High Performance Tomography

Computed Tomography (CT) Image Reconstruction is an important technique used in a wide range of applications, ranging from explosive detection, medical imaging to scientific imaging. Among available reconstruction methods, Model Based Iterative Reconstruction (MBIR) produces higher quality images and allows for the use of more general CT scanner geometries than is possible with more commonly used methods. The high computational cost of MBIR, however, often makes it impractical in applications for which it would otherwise be ideal. This dissertation describes the concept of a super-voxel (SV) that significantly reduces the computational cost of MBIR while retaining its benefits. It describes how scanner data can be organized into super- voxels (SV) that, dramatically increases locality and prefetching, regularizes data access pattern, enables massive parallelism, and ensures fast convergence. Results indicate that the super-voxel algorithm has an average speedup of 9776 compared to the fastest state-of-the-art 3D image reconstruction implementation on a 69632-core distributed system