Affiliations: The CT Laboratory, School of Mathematical Sciences,
Capital Normal University, Beijing, China
Note: [] Corresponding author: Xing Zhao, The CT Laboratory, School of
Mathematical Sciences, Capital Normal University, Beijing, 100048, China. Tel.:
+86 10 68907074; E-mail: [email protected]
Abstract: As a whole process, we present a concept that the complete
reconstruction of CT image should include the computation part on GPUs and the
data storage part on hard disks. From this point of view, we propose a
Multi-Thread Scheduling (MTS) method to implement the 3D CT image
reconstruction such as using FDK algorithm, to trade off the computing and
storage time. In this method we use Multi-Threads to control GPUs and a
separate thread to accomplish data storage, so that we make the calculation and
data storage simultaneously. In addition, we use the 4-channel texture to
maintain symmetrical projection data in CUDA framework, which can reduce the
calculation time significantly. Numerical experiment shows that the time for
the whole process with our method is almost the same as the data storage time.
Keywords: FDK algorithm, complete reconstruction, Multi-Thread Scheduling, multi-GPU, large data
