An extension to 3D topological thinning method based on LUT for colon centerline extraction

Topological thinning is a valid but time-consuming method to calculate the centerline of human colon or other hollow organs accurately. An optimized 3D topological thinning method based on Look-up Table (LUT), which was proposed by Sadlier, proves to be effective in improving the efficiency on many occasions. However, it is still inefficient when processing some complex datasets. In this paper, we first analyze the reason causing the unstable performance, and then present an extension to Sadlier's method, which enables the rapid execution of the extraneous loops removing by avoiding unnecessary global connectivity testing. To reach this purpose, a min-heap structure is introduced to select a seed from the candidate voxels set of the final centerline, and region growing technique is used to find the voxels in the same branch with the seed. The comparison among the standard topological thinning, LUT method and the extension to LUT method indicates the extension achieves the most efficient performance.