard, it is unlikely that these results are anatomically accurate. Webb and
coworkers used phantoms which were composed of lucite cylinders whose lumen diameters ranged from 3.1 to 12.1 mm
and measured the wall thicknesses and lumen diameters using a variety of window levels when the cylinders were immersed
in water or air (3). When the cylinders were immersed in water, a window level of 150 HU appeared to provide the most
accurate results for lumen diameter and wall thickness measurements. However, when surrounded by air, a window level
of 450 HU produced the most accurate measurements of airway wall thickness.
McNamara and coworkers attempted to more closely simulate airways in the lung parenchyma by using a phantom made
of cylinders of sweet potato embedded within plastic sponges.
The range of lumen diameters in the 13 phantom airways was
1.0 to 5.4 mm and wall thicknesses ranged from 0.5 to 2.3 mm. They measured the phantom airway dimensions using an
optical micrometer to the nearest 0.01 mm which were then
scanned using 1.5-mm slice thickness CT (2). Figure 3 shows
the results of the phantom study of McNamara and coworkers.
The mean difference between the measured and CT estimates
of airway lumen diameter is shown for the 13 phantom airways versus the range of window levels and window widths
that were studied. It is apparent that window width has little
influence on the measurements, but the window level has a
dramatic effect. Using a window level of less than ~ 500 HU
led to an underestimation of airway lumen diameter, whereas
a window level greater than ~ 400 HU led to an overestimation of lumen diameter. As in the study of Webb and coworkers (3), a window level of 450 HU appears to give the closest
estimates of true airway lumen diameter. Of note was that the
error produced by using the incorrect window level was a constant in absolute terms and therefore was a smaller percentage
error as airway lumen size or wall thickness increased. These
data indicate that the error produced by the use of inappropriate window level is constant and therefore, related to edge detection. Clearly, this error is related to volume averaging in
which the tissue density represented by the pixel's HU value is
an average of the density of the tissues, which fall within that
pixel. For example, when a pixel contains both airway wall
and lumen, the HU value will be dependent upon the relative
amounts of wall or lumen in that pixel. When the window level
is too high, pixels that contain substantial portions of airway
wall are viewed as containing air and when the window level
is too low, pixels containing only a small amount of airway tissue are viewed as being entirely composed of airway wall (Figure 4).
|
|
Although these studies indicated that window level was more important than window width, the results of a study by Bankier and coworkers (5) suggested that window widths rather than window levels affected airway measurements. They studied cadaveric lungs that were fixed in inflation and then scanned using 1.5-mm and 10-mm collimation. The lungs were then cut in the same plane in which they were scanned and the dimensions of cross-sectioned airways were measured using planimetry. They measured the wall thickness along the shortest diameter of the airway and found that window levels did not influence measurements and that window widths between 1,000 and 1,500 HU yielded accurate wall thickness measurements. Widths less than 1,000 HU were associated with overestimation of wall thickness. This raises the possibility that the measurement of airway wall thickness in situ is affected by different display parameters than the measurement of airway wall area in phantoms.
Airway Orientation
In studies in which measurements of airway lumen and wall area have been made, the analyses have been restricted to airways that appear to have been cut in cross section based on the apparent "roundness" of the airway lumen. The long to short diameter ratio of the airway lumina is the most common way of assessing roundness and an upper limit of 1.5 is commonly used. This assumes that airway lumina are always round, even during airway narrowing induced by airway smooth muscle (ASM) contraction, and this assumption may not be correct. If angled orientation was the only cause of a deviation of the airway's long/short diameter ratio from 1, then including airways with a ratio of up to 1.5 will mean that airways imaged up to an angle of 48° will be accepted for measurement. Measuring airway lumen and airway walls when they are not perpendicular to the scanning plane may lead to significant errors, the magnitude of which will depend on how acutely the airways are angled, the collimation, and the FOV. The larger the angle and FOV and the thicker the collimation, the greater the overestimation of airway wall area.
Webb and coworkers are the only investigators who have reported data examining quantitative measurements from CT when airways are imaged at an angle other than 90° (3). They studied four phantom airways made of lucite (lumen diameters between 3.1 and 12.7 mm and outer diameters between 6.7 and 15.8 mm), oriented at angles between 0° and 90° and surrounded by either air or water. They used a pixel size of 1.1 × 1.1 mm and collimations of 1.5, 5, and 10 mm. Their data show that airway angle, airway size, and collimation interact to produce an error in lumen diameter measurement. It is important to note that they only assessed the effect of these parameters on the short diameter of the lumen and did not consider the effect on lumen area, the long axis of the lumen, or airway wall area. The short axis is underestimated when the airway angle and collimation are greater and the airway is smaller. This effect is clearly the result of volume averaging, which is greater under the aforementioned conditions (Figure 5). It is not possible, however, to infer from these results how airway angle, collimation, and airway size affect quantitative airway measurements. First, airways are not surrounded by either water or air, lung parenchyma has a density somewhere between the two. More importantly, lumen areas and not the short diameters are usually measured in quantitative studies; it would seem more likely that lumen area is overestimated with greater airway angulation, in comparison to the underestimation of the short diameter with increasing airway angle.
|
Resolution
The ability of HRCT to accurately measure small structures such as small airway lumen and thin airway walls is dependent on resolution and hence pixel size. The morphometrically measured airway wall thickness in 1.5-mm-diameter airways is about 0.15 to 0.2 mm; it is understandable that the lower limit of HRCT accuracy will be somewhere above this, because the smallest pixel size is 0.25 × 0.25 mm. The accuracy of lumen measurements has been addressed in two studies in which Plexiglas phantoms were used (6, 7) as well as in the study of the sweet potato phantom mentioned previously. These data suggest that at the appropriate window level and width, the smallest visible airways on which reasonably accurate measurements of lumen diameter can be made, are in the range of 1.5 to 2 mm diameter. This represents the upper size limit of the so-called "small airways" as defined physiologically. This remarkable degree of precision allows a whole new approach to the study of small airways disease.
Assessing Acute Changes in the Airways
The use of HRCT scanning allows the measurement of airway wall dimensions before and after an intervention such as the administration of a bronchoconstricting agonist. The ability to make accurate measurements of changes in airway wall or lumen dimensions after an intervention is dependent on the ability to match airways pre- and postintervention. It is insufficient to simply use the same slice number because very small changes in lung volume can alter the relationship between the slice number and lung anatomy, even if the subject's position remains constant. This problem has been approached by obtaining a series of scans through an area of interest before and after an intervention and then using anatomical features such as vascular or bronchial bifurcations, fissures, and mediastinal structures to match scan levels pre- and postintervention. Although this appears to be a reasonable approach, the error produced by estimating matched slices in such a manner has not been calculated. The ability to match the levels pre- and postintervention has been greatly aided by the development of volumetric or spiral CT. Using spiral CT, data from a large portion of lung can be acquired during a single breath hold while still using 1-mm collimation. The contiguous 1-mm cross sections can then be "stacked" one on top of each other generating a complete data volume on which reconstructions can be made in all three planes. The use of the volumetric scanning allows reconstruction of images at any desired level within the imaged volume of lung and therefore allows many more airways to be matched pre- and postintervention. Another technical issue that has to be addressed when scans are to be compared is how to ensure comparable lung volume. Lung volume influences airway wall and lumen dimensions in a predictable manner (8) and if the intervention itself has the capacity to alter lung volume, changes in these dimensions could be due to changes in lung volume rather than to actual changes in the airways themselves. To circumvent this, Okazawa and coworkers (9) used a spirometer in the CT suite to ensure that lung volume was comparable before and after a bronchoconstricting stimulus.
Reconstruction Algorithm
The use of a high-spatial frequency reconstruction algorithm is recommended for HRCT data because it increases spatial resolution. However, image noise becomes more apparent with high-spatial frequency reconstruction which may impair the observer's ability to resolve small airways. This may become important when using visual techniques for measurement of airway dimensions. Image noise is inversely proportional to the number of photons emitted (or radiation dose used). Increasing the kilovolt peak (kVp) or milliamperes (mA) will decrease inherent noise at the expense of increasing the radiation dose. However, a systematic study of the effect of reconstruction algorithm and noise on the accuracy of airway dimensions has not been done.
Advanced Image Reconstruction Techniques
Using volumetric imaging, Ferretti and colleagues have developed a reconstruction technique which produces simulations of an internal view of the major airways that can be seen at bronchoscopy (10); they refer to this as "virtual bronchoscopy." The resolution of the scans is sufficient to view the segmental bronchi, and surface shading allows the contours of the airways' inner walls to be visualized as if illuminated from the light source of a bronchoscope. Although this technique provides attractive images and may help to plan a surgical approach to lobectomy or pneumonectomy, it has not been used to quantify airway dimensions. A sophisticated image analysis technique was developed by Wood and coworkers to make quantitative measurements of airway wall and lumen areas from spiral CT data (11). The first step in their analysis was to convert the asymmetric CT voxels into cubic dimensions (isotropic voxels). Because volumetric CT data generated using a 20-cm FOV, 512 × 512 matrix and 1-mm collimation have voxel dimensions of 0.39 × 0.39 × 1.00 mm, voxels were converted to approximately 0.4 × 0.4 × 0.4 mm voxels by interpolation in the longitudinal axis. This manipulation allows the images to be reconstructed in any orientation. They then defined the central axis of the airway and reconstructed the airway lumen in a plane perpendicular to this axis. This analysis technique overcomes the major limitation to the use of HRCT in quantitative analysis, which is that accurate or true airway lumen and airway wall area can only be measured from airways which are oriented approximately perpendicular to the plane of scanning. They used CT data from a phantom study to validate their technique; their data showed that the measurements were accurate in airways; larger than 2 mm lumen diameter. There was an overestimation of airway diameter of approximately 20% in 2-mm-diameter airways; this equates to an approximately 40% error in lumen area and 50% error in 1-mm airways equating to approximately 140% error in lumen area.
Quantitative Image Analysis Techniques
Numerous image analysis techniques have been developed to make measurements of airway dimensions. Although these have been used almost exclusively for research purposes, they will, with continuing refinement, eventually be of benefit in the clinical practice of radiology. All, except the algorithm developed by Wood and coworkers described previously, are based on identifying airways that appear to have been imaged cross-sectionally and, therefore, appear round. Development of analysis algorithms requires validation using data from phantom studies; this allows the inherent accuracy of the "whole system" (scanner, technique, reconstruction algorithm, and analysis algorithm) to be assessed. When applied to scans of real lungs the accuracy will be less than the inherent accuracy of the "whole system," owing to the effects of varying airway tissue density (especially at the adventitial-parenchymal border and at mucosal folds), the presence of adjacent blood vessels that cannot be separated from the airway wall, and the inevitable tangential slicing. The accuracy of any analysis techniques could be more realistically assessed if excised lungs could be scanned and the HRCT measurements of airway dimensions could be directly compared with morphometric measurements. Studies of the reproducibility of airway measurements are also needed if longitudinal studies are to be performed. This information is important when examining whether there is heterogeneity of airway response to pharmacologic stimuli. To conclude that a heterogeneous response is present, the variable response to the intervention must be greater than the measurement variability.
McNamara and coworkers (2) and Webb and coworkers
(3) generated images at the "optimal" window level and width
and measured the airway lumen and adventitial interface by
eye on enlargements of the images. This analysis technique is
time-consuming and potentially prone to significant interobserver differences. Herold and coworkers photographed 2-mm
scans at 450 HU and digitized the images using a video camera and viewing box (7). They used a digital image analysis
program which required the operator to define a "seed point"
at the lumen-wall interface; the program then drew an isocontour line to define the lumen. This method is also dependent
on the operator's ability to determine the interface and could
be prone to interobserver differences when applied to the
study of airways in vivo although it appeared to be accurate
for Plexiglas airways of 1 mm diameter or greater.
Amirav and colleagues developed a more operator-independent algorithm to measure the airway lumen (6). Their algorithm is an edge detection method based on the "full width at half maximum" principle. They first estimated the lumen perimeter by a hand drawn line that was then repeatedly smoothed. Multiple lines were then generated perpendicular to the smoothed perimeter, radiating outward away from the lumen into the airway wall and parenchyma. The profile of HU along this line has a minimum in the lumen and a maximum in the soft tissue. The middle value ("half maximum") is calculated and the point on the hand drawn line is then moved to the "half maximum" point. This is repeated for all points radiating from the hand drawn line that now defines the airway lumen perimeter. The advantages of this method are that it is relatively operator-independent and very fast. Studies of a Plexiglas phantom (3-mm collimation) showed it to be highly accurate for "airways" 2 mm diameter or greater. The algorithm's measurement variability was also estimated by the coefficient of variation (CV = SD/mean) from repeated measurements of the same phantom airway. The CV was 17.8% for 1-mm- and 2.2% for 6-mm-diameter phantom airways, respectively. Variability was similar when making in vivo measurements in ventilated pigs, in which all respiratory motion artifacts could be eliminated; the CV were 16.6% and 4.4% for 1.3-mm- and 6.3-mm-diameter airways, respectively.
McNitt-Gray and coworkers (12) tested an analysis method
in which a threshold number was used to detect the airway lumen area; all pixels with values below this threshold being designated as lumen. They found that a threshold value of 500
HU yielded the most accurate measurements of the lumen of
a bronchial phantom. This threshold is consistent with the
findings of other studies (2, 3).
| |
NORMAL STRUCTURE AND FUNCTION |
|---|
|
TOP
INTRODUCTION
TECHNICAL CONSIDERATIONS
NORMAL STRUCTURE AND FUNCTION
DESCRIPTIVE STUDIES IN ASTHMA
FUTURE DIRECTIONS FOR THE...
SUMMARY
REFERENCES
|
|---|
Animal Studies
There have been five studies in which quantitative methods have been used to examine airway narrowing in animals using CT (2, 6, 13). The main conclusions of those studies are that (1) it is possible to directly measure the magnitude of airway narrowing in vivo, (2) the magnitude of narrowing does not correlate with functional indices of airway narrowing, (3) the magnitude varies greatly between similar and different sized airways (heterogeneity of response), (4) airway diameter changes anisotropically with lung volume, (5) the magnitude of narrowing during carbachol challenge and lung deflation appears to be greater in intermediate-sized airways, and (6) airway wall area decreases during airway narrowing caused by pharmacologically induced ASM contraction. The results of these animal studies are important because they highlight the possible information, limitations, and difficulties that can be expected in similar human studies.
Heterogeneity in the distribution of airway obstruction, during agonist-induced airway narrowing, was observed in anesthetized and ventilated pigs (6) and dogs (7, 13). Although a dose-response relationship between log dose of methacholine and the magnitude of lumen narrowing was demonstrated, no correlation was found between lung inflation pressures (a measurement which has been used to estimate the increase in pulmonary resistance) and the magnitude of lumen narrowing determined by CT. The investigators did not offer any explanation for the lack of correlation but possible explanations include too small a sample of airways given the heterogeneity of airway narrowing and the fact that they scanned only a limited lung region (the diaphragmatic lobes). Another important finding reported in these studies was that the magnitude of airway narrowing and its heterogeneity were not different when histamine was administered by nebulizer or by the intravenous route (13). This suggests that the heterogeneity is predominantly due to local mechanisms in or around the airways rather than to the variable deposition of agonist.
HRCT has also been used to examine the relationship between changes in airway diameter and changes in lung volume (8). This relationship had been previously studied using tantalum bronchography, and the data suggested that airway diameter changed as the cube root of lung volume, i.e., the change in airway diameter or circumference could be predicted for a given change in lung volume. HRCT has theoretical advantages over the older method; it may be used to study humans because it is noninvasive, more objective measurements can be made using advanced image analysis algorithms, and it avoids the use of a bronchographic contrast material which could itself cause airway narrowing due to mucosal irritation. In addition, if airways narrow by assuming an oval or flattened shape rather than concentrically, the unidimensional estimate of diameter, which is provided by bronchography, could over- or underestimate true lumenal area. Brown and Mitzner (8) used HRCT to construct airway pressure-diameter curves in dogs and concluded that airway diameter changed anisotropically with lung volume and that the relationship was affected by ASM tone. Their results showed that airway diameter increases substantially with relatively small increases in lung volume and reaches a plateau at transmural pressures of only 5 to 7 cm H2O in lungs pretreated with atropine. After methacholine induced airway narrowing, airways continued to dilate even up to a transmural pressure of 30 cm. It is important that these studies are confirmed in humans because the relationship between airway circumference and lung volume is vital to making the link between mechanics data from studies of in vitro smooth muscle preparations, mathematical modeling of the tracheobronchial tree, and in vivo studies of airway narrowing and hyperresponsiveness.
McNamara and coworkers studied airway narrowing in excised dog lungs (2). They found that (1) in response to carbachol, the airways 2 to 6 mm in diameter tended to narrow more than larger airways, (2) airway narrowing in response to lung deflation tended to be greater in intermediate-sized airways (2 to 4 mm diameter, Figure 6), (3) airways in lobes held at lower transpulmonary pressure narrowed more in response to carbachol, and (4) the airway wall area decreased during airway narrowing induced by agonist but not during lung deflation alone.
|
The mechanism of the greater airway narrowing observed in intermediate-sized airways is unclear. Possible reasons include preferential deposition by the nebulized agonist in these airways (although the results of Brown and coworkers [13] cited previously would suggest otherwise), or differences in the proportion, contractility, or loads on ASM in this sized airway. The observation of a decrease in airway wall area during ASM contraction is even harder to explain on the basis of recognized airway anatomy and physiological principles. Although compression of airway structures internal to the airway smooth muscle layer might expel bronchial capillary blood or interstitial fluid, the displaced blood or fluid should relocate to the outer airway wall compartment where one would expect a negative pressure to develop; such a relocation of fluid would not result in a net change in airway wall area. Airway lengthening secondary to hyperinflation is another possible explanation for an apparent decrease in airway wall area, but in these studies hyperinflation did not occur as determined by the lack of change in the lung parenchymal area on HRCT. The authors also considered the possibility that this unusual observation could be due to a measurement artefact. Interestingly, similar results have been reported in rabbits in which airway wall area measured by morphometry was found to decrease after prolonged bronchoconstriction (~ 30 min) but not after bronchoconstriction for a shorter duration (14).
Human Studies
To date there are only two reports of the use of quantitative HRCT techniques to study the effects of airway smooth muscle contraction on normal and asthmatic human airways (9, 15). Okazawa and coworkers (9) found that after nebulized methacholine challenge, airways narrowed throughout the bronchial tree; similar to the canine studies the percent narrowing was greater in the intermediate-sized airways (2 to 4 mm diameter, Figure 7). The magnitude and distribution of airway narrowing was not different in the asthmatics and the normal subjects, although the concentration of methacholine aerosol delivered to the normal subjects was approximately 25-fold greater than the asthmatics. Another similarity to the results of canine studies was that airway wall area decreased during airway narrowing in the normal subjects. However in asthmatics there was no change in airway wall area and the difference between the behavior of the normal and asthmatic airways, with respect to wall area, was significant in airways less than 6 mm initial diameter (Figure 8).
|
|
As in the animal studies, the explanation for the changes in airway wall area is unclear, but the differences that were observed between normals and asthmatics could have important implications for the mechanism of airway hyperresponsiveness in asthmatics. If the relatively greater airway wall area in asthmatic subjects was caused by an increase in fluid volume in the tissue compartment external to the ASM, this could reduce the stress applied to the ASM at any transmural pressure. This would allow the ASM to shorten more and produce a greater degree of narrowing for the same amount of force generation by the ASM. This mechanism has been considered to be a potentially important factor contributing to airway hyperresponsiveness. Results of previous studies in which 131Xe was used to assess ventilation distribution suggested heterogeneity of airway narrowing during methacholine challenge in normal and asthmatic subjects. However, the heterogeneity in the distribution of airway narrowing which has been found in animals using HRCT has not been studied in normal or asthmatic human subjects.
Goldin and coworkers (15) studied six normals and 15 asthmatics using "thin-section CT" and measured airway lumen areas using a quantitative analysis method (12). Despite only a 3% decrease in FEV1 after methacholine challenge in normal subjects, they measured a 30% decrease in lumen area in 1.6 to 2.5 mm diameter airways in which the greatest narrowing occurred. The investigators, however, did not ensure similar lung volumes during HRCT between pre- and postmethacholine scans, which were performed at FRC. In normals however, it is likely that the differences in lung volumes were negligible because the mean change in FEV1 was only 3%. Also, although they describe their technique as "thin-section CT," they used 3-mm rather than 1- to 1.5-mm collimation, which could theoretically lead to underestimation of airway lumen areas for reasons described previously.
| |
DESCRIPTIVE STUDIES IN ASTHMA |
|---|
|
TOP
INTRODUCTION
TECHNICAL CONSIDERATIONS
NORMAL STRUCTURE AND FUNCTION
DESCRIPTIVE STUDIES IN ASTHMA
FUTURE DIRECTIONS FOR THE...
SUMMARY
REFERENCES
|
|---|
There have been a variety of reports, describing rather than quantifying HRCT findings in asthmatic subjects (16). The commonest abnormalities found were bronchial wall thickening and bronchial dilatation. The prevalence of abnormalities in such series, which are based on subjective interpretation, is dependent on patient selection (e.g., smokers or nonsmokers), standardization of diagnostic criteria, technical specifications of the scanners used, window settings used for photographing scans, and the presence and nature of a control group. The use of healthy control subjects is important and in three studies such a group has been included (16, 18, 20). Direct comparisons between such studies and extrapolation to the general asthmatic populations are generally not possible because of differences between studies with respect to those factors listed previously. The relatively high interobserver variability reported by Grenier and coworkers (17) suggests inherent differences in diagnostic criteria exist between observers. These observational studies, however, provide some interesting insights into the structural abnormalities present in asthmatic patients in vivo.
Bronchial Wall Thickening
There are several possible causes of airway wall thickening in asthma, a structural change that may have important functional consequences. Thickening may result from edema and a transient inflammatory infiltrate but may also be due to an increase in the mass of ASM, extracellular matrix, and elastic and collagenous tissue. Modeling studies suggest that increases in airway wall thickness may potentiate the amount of airway narrowing for any given amount of ASM shortening without reducing resting maximal flow rates below normal. Furthermore, an increase in ASM mass, if accompanied by preservation of contractile capacity, could be an important contributor to excessive narrowing in asthmatic airways. There is still some debate, however, as to the degree to which each of these airway wall compartments contributes to wall thickening and how they may influence airway function. Unfortunately, it is difficult to define bronchial wall thickness in a normal population in CT studies because such measurements require standardization for airway size and lung inflation and possibly also lung size, age, and sex. Airway wall thickness has been related to airway size in histological studies by expressing the airway wall area as a function of airway internal or basement membrane perimeter and by calculating wall area as a percentage of idealized lumenal area plus wall area. However, there are no comparable HRCT data on wall area or fractional wall areas in normals and therefore interpretations of abnormalities in airway wall thickness in clinical HRCT have been largely subjective. In fact on HRCT, a reliable yardstick of airway size such as basement membrane perimeter is not possible because the airway wall compartments cannot be distinguished. An airway may appear thickened simply because it is narrowed. It is likely that to adequately describe the quantitative changes in wall dimensions, a frequency distribution of airway wall areas from airways sampled in a defined manner will have to be employed.
Nonetheless, the results of CT studies are consistent with the increased airway wall thickness described in surgical and postmortem studies, although there are large differences in the reported prevalence, probably related to the methodological considerations discussed previously. In the studies described subsequently, bronchial wall thickening was subjectively interpreted by radiologists blinded to the clinical status of the patients. Lynch and coworkers (18) reported results of a study in which the presence of abnormalities in three HRCT sections was determined by the consensus interpretation of two independent radiologists. Bronchial thickening was present in 44 of 48 (92%) asthmatics compared with 5 of 27 (19%) healthy control subjects. Angus and coworkers reported bronchial wall thickening in 16 of 17 asthmatic subjects who had clinical evidence of allergic bronchopulmonary aspergillosis (ABPA) and in nine of 11 asthmatic subjects who did not have ABPA (19). Park and coworkers reported bronchial thickening in 44% of HRCT scans from 39 asthmatic patients compared with 4% of scans from 14 normal subjects (16). One study differs notably from those previously described, in that objective quantitative image analysis methods were used to measure differences between asthmatic and normal subjects. Okazawa and coworkers (9) found an increase in airway wall area of airways less than 6 mm in diameter in six clinically stable asthmatics compared with similar sized airways in six age-matched healthy control subjects.
Asthmatic airway wall thickening does not appear to be reversible, at least in the short term. Paganin and coworkers (20) reported results of a HRCT study of 57 randomly selected asthmatics in which airway wall thickness was subjectively assessed by three independent radiologists by consensus. Airway wall thickening was not found in any of the 10 normal control subjects and was found in many fewer patients (16%) than in the aforementioned studies. Ten asthmatics who were having acute exacerbations, were also scanned 2 wk apart, between which they received methylprednisolone 2 mg/kg of body weight systemically and intravenous salbutamol, 0.05 mg/kg of body weight. Airway wall thickening was present in four of these patients but did not change after treatment. This study was short-term and although the sensitivity of subjective methods is unknown, differences in the number of asthmatics and normals who had airway wall thickening were found. However, subjective interpretation may not be able to detect small changes in individuals after treatment.
CT data have also provided some support for a relationship
between airway wall thickness and airway hyperresponsiveness; Boulet and coworkers found a negative correlation between the thickening of the intermediate stem bronchus and
the provocative dose of methacholine causing a 20% fall in
FEV1 (PD20) (22). Although there was no correlation with the
FEV1, these results suggest that either structural changes in
the large airways are an important determinant of airway hyperresponsiveness or that the changes measured in the airway
wall of the central airways reflect similar changes throughout
the whole bronchial tree. Airway diameter was measured using electronic calipers that were part of the standard set of
analysis instruments accompanying the CT scanner. Such a visual measurement method is potentially subject to significant variability and analyses were limited to the intermediate bronchus owing to poor reproducibility in smaller airways. In addition, images were viewed at a window level of 600 HU which
may have led to quantitative inaccuracies.
Awadh and colleagues (23) studied three groups of asthmatics and a control group: asthmatics who had a history of previous life-threatening attack (defined by intubation or hypercapnia), asthmatics who were taking 1,000 µg/d of inhaled corticosteroid, and asthmatics who were taking less than 1,000 µg/d. HRCT were performed using 1-mm slices, at five selected levels at "end-inspiration." Airway wall thickness was measured in the shortest diameter and standardized for airway size by dividing by the diameter of the outer airway wall. In asthmatics who had a history of life-threatening attacks and asthmatics who were taking 1,000 µg/day of inhaled corticosteroid, the mean airway wall thickness was similar but was greater than in both the control subjects and in the asthmatics requiring less than 1,000 µg/day of inhaled corticosteroid. It should be noted that the thickness/outer wall diameter ratio that they used would be sensitive to the degree of narrowing or dilatation of the airways. Airway walls must thicken during airway narrowing if the wall area is conserved; thus for a given wall area, ASM contraction thickens the wall while reducing the outer diameter. A more reliable measure would be the airway wall area. The results of this study could therefore have been affected by any differences in lung inflation and the presence of reversible airway narrowing.
Airway Narrowing in Asthmatics Measured by HRCT
The two most common reasons for measuring airway size would be to estimate the distribution of airway lumen areas in a subject group (for clinical research studies) or to detect abnormalities such as airway dilatation or constriction, in clinical studies. The distribution of lumen areas in randomly selected airways could be used to estimate the overall airway lumen area in a particular subject and could be compared between groups. The frequency distribution of airway lumen areas has been found to be shifted to the left (a preponderance of smaller lumen) in asthmatics compared with age-matched normals (9). Such a shift could be due to both an increased ASM tone at the time of scanning and/or to structural changes in the airway wall leading to increased airway wall thickness which encroaches into the airway lumen.
Another method of measuring airway lumen area is to compare it with the accompanying blood vessel and calculate a ratio. This method is commonly used in clinical practice to detect airway dilatation and "bronchiectasis" defined as the presence of airways larger than their accompanying vessel. In the study by Park and coworkers (16), the mean inner bronchial/arterial diameter ratio of asthmatics who had a normal FEV1 was 0.60 ± 0.16. In those whose FEV1 was decreased but above 60% predicted, the ratio was also 0.60 ± 0.18 compared with the ratio of 0.65 ± 0.16 in normals. In asthmatic subjects whose FEV1 was less than 60% predicted, the mean ratio (0.48 ± 0.11) was lower than that in normals. In this study, there was no difference in the ratios measured in different lobes or segments. Berend and coworkers examined inner bronchial diameter/arterial ratios by quantitative histology (24) and also reported a mean ratio of 0.62 ± 0.02 and no differences between lobes and segments, which is remarkably similar to that found in the HRCT study.
Kim and coworkers used HRCT to examine the arterial/ outer bronchial diameter ratio in patients who did not have cardiopulmonary disease and found a wide range of ratios (25). Additionally they found statistically significant differences between segments and between lobes. The mean arterial/outer bronchial diameter ratio was 0.98 ± 0.14 (outer bronchial/arterial diameter ratio = 1.02). The inner bronchial/ arterial ratio in this study can be calculated as 0.65 by taking the ratios minus 2 × the wall thickness, which is similar to the normal ratios reported by Park and coworkers (16).
Bronchial Dilatation in Asthmatics
Seemingly contradictory to the airway narrowing reported in the previously cited studies, the presence of bronchial dilatation (Figure 9) has also been reported in many descriptive studies of airway abnormalities in asthmatics (16, 18, 19, 21, 26). The reported prevalence of bronchial dilatation in asthmatic patients who do not have clinical evidence of ABPA varied between 18 and 77% in these studies. In three studies HRCT data from healthy control subjects were compared with the scans of asthmatics. Lynch and coworkers (18) found that 36% of all airways examined in CT scans from 48 asthmatic patients had bronchial dilatation defined as an internal diameter greater than the accompanying artery, and 77% of these asthmatic patients had at least one dilated bronchus. They also found bronchial dilatation in 19% of all airways examined in 27 healthy control subjects; 59% of subjects had at least one dilated bronchus. Park and coworkers (16) reported bronchial dilatation in 31% of CT readings by two independent observers in 39 patients who had asthma compared with 4% of CT readings from 14 normal subjects. Paganin and coworkers examined the prevalence of "bronchiectasis," defined by the criteria of Naidich and colleagues (27) in 126 randomly selected asthmatic patients. Bronchiectasis was found in approximately 85% (exact numbers unspecified) of nonatopic asthmatic patients who had clinically severe disease, as judged by a clinical asthma severity score compared with approximately 50% in those who had clinically mild disease. Bronchiectasis was present in approximately 50% of atopic asthmatic patients who had clinically severe disease and in approximately 20% who had clinically mild disease. No abnormalities were found in 10 healthy control subjects (21).
|
In two additional studies, bronchial dilatation was reported by blinded CT readers using subjective interpretation but normal control subjects were not included. However, 14 of 17 (83%) asthmatic patients who had ABPA had bronchial dilatation compared with 2 of 11 (18%) asthmatics who did not have ABPA (19). In another study, 41% of airways in eight patients who had ABPA were judged to be dilated compared with 15% of airways in asthmatics who did not have ABPA (26).
Because bronchial dilatation in asthma has only been recognized recently due to the introduction of HRCT, its functional significance is yet to be studied. The fact that 4 to 19% of airways are bigger than the accompanying artery in healthy control subjects (16, 18) illustrates the wide variation in this ratio in normals (25). However, the evidence suggests that this ratio tends to be higher in asthmatics; there is clearly much variability between airways in their degree of dilatation or narrowing because dilatation and narrowing have been reported in the same patient. This heterogeneity could be between parallel airways, or the dilatation and narrowing could be serially distributed along the airway. Parallel heterogeneity in the distribution of airway narrowing has been proposed to explain the poor correlation between changes in lung resistance and the degree of airway narrowing seen on HRCT (13). To date, however, there have been no studies on serial heterogeneity of airway caliber.
An alternative explanation for airway dilatation is a narrowing of the accompanying artery resulting in an apparent
increase in the bronchial diameter. Pulmonary arteries are responsive to changes in alveolar oxygen tension and an increase in the bronchial/arterial diameter ratio has been observed in HRCT studies performed at altitude compared with
at sea level, presumably owing to the reduced ambient oxygen
tension at altitude (28). In addition, heterogenous changes
have been observed in relatively large diameter arteries in response to hypoxia in HRCT studies (29). Heterogeneity in the
bronchial/arterial diameter ratio is likely to be associated with
abnormalities in ventilation perfusion distribution, which are
common in asthmatic subjects, even during clinically stable periods. Photographing HRCT scans at lower window levels
also increases the measured bronchial/arterial diameter ratios
(28) by increasing the apparent lumen diameter while reducing the measured arterial diameter. This could explain some of
the high prevalence of apparent bronchial dilatation reported
in asthmatics and normals because a window level of 700
HU is commonly used in clinical reporting.
Emphysema and Gas Trapping
The extent and severity of emphysema can be assessed on CT
using either visual scoring systems or by using a computer program that highlights all pixels within a given range (Density
mask) (30). The optimal threshold for the detection of emphysema using the Density mask is 910 HU on conventional 7- to
10-mm collimation CT scans (30) and 950 HU on high-resolution CT (31). Although it provides better depiction of the
parenchymal abnormalities, high-resolution technique is not
essential for the assessment of emphysematous areas as the results of studies demonstrate a good correlation between quantitative histology and CT using 10-mm collimation (30, 32).
In asthmatics, low attenuation areas on CT have been reported and could be the result of a regional reduction in pulmonary blood flow secondary to hypoxic pulmonary vasoconstriction (16), gas trapping, emphysema, or a combination of all three processes. Because it is impossible to examine the gross pathology of resected lung in asthmatics as has been done in smokers who have emphysema, it is unclear to what extent these processes contribute to low attenuation areas in asthma. Paganin and coworkers, using a visual scoring system for emphysema on CT, have reported the presence of emphysema in nonsmoking asthmatic patients (20, 21).
Newman and coworkers calculated the percent of the lung
in single slices whose pixels had an attenuation value of less
than 900 HU, using a CT slice at the level of the transverse
aorta and another just above the diaphragm (33). They scanned
18 asthmatics and 22 normal controls at both full inspiration
and full expiration using 1.5-mm and 10-mm thickness. They
reported significantly more extensive low attenuation areas in
the diaphragmatic 1.5-mm slice of expiratory scans in asthmatics (mean ± SEM: 10 ± 1.4%) compared with the normal control subjects (1 ± 1.3%). The percentage of low attenuation
areas was less in the 10-mm slices but there were still significant differences between asthmatics and normals on the expiratory scans; the differences between asthmatics and normals
disappeared on the inspiratory scans. In this study, asthmatics were older than the normal control subjects (53.5 yr versus
32.2 yr) which could have biased the results since in previous
studies performed at near full inspiration, low attenuation regions defined using a Density mask of 950 HU, are more extensive with increasing age (34). They also reported positive
correlations between the extent of low attenuation areas with
the absolute TLC (r = 0.64), residual volume (RV) (r = 0.78),
and FRC (r = 0.74) but did not perform a multivariate analysis. Furthermore, normalization of the extent of low attenuation areas for lung size should probably have been made
because these two parameters were correlated (34). Nonetheless, the absence of a difference on inspiratory scans strongly
suggests that in the asthmatics, there is little or no emphysema
and that substantial gas trapping may occur during expiration because of severe airway narrowing and/or airway closure
(Figures 10A and 10B). The gas trapping seen on CT was also
evident in the asthmatics' lung function; the RV was 153 ± 11% of predicted.
|
Kinsella and coworkers (35) evaluated the extent of "emphysema" in 10-mm slice thickness CT using a previously described visual scoring system (36). They studied 10 asthmatics who were nonsmokers but had suffered from asthma for at least 10 yr and compared them with an age- and sex-matched group of chronic cigarette smokers who had fixed airflow obstruction. The mean TLC of asthmatics was 109% predicted and was increased (greater than 120% predicted) in four asthmatics; the mean TLC of smokers was 105% of predicted and increased in three smokers. A median of 0% (range 0 to 4%) of the lung was affected by "emphysema" in asthmatics and a median of 10% (range 1 to 60%) was affected in the smokers.
Gevenois and coworkers measured the extent of low attenuation lung regions in 10 mild asthmatics who were asymptomatic and did not require anti-inflammatory medication,
seven severe asthmatics who had airflow obstruction and hyperinflation (TLC > 115% of predicted), and 42 normal control subjects (34). All subjects were nonsmokers. They acquired eight 1-mm-thick scans from the aortic arch to the
diaphragm and used a CT Density mask of 950 HU. They
did not find any differences in the proportion of low-density lung regions between groups. They also did not find any
changes in the relative size of low-density lung units after allergen challenge although hyperinflation had occurred, probably because CT scans were taken at or near TLC. The gas
trapping observed in the expiratory CT scans in the study by
Newman and coworkers (33) would suggest that differences
may have been apparent in this study if expiratory scans had
been taken.
In the study by Goldin and coworkers (15), the frequency distributions of lung parenchymal attenuation values measured by 3-mm CT slices were assessed before and after methacholine challenge. They found that the distributions of attenuation values were shifted to the left during airway narrowing in both scans performed at FRC and at RV, i.e., lungs became less dense. At FRC, this may have been purely the result of hyperinflation. However, at RV, this decrease in attenuation and its patchy distribution supports the presence of gas trapping due to airway closure during ASM activation. Taken together, these studies suggest that in asthmatics hyperinflation and gas trapping can lower the average lung attenuation but that areas of pathologically low attenuation, such as caused by emphysema, do not occur.
Airway Narrowing during Asthma Attacks
There have only been three studies in which the site and magnitude of airway narrowing have been examined in asthmatics. Paganin and coworkers examined the sites of airway narrowing in 10 asthmatic subjects during exercise challenge using 1-mm thickness HRCT scans (37). They were able to compare airways before and after exercise challenge and bronchodilator inhalation by using anatomical landmarks. They found a significant reduction in airway lumen area in those patients who had a reduction in FEV1 of greater than 20%, but no significant changes in those in whom the reduction in FEV1 was only 10 to 15%. In addition to between-subject variability, they also found that the magnitude of airway narrowing varied within patients; some airways appeared to close whereas others dilated. However, in their image analysis they used variable window levels (4) and for reasons discussed previously, this would bias their results.
Okazawa and coworkers (9) measured the change in airway wall and lumen area in six asthmatics and six normals after methacholine challenge using 1.5-mm slice HRCT scans. Airway dimensions in asthmatics and normal control subjects were measured from HRCT data using a validated quantitative image analysis method (2). There were similar decreases in FEV1 after methacholine challenges in asthmatics (31%) and in normals (37%); the distribution of airway narrowing was similar in the two groups, greatest in 2- to 4-mm airways of approximately 40% in both asthmatics and normals. They also found that airway wall area decreased in the normals after methacholine challenge but did not change in asthmatics. The same group reported a decrease in the airway wall area after methacholine challenge in excised dog lungs, studied using identical methods (2). The potential significance of this finding to airway mechanics has been previously discussed.
Goldin and coworkers also measured changes in airway lumen area in normal and asthmatic subjects, before and after nebulized methacholine challenge (15), also using validated analysis methods. However, as previously mentioned, they did not standardize lung volumes during the pre- and postmethacholine lung scans. Because the mean decrease in FEV1 during the challenge was 26%, significant hyperinflation almost certainly occurred, partially dilating the airways. This would, theoretically, have led to an underestimation of the narrowing. Despite this, they reported a mean decrease of 95% of the lumen areas in the smallest airways that were visible on CT, which were 1.6 to 2.5 mm in diameter.
| |
FUTURE DIRECTIONS FOR THE USE OF HRCT IN AIRWAY RESEARCH |
|---|
|
TOP
INTRODUCTION
TECHNICAL CONSIDERATIONS
NORMAL STRUCTURE AND FUNCTION
DESCRIPTIVE STUDIES IN ASTHMA
FUTURE DIRECTIONS FOR THE...
SUMMARY
REFERENCES
|
|---|
From a methodological standpoint, there are still large improvements that can be made in quantitative image analysis techniques. Although these quantitative image analysis techniques will be primarily used for research purposes initially, they could be of great benefit in everyday clinical use of HRCT, especially if computer-based algorithms requiring minimal technician or radiologist input can be developed. For example, automated image analysis could determine the extent of emphysema and the amount of lung tissue loss within each lung region, with adjustment for age, predicted lung volume, and pleural pressure gradient. From the point of view of examining airway lumen area and wall thickness as well as bronchial/arterial diameter ratios, semi-automated computer algorithms could be developed to estimate these variables and compare them to a reference distribution from normal subjects. This could also allow comparisons of these measurements to clinical and physiological variables including flow rates, measures of bronchial responsiveness, and symptoms. Such an automated approach to quantification of airway measurements would improve on subjective interpretation.
The use of phantoms made of artificial materials to simulate airway tissue is unlikely to be representative of tissues in vivo. Phantom airways do not have mucosal folds nor do they contain any surface lining fluid or mucus and therefore they have a sharper air-soft tissue interface. Most of the airways examined in axial CT slices are also more likely to be running obliquely to the plane of the section, rather than perpendicularly owing to the anatomy of the lung. However, phantom "airways" are aligned accurately to be perpendicular to the plane of the imaging slice, which eliminates the possible effects of oblique imaging. There are limited data on the effects of an oblique airway on quantitative measurement (3). Although airway wall area is an important parameter in mechanical models of airway narrowing, few HRCT studies have examined this variable; in fact there are no published data on the accuracy of airway wall measurements using HRCT. If future quantitative HRCT studies are to be conducted on airways, HRCT-histologic correlation studies similar to those in emphysema are required.
The reproducibility of in vivo airway measurements by the different quantitative HRCT methods has also not been established. Reproducibility is important when assessing the heterogeneity of airway wall thickness and the response to agonist challenge in asthmatics and normals. An apparent heterogeneity in the distribution of airway narrowing could, in part, be due to variation in the measurement itself. Knowledge of the reproducibility of measurements would allow a proper statistical evaluation of heterogeneity. These studies are also important before HRCT can be used as a tool to examine the longitudinal effect of therapeutic interventions designed to affect airway lumenal or wall dimensions.
HRCT also offers great potential in the study of relatively acute changes in airway size. Although the relationship of airway size and lung volume has been examined in animals using tantalum bronchography or HRCT, there are few data on the changes in airway lumen area as a function of lung volume or transpulmonary pressure in human lungs in health or disease. In addition, the recent considerable interest in the effects of a deep inspiration on airway caliber can be examined relatively noninvasively using HRCT.
| |
SUMMARY |
|---|
|
TOP
INTRODUCTION
TECHNICAL CONSIDERATIONS
NORMAL STRUCTURE AND FUNCTION
DESCRIPTIVE STUDIES IN ASTHMA
FUTURE DIRECTIONS FOR THE...
SUMMARY
REFERENCES
|
|---|
Observational studies suggest a high prevalence of airway abnormalities in clinically stable asthmatics and complement similar findings in studies of the distribution of ventilation, airway closure, and gas exchange in asthmatic subjects. The findings of airway wall thickening support histopathological studies but observations of emphysema, bronchial dilatation, and emphysema are more difficult to interpret in view of the lack of correlation studies. This may be an area of future research.
| |
Footnotes |
|---|
Correspondence and requests for reprints should be addressed to Peter D. Paré, Pulmonary Research Laboratory, St. Paul's Hospital, Vancouver, BC, V6Z 1Y6 Canada.
(Received in original form May 19, 1998 and in revised form September 24, 1998).
G. King was supported by an MRC/CLA Fellowship and an Astra/MRC/PMAC Canada Fellowship.Acknowledgments: Supported by Medical Research of Canada.
| |
References |
|---|
|
TOP
INTRODUCTION
TECHNICAL CONSIDERATIONS
NORMAL STRUCTURE AND FUNCTION
DESCRIPTIVE STUDIES IN ASTHMA
FUTURE DIRECTIONS FOR THE...
SUMMARY
REFERENCES
|
|---|
1.
Mayo, J. R.,
W. R. Webb,
R. Gould,
M. G. Stein,
I. Bass,
G. Gamsu, and
H. I. Goldberg.
1987.
High-resolution CT of the lungs: an optimal approach.
Radiology
163:
507-510
2.
McNamara, A. E.,
N. L. Müller,
M. Okazawa,
J. Arntorp,
B. R. Wiggs, and
P. D. Paré.
1992.
Airway narrowing in excised canine lungs measured by high-resolution computed tomography.
J. Appl. Physiol
73:
307-316
3. Webb, W. R., G. Gamsu, S. D. Wall, C. E. Cann, and E. Proctor. 1984. CT of a bronchial phantom: factors affecting appearance and size measurements. Invest. Radiol 19: 394-398 [Medline].
4. Senéterre, E., F. Paganin, J. M. Bruel, F. P. Michel, and J. Bousquet. 1994. Measurement of the internal size of bronchi using high resolution computed tomography (HRCT). Eur. Respir. J. 7: 596-600 [Abstract].
5.
Bankier, A. A.,
D. Fleischmann,
R. Mallek,
A. Windisch,
F. W. Winkelbauer,
M. Kontrus,
L. Havelec,
C. J. Herold, and
P. Hubsch.
1996.
Bronchial wall thickness: appropriate window settings for thin-section
CT and radiologic-anatomic correlation.
Radiology
199:
831-836
6.
Amirav, I.,
S. Kramer,
M. Grunstein, and
E. Hoffman.
1993.
Assessment
of methacholine-induced airway constriction with ultrafast high-resolution computed tomography.
J. Appl. Physiol
75:
2239-2250
7.
Herold, C.,
R. Brown,
W. Mitzner,
J. Links,
C. Hirshman, and
E. Zerhouni.
1991.
Assessment of pulmonary airway reactivity with high-resolution CT.
Radiology
181:
369-374
8. Brown, R., and W. Mitzner. 1996. Effect of lung inflation and airway muscle tone on airway diameter in vivo. J. Appl. Physiol 79: 1581-1588 .
9. Okazawa, M., N. L. Müller, A. E. McNamara, S. Child, L. Verburgt, and P. D. Paré. 1996. Human airway narrowing measured using high resolution computed tomography. Am. J. Respir. Crit. Care Med 154: 1557-1562 [Abstract].
10. Ferretti, G., D. Vining, J. Knoplioch, and M. Coulomb. 1996. Tracheobronchial tree: three-dimensional spiral CT with bronchoscopic perspective. J. Comput. Assist. Tomogr. 20: 777-781 [Medline].
11.
Wood, S. A.,
E. A. Zerhouni,
J. D. Hoford,
E. A. Hoffman, and
W. Mitzner.
1995.
Measurement of three-dimensional lung tree structures
by using computed tomography.
J. Appl. Physiol
79:
1687-1697
12. McNitt-Gray, M. F., J. G. Goldin, T. D. Johnson, D. P. Tashkin, and D. R. Aberle. 1997. Development and testing of image-processing methods for the quantitative assessment of airway hyperresponsiveness from high-resolution CT images. J. Comput. Assist. Tomogr 21: 939-947 [Medline].
13.
Brown, R.,
C. Herold,
C. Hirshman,
E. Zerhouni, and
W. Mitzner.
1993.
Individual airway constrictor response heterogeneity to histamine assessed by high-resolution computed tomography.
J. Appl. Physiol.
74:
2615-2620
14.
Sasaki, F.,
L. Saitoh,
L. Verburgt, and
M. Okazawa.
1996.
Airway wall
dimensions during carbachol-induced bronchoconstriction in rabbits.
J. Appl. Physiol
81:
1578-1583
15.
Goldin, J. G.,
M. F. McNitt-Gray,
S. M. Sorenson,
T. D. Johnson,
B. Dauphinee,
E. C. Kleerup,
D. P. Tashkin, and
D. R. Aberle.
1998.
Airway hyperreactivity: assessment with helical thin-section CT.
Radiology
208:
321-329
16.
Park, C. S.,
N. L. Müller,
S. A. Worthy,
J. S. Kim,
N. Awadh, and
M. Fitzgerald.
1997.
Airway obstruction in asthmatic and healthy individuals: inspiratory and expiratory thin-section CT findings.
Radiology
203:
361-367
17. Grenier, P., I. Mourey-Gerosa, K. Benali, M. Brauner, A. Leung, S. Lenoir, M. Cordeau, and B. Mazoyer. 1996. Abnormalities of the airways and lung parenchyma in asthmatics: CT observations in 50 patients and inter- and intraobserver variability. Eur. Radiol 6: 199-206 [Medline].
18.
Lynch, D. A.,
J. D. Newell,
B. A. Tschomper,
T. M. Clink,
L. S. Newman, and
R. Bethel.
1993.
Uncomplicated asthma in adults: comparison of CT appearance of the lungs in asthmatic and healthy subjects.
Radiology
188:
829-833
19.
Angus, R.,
M.-I. Davies,
M. Cowan,
C. McSharry, and
N. Thomson.
1994.
Computed tomographic scanning of the lung in patients with allergic bronchopulmonary aspergillosis and in asthmatic patients with a
positive skin test to Aspergillus fumigatus.
Thorax
49:
586-589
20. Paganin, F., V. Trussard, E. Senterre, P. Chanez, J. Giron, P. Godard, J. Senac, F. B. Michel, and J. Bousquet. 1992. Chest radiography and high resolution computed tomography of the lungs in asthma. Am. Rev. Respir. Dis 146: 1084-1087 [Medline].
21. Paganin, F., E. Séneterre, P. Chanez, J. P. Daurés, J. M. Bruel, F. B. Michel, and J. Bousquet. 1996. Computed tomography of the lungs in asthma: influence of disease severity and etiology. Am. J. Respir. Crit. Care Med. 153: 110-114 [Abstract].
22. Boulet, L.-P., M. Bélanger, and G. Carrier. 1995. Airway responsiveness and bronchial-wall thickness in asthma with or without fixed airflow obstruction. Am. J. Respir. Crit. Care Med 152: 865-871 [Abstract].
23.
Awadh, N.,
N. L. Müller,
C. S. Park,
R. T. Abboud, and
J. M. Fitzgerald.
1998.
Airway wall thickness in patients with near fatal asthma and
control groups: assessment with high resolution computed tomographic scanning.
Thorax
53:
248-253
24.
Berend, N.,
A. J. Woolcock, and
G. E. Marlin.
1979.
Relationship between bronchial and arterial diameters in normal human lungs.
Thorax
34:
354-358
25. Kim, S., J. Im, I. Kim, S. Cho, S. Cha, K. Park, and D. Kim. 1995. Normal bronchial and pulmonary arterial diameters measured by thin section CT. J. Comput. Assist. Tomogr 19: 365-369 [Medline].
26. Neeld, D., L. Goodman, J. Gurney, P. Greenberger, and J. Fink. 1990. Computerized tomography in the evaluation of allergic bronchopulmonary aspergillosis. Am. Rev. Respir. Dis 142: 1200-1205 [Medline].
27. Naidich, D., D. McCauley, N. Khouri, F. Stitik, and S. Siegelman. 1982. Computed tomography of bronchiectasis. J. Comp. Assist. Tomogr 6: 437-444 [Medline].
28. Kim, J. S., N. L. Müller, C. S. Park, D. A. Lynch, L. S. Newman, P. Grenier, and C. J. Herold. 1997. Broncho-arterial ratio on thin-section CT: comparison between high altitude and sea level. J. Comput. Assist. Tomogr 21: 306-311 [Medline].
29.
Herold, C.,
R. Wetzel,
J. Robotham,
S. Herold, and
E. Zerhouni.
1992.
Acute effects of increased intravascular volume and hypoxia on the
pulmonary circulation: assessment with high-resolution CT.
Radiology
183:
655-662
30.
Muller, N. L.,
C. A. Staples,
R. R. Miller, and
R. T. Abboud.
1988.
"Density mask": an objective method to quantitate emphysema using computed tomography.
Chest
94:
782-787
31. Geve