Vol.:(0123456789)1 3

Chinese Journal of Academic Radiology (2019) 1:43–48 https://doi.org/10.1007/s42058-019-00007-0

REVIEW

Progress in the imaging of COPD: quantitative and functional evaluation

Li Fan1 · Xiuxiu Zhou1 · Yi Xia1 · Yu Guan1 · Di Zhang1 · ZhaoBin Li2 · Shiyuan Liu1

Received: 22 November 2018 / Revised: 6 April 2019 / Accepted: 22 April 2019 / Published online: 20 May 2019 © Springer Nature Singapore Pte Ltd. 2019

AbstractChronic obstructive pulmonary disease (COPD) is a common, preventable, and treatable disease, which has caused serious social and economic burden. The main characteristic of COPD is the heterogeneity of disease, manifesting as emphysema, functional small airways disease (fSAD) and large airway diseases. Imaging plays an important role in the evaluation of COPD. In this article, we summarized the recent advances in the imaging of COPD, especially in the quantitative and func-tional evaluations of the disease. Imaging provides the detailed anatomical, quantitative and function information, illustrates regional heterogeneity and spatial distribution, as well as provides microstructural assessment on the alveolar level. Espe-cially, air trapping index (ATI), parametric response mapping (PRM), gas exchange and texture analysis facilitate the early diagnosis, phenotype classification, severity and therapeutic effect evaluation.

Keywords Chronic obstructive pulmonary disease · Tomography · X-ray computed · Quantitative imaging · Magnetic resonance imaging

Introduction

Chronic obstructive pulmonary disease (COPD) is a com-mon, preventable, and treatable disease characterized by per-sistent respiratory symptoms and airflow limitation that is due to airway and/or alveolar abnormalities, usually caused by significant exposure to noxious particles or gases [1]. COPD is the third cause of worldwide death, accounting for the one-third of death in China [2], which has caused serious social and economic burden. The main characteris-tic of COPD is the heterogeneity of disease, manifesting as emphysema, functional small airways disease (fSAD) and large airway diseases. Moreover, emphysema and fSAD usually coexist. How to differentiate the emphysema from

fSAD is very important to improve the classification of phe-notypes, instruct the individualized therapy and improve the prognosis. Pulmonary function test (PFT) is the clini-cal established approach for the diagnosis of COPD. The diagnosis is based on spirometry, a ratio of forced expira-tory volume in the first second (FEV1)/forced vital capac-ity (FVC) is less than 70% (FEV1/FVC < 70%), and FEV1 increases less than 200 ml after the use of a bronchodilator [3]. However, PFT reflects the whole lung function change, which cannot reflect the regional change and is not sensitive to the early change of COPD.

Imaging plays an important role in the evaluation of COPD [4]: with the advent of CT in 1970s, pulmonary parenchyma and large airway were evaluated; in the fol-lowing 30 years, CT lung density analysis was performed; since 2000s, the phenotype of COPD and airway quantitative analysis has been studied; and the latest CT-based and MR-based technologies make it possible to evaluate the morpho-logical and functional change of COPD in the early stage. According to the dominancy of emphysema and the presence of bronchial wall thickening, COPD is classified into three phenotypes: A phenotype, no or mild emphysema with or without bronchial wall thickening; E phenotype, apparent emphysema without bronchial wall thickening; M pheno-type, apparent emphysema with bronchial wall thickening

* ZhaoBin Li lizhaobin79@163.com

* Shiyuan Liu liushiyuan@smmu.edu.cn

1 Department of Radiology, Changzheng Hospital, Second Military Medical University, No. 415 Fengyang Road, Shanghai 200003, China

2 Department of Radiation Oncology, Shanghai Jiao Tong University Affiliated Sixth People’s Hospital, No. 600 Yishan Road, Shanghai 200233, China

44 Chinese Journal of Academic Radiology (2019) 1:43–48

1 3

[5]. In this article, we summarized the progress in the imag-ing of COPD, especially on quantitative, functional evalua-tion and texture analysis.

Quantitative imaging

Quantitative evaluation of COPD mainly focuses on CT lung density, small airway and small pulmonary vessels, which can provide detailed anatomical and quantitative information.

Lung density analysis

In clinical routine work, chest CT is performed on the end of inspiration. According to the density of pulmonary parenchyma on inspiratory phase CT images, the pulmo-nary parenchyma is classified into 3 regions, E (emphysema) region (− 950 to − 1024 HU), H (hyperinflated) region (− 900 to − 949 HU) and N (normal) region (− 400 to − 899 HU) [6]. The pixel percentage of low-density region with attenuation less than − 950 HU is defined as LAA% − 950 or emphysema index (EI = TEV/TLV). It has been reported that healthy non-smokers showed emphysema-like change with EI > 0% [7]. However, the normal limit of EI among the healthy non-smokers varied from less than 1–6.8% in previ-ous studies with different age ranges [7–11]. In the state-ment of the Fleischner society on CT-definable subtypes of COPD, the emphysema was defined as the pixels < − 950

HU more than 6% by CT quantitative evaluation [12]. It has been reported that TEV and EI were negatively correlated with FEV1/FVC [13]. At present, evaluation of lung density on the lobar level is possible, which evaluates the heteroge-neity of COPD.

Air trapping is the retention of air in the lung distal to an obstruction (usually partial), which is seen on end-expiration CT scans manifesting as parenchymal areas with less than normal increase in attenuation and lack of volume reduc-tion [14]. Detailed analysis of density change on inspiration and expiration CT of COPD can provide new insights into pulmonary functional impairment in each lung area. Paired inspiratory and expiratory phases could obtain more param-eters, including expiratory to inspiratory ratio of mean lung density (E/I MLD), air trapping index (ATI), relative vol-ume change − 856 to − 950 (RVC856–950), expiratory − 856 (Exp-856) and parametric response mapping (PRM) (Fig. 1). The expiration images were deformed to match the inspira-tion images per pixel with a non-rigid method, then sub-traction was performed between inspiration and expiration CT images, and subtraction value per pixel was obtained. ATI is defined as the volume of voxels having a subtraction value lower than a threshold for air trapping [15]. The sub-tracted images could evaluate the heterogeneity of COPD comprehensively, including evaluation the air trapping in non-emphysema area. ATI, in compassion with EI, showed more significant correlation with PFT. Lee et al. found the optimal CT threshold of subtraction method for air trapping was 60 HU; ATI with 60 HU thresholds was comparable

Fig. 1 CT quantitative analysis of air trapping, volume and mean lung density on the lobar level with paired inspiratory and expiratory CT scan-ning

45Chinese Journal of Academic Radiology (2019) 1:43–48

1 3

to E/I MLD and Exp-856 [15]. RVC is defined as the dif-ference between the expiratory and inspiratory values for relative lung volumes, which is the limited lung volume with attenuation between − 856 and – 950 HU. It has been proven that in COPD patients no matter with emphysema or not, RVC856-950 showed the most significant relation to emphysema [16]. Exp-856 is defined as the percent of the lung voxels with attenuation < − 856 HU on the expiratory CT images [17]. In the previous study with paired respira-tory phases, Exp-856 was used to evaluate the air trapping, but could not differentiate the cause from emphysema or functional small airways disease (fSAD). fSAD is defined as non-emphysematous airflow obstruction [18, 19]. PRM is a postprocessing approach to evaluate fSAD, which is based on the voxel density change between inspiratory and expira-tory CT. Boes [18] found fSAD was a transitional phase from normal parenchyma to emphysema. PRM combines the subjective and objective evaluation of COPD, and helps to assess the phenotype, severity, spatial heterogeneity, lung density changes and longitudinal surveillance. Based on top-ological techniques, Hoff [19] generated 3D maps of local topological features from 3D PRMfSAD classification maps and found that the surface area of fSAD (SfSAD) was the most robust and significant independent indicator of clini-cally meaningful measures of COPD.

Besides, dynamic imaging evaluates the morphological and functional change during the whole respiratory cycle and the dynamic progression of disease. Respiratory gated chest 4D-CT allows for the morphological and functional information, providing new insights into functional impair-ment and individual treatment planning [20].

Small airway quantitative analysis

Small airway remodeling is one of the major characteristics of COPD. Direct measurement of small airway is applied in the 6th or proximal generation of airways, including wall thickness (WT), wall area (WA), lumen area (LA), wall area percentage (WA%), and inner perimeter 10 mm (Pi10). Guan Y et al. found WA% of 6th order bronchus has a significant positive correlation with EI, indicating that the extent of distal bronchial wall thickening is more rel-evant with emphysema [21]. Previous studies focused on cross-sectional CT images of small airways. Oguma T was the first to assess the longitudinal structure of the airway lumen, and found longitudinal airway lumen shape irregu-larity was significantly greater in patients with COPD than patients with bronchial asthma and healthy controls [22]. Indirect measurement of small airway is to analyze ATI and PRMfSAD using paired inspiratory and expiratory scanning [6, 16, 18]. It has been reported that PRMfSAD was negatively correlated with total airway count (TAC) [23], which was independently associated with lung function decline. TAC

may reflect the airway-related disease changes that accumu-late in the “quiet” zone in early/mild COPD, indicating that TAC may be a biomarker to predict the accelerated COPD progression [23].

Small pulmonary vessels quantitative analysis

Small pulmonary vessels remodeling in COPD could be evaluated with the percentage of small pulmonary vessels (% cross-sectional area [CSA]), as a parameter of pulmonary perfusion, which correlates inversely with airflow limita-tion as well as with pulmonary hypertension and radio-graphic emphysema in COPD [24]. Takayanagi et al. found the WA% at the distal bronchi and the %CSA < 5 did not change in parallel with parameters of LAA over the same period, suggesting airway disease and vascular remodeling may be reversible to some extent by smoking cessation and appropriate treatment [25]. Therefore, in the early stage of COPD, we should pay more attention to the pulmonary vas-cularity and airway disease rather than parenchymal destruc-tion. Moreover, CT measurements of small vessels support a distinct vessel-related phenotype in COPD patients with severe pulmonary hypertension (HP), combining %CSA < 5, PaO2 and WT best prediction of severe PH in patients with COPD [26].

Functional imaging

Perfusion imaging

CT and MR perfusion imaging have been widely used in the assessment of focal pulmonary parenchyma perfusion in a variety of patients with pulmonary embolism, lung cancer and emphysema. Previous studies focused on the early diagnosis, severity assessment, phenotype classifi-cation, differential diagnosis of high-risk population from the mild COPD patients and correlation with PFT and CT quantitative parameters. Compared with CT, MR perfusion imaging shows higher potential to distinguish controls from mild COPD and appears more sensitive in identifying abnor-malities among smokers with normal PFT [27]. CT and MRI perfusion parameters, such as positive enhancement integral (PEI), maximum slope of increase (MSI), maximum slope of decrease (MSD), signal enhancement ratio (SER), and signal intensity ratio (RSI) of perfusion defects to normal lung, were positively correlated with PFT parameters; and negatively correlated with EI [5, 13, 28].

Ventilation imaging

A series of ventilation imaging approaches have been used to evaluate the ventilation status of COPD, including the static

46 Chinese Journal of Academic Radiology (2019) 1:43–48

1 3

imaging, dynamic imaging, hyperpolarized inert gas imag-ing, oxygen enhanced imaging and fluorinated gas imaging [29–33]; which focused on subjective and semiquantitative evaluation, such as ventilation defect, homogeneity and so on. At present, objective and regional evaluation come true. Lobar ventilation and ADC values obtained from hyperpo-larized 129Xe MR imaging demonstrated correlation with quantitative CT percentage emphysema on a lobar basis and with PFT results [34]. Capaldi et al. compared MR imaging and CT PRM measurements of gas trapping and emphy-sema in ex-smokers both with and without COPD, and found ventilation defect percent (VDP) was related to PRM gas trapping in patients with mild-to-moderate COPD, whereas in patients with severe COPD, VDP correlated with both PRM gas trapping and PRM emphysema [35] (Fig. 2). It has been reported that xenon-enhanced dual-energy CT has been used to evaluate the regional emphysema, air trapping and xenon-ventilation abnormalities in COPD patients, and found the combination of attenuation and xenon can predict more accurate PFTs [36, 37].

Gas exchange imaging

Diffusing capacity of the lungs for carbon monoxide (DLCO) is a direct marker of global gas exchange. Hyper-polarized (HP) 129Xe gas is liposoluble, and can transfer from alveoli to interstitial barrier tissues and capillary RBCs, transferring from the gas phase (GP) signal to the dissolved phase (DP) signal. Pulmonary interstitial barrier tissues are composed of alveolar epithelial cells, interstitial tissues, capillary endothelial cells, and plasma. Therefore, there are three compartments, including the alveoli (GP), pulmonary barrier tissues (TP) and capillary red blood cells (RBCs). The 129Xe spectrum exhibits three resonances in lung at 0 ppm, 198 ppm and 217 ppm corresponding to 129Xe in airspaces, dissolved in barrier and RBCs [30, 38]. TP/GP, RBC/GP, RBC/TP and DP/GP could quantify the alveolar capillary membrane thickness indirectly. Wang JM et al. found 129Xe MRI-derived barrier uptake, RBC transfer and RBC/barrier ratio correlated well with DLCO [38]. The direct regional gas exchange measurements are sensitive to

Fig. 2 [35] Ventilation and PRM in a 55-year-old man without COPD (FEV1, 83% of predicted value; FEV1/FVC, 77%; residual volume to total lung capacity ratio [RV/TLC], 45%), a 69-year-old man with GOLD I disease (FEV1, 89% of predicted value; FEV1/FVC, 69%; RV/TLC, 39%; DLCO, 67% of predicted value), an 84-year-old man with GOLD II disease (FEV1, 52% of predicted value; FEV1/FVC, 44%; RV/TLC, 62%; DLCO, 47% of predicted value), and a 67-year-old woman with GOLD III disease (FEV1, 33% of pre-dicted value; FEV1/FVC, 39%; RV/TLC, 72%; DLCO, 28% of predicted value). First row: 3He MR images coregistered with 1H MR images (gray-scale) show static ventilation (blue areas). Second row: 3He MR imaging ADC maps show that the ex-smokers with more advanced COPD (GOLD II/III disease) have elevated ADC values. Third row: CT attenu-ation masks show areas of less than 2950 HU (yellow areas). Fourth row: PRMs show areas of healthy tissue (green), gas trapping (yellow), and emphy-sema (red). Permission has been obtained from Radiology [35]

47Chinese Journal of Academic Radiology (2019) 1:43–48

1 3

disease at the alveolar level, helping to the early diagnosis and intervention.

Texture analysis

In the statement of Fleischner society on CT-definable sub-types of COPD, emphysema is classified into three sub-types, centrilobular emphysema, pan-lobular emphysema and para-septal emphysema [12]. The different subtypes of emphysema and different HRCT phenotypes of COPD have different clinical managements; the classification of subtypes and phenotypes is very important. Automated texture-based quantification of emphysema subtypes has been successfully implemented via supervised learning of these three emphysema subtypes. Yang et al. extracted three types of texture features, including frequency histograms of textons, soft histograms of intensities and difference of Gaussian (DoG) responses, and joint histograms of local binary patterns (LBP) and intensities; and demonstrated that unsupervised learning on a large heterogeneous database of CT scans can generate texture prototypes that are visually homogeneous and distinct, reproducible across subjects, and capable of predicting the three standard radiological sub-types accurately, which open the way to new interpretations of lung CT scans with finer subtyping of emphysema [39]. It has been reported that hybrid airway segmentation using multi-scale tubular structure filters and texture analysis on 3D chest CT scans has been performed successfully in the Korean Obstructive Lung Disease (KOLD) Cohort [40]. The method limitations were higher false-positive rates than those of the other methods and risk of leakage. In future studies, application of a convolutional neural network will help overcome these shortcomings [40].

Conclusion

Imaging plays a great role in the assessment of COPD, which provides the detailed anatomical, quantitative and functional information, illustrates the regional heterogeneity and spatial distribution, as well as provides microstructural assessment at the alveolar level. Especially, ATI, PRM, gas exchange and texture analysis facilitate the early diagnosis, phenotype classification, severity and therapeutic effect evaluation in the near future.

Funding This work was supported by the National Natural Sci-ence Foundation of China [Grant numbers 81871321, 81370035]; the National Key R&D Program of China [Grant numbers 2016YFE0103000, 2017YFC1308703].

Compliance with ethical standards

Conflict of interest All authors declare that they have no conflict of interest.

References

1. Vogelmeier CF, Criner GJ, Martinez FJ, et al. Global Strategy for the Diagnosis, Management, and Prevention of Chronic Obstruc-tive Lung Disease 2017 Report. GOLD Executive Summary. Am J Respir Crit Care Med. 2017;195(5):557-582.

2. Wang C, Xu J, Yang L, et al. Prevalence and risk factors of chronic obstructive pulmonary disease in China (the China Pulmonary Health [CPH] study): a national cross-sectional study. Lancet. 2018;391(10131):1706–17.

3. Pauls S, Gulkin D, Feuerlein S, et al. Assessment of COPD sever-ity by computed tomography: correlation with lung functional testing. Clin Imaging. 2010;34:172–8.

4. Washko GR, Coxson HO, O’Donnell DE, et al. CT imaging of chronic obstructive pulmonary disease: insights, disappointments, and promise. Lancet Respir Med. 2017;5(11):903–8.

5. Fan L, Xia Y, Guan Y, et al. Characteristic features of pulmonary function test, CT volume analysis and MR perfusion imaging in COPD patients with different HRCT phenotypes. Clin Respir J. 2014;8(1):45–54.

6. Kim EY, Seo JB, Lee HJ, et al. Detailed analysis of the density change on chest CT of COPD using non-rigid registration of inspi-ration/expiration CT scans. Eur Radiol. 2015;25(2):541–9.

7. Marsh S, Aldington S, Williams MV, et al. Utility of lung den-sity measurements in the diagnosis of emphysema. Respir Med. 2007;101:1512–20.

8. Gevenois PA, Scillia P, de Maertelaer V, et al. The effects of age, sex, lung size, and hyperinflation on CT lung densitometry. Am J Roentgenol. 1996;167:1169–73.

9. Irion KL, Marchiori E, Hochhegger B, et al. CT quantification of emphysema in young subjects with no recognizable chest disease. Am J Roentgenol. 2009;192:W90–6.

10. Zach JA, Newell JD Jr, Schroeder J, et al. Quantitative computed tomography of the lungs and airways in healthy nonsmoking adults. Invest Radiol. 2012;47:596–602.

11. Park KJ, Bergin CJ, Clausen JL. Quantitation of emphysema with three-dimensional CT densitometry: comparison with two-dimen-sional analysis, visual emphysema scores, and pulmonary function test results. Radiology. 1999;211:541–7.

12. Lynch DA, Austin JH, Hogg JC, et al. CT-definable subtypes of chronic obstructive pulmonary disease: a statement of the Fleisch-ner society. Radiology. 2015;277:192–205.

13. Xia Y, Guan Y, Fan L, et al. Dynamic contrast enhanced magnetic resonance perfusion imaging in high-risk smokers and smoking-related COPD: correlations with pulmonary function tests and quantitative computed tomography. COPD. 2014;11(5):510–20.

14. Hansell DM, Bankier AA, MacMahon H, et  al. Fleischner Society: glossary of terms for thoracic imaging. Radiology. 2008;246:697–722.

15. Lee SM, Seo JB, Lee SM, et al. Optimal threshold of subtraction method for quantification of air-trapping on coregistered CT in COPD patients. Eur Radiol. 2016;26(7):2184–92.

16. Hersh CP, Washko GR, Estépar RS, et al. Paired inspiratory -expiratory chest CT scans to assess for small airways disease in COPD. Respir Res. 2013;14:42.

17. Busacker A, Newell JD Jr, Keefe T, et al. A multivariate analysis of risk factors for the air-trapping asthmatic phenotype as meas-ured by quantitative CT analysis. Chest. 2009;135(1):48–56.

48 Chinese Journal of Academic Radiology (2019) 1:43–48

1 3

18. Boes JL, Hoff BA, Bule M, et al. Parametric response map-ping monitors temporal changes on lung CT scans in the sub-populations and intermediate outcome measures in COPD Study (SPIROMICS). Acad Radiol. 2015;22(2):186–94.

19. Hoff BA, Pompe E, Galbán S, et al. CT-Based Local Distri-bution Metric Improves Characterization of COPD. Sci Rep. 2017;7(1):2999.

20. Ley-Zaporozhan J, Ley S, Mews J, et al. Changes of Emphysema Parameters over the Respiratory Cycle During Free Breathing: preliminary Results Using Respiratory Gated 4D-CT. COPD. 2017;14(6):597–602.

21. Guan Y, Fan L, Xia Y, et al. CT quantitative analysis of small airway remodeling and lobe-based emphysema of chronic obstruc-tive pulmonary disease and its correlation with pulmonary func-tion test. Chin J Med Imaging Technol. 2015;31(2):181–4 (In Chinese).

22. Oguma T, Hirai T, Fukui M, et al. Longitudinal shape irregularity of airway lumen assessed by CT in patients with bronchial asthma and COPD. Thorax. 2015;70(8):719

23. Kirby M, Tanabe N, Tan WC, et al. Total Airway Count on Com-puted Tomography and the Risk of Chronic Obstructive Pulmo-nary Disease Progression. Findings from a Population-based Study. Am J Respir Crit Care Med. 2018;197(1):56-65.

24. Matsuoka S, Washko GR, Dransfield MT, et al. Quantitative CT measurement of cross-sectional area of small pulmonary vessel in COPD: correlations with emphysema and airflow limitation. Acad Radiol. 2010;17(1):93–9.

25. Takayanagi S, Kawata N, Tada Y, et al. Longitudinal changes in structural abnormalities using MDCT in COPD: do the CT meas-urements of airway wall thickness and small pulmonary vessels change in parallel with emphysematous progression? Int J Chron Obstruct Pulmon Dis. 2017;12:551–60.

26. Coste F, Dournes G, Dromer C, et al. CT evaluation of small pulmonary vessels area in patients with COPD with severe pul-monary hypertension. Thorax. 2016;71(9):830–7.

27. Fan L, Xia Y, Guan Y, et al. Capability of differentiating smokers with normal pulmonary function from COPD patients: a compari-son of CT pulmonary volume analysis and MR perfusion imaging. Eur Radiol. 2013;23(5):1234–41.

28. Guan Y, Xia Y, Fan L, et al. Quantitative assessment of pulmonary perfusion using dynamic contrast-enhanced CT in patients with chronic obstructive pulmonary disease: correlations with pulmo-nary function test and CT volumetric parameters. Acta Radiol. 2015;56(5):573–80.

29. Li H, Zhang Z, Zhao X, et al. Quantitative evaluation of radiation-induced lung injury with hyperpolarized xenon magnetic reso-nance. Magn Reson Med. 2016;76(2):408–16.

30. Norquay G, Leung G, Stewart NJ, et al. 129 Xe chemical shift in human blood and pulmonary blood oxygenation measurement in humans using hyperpolarized 129 Xe NMR. Magn Reson Med. 2017;77(4):1399–408.

31. Horn FC, Rao M, Stewart NJ, et al. Multiple breath washout of hyperpolarized 129 Xe and 3He in human lungs with three-dimen-sional balanced steady-state free-precession imaging. Magn Reson Med. 2017;77(6):2288–95.

32. Couch MJ, Ball IK, Li T, et al. 19 F MRI of the Lungs Using Inert Fluorinated Gases: Challenges and New Developments. J Magn Reson Imaging. 2018 Sep 24 https ://doi.org/10.1002/jmri.26292 . [Epub ahead of print].

33. Fuseya Y, Muro S, Sato S, et al. Complementary regional hetero-geneity information from COPD patients obtained using oxygen-enhanced MRI and chest CT. PLoS One. 2018;13(8):e0203273.

34. Matin TN, Rahman N, Nickol AH, et al. Chronic obstructive pul-monary disease: lobar analysis with hyperpolarized 129Xe MR Imaging. Radiology. 2017;282(3):857–68.

35. Capaldi DPI, Zha N, Guo F, et al. Pulmonary imaging biomarkers of gas trapping and emphysema in COPD: 3He MR imaging and CT parametric response maps. Radiology. 2016;279:597–608.

36. Sugino K, Kobayashi M, Nakamura Y, et al. Xenon-Enhanced Dual-Energy CT Imaging in Combined Pulmonary Fibrosis and Emphysema. PLoS One. 2017;12(1):e0170289.

37. Lee SM, Seo JB, Hwang HJ, et al. Assessment of regional emphy-sema, air-trapping and Xenon-ventilation using dual-energy computed tomography in chronic obstructive pulmonary disease patients. Eur Radiol. 2017;27(7):2818–27.

38. Wang JM, Robertson SH, Wang Z, et al. Using hyperpolarized 129Xe MRI to quantify regional gas transfer in idiopathic pulmo-nary fibrosis. Thorax. 2018;73(1):21–8.

39. Yang J, Angelini ED, Smith BM, et al. Explaining Radiological Emphysema Subtypes with Unsupervised Texture Prototypes: MESA COPD Study. Med Comput Vis Bayesian Graph Models Biomed Imaging. 2016;2017(2017):69–80.

40. Lee M, Lee JG, Kim N, et al. Hybrid Airway Segmentation Using Multi-Scale Tubular Structure Filters and Texture Analysis on 3D Chest CT Scans. J Digit Imaging. 2018 Nov 21. https ://doi.org/10.1007/s1027 8-018-0158-8.

Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.