Tetrahedral Image-To-Mesh Conversion for Anatomical Modeling and Surgical Simulations

We present an Image-To-Mesh Conversion method for building a realistic biomechanical model particularly targeted for surgical simulations. Our implementation generates tetrahedral meshes that conform to the physical boundaries of multilabel segmented images. Our approach, initially creates a Body-Centered Cubic (BCC) lattice that is a coarse approximation of the object boundaries, and then subdivides the lattice using a red-green refinement strategy that guarantees the high quality of the new elements. In a later step, our method deforms the lattice surfaces to their corresponding tissue boundaries using a point-based registration scheme. As a result, the final mesh is smooth and accurately represents the object boundaries allowing a faithful response of the biomechanical properties of the tissues involved in a surgical simulation. Besides, the generated mesh is adaptive with smaller elements in areas where more detail is desired and larger elements in the remainder of the image regions. We evaluate our method qualitatively and quantitatively on isotropic and anisotropic segmented volumetric images. The described implementation will be available within two popular open source software: the 3D Slicer for visualization and image analysis, and the SOFA framework for real-time medical simulations.