| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Thoracic Imaging |
1 From the Department of Diagnostic Radiology, Fujisawa City Hospital, Fujisawa City, Kanagawa, Japan (S.M., H.A., K.K.); and Department of Radiology, St Marianna University School of Medicine, Kawasaki City, Kanagawa, Japan (Y.K., Y.N., H.N.). Received September 11, 2003; revision requested November 24; final revision received March 23, 2004; accepted April 15. Address correspondence to S.M., Department of Radiology, St Marianna University School of Medicine, 2–16-1 Sugao, Miyamae-Ku, Kawasaki City, Kanagawa 216-8511, Japan (e-mail: shinma@d9.dion.ne.jp).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Institutional review board did not require its approval or informed patient consent. Airway dimensions were analyzed in 52 patients (30 men and 22 women) without known cardiopulmonary disease. Contiguous 2-mm CT sections were obtained after reconstruction, extending from origin of right posterior basal segmental bronchi to posterior subsegmental bronchi. Following parameters were determined with semiautomatic image-processing program: luminal area (LA), total airway area (TA), short axis of lumen (LSD), and short axis of total airway (TSD). In airways in which adjacent vessel or branching of small bronchus abutted boundary of airway, extrapolated line was traced by one radiologist. Airway wall area (WA) was calculated as TA – LA, and WT was calculated as (TSD – LSD)/2. Relative WA (WA% = [WA/TA] · 100) and ratio of airway WT to total diameter (D) (WT/D = WT/TSD) were calculated. Linear regression analysis and Spearman rank correlation were used to evaluate relationship between airway parameters (LA, WA%, and WT/D ratio) and distance from origin of segmental bronchi.
RESULTS: LA decreased as CT proceeded from hilum to periphery (r = –0.765, P < .001). In 308 (32.7%) of 943 bronchi, however, LA increased as CT proceeded from hilum to periphery. LA increased by 10% or more in 101 (10.7%) of 943 bronchi. Mean changes in WA% and WT/D ratio between two contiguous sections were 0.66 ± 5.05 (standard deviation) and 0.003 ± 0.024, respectively. WA% changed by more than 5% between two contiguous sections in 274 (29.0%) of 943 bronchi. WT/D ratio changed by more than 0.02 between two contiguous sections in 338 (35.8%) of 943 bronchi.
CONCLUSION: Variation of airway lumen and WT is found in asymptomatic subjects without known cardiopulmonary disease.
© RSNA, 2004
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Anatomically, lobar and segmental airways have mean diameters of 5–8 mm, and subsegmental airways have mean diameters of 1.5–3 mm. Bronchial WT is approximately proportional to bronchial diameter (13). For quantitative analysis of airway WT, researchers have used the ratio of airway WT to outer diameter (1,6,8,11), defined as wall area (WA) divided by total airway area, multiplied by 100 (3,8,9,11). The percentage WA reportedly has a negative correlation with bronchial diameter (10,14). To our knowledge, however, serial changes in these parameters on transverse CT sections of a single airway in subjects without cardiopulmonary disease have not been assessed previously.
In studies of airway size, heterogeneity in the distribution of agonist-induced bronchoconstriction has been observed (1,2,14–18). In a longitudinal study, individual airways within a subject varied in size by as much as twofold over time (19). The ability to make accurate measurements of such changes in airway dimensions longitudinally or after intervention is dependent on the ability to match CT sections of the same airways at the same levels. Reproducibility of airway measurements is important when assessing heterogeneity of airway dimensions. In addition, reproducibility of airway measurements and knowledge of serial changes in airway caliber in subjects without cardiopulmonary disease would allow accurate statistical analysis of airway heterogeneity. To date, however, there have been no studies of the variation of single-airway caliber in subjects without cardiopulmonary disease, to our knowledge.
The recent introduction of multi–detector row spiral CT offers a further increase in performance, particularly the ability to scan larger anatomic volumes with high spatial resolution. Serial thin sections of the entire thorax can be obtained during a single suspended breath hold. Thus, the purpose of this study was to retrospectively analyze serial changes in the airway lumen and WT at multi–detector CT in asymptomatic subjects.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Simultaneously, the bronchial morphologic abnormality was also evaluated visually. In these 102 patients without obvious pulmonary abnormality, there was no patient with an internal bronchial diameter greater than that of the adjacent pulmonary artery. Fifty of the 102 patients included in this study initially were excluded because of (a) history of cardiopulmonary disease, (b) clinical pulmonary symptoms at the time the study was conducted, (c) current or previous smoking habit, and (d) age older than 61 years. Age was limited to reduce the effects of aging on bronchial measurement (20).
The final study population comprised 52 patients (30 men and 22 women; mean age, 46.7 years ± 13.2 (standard deviation); age range, 22–60 years; men had a mean age of 45.4 years ± 13.0 and an age range of 22–60 years; women had a mean age of 48.4 years ± 13.4 and an age range of 23–60 years). According to results of the t test, no significant difference in age was found between men and women (P = .43).
Indications for CT were (a) assessment of lung nodules (n = 22), including suspected malignant lesions and pulmonary vascular malformations on the basis of chest radiographs, but the node could not be recognized on CT images; and (b) screening for lung metastasis (n = 30) in patients known to have an extrapulmonary malignant lesion. Twenty-eight patients underwent CT for assessment of metastasis before surgery or chemotherapy. Two patients underwent radiation therapy for extrathoracic malignancy. No patients were treated with chemotherapy before CT examination.
Multi–Detector Row Spiral CT Protocol
CT was performed by using a multi–detector row spiral CT scanner (Aquillion; Toshiba, Tokyo, Japan) with four detector arrays. Patients were scanned in the supine position within one breath hold at deep inspiration. The scans were obtained with 4 x 2-mm collimation (four detectors with 2-mm section thickness), with a table feed of 11 mm per 0.5-second scanner rotation (ie, pitch of 5.5). Scanning was performed at 120 kV and 125 mAs, regardless of patient size, by using a 512 x 512 matrix. Images were viewed at settings appropriate for lung parenchyma (window width, 1500 HU; window level, –450 HU) (15,21), and images containing the right posterior basal bronchus were selected by two radiologists (S.M., K.K.) in consensus. These sites were chosen because they are more convenient for obtaining a cross-sectional view of the bronchus. Only the right posterior basal bronchus was evaluated, because transmitted cardiac motion artifacts may obscure detail in the left lower lobe (22).
Although the apical bronchus is also convenient for obtaining a cross-sectional view of the bronchus, we required many contiguous CT sections to analyze serial changes in the airway lumen and WT. Therefore, we chose the posterior basal bronchus, of which many more contiguous CT sections could be obtained. With the bronchus identified, the raw data were reconstructed retrospectively with a section thickness of 2 mm and a field of view of 9 cm by using a lung algorithm (FC53; Toshiba). Fourteen to 27 contiguous sections with a thickness of 2 mm were obtained, extending from the origin of the posterior segmental bronchi or common trunk of the lateral and posterior segment to the posterior subsegmental bronchi. The end point of measuring subsegmental bronchi was where the luminal area became less than 4 mm2, as dictated by results of validation analysis.
Validation of Airway Analysis
We performed validation analysis by using a phantom to determine the most appropriate window setting and to test analysis software. The phantom consisted of a polystyrene foam block and five plastic circular cylinders, which represented the lung parenchyma and airways, respectively. The actual size of the cylinders was measured with an optical micrometer caliper to the nearest 0.01 mm. Cylinder WT ranged from 0.69 to 1.15 mm, and luminal area ranged from 2.5 to 51.5 mm2. The actual percentage WAs of the phantoms were calculated from the actual luminal area and WT and ranged from 39.5% to 75.3%.
The percentage error, as measured by the difference between nominal and measured area, varied according to the window level. Percentage differences of luminal area and percentage WA were smallest at a window level of –500 HU in all plastic right circular cylinders between –350 HU and –700 HU; therefore, this window level was used in all measurements. The percentage errors of luminal area and percentage WA were greatest for the smallest cylinder (2.5 mm2) (2.1% and –2.3%, respectively). The results showed that our system of measurement was accurate within a luminal area range of 4–54 mm2 (percentage error ranged from –0.1% to 1.0% in luminal area and –0.7% to 1.1% in percentage WA). Therefore, bronchi with a luminal area less than 4 mm2 were excluded.
Airway Analysis
For analysis of airway dimensions, window width was set to the minimum value of 2 HU to obtain the thresholding image (23). A window level of –500 HU was chosen, as dictated by results of validation analysis. Bronchial lumen was displayed in black, and bronchial wall was white (Fig 1). Images were transferred to a personal computer for quantitative analysis, and measurement of airway dimensions was performed by using a semiautomatic image-processing program (NIH Image version 1.62, developed at the U.S. National Institutes of Health, Bethesda, Md), which calculates cross-sectional areas of bronchus and lengths of the major and minor axes of the best-fitting ellipse. By using the outlining tool, bronchial lumen segmented by means of thresholding was measured automatically, and luminal area (LA) and short and long axes of the lumen were determined.
|
|
|
To evaluate oblique airway orientation, the long-to-short axis ratio was calculated by dividing the long axis of the lumen by the short axis of the lumen. Airways with an obliquity of greater than 1.5 were excluded from the analysis of percentage WA derived from area measurements because of potential inaccuracies in measurement caused by volume averaging of the oblique sections (15).
Airway WA was calculated as total airway area minus LA, and airway WT was calculated as short axis of the total airway minus short axis of the lumen, divided by 2. We used two previously reported indexes: percentage WA (relative WA [WA%] = WA divided by total airway area, times 100) (3,8,9,11) and ratio of airway WT to total diameter (D) (WT/D, calculated by dividing WT by short axis of total airway) (1,6,8,11).
To compensate for individual airway size, those parameters were also expressed as a percentage of the most proximal portion of the segmental bronchi. To analyze variability in airway lumen and bronchial wall, the percentage change in LA between contiguous sections was calculated as (LAn – LAn–1)/(LAn–1 · 100), where n is the section number. Changes in WA% and WT/D ratio between two contiguous sections were also calculated by subtracting these parameter values of the first contiguous section and the next contiguous section. Percentage change in LA and change in WA% and WT/D ratio were also evaluated in each subject. Furthermore, changes in LA between contiguous sections were assessed visually by two radiologists (S.M., K.K.) in consensus.
To determine how percentage change in LA and change in WA% and WT/D ratio varied with airway size, airways were divided into three size categories: small (luminal area < 10 mm2), medium (luminal area, 10–20 mm2), and large (luminal area > 20 mm2). In addition, since the relationship between airway dimension and age has been reported (20), we divided our subjects into two groups of subjects—those 40 years old and older versus those younger than 40 years—to evaluate the relationship between age, percentage change in LA, and changes in WA% and WT/D ratio.
Reproducibility of Airway Dimensions
All airway dimensions were measured by one observer (S.M.). Intraobserver error was tested by having this observer measure LA, WA%, and WT/D ratio in 10 randomly selected subjects two times, separated by an interval of 2 weeks. Interobserver error was determined by having two observers (S.M., K.K.) measure LA, WA%, and WT/D ratio in 10 randomly selected subjects. Analysis of intra- and interobserver reproducibility was conducted by using Bland-Altman analysis (24).
Statistical Analysis
All statistical analyses were performed by using Stat View version 5.0 software (SAS Institute, Cary, NC). Data were expressed as mean ± standard deviation. Linear regression analysis and Spearman rank correlation analysis were used to evaluate the relationship between airway parameters (LA, percentage LA, WA%, percentage WA%, WT/D ratio, and percentage WT/D ratio) and distance from the origin of segmental bronchi. Comparison of these parameters among the three groups classified according to airway size (airway luminal area) was performed by using one-way analysis of variance, with multiple comparisons determined with the Scheffé test. Correlation of the long-to-short axis ratio with percentage change in LA and changes in WA% and WT/D ratio was also assessed with linear regression analysis. Comparison of these parameters between the two groups classified according to age was performed by using the Mann-Whitney U test. A P value of less than .05 was considered to indicate a statistically significant difference.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
|
|
|
In three sections, two contiguous sections with an obliquity grater than 1.5 were seen. Thus, those parameters could not be calculated—in total, 13 sections. Finally, percentage change in LA and change in WA% and WT/D ratio were obtained in 943 sections. The results of parameter measurements are shown in Table 2. LA and percentage LA decreased as CT proceeded from hilum to periphery (r = –0.765 and P < .001 versus r = –0.851 and P < .001) (Fig 3). The mean percentage change in LA between two contiguous sections was –6.8 ± 15.6 (range, –67.3 to 48.2). Of the 943 bronchi used for analysis, 308 (32.7%) showed increased LA as CT sections proceeded from hilum to periphery. Furthermore, in 101 (10.7%) of the 943 bronchi, LA increased by 10% or more. The frequency distribution of percentage change in LA is shown in Figure 4. Forty-two (80.8%) of 52 subjects had bronchi in which LA increased more than 10% on the peripheral side. In the visual evaluation, the airway lumen dilated on the peripheral side in 95 (9.4%) of 1008 bronchi.
|
|
|
|
|
|
|
|
|
|
The percentage change in LA and changes in WA% and WT/D ratio in airways of different sizes are summarized in Table 3. The percentage change in LA was significantly greater in small airways (<10 mm2) than in larger airways. The changes in WA% and WT/D ratio were also significantly greater in small airways than in large airways. There was no significant difference in percentage change in LA and changes in WA% and WT/D ratio between medium and large airways. Standard deviations of percentage change in LA and changes in WA% and WT/D ratio were slightly larger in small airways than in larger airways. There was no significant difference between the two age groups of 40-year-old subjects and older versus subjects younger than 40 years and percentage change in LA and changes in WA% and WT/D ratio (P = .77, .62, and .66, respectively).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Although bronchial WT is approximately proportional to bronchial diameter (13), the change in WA% between two contiguous sections within the same bronchus was greater than 5% in 29.0% of bronchi, and the change in WT/D ratio between two contiguous sections within the same bronchus was greater than 0.02 in 35.8% of bronchi. There are several factors that may explain this variation of airway caliber and WA% within individual bronchi.
Many factors affect the accuracy of airway dimension measurements on CT images. The window level has been shown to be important in quantitative studies of airway lumen and wall dimensions (15,21,23,25). We used a single threshold (–500 HU) to measure airway dimensions. McNitt-Gray et al (4) reported that a threshold value of –500 HU yielded the most accurate measurements of the airway lumen in a bronchial phantom. In our validation phantom analysis, this threshold value was found to be suitable for measurement of the bronchi. For small bronchi, however, volume averaging at the air-bronchial wall interface results in pixels with attenuation that is less than the chosen threshold (4).
In the present study, the percentage change in LA and changes in WA% and WT/D ratio were significantly greater in small airways than in larger airways. The measurement variability of airway luminal area is reportedly greater for smaller airways (17). There was no obvious correlation between measurement error and airway size in the present reproducibility analysis. Thus, the additional manual editing we performed had little influence on variation of airway luminal area and WT.
Many studies have shown that oblique airway orientation affects measurement of airway lumen and WT (21,26,27). Measurement of airway dimensions on CT images has previously been restricted to airways that appear to have been cut in cross-section, on the basis of the apparent roundness of the airway lumen. Measurement of airway lumen and WA when they are not perpendicular to the scanning plane leads to marked errors, the magnitude of which depends on the acuteness of the angle (26,27).
Webb et al (21) studied an airway phantom in which the airways were oriented at various angles. They found that airway angles interact to produce error in measurement of airway lumen. King et al (26) introduced an automated CT image analysis algorithm that was not affected by oblique airway orientation. They reported that the manual method led to underestimation of airway lumen area and overestimation of airway WA in direct relation to airway size and airway angle of orientation. Although airways with a long-to-short axis ratio of up to 1.5 were not included in the present study and the mean ratio was 1.22, airway orientation seemed to affect measurement of the airway lumen and WA to some extent. However, there was no significant correlation between long-to-short axis ratio and percentage change in LA or changes in WA% or WT/D ratio. Therefore, the influence of airway orientation on variation of airway dimensions is apparently limited.
The bronchial wall consists of epithelium, smooth muscle, interstitial connective tissue, and cartilage. The proportions of these elements vary at different levels of the bronchus. Therefore, variation within individual airway lumens and WA% may be based on variation of bronchial morphology. Carroll et al (28) assessed variability of intra-airway structure by using an inflation-fixed lung and found that intra-airway variability for measurements of airway dimensions was low. However, they used three transverse sections, only 20 µm apart; thus, the result was not representative of the entire bronchial tree.
An interesting feature of published thin-section CT scan data has been the extremely heterogeneous constrictor responses of individual airways to both histamine and methacholine challenge in dogs and humans (1,2,14–18). Although mechanisms for this airway response variability are not well understood, the present results indicate that variation of airway lumen and WA is present in asymptomatic subjects without constricted bronchi. Thus, quantitative evaluation of the degree of heterogeneous constrictor responses should include consideration of variation of airway lumen.
The ability to measure airway lumen and WA accurately before and after intervention or with different physiologic conditions, such as a bronchoconstricting agonist, is dependent on the ability to match CT scanning level. It is insufficient to simply use the same CT section because very small changes in lung volume can alter the relationship between the section number and lung anatomy. Desai et al (29) investigated reproducibility of measurements of bronchial circumference and luminal area after repositioning the scan. In their study, changes in luminal area after repositioning were not small. Therefore, variation in an individual bronchus may have additional effects on quantitative evaluation of slight changes in CT sections acquired before and after intervention.
Moreover, it is possible that airways measured by using only a few CT sections are not representative of most conducting airways. Wood et al (30) developed image analysis techniques for accurate quantitative measurement of airway wall and lumen areas from spiral CT data. Their analysis technique overcomes the major limitation in the use of thin-section CT in quantitative analysis, which is that accurate measurement of luminal area and airway WA can be obtained by using airways oriented approximately perpendicular to the scanning plane. This method represents a substantial improvement in quantitative analysis of entire airways. Nevertheless, most quantitative analysis of airway dimensions has been based on identification of cross-sections of airways that appear to be round.
Although Brown et al (19) observed variability in the size of individual airways in a longitudinal study, they also found that the composite average size of all airways was unchanged over time. Furthermore, Nakano et al (8) analyzed the airway wall and luminal dimensions of the right upper lobe apical segmental bronchus to examine the relationship with clinical indexes in chronic obstructive pulmonary disease. They found a highly significant relationship between WA% in the right apical bronchus and average WA% in all other airways that were imaged in cross-section. These findings suggest that the extent of individual airway variability is disregarded when mean values of individual airways are used.
In many previous studies, airway analysis was performed by using mean values of bronchial measurements, but structure or reaction of local bronchi could not be assessed. More detailed and accurate analysis is required to evaluate the local bronchial reaction. In future investigations, we plan to assess approaches to dealing with the bronchial variation observed in the present study.
We assessed only one segmental bronchus; thus, results of our current study might not be representative of all segment bronchi. However, some researchers have reported that WT/D ratio or the bronchial lumen ratio revealed no statistically significant differences between segments, lobes, and lungs (20,31). Therefore, we speculated that the variation of airway lumen and WT may exist, even in other bronchi. Further investigations will be required. The posterior basal bronchus of the lower lobes that we used may be frequently involved in minor infection or subclinical aspiration. These subclinical airway diseases may affect our results.
The present study has several limitations. First, our criteria for choosing asymptomatic subjects may not have been sufficiently strict because we did not perform pulmonary function tests in our subjects. Thus, subjects with subclinical airway disease may have been included in this study erroneously. There is reportedly a correlation between pulmonary function testing and bronchial WA or airway lumen (3,5,6,8,9), and inclusion of subjects with subclinical airway disease could have affected our results. In addition, subjects might have a history of pulmonary disease such as an infectious disease, allergic history, and work history that may cause pulmonary disease but were not recorded.
Second, the CT scans we obtained were not gated to spirometry, and we did not investigate the effect of the lung volume on airway dimensions; however, lung volume influences airway wall and lumen dimensions in a predictable manner (32). We instructed subjects to take one breath and hold it at deep inspiration during scanning, but the degree of inspiration may have differed slightly between subjects. Moreover, in healthy subjects, deep inspiration affects airway dimensions (33,34). Respiratory motion affects image quality, as well as cardiac motion (22), and it might affect our results.
Third, collimation affects the quality of the CT scan and the accuracy of airway wall dimension measurements (21). In this study, the scans were obtained with 4 x 2-mm collimation because data were obtained in routine clinical practice. Measured airways were not completely perpendicular to the scanning plane. Thus, the airway WA might have been overestimated in comparison to that which might have resulted from use of a thinner collimation, such as 1 mm. In addition, a helical pitch of 5.5 was used in our study. The helical pitch also affects image quality of multi–detector row spiral CT (35). Although there is no report on the relationship between bronchial measurement and helical pitch, our results might be affected.
In conclusion, we observed variation of airway lumen and WA on cross-sectional CT images in asymptomatic subjects. Although thin-section CT has been used with increasing frequency to study airway dimensions in both normal and diseased lungs, the present results suggest that variation of airway lumen and WA, even in asymptomatic subjects without known cardiopulmonary disease, should not be disregarded in quantitative analysis of airway dimensions.
|
FOOTNOTES |
|---|
Authors stated no financial relationship to disclose.
Author contributions: Guarantor of integrity of entire study, S.M.; study concepts and design, S.M.; literature research, S.M.; clinical studies, S.M., K.K., H.A.; data acquisition, S.M., K.K.; data analysis/interpretation, S.M., K.K., Y.K., H.N.; statistical analysis, S.M., Y.K.; manuscript preparation, S.M., Y.K.; manuscript definition of intellectual content, all authors; manuscript editing, S.M., Y.K., Y.N.; manuscript revision/review and final version approval, all authors
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  M. Cazzola, W. MacNee, F. J. Martinez, K. F. Rabe, L. G. Franciosi, P. J. Barnes, V. Brusasco, P. S. Burge, P. M. A. Calverley, B. R. Celli, et al. Outcomes for COPD pharmacological trials: from lung function to biomarkers Eur. Respir. J., February 1, 2008; 31(2): 416 - 469. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Hasegawa, Y. Nasuhara, Y. Onodera, H. Makita, K. Nagai, S. Fuke, Y. Ito, T. Betsuyaku, and M. Nishimura Airflow Limitation and Airway Dimensions in Chronic Obstructive Pulmonary Disease Am. J. Respir. Crit. Care Med., June 15, 2006; 173(12): 1309 - 1315. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. A. de Jong, N. L. Muller, P. D. Pare, and H. O. Coxson Computed tomographic imaging of the airways: relationship to structure and function Eur. Respir. J., July 1, 2005; 26(1): 140 - 152. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
