Automatic finite elements mesh generation from planar contours of the brain: an image driven 'blobby' approach

Abstract

In this paper, we address the problem of automatic mesh generation for finite elements modeling of anatomical organs for which a volumetric data set is available. In the first step a set of characteristic outlines of the organ is defined manually or automatically within the volume. The outlines define the "key frames" that will guide the procedure of surface reconstruction. Then, based on this information, and along with organ surface curvature information extracted from the volume data, a 3D scalar field is generated. This field allows a 3D reconstruction of the organ: as an iso-surface model, using a marching cubes algorithm; or as a 3D mesh, using a grid "immersion" technique, the field value being used as the outside/inside test. The final reconstruction respects the various topological changes that occur within the organ, such as holes and branching elements.