Accelerating Image Reconstruction in Dual-Head PET System by GPU and Symmetry Properties

Abstract

Positron emission tomography (PET) is an important imaging modality in both clinical usage and research studies. We have developed a compact high-sensitivity PET system that consisted of two large-area panel PET detector heads, which produce more than 224 million lines of response and thus request dramatic computational demands. In this work, we employed a state-of-the-art graphics processing unit (GPU), NVIDIA Tesla C2070, to yield an efficient reconstruction process. Our approaches ingeniously integrate the distinguished features of the symmetry properties of the imaging system and GPU architectures, including block/warp/thread assignments and effective memory usage, to accelerate the computations for ordered subset expectation maximization (OSEM) image reconstruction. The OSEM reconstruction algorithms were implemented employing both CPU-based and GPU-based codes, and their computational performance was quantitatively analyzed and compared. The results showed that the GPU-accelerated scheme can drastically reduce the reconstruction time and thus can largely expand the applicability of the dual-head PET system.

Comparison of GPU reconstruction based on different symmetries for dual‐head PET

“…Based on LOR symmetry, the time cost of the backprojection is larger than that of the forward projection as indicated in Table . Similarly, as addressed by Chou, the acceleration ratio factor for forward projection is also slightly reduced to ~120 when comparing the GPU with CPU, but the acceleration ratio for the backprojection remains essentially identical at ~36 . The voxel‐based symmetry has been used to reduce the time of the MC simulation in SRM generation.…”

“…The LOR‐based GPU reconstruction has been investigated by Chou et al. in reference . Here, we describe it briefly.…”

“…Both of them are able to reduce the size of the SRM. For the second one, the Graphics Processing Unit (GPU) technology, which has great popularity in parallel computation, is suitable to accelerate the iterative reconstruction …”

“…For the dual‐head PET reconstruction, Chou et al . have proposed a GPU acceleration strategy based on the Compute Unified Device Architecture (CUDA).…”

“…For the second one, the Graphics Processing Unit (GPU) technology, which has great popularity in parallel computation, is suitable to accelerate the iterative reconstruction. 17,18 For the dual-head PET reconstruction, Chou et al 17 have proposed a GPU acceleration strategy based on the Compute Unified Device Architecture (CUDA). In their strategy, LORbased symmetry has been utilized.…”