Use of CUDA with high-end graphics card for CT performance improvements

Abstract

Computed Tomography has a long history of about 40 years as the first non-destructive investigation method that overcomes the drawbacks of conventional X-ray imaging. Nowadays, it is widely used in a wide variety of applications in several fields and industries such as healthcare, manufacturing, etc. As these applications require interactive 3D image generation and real-time feedback, the traditional host-based reconstruction method is unable to meet such demands due to the high computational cost of the algorithm. In this project, improvement is proposed to the traditional host-based reconstruction program by exploring the Compute Unified Device Architecture (CUDA) from NVIDIA.

In order to design and develop a reconstruction module which is fast enough at reasonable cost, some research has been done on CUDA and similar effort to speed up the process. The strengths and weaknesses of CUDA in comparison to other alternatives were used to evaluate the feasibility and refine the objective of this project.