Size variation and collapse of emphysema holes at inspiration and expiration CT scan: evaluation with modified length scale method and image co-registration

Abstract

A novel approach of size-based emphysema clustering has been developed, and the size variation and collapse of holes in emphysema clusters are evaluated at inspiratory and expiratory computed tomography (CT). Thirty patients were visually evaluated for the size-based emphysema clustering technique and a total of 72 patients were evaluated for analyzing collapse of the emphysema hole in this study. A new approach for the size differentiation of emphysema holes was developed using the length scale, Gaussian low-pass filtering, and iteration approach. Then, the volumetric CT results of the emphysema patients were analyzed using the new method, and deformable registration was carried out between inspiratory and expiratory CT. Blind visual evaluations of EI by two readers had significant correlations with the classification using the size-based emphysema clustering method (r-values of reader 1: 0.186, 0.890, 0.915, and 0.941; reader 2: 0.540, 0.667, 0.919, and 0.942). The results of collapse of emphysema holes using deformable registration were compared with the pulmonary function test (PFT) parameters using the Pearson's correlation test. The mean extents of low-attenuation area (LAA), E1 (<1.5 mm), E2 (<7 mm), E3 (<15 mm), and E4 (≥15 mm) were 25.9%, 3.0%, 11.4%, 7.6%, and 3.9%, respectively, at the inspiratory CT, and 15.3%, 1.4%, 6.9%, 4.3%, and 2.6%, respectively at the expiratory CT. The extents of LAA, E2, E3, and E4 were found to be significantly correlated with the PFT parameters (r=-0.53, -0.43, -0.48, and -0.25), with forced expiratory volume in 1 second (FEV₁; -0.81, -0.62, -0.75, and -0.40), and with diffusing capacity of the lungs for carbon monoxide (cDLco), respectively. The fraction of emphysema that shifted to the smaller subgroup showed a significant correlation with FEV₁, cDLco, forced expiratory flow at 25%-75% of forced vital capacity, and residual volume (RV)/total lung capacity (r=0.56, 0.73, 0.40, and -0.58). A detailed assessment of the size variation and collapse of emphysema holes may be useful for understanding the dynamic collapse of emphysema and its functional relation.

Keywords: chronic obstructive pulmonary disease; computed tomography; emphysema size; emphysema size change; length scale analysis; quantitative imaging.

Figure 1

Flow diagram of size-based emphysema clustering using length scale analysis. Abbreviations: CT, computed tomography; LPF, low-pass filter.

Figure 2

Results of the size-based emphysema clusters of COPD patients. Notes: (A) Volumetric chest CT; (B) emphysema index; and (C) size-based emphysema cluster. Abbreviations: CT, computed tomography; COPD, chronic obstructive pulmonary disease.

Figure 3

Process of the deformation of emphysema clusters according to the specific sizes (<1.5, <7, <15 and >15 mm) by using co-registration information. Abbreviation: CT, computed tomography.

Figure 4

Iterative process of size-based emphysema clustering. Notes: (A) 1st line: results of the emphysema mask by >15 mm; 2nd line: results of the emphysema mask by <15 mm; 3rd line: results of the emphysema mask by <7 mm; 4th line: results of the emphysema mask by <1.5 mm. (B) Results of 4 subgroups’ emphysema mask. Abbreviation: CT, computed tomography.

Figure 5

Results of the artificial phantom study. Notes: Diameter of the phantom sphere: (A–F) 4, 6, 8, 10, 12, 14 mm; (G–H) 6, 8, 10, 12, 14 mm; color of phantom sphere: 4 mm cyan, 6 mm magenta, 8 mm yellow, 10 mm blue, 12 mm green, 14 mm red; nonattached phantom spheres, classified nonattached phantom spheres with predefined color; slightly attached phantom spheres (0 mm [0 voxel] interval in the x-axis direction), classified slightly attached phantom spheres with predefined color; strongly attached phantom spheres (2 mm [3 voxel] intervals in the x-axis direction), classified strongly attached phantom spheres with predefined color; complex attached phantom spheres (1.35 mm [2 voxel], 1.35 mm [2 voxel], and 0 mm [0 voxel] intervals in the x-, y-, and z-axis directions), classified complex attached phantom spheres with predefined color.

Figure 6

Differentiation of the changed emphysema holes by subgrouping. Notes: The first row shows the mean non-emphysema (light gray) and emphysema fractions (dark gray) in the inspiratory CT of the study population. The second row shows the changed fraction of each volume in the first row. The third column of the second row (sky blue) shows the shifting component portion to the smaller emphysema holes (mid-gray). The next left column (mid-gray color) shows the no-change portion of the emphysema holes, and the last column of the second row (orange) shows the increased portion of the emphysema holes. The third row shows the emphysema subgroup (the dashed line indicates the identical portion of the second row), and the fourth row shows the changed group of each subgroup. The fifth row represents the decreasing and increasing of each sub-group. (eg, the first blue, green, and red columns of the fifth row show the decreased, no-change, and increased portion of subgroup E1, respectively). Abbreviation: CT, computed tomography.