Abstract
In the medical field, three-dimensional (3D) digital images are commonly used with recent advances of imaging techniques such as magnetic resonance imaging (MRI) and X-ray computed tomography. In most cases, objects are visualized in 3D shapes by surface rendering or volume rendering using the information of isosurfaces. However, objects of interest are frequently occluded by other objects or by themselves. Volume clipping is a common technique to reduce the occurrence of occlusion, but the procedure is difficult to apply for complicated shapes. The Region-based Contour Tree (RBCT) describes a structure of isosurfaces in digital images as a tree structure. Using RBCT, we can extract a region surrounded by a single isosurface, and the extracted region is not occluded by other regions. Nevertheless, self-occlusion cannot be eliminated. In order to eliminate self-occlusion, we propose a method to divide isosurfaces using RBCT. We introduce a “region-dividing plane” that divides a region into two parts: the part of interest and remainder, which is hidden by self-occlusion. Using RBCT interactively combined with the region-dividing, we can efficiently extract regions of interest from a 3D image without self-occlusion even if the shapes of regions are complicated.
