Reconstruction of the Human Airways - Computational Fluid and Particle Dynamics in the Human Respiratory System - page 66

Fig. 3.16 Adaptive
cylindrical regions of interest
(numbered grey cylinders)
that follows the airway tree
branches as the fuzzy
connectivity segmentation
progresses. This leads to
faster segmentation time as
the segmentation process is
kept close to the airway and
fewer background voxels
need be analysed. In addition
any segmentation leaks can
be detected and dealt with
early segmentation leakage
After segmentation the model can be smoothed to reduce any artificial regions
created due to the effects from noise found in the scanned images. In addition, if
the model is too rough it will create problems during the surface fitting and meshing
process. On the other hand if it is too smooth some important topological features
will be omitted. The outputted STL model from the particular segmentation software
used lacked geometric topological relations for CFD meshing. The model needs to
go through CAD-based modelling process which is a reverse-engineering process.
The STL model is full of triangulated faces which have a NURBS surface definition
applied to it. 1,084 NURBS patches were generated on the airway tree and the surface
model is saved as IGES file (Fig. 3.17 ).
CAD-based Method: An alternative method for airway reconstruction is to use ex-
isting data available in the literature. A widely used model is Weibel's symmetric
model from Weibel (1963) with an updated set of data given in Weibel (Weibel 1997).
Fig. 3.17 Tracheobronchial airway tree with NURBS surface model applied from a triangulated
STL file