High Quality Surface Mesh Generation for Multi-physics Bio-medical Simulations

Manual surface reconstruction is still an everyday practice in applications involving complex irregular domains necessary for modeling biological systems. Rapid development of biomedical imaging and simulation, however, requires automatic computations involving frequent re-meshing of (r)evolving domains that human-driven generation can simply no longer deliver. This bottleneck hinders the development of many applications of high social importance, like computational physiology or computer aided medicine. Therefore, automated meshing algorithms must be available before problems like interaction of blood flow with largely deforming vessel walls in the coronary arteries or multi-scale models of tumor growth can be efficiently studied. While many commercial packages offer mesh generation options, these usually depend on high quality input surface description, which is rarely available when depending on image segmentation results. We propose an efficient and relatively simple approach to automatically recover a high quality surface mesh from low-quality, oversampled and possibly non-consistent inputs that are often obtained via 3-D acquisition systems like magnetic resonance imaging (MRI), microscopy or laser scanning. The presented technique is particularly well suited to generate geometrical descriptions for multi-physics simulations over complex domains encountered in biomedical problem solving. As opposed to the majority of the established meshing techniques our procedure is easy to implement and very robust against damaged or partially incomplete, inconsistent or discontinuous inputs.