A triangle surface mesh generated from a 3D well-composed image of a human brain.
Emerging technologies on image-guided procedures, interventions, and 3D modeling are all benefiting from a class of algorithms on digital topology and computational geometry that generates well-composed, high-quality 3D meshes suited for next-generation applications in 3D imaging. A well-composed 3D boundary from a given binary 3D volume satisfy a manifold property and is represented by a topological 3D surface. This property enables us to compute differential quantities, such as 3D surface curvature or Jacobian (a local volume measurement of shrinkage or expansion) directly from the voxel data. In particular, 3D models of guaranteed high-quality from 3D imaging can be more robustly computed if these images are well-composed, below is an example of a well-composed 3D brain model.
Segmentation algorithms generate 3D binary images from 3D medical scanner devices, e.g., MRI imaging systems and by itself cannot guarantee well-composedness of a 3D surface model generated by using triangulation algorithms. Thus, a segmented 3D volume needs to be transformed into a well-composed 3D binary volume before a model generation step (i.e., triangulation) so it can be used more accurately and reliably for quantitative measurements derived from 3D models; for example, the computation of surface curvature and local volume changes. Basically, this transformation accounts for switching object-voxels to background voxels (or vice-versa) to satisfy its topological property.
Author(s): Marcelo Siqueira, Federal University of Rio Grande do Norte (UFRN) et. al.
March 2008, Journal of Mathematical Imaging and Vision. DOI: 10.1007/s10851-007-0054-1