Surface Segmentation: The Case of Bronchus Anatomical Structure

Surface Segmentation: The Case of Bronchus Anatomical Structure

S. Zimeras (University of the Aegean, Greece)
DOI: 10.4018/978-1-5225-1724-5.ch009


Segmentation is a powerful procedure that could be used to extract relevant information of the images based on advanced techniques (like active contours, region growing, Markov random fields, and medical atlas analysis). For the procedures, the main task is the contour, or volume or surface representation of specific parts of the organs that could be used for the benefit of the patients under doctor evaluation. So, in real cases, the proposed process must be quick, accurate and easy to implement. The segmentation of the organ is another problem that must be considered. More complicated, more demanding the segmentation process. In our case (bronchus segmentation) a quick, effective and easy to implement procedure is proposed based on the combination of boundary tracking and region growing techniques.
Chapter Preview


Based on the bronchus morphology, the segmentation procedure includes linear interpolation or B-splines between gap contour reconstructions. The common procedure is the linear interpolation where in the plane of slice i, the surface created between slice pairs 1 - i and I will usually not agree in the normal surface with the surface created between slices i and 1 + i. To avoid discontinuities in the normal surface, more than just two slices, at a given time, must be considered.

Assume that we begin with k sets of constraints, one set for each 2D data slice. Instead of considering the contours in pairs, we place the constraints for all of the k slices into 3D simultaneously. Specifically, the constraints of slices i are placed in the plane z = si, where s is the spacing between planes. Once the constraints from all slices have been placed in 3D, we invoke the implicit function interpolation once to create a single implicit function in 3D for the complete set of contours (Karangelis, 2004).

Figure 1.

Implicit surface interpolation

In cases where key contours between single contour shapes and split contours is usually not an option, advanced interpolation methods are required to fill the missing gaps. The main challenge for the contour interpolation of surface reconstruction algorithm is how to handle the bifurcation problem. For the implicit surfaces, the definition of bifurcation is a property that can be handled with flexibility. The degree of bifurcation is highly related to the degree of continuity of the implicit functions and the distance between the single and contour pairs.

Complete Chapter List

Search this Book: