Generating Labeled Voxelizations for Numerical Simulation

Recent advances in imaging systems such as MRI, CT, and PET enable rapid acquisition of biological data. These data are suitable for a variety of simulation techniques, including characterizing the response of biological systems to directed energy. However, these simulations are hampered by the challenge of converting the data into voxelized models. The data requires segmentation to demarcate various biological structures into valid, labeled regions where each voxel is associated with a particular biological feature. In the proposed work we investigate various data forms and segmentation techniques to perform this voxelization, with the goal of creating a software application for automatic and/or semiautomatic voxel labeling. We expect to develop these tools in an open source framework with the ultimate goal of commercializing these techniques by embedding them in Kitware's VolView volume visualization system.