Measurement, Evaluation and Analysis of Wall Thickness of 3D Airway Trees across Bifurcations

Airway segmentation and analysis is an essential step in the understanding of pulmonary anatomy and early detection of lung diseases. In this paper, we evaluate a newly developed method for segmenting the inner and outer surfaces of the airway tree. This is the first approach reported to validate the segmentation of the double wall surfaces for the entire airway tree including bifurcations. Two approaches help analyze the tree topology: (1) 3D thinning algorithm which produces a one-voxel wide centerline, and (2) Delaunay triangulation based generation of both the inner and outer 3D medial axes (sheets). The two airway wall surfaces are identified using a 3D optimal graph search approach. Once segmented, the bifurcation/carina areas are defined based on (1) the surface point proximity to the branch points identified on the centerline, and (2) the surface point distance from the 3D outer medial sheets. The wall thickness measurements for both the bifurcation/ nonbifurcation and carina/non-carina points are grouped and reported as a function of tree generation. The accuracy of wall thickness measurement is assessed on CT-scanned double-wall physical phantoms. The airway wall thickness accuracy is similar for the bifurcation/carina and nonbifurcation/carina areas. The measurements on normal human in vivo data reflect reasonable and consistent differences of wall thickness from generation to generation.