Elsevier

Academic Radiology

Volume 20, Issue 5, May 2013, Pages 527-536
Academic Radiology

Original investigation
Registration-Based Lung Mechanical Analysis of Chronic Obstructive Pulmonary Disease (COPD) Using a Supervised Machine Learning Framework

https://doi.org/10.1016/j.acra.2013.01.019Get rights and content

Rationale and Objectives

This study evaluated the performance of computed tomography (CT)-derived biomechanical based features of lung function and the presence and severity of chronic obstructive pulmonary disease (COPD). It performed well when compared to CT-derived density and textural features of lung function and the presence and severity of COPD.

Materials and Methods

A total of 162 subjects (Global Initiative for Chronic Obstructive Lung Disease [GOLD] stages 0–4 and nonsmokers) subjects with CT scan performed at total lung capacity or expiration to functional residual capacity were evaluated. CT-derived biomechanical, density, and textural feature sets were compared to forced expiratory volume in 1 second (FEV1)%, FEV1/forced vital capacity, and total St. George's respiratory questionnaire scores. The ability of these feature sets to assess the presence and severity of COPD was also evaluated. Optimal features are selected by linear forward feature selection and the classification is done using k nearest neighbor learning algorithm.

Results

The proposed biomechanical features showed good correlations with the pulmonary function tests and health status metrics. In COPD versus non-COPD classification, biomechanical feature set achieved an area under the curve (AUC) of 0.85 performing well in comparison to density (AUC = 0.83) and texture (AUC = 0.89) feature sets. Classifying the subjects into the severity of GOLD stage using biomechanical features (AUC = 0.81) performed better than the density- and texture-based feature sets, AUC = 0.76 and 0.73, respectively. The biomechanical features performed better alone than in combination with the other two feature sets.

Conclusion

This study shows the effectiveness of CT-derived biomechanical measures in the assessment of airflow obstruction and quality of life in subjects with COPD. CT-derived biomechanical features performed well in assessing the presence and severity of COPD.

Section snippets

COPDGene Data

A database of 162 subjects with varying distribution of COPD severity and nonsmoker subjects without COPD were used in this study. All the subjects were approved by the institutional review boards and provided written consent for participation in the study. The distribution of the subjects was: 27 nonsmokers, 30 Global Initiative for Chronic Obstructive Lung Disease (GOLD)0, 29 GOLD1, 29 GOLD2, 28 GOLD3, and 19 GOLD4. These subjects were selected from the Iowa Cohort of the COPDGene study. The

Correlation Results

Pearson's linear correlation coefficient, r, and corresponding P values were calculated for the PFT measurements and the optimal set of CT-based features from each feature set. Spearman's correlation coefficient, ρ, was reported to find the correlation between SGRQ total score and CT-based features. The results are shown in Table 2 and all the correlations are statistically significant, P < .001. The air-trapping measure in the density-based feature set showed a strong negative correlation with

Discussion

The two classification experiments conducted in this study show that the CT-derived estimates of regional lung tissue biomechanics serve as useful quantitative imaging metrics to characterize patients with COPD. The relationship between the CT-derived biomechanical features and clinical diagnostic measures (PFT measurements, SGRQ scores) are shown in Table 2. The Jacobian features, which capture local volume changes, have shown strong correlations with FEV1% predicted (r = 0.80) and FEV1/FVC

Conclusion

This study constructed CT-derived density, texture, and biomechanical feature sets and showed these individual feature sets had statistically significant correlations with measures of airflow obstruction and patient quality of life in subjects with COPD. It also showed that these feature sets could be used individually and in combination to classify subjects with and without COPD and to classify subjects by their severity of COPD using the GOLD staging system for COPD. The biomechanical feature

Acknowledgments

The authors thank the COPDGene research group for providing the data required for this study. This work was supported by Grants HL079406 and HL064368 from the National Institutes of Health. E.A.H. and J.M.R. are founders and shareholders of VIDA Diagnostics, Inc., and J.D.N. is a paid consultant for VIDA Diagnostics Inc.

References (41)

  • Y. Yin et al.

    Mass preserving nonrigid registration of CT lung images using cubic B-spline

    Med Phys

    (2009)
  • E. Spaggiari et al.

    Early smoking-induced lung lesions in asymptomatic subjects. Correlations between high resolution dynamic CT and pulmonary function testing

    La Radiol Med

    (2005)
  • P.A. Gevenois et al.

    Comparison of computed density and macroscopic morphometry in pulmonary emphysema

    Am J Resp Crit Care Med

    (1995)
  • G.A. Gould et al.

    CT measurements of lung density in life can quantitate distal airspace enlargement—an essential defining feature of human emphysema

    Am Rev Respir Dis

    (1988)
  • G.A. Gould et al.

    Lung CT density correlates with measurements of airflow limitation and the diffusing capacity

    Eur Respir J

    (1991)
  • K.L. Boedeker et al.

    Emphysema: effect of reconstruction algorithm on CT imaging measures

    Radiology

    (2004)
  • J.P. Sieren et al.

    Reference standard and statistical model for intersite and temporal comparisons of CT attenuation in a multicenter quantitative lung study

    Med Phys

    (2012)
  • R. Uppaluri et al.

    Computer recognition of regional lung disease patterns

    Am J Resp Crit Care Med

    (1999)
  • R. Uppaluri et al.

    Interstitial lung disease: a quantitative study using the adaptive multiple feature method

    Am J Resp Crit Care Med

    (1999)
  • L. Sorensen et al.

    Texture-based analysis of COPD: a data-driven approach

    IEEE Trans Med Imaging

    (2012)
  • Cited by (54)

    • A Novel CT-Based Radiomics Features Analysis for Identification and Severity Staging of COPD

      2022, Academic Radiology
      Citation Excerpt :

      Radiomic features, roughly subdivided into statistical, including histogram-based and texture-based; transform-based; and shape-based, can capture tissue and lesion characteristics such as heterogeneity. For example, image texture analysis, which uses more information than densitometry, including the spatial relationships between different intensities on the image, has been proposed as a possibly stronger alternative for emphysema quantification (26-27) that may in turn lead to improvements in diagnosing of COPD. We believe that the impact of radiomics is increased when the wealth of information that it provides—typically hundreds, or even thousands of features, a fraction of which will contribute to a disease-specific radiomic signature—is processed using machine learning techniques.

    • Quantitative CT metrics are associated with longitudinal lung function decline and future asthma exacerbations: Results from SARP-3

      2021, Journal of Allergy and Clinical Immunology
      Citation Excerpt :

      Finally, in this study we demonstrated that lung deformation measurements (Jacobian determinant values), as well as lung deformation spatial heterogenity (Jacobian gradient values), were independently associated with several asthma-related outcomes. The Jacobian determinant is a biomechanical measurement calculated from image coregistration and reflects the local volume change that occurs at a point from expiration to inspiration.56-58 Although lung deformation measurements are less intuitive, Jacobian measurements have previously been found to be of importance in both COPD and asthma.56,57,59-62

    • MRI and CT lung biomarkers: Towards an in vivo understanding of lung biomechanics

      2019, Clinical Biomechanics
      Citation Excerpt :

      Choi et al., 2017) In addition, airway measurements such as total airway length and branch count made using CT have been correlated with pulmonary function tests of lung biomechanics (Karayama et al., 2017; Pu et al., 2012). Using multi-volume CT, Jacobian biomarkers of lung deformation have been evaluated in patients with obstructive lung disease including severe asthmatics (Choi et al., 2013; Jahani et al., 2017) and COPD participants in the COPDgene study (Bhatt et al., 2017; Bodduluri et al., 2013; Bodduluri et al., 2017). Differences and abnormal lung biomechanical properties have been quantified in severe asthmatics including a diminished volume change during breathing, suggestive of gas-trapping (Choi et al., 2013).

    View all citing articles on Scopus
    View full text