Parallel Regional Growth Marching Cubes: Efficient Surface Reconstruction Based On Mpi

Xiaohui Wei,Xinyan Bao,Xiaoli Pang,Haolong Cui
DOI: https://doi.org/10.1007/978-3-319-71589-6_39
2017-01-01
Abstract:Three-dimensional surface reconstruction from medical images is an important task in clinical medical practice. Of all the methods for extracting surfaces from scalar fields, marching cubes algorithm is most popular due to its simplicity. In this contribution, we developed a robust and efficient algorithm for reconstructing surfaces from tomography images such as CT and MRI, in which we combine the regional growth marching cubes algorithm with message passing interface to highly parallelize the extracting process. The scalar field is first partitioned into parts and for each part, a process is created to run the regional growth marching cubes on the part. The parallel processing is implemented using MPI. Experiments show that the method has a higher efficiency in time than both marching cubes and regional growth marching cubes algorithms.
What problem does this paper attempt to address?