Effect of threshold on the correlation between airflow obstruction and low attenuation volume in smokers assessed by inspiratory and expiratory MDCT

Abstract

Background

The estimation of emphysematous changes is very sensitive to computed tomography (CT) threshold level. In clinical practice, the predetermined threshold is usually set at −950 Hounsfield units (HU) for the detection of low attenuation volume (LAV). However, threshold levels that are tightly connected to pulmonary function abnormalities have not been determined.

Purpose

To determine the threshold level for calculating an LAV that closely reflects airflow limitation in patients with chronic obstructive pulmonary disease (COPD).

Material and Methods

Seventy-six consecutive non-COPD smokers and COPD patients underwent paired inspiratory and expiratory multidetector CT (MDCT). LAV% was segmented every 10 HU between −1000 and −750 HU to examine the correlation between LAV% and indexes of obstructive impairment.

Results

LAV% gradually increased as the threshold level increased on both inspiratory and expiratory images. LAV% on inspiratory images was higher than that on expiratory images at all threshold levels between −1000 and −750 HU. The threshold level that correlated with obstructive impairment differed between the two images: −930 HU on inspiratory and −870 or −880 HU on expiratory images.

Conclusion

LAV% dramatically changed according to the threshold level on both inspiratory and expiratory images, indicating that LAV% is dependent on the attenuation threshold level in patients with COPD. The threshold linking LAV% to airflow limitation was higher on expiratory than on inspiratory images.

Keywords

Chronic obstructive pulmonary disease multidetector computed tomography low attenuation volume threshold

Introduction

Chronic obstructive pulmonary disease (COPD) is characterized by airflow limitation that is not fully reversible (1). Inflammation in COPD may start in the small airways (2,3) and progress first to lung parenchyma with alveolar destruction (lung emphysema) and then to the central airways, resulting in bronchiolitis (airway diseases) (1). Pulmonary function tests (PFTs) are the gold standard for the diagnosis and staging of COPD. However, recent advances in multidetector computed tomography (MDCT) have made it possible to detect small units of lung emphysema at the level of subsegmental bronchi (4). Quantitative CT indices have been associated, at least in part, with lung function abnormalities in COPD patients. Densitometric parameters, including low attenuation area (LAA) and low attenuation volume (LAV), correlate with airflow limitation (5). Furthermore, LAA and LAV on expiratory CT images are reportedly stronger predictors of lung function than those on inspiratory images (6).

However, the threshold level for calculating an LAV that closely reflects airflow limitation has not been determined for either inspiratory or expiratory MDCT images. Reproducibility in estimating the magnitude of emphysematous changes can be very sensitive to differences in CT devices, techniques, and reconstruction algorithms (7). Some studies have found that the optimal threshold level for detecting LAV is −950 Hounsfield units (HU) or lower (8). Because these threshold levels were derived from microscopic or macroscopic studies, threshold levels that more closely reflect pulmonary function abnormalities have not been determined, especially in patients with mild to moderate COPD or smokers who have emphysematous changes but do not have airflow limitation. Therefore, the aim of this study was to determine LAV% in patients with COPD using paired inspiratory and expiratory MDCT images and to examine the effects of threshold levels on the relationship between LAV% and airflow limitation.

Material and Methods

Subjects

From July 2009 to November 2011, 76 consecutive patients diagnosed with, or suspected of having, clinically stable COPD underwent PFTs the same day that MDCT was performed at Chiba University Hospital. COPD was diagnosed on the basis of past history, physical examination and spirometric data according to the guidelines of the Global Initiative for Chronic Obstructive Lung Disease (GOLD) (1). Patients with COPD were classified according to severity as GOLD stages 1 to 4. All patients had a history of smoking. Exclusion criteria included: (i) obvious abnormal lung parenchymal lesions, other than emphysema; (ii) other potentially confounding abnormalities including pneumothorax, pleural effusion, cardiac failure, or postoperative status; or (iii) excessive image noise preventing image analysis. This study was approved by the Institutional Review Board (# 857) of Chiba University Hospital, and informed consent was obtained from each participant.

MDCT scanning

CT images were obtained using a 64-multi-detector CT (Aquilion-ONE; Toshiba Medical, Tokyo, Japan). MDCT images were obtained during a breath-hold in both deep inspiration and deep expiration with the patient in the supine position. None of the patients were given contrast medium. Before scanning, each patient was carefully instructed on how to breathe during scanning. CT parameters in this study were as follows: collimation, 120 kV; 200 mA (CT-AEC); gantry rotation time, 0.5 s; and beam pitch, 0.83. All images were reconstructed using a standard reconstruction algorithm with a slice thickness of 0.5 mm and a reconstruction interval of 0.5 mm. The voxel size was 0.63 × 0.63 × 0.5 mm.

Imaging analysis

Reconstruction images were transferred to a commercially available workstation (Aze, AZE Ltd., Tokyo, Japan). Whole lung volume was obtained between −1024 and −500 HU. A pulmonologist (NK, with 13 years' experience in interpreting thoracic CT) used minimal intervention to exclude the trachea and large bronchi near the hilum. For lung segmentation, lung volumes with attenuation values lower than thresholds ranging from −1000 to −750 HU were obtained on both inspiratory and expiratory images. The percentage of LAV to whole lung volume was defined as low attenuation volume percentage (LAV%). Fig. 1 shows an example of the wide variation in LAV% according to threshold level in a patient with COPD.

Fig. 1.

Coronal CT image obtained in a 75-year-old man with COPD. In each figure: light blue denotes areas with attenuation at each threshold level; black shows areas with attenuation of −500 with respect to each threshold level; and gray indicates vascular and other non-parenchymal structures. Low attenuation volume percentage (LAV%) dramatically changed depending on the threshold level on both inspiratory and expiratory images (upper panels: inspiratory images; lower panels: expiratory images).

Mean lung density (MLD) was also quantified in both scans. The ratio of MLD in expiration to MLD in inspiration (MLD ratio) was obtained by dividing the mean lung density at full expiration by that at full inspiration. The lowest fifth percentile of whole lung density in both scans was also measured.

A visual emphysema score was determined independently by three pulmonologists (NK, ST, TS) using the modified Goddard scoring system (9,10). Six images were analyzed from three slices in the lungs and an average score of all images was taken as the representative value of the severity of emphysema in each subject. Each image was classified as normal (score 0), <5% affected (score 0.5), <25% affected (score 1), <50% affected (score 2), <75% affected (score 3), or >75% affected (score 4). The three observers were blinded to baseline patient characteristics and the results of lung function tests. They repeated visual scoring 2 months after the first rating session to measure the inter- and intra-operator agreement of emphysema extent. Final evaluations were achieved by consensus (9). The relationship between these parameters and obstructive impairment was then assessed.

Pulmonary function tests

PFTs were performed on the same day as the MDCT scans. Spirometric measurements were performed according to the guidelines of the American Thoracic Society (11). Spirometric measurements were performed using Fudac-60 (Fukuda Denshi; Tokyo, Japan). Vital capacity, forced vital capacity (FVC), and forced expiratory volume in 1 s (FEV₁) were measured, and the percentage of FEV₁ predicted values (FEV₁%predicted) was calculated.

Statistical analysis

Correlations between PFTs and CT measurements were examined using Spearman rank correlation analysis, with data presented as means ± standard deviations. A comparison of ΔLAV% (100 × [expiratory LAV% – inspiratory LAV%]/inspiratory LAV%) between non-COPD smokers and patients with severe COPD (stages 3 and 4) were performed using the Mann-Whitney U test. Agreement in the inter- and intra-operator agreement of the visual score of emphysema extent was measured using the Bland-Altman method (12,13). A Kendall t test was performed to preliminarily assess correlation in both inter- and intra-operator agreement. The level of significance was set at P < 0.05 for all statistical analyses, which were performed using JMP 10.0 software (SAS Institute, Cary, NC, USA).

Results

Characteristics of the study subjects

Clinical characteristics including the results of the PFTs are shown in Table 1. The average FEV₁/FVC and FEV₁% predicted were 58.0% and 69.2%, respectively. Patients were classified into the following GOLD stages: normal, n = 18; stage 1, n = 19; stage 2, n = 19; stage 3, n = 15; and stage 4, n = 5. Seventy-four percent of the subjects were non-COPD smokers and COPD patients with mild/moderate degree of airflow limitation.

Table 1.

Patient characteristics and results of PFTs (n = 76).

Mean (SD)	Range
Age (years)	69 (9.0)	44–84
Smoking index (pack-year)	45.3 (22.9)	4–156
BMI	22.5 (3.5)	16.2–33.4
Stage normal/1/2 / 3/4 (Number)	18/19 /19/15 /5
FEV₁ (L)	1.81 (0.74)	0.5–3.3
FEV₁/FVC	58.0 (14.4)	26.5–78.6
FEV₁%pred (%)	69.2 (23.8)	25.2–124
V₅₀/V₂₅	3.34 (1.31)	0.8–7.0
%VC	96.3 (18.8)	49.1–141.3

SD; standard deviation.

LAV% on inspiratory and expiratory MDCT images

The CT indexes derived from inspiratory and expiratory images are summarized in Tables 2 and 3. LAV% gradually increased as the threshold level increased, both on the inspiratory and expiratory images. LAV% on inspiratory images was significantly higher than on expiratory images at all threshold levels between −1000 and −750 HU (Fig. 2). Table 2 shows LAV% according to threshold levels between −950 and −850 HU. The largest difference in LAV% between inspiratory and expiratory images (expiratory LAV% – inspiratory LAV%) was observed around −850 HU. To evaluate the change in LAV% from the inspiratory to the expiratory phase, we calculated ΔLAV% (100 × [expiratory LAV% − inspiratory LAV%]/inspiratory LAV%). ΔLAV% was significantly larger in non-COPD smokers compared to patients with severe COPD (stages 3 and 4) (Fig. 3).

Fig. 2.

Distribution of LAV% according to the attenuation threshold on inspiratory and expiratory images.

Fig. 3.

The change in LAV% from the inspiratory to the expiratory phase (ΔLAV%) was larger in non-COPD smokers compared to patients with severe COPD. The horizontal line is the median value, the box is the interquartile range, and the whiskers indicate the range, excluding outlying and extreme values (i.e. points with values >1.5 box lengths from the upper or lower limits of the box).

Table 2.

LAV% according to the threshold on inspiratory and expiratory images.

Threshold (HU)	Mean (SD)	Range
Inspiration (−950 HU)	7.6 (11.9)	0.0–48.2
Expiration (−950 HU)	5.2 (10.0)	0.0–46.3
Inspiration (−930 HU)	14.0 (15.7)	0.2–60.1
Expiration (−930 HU)	8.5 (13.4)	0.0–58.0
Inspiration (−910 HU)	25.5 (18.8)	1.0–69.1
Expiration (−910 HU)	13.1 (16.8)	0.0–66.8
Inspiration (−890 HU)	41.4 (20.2)	5.7–76.2
Expiration (−890 HU)	19.9 (20.3)	0.1–73.6
Inspiration (−870 HU)	57.0 (18.1)	17.2–82.4
Expiration (−870 HU)	28.8 (23.0)	0.4–79.1
Inspiration (−850 HU)	69.1 (14.2)	32.1–87.1
Expiration (−850 HU)	38.9 (24.0)	1.8–83.5

SD; standard deviation.

Table 3.

CT indexes derived from inspiratory and expiratory scans.

CT measurements	Mean (SD)	Range
Inspiratory MLD	−877.3 (23.7)	−933.9– −826.9
Expiratory MLD	−834.3 (43.6)	−930.5– −727.4
MLD ratio	0.95 (0.04)	0.85–1.0
Inspiration – 5th percentile	−942.1 (28.8)	−996.7– −892.0
Expiration – 5th percentile	−919.2 (42.8)	−997.5– −832.7
Visual score	1.1 (1.2)	0.0–4.0

MLD, mean lung density.

Correlations between LAV% and obstructive impairment

On both inspiratory and expiratory CT images, LAV% showed a negative correlation with FEV₁/FVC and FEV₁% spread at all threshold levels (Table 4). A closer correlation between LAV% and indexes of obstructive impairment was observed when the attenuation value was −930 HU on inspiratory CT images (Fig. 4). In contrast, on expiratory CT images, LAV% showed a close negative correlation with airflow limitation at a threshold attenuation value of −870 or −880 HU (Fig. 5), both of which were higher than the values found on the inspiratory CT images.

Fig. 4.

Inspiratory CT images showed a closer correlation to obstructive impairment when the attenuation value was −930 HU.

Fig. 5.

Expiratory CT images showed a closer correlation to obstructive impairment when the attenuation value was −870 HU.

Table 4.

Correlation between LAV% at different levels of threshold and obstructive impairment.

Threshold (HU)	FEV₁/FVC	FEV₁%pred
Inspiration (−950 HU)	−0.65	−0.57
Expiration (−950 HU)	−0.63	−0.57
Inspiration (−930 HU)	−0.69	−0.58
Expiration (−930 HU)	−0.68	−0.60
Inspiration (−910 HU)	−0.69	−0.55
Expiration (−910 HU)	−0.71	−0.62
Inspiration (−890 HU)	−0.64	−0.50
Expiration (−890 HU)	−0.73	−0.63
Inspiration (−870 HU)	−0.59	−0.45
Expiration (−870 HU)	−0.73	−0.64
Inspiration (−850 HU)	−0.54	−0.42
Expiration (−850 HU)	−0.71	−0.63

All correlations were statistically significant (P < 0.001).

Correlation between MLD, lowest fifth percentile, visual score, and obstructive impairment

Both inspiratory and expiratory MLD correlated with indexes of obstructive impairment (Table 5). Expiratory images correlated much more closely with indexes of obstructive impairment than did inspiratory images. A negative correlation was also observed between the MLD ratio and indexes of obstructive impairment. An inverse correlation was found between the lowest fifth percentile on both images and airflow limitation, and a negative correlation was observed between visual score and airflow limitation. Visual scoring also correlated significantly with the results of LAV% (r = 0.34 [P < 0.05] − 0.85 [P < 0.001]). Inter-operator agreement, which was evaluated with the Bland-Altman method, is shown in Fig. 6. The kappa coefficients of inter-operator agreement for visual scoring ranged between 0.69 and 0.83. The kappa coefficients of intra-operator agreement for visual scoring ranged between 0.95 and 0.98.

Fig. 6.

(a–c) Bland-Altman plot showing inter-observer variability of visual scoring for the extent of emphysema. The solid line represents the mean value of the differences in measurements between the two observers. The dashed lines represent the limits of agreement.

Table 5.

Correlation coefficients between CT measurements and obstructive impairment.

CT measurements	FEV₁/FVC	FEV₁ %pred
Inspiratory MLD	0.63	0.49
Expiratory MLD	0.74	0.57
MLD ratio	−0.62	−0.51
Inspiration – 5th percentile	0.63	0.53
Expiration – 5th percentile	0.61	0.56
Visual score	−0.52	−0.47

All statistic values are significant (P < 0.001).

Discussion

In this study, LAV% gradually increased as the threshold level increased between −1000 to −750 HU on both inspiratory and expiratory images (Figs. 1 and 2). LAV% was higher on inspiratory images at all thresholds examined. Inspiratory HRCT or the MDCT scan is the standard technique employed for making detailed morphological assessments of pulmonary parenchyma at high spatial resolutions. This study indicated that both inspiratory and expiratory LAV% depend on the attenuation threshold level in patients with COPD, including those with mild/moderate airflow limitation, and non-COPD smokers. Furthermore, changes in LAV% from the inspiratory to the expiratory phase (ΔLAV%) were greater in non-COPD smokers than in those with severe COPD, suggesting that trapped air might move less during the respiratory phase in patients with advanced emphysema.

An understanding of the pathophysiology of COPD is essential for proper diagnosis and quantification of emphysematous lesions. Emphysematous lesions can be visually measured as the relative lung area occupied by pixels with attenuation values below a predetermined threshold. This visual scoring method continues to be used because it does not require any special software. In the present study, we quantified emphysematous changes by visual scores and their correlation with LAV%. Moreover, inter-observer performance was relatively high. In 2012, Mascalchi et al. reported that inter- and intra-operator agreement of emphysema using a visual scale was low and decreased with increasing emphysema extent (12). The discrepancy in the two sets of data may be due to our method of evaluating three slices using a six-level visual scale for each patient (9,10). Mascalchi et al. assessed an average of 22 ± 2 sections using a visual scale that defined emphysema extent using a range of 0–100. Their explanation for the low level of agreement rested on the difficulty in differentiating between extent of emphysema and low attenuation areas due to air trapping in inspiratory CT scans. This area remains controversial and requires further investigation for clarification.

If the computer-based scoring is selected for use, the predetermined threshold is usually set at −950 HU in clinical practice, since the relative area of lung with attenuation values lower than −950 HU, at full inspiration on incremental thin-section CT images, is representative of microscopic and macroscopic emphysema (8). However, technical progress may have changed this threshold: Müller and colleagues examined conventional 1-cm thick CT scans to assess correlation with the visual pathologic grade of emphysema, and found that the highest correlation coefficient was obtained at −910 HU (14). On MDCT images obtained at full inspiration, the relative area of lung with attenuation pixels lower than −960 HU has proven to be an appropriate measure (15).

The current study shows the influence of the threshold when evaluating LAV% on both inspiratory and expiratory images. On both images, LAV% dramatically changed according to the threshold between −1000 and −750 HU. Even when a lower threshold was used for assessing LAV%, significant differences were observed between the two images.

Healthy lungs exhibit a homogeneous attenuation ranging between −700 and −900 HU at full inspiration. The term “emphysema” includes various degrees of disappearance of alveoli, pulmonary interstitial tissue, and pulmonary vessels, and also included low attenuation clusters of various sizes (16,17). Static parenchymal alterations (anatomical emphysema) and dynamic functional alterations (air trapping) could influence attenuation values, thus the definitive threshold to assess “emphysema” or low attenuation areas as a link with obstructive disorders remains undefined.

We also found that the change in LAV% in smokers without COPD was higher than that of COPD patients with severe and very severe stages of the disease. Severe airflow limitation under advanced emphysema could lead to less change in functional alternations. Our results agree with previous findings that the change in the mean lung attenuation differentiated COPD patients from systemic sclerosis patients with and without pulmonary involvement by Camiciottoli et al. (18). They confirmed that the clinical utility of paired inspiratory/expiratory images for differentiating patients with obstructive and restrictive lung disease from normal subjects was better compared with data obtained only in inspiratory scans.

A significant correlation between quantitative parameters derived from MDCT scans obtained at deep inspiration and expiration and indexes of obstructive impairment was observed in this study. Inspiratory LAV% showed a close negative correlation with FEV₁/FVC and FEV₁ % predicted at attenuation values lower than −930 HU, while expiratory LAV% showed a close negative correlation with these parameters at attenuation values lower than −870 or −880 HU. An overestimation of emphysema may occur on inspiratory images when part of the hyperaeration observed at inspiration is no longer observed at expiration. Some studies have found expiratory images to be superior for quantitating emphysema and showing better correlation with indexes of obstructive impairment than inspiratory images (19). LAV% on expiratory images did correlate more closely with airflow limitation than inspiratory images in the present study. Other studies found that inspiratory CT parameters were significant predictors of PFT abnormalities or pathological findings rather than expiratory scans, and that expiratory CT parameters reflected airflow obstruction and air-trapping (20 –23). For example, Camiciottoli et al. showed that a relative area with an attenuation volume < −950 HU at 90%VC (RAI₉₅₀) independently predicted the percentage of predicted diffusing capacity of the lung for carbon monoxide (DLco%), and that inspiratory measurements reflect the extent of emphysema. They also showed that mean CT lung density at 10% of VC (MeanCTexp) independently predicted FEV1/FVC, percentage of predicted residual volume (RV%) and MRC (Medical Research Council) dyspnea scale. They confirmed that expiratory measurements reflect airflow limitation and lung hyperinflation with attendant dyspnea perception. Moreover, our study indicated that LAV% gradually increased as the threshold level increased on both inspiratory and expiratory images. By using a higher threshold, we could measure not only the volume of emphysema, but also the area of air-trapping or lung parenchyma, in which small pulmonary vessels are destroyed and detected at relatively lower lung density compared to that of normal lung parenchyma. Previous study reported that the relative volume change calculated on paired inspiratory and expiratory MDCT images using a threshold of −860 HU in emphysematous lungs correlated closely with parameters of airway dysfunction in COPD, regardless of the degree of emphysema (24). However, in our study LAV% differed between inspiratory and expiratory images, and the threshold level that closely correlated with indexes of obstructive impairment was different between inspiratory and expiratory images.

We should mention some limitations of this study. First, we did not evaluate pathological findings in the subjects. In an earlier study, a low attenuation area of less than −910 HU was a better predictor of lung emphysema based on pathological findings (14). However, it is unknown to what degree emphysema was detected in pathological specimens associated with airflow limitation, and emphysematous lesions are heterogeneously distributed in the lungs in COPD. Second, the technique used to determine the threshold is very sensitive to CT devices, scanning techniques, slice thickness, reconstruction algorithm, and other technical conditions (7,25 –27). However, we were able to obtain reliable data from a single-center prospective study, using a CT produced by one manufacturer. Third, full inspiration and expiration are more subject-dependent than operator-dependent. Moreover, this study focused primarily on emphysematous changes and not on airway abnormalities. Airway abnormalities, typically in the airway-dominant COPD phenotype, can influence airflow limitation and air-trapping (28,29). McDonough et al. reasoned that the narrowing and disappearance of small airways before the onset of emphysematous destruction could explain the increased peripheral airway resistance in COPD (30). Further studies are needed to evaluate the relationship between airway abnormalities and airflow limitation according to the severity of lung emphysema.

In conclusion, LAV% dramatically changed according to the threshold level on both inspiratory and expiratory images. The linkage of threshold defining LAV% to airflow limitation was higher on expiratory images than on inspiratory images. Therefore, when LAA or LAV is evaluated at any threshold level on CT images, it is of clinical importance in terms of an association with physiological obstructive impairment.

Footnotes

Acknowledgements

We thank Mr. Hirotaka Sato and Mrs. Masako Suzuki who have enriched our research for many years.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

References

The Global Strategy for the Diagnosis, Management and Prevention of COPD, Global Initiative for Chronic Obstructive Pulmonary Disease (GOLD) 2011. See http://www.goldcop.org/.

Hogg

Macklem

Thurlbeck

. Site and nature of airways obstruction in chronic obstructive lung disease. N Engl J Med 1968; 278: 1355–1360.

Hogg

Chu

Utokaparch

. The nature of small-airway obstruction in chronic obstructive pulmonary disease. N Engl J Med 2004; 350: 2645–2653.

Nakano

Muro

Sakai

. Computed tomographic measurements of airway dimensions and emphysema in smokers: correlation with lung function. Am J Respir Crit Care Med 2000; 162: 1102–1108.

Park

Bergin

Clausen

. Quantitation of emphysema with three-dimensional CT densitometry: comparison with two-dimensional analysis, visual emphysema scores, and pulmonary function test results. Radiology 1999; 211: 541–547.

Zaporozhan

Ley

Eberhardt

. Paired inspiratory/expiratory volumetric thin-slice CT scan for emphysema analysis: comparison of different quantitative evaluations and pulmonary function test. Chest 2005; 128: 3212–3220.

Yuan

Mayo

Hogg

. The effects of radiation dose and CT manufacturer on measurements of lung densimetory. Chest 2007; 132: 617–623.

Gevenois

de Maertelaer

De Vuyst

. Comparison of computed density and macroscopic morphometry in pulmonary emphysema. Am J Respir Crit Care Med 1995; 152: 653–657.

Makita

Nasuhara

Nagai

. Characterisation of phenotypes based on severity of emphysema in chronic obstructive pulmonary disease. Thorax 2007; 62: 932–937.

10.

Goddard

Nicholson

Laszlo

. Computed tomography in pulmonary emphysema. Clin Radiol 1982; 33: 379–387.

11.

American Thoracic Society/European Respiratory Society Task Force. Standards for the Diagnosis and Management of Patients with COPD. Version 1.2 , New York, NY: American Thoracic Society, 2004. See http://www.thoracic.org/go/copd.

12.

Mascalchi

Diciotti

Sverzellati

. Low agreement of visual rating for detailed quantification of pulmonary emphysema in whole-lung CT. Acta Radiol 2012; 53: 53–60.

13.

Bland

Altman

. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1: 307–310.

14.

Mlüller

Staples

Miller

. “Density mask”. An objective method to quantitate emphysema using computed tomography. Chest 1988; 94: 782–787.

15.

Madani

Zanen

de Maertelaer

. Pulmonary emphysema: objective quantification at multi-detector row CT-comparison with macroscopic and microscopic morphometry. Radiology 2006; 238: 1036–1043.

16.

Mishima

Hirai

Itoh

. Complexity of terminal airspace geometry assessed by lung computed tomography in normal subjects and patients with chronic obstructive pulmonary disease. Proc Natl Acad Sci USA 1999; 96: 8829–8834.

17.

Matsuoka

Kurihara

Yagihashi

. Morphological progression of emphysema on thin-section CT: Analysis of longitudinal change in the number and size of low-attenuation clusters. J Comput Assist Tomogr 2006; 30: 669–674.

18.

Camiciottoli

Diciotti

Bartolucci

. Whole-lung volume and density in spirometrically-gated inspiratory and expiratory CT in systemic sclerosis: correlation with static volumes at pulmonary function tests. Sarcoidosis Vasc Diffuse Lung Dis 2013; 30: 17–27.

19.

Kauczor

Hast

Heussel

. CT attenuation of paired HRCT scans obtained at full inspiratory/expiratory position: comparison with pulmonary function tests. Eur Respir J 2002; 12: 2757–2763.

20.

Gevenois

De Vuyst

. Pulmonary emphysema: quantitative CT during expiration. Radiology 1996; 199: 825–829.

21.

Eda

Kubo

Fujimoto

. The relations between expiratory chest CT using helical CT and pulmonary function tests in emphysema. Am J Respir Crit Care Med 1997; 155: 1290–1294.

22.

Kubo

Eda

Yamamoto

. Expiratory and inspiratory chest computed tomography and pulmonary function tests in cigarette smokers. Eur Respir J 1999; 13: 252–256.

23.

Camiciottoli

Bartolucci

Maluccio

. Spirometrically gated high-resolution CT findings in COPD: lung attenuation vs lung function and dyspnea severity. Chest 2006; 129: 558–564.

24.

Matsuoka

Kurihara

Yagihashi

. Quantitative assessment of air trapping in chronic obstructive pulmonary disease using inspiratory and expiratory volumetric MDCT. Am J Roentgenol 2008; 190: 762–769.

25.

Boedeker

McNitt-Gray

Rogers

. Emphysema: reconstruction algorithm on CT imaging measures. Radiology 2004; 232: 295–301.

26.

Kemerink

Lamers

Thelissen

. CT densimetry of the lungs: scanner performance. J Comput Assist Tomogr 1996; 20: 24–33.

27.

Madani

De Maertelaer

Zanen

. Pulmonary emphysema: radiation dose and section thickness at multidetector CT quantification-comparison with macroscopic and microscopic morphometry. Radiology 2007; 243: 250–257.

28.

Hasegawa

Nasuhara

Onodera

. Airflow limitation and airway dimensions in chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2006; 173: 1309–1315.

29.

Matsuoka

Kurihara

Yagihashi

. Airway dimensions at inspiratory and expiratory multisection CT in chronic obstructive pulmonary disease : correlation with airflow limitation. Radiology 2008; 248: 1042–1049.

30.

McDonough

Yuan

Suzuki

. Small-airway obstruction in chronic obstructive pulmonary disease. N Engl J Med 2011; 365: 1567–1575.