GPU-accelerated SART reconstruction using the CUDA programming environment

Abstract

The Common Unified Device Architecture (CUDA) introduced in 2007 by NVIDIA is a recent programming model making use of the unified shader design of the most recent graphics processing units (GPUs). The programming interface allows algorithm implementation using standard C language along with a few extensions without any knowledge about graphics programming using OpenGL, DirectX, and shading languages.We apply this novel technology to the Simultaneous Algebraic Reconstruction Technique (SART), which is an advanced iterative image reconstruction method in cone-beam CT. So far, the computational complexity of this algorithm has prohibited its use in most medical applications. However, since today's GPUs provide a high level of parallelism and are highly cost-efficient processors, they are predestinated for performing the iterative reconstruction according to medical requirements.In this paper we present an efficient implementation of the most time-consuming parts of the iterative reconstruction algorithm: forward-and back-projection. We also explain the required strategy to parallelize the algorithm for the CUDA 1.1 and CUDA 2.0 architecture. Furthermore, our implementation introduces an acceleration technique for the reconstruction compared to a standard SART implementation on the GPU using CUDA. Thus, we present an implementation that can be used in a time-critical clinical environment.Finally, we compare our results to the current applications on multi-core workstations, with respect to both reconstruction speed and (dis-)advantages. Our implementation exhibits a speed-up of more than 64 compared to a state-of-the-art CPU using hardware-accelerated texture interpolation.

Cache-aware GPU memory scheduling scheme for CT back-projection

“…The inputs were 364 1024×768 projections on a half circal trajectory. We reconstructed a 3D volume within the FOV with two different resolution, 256 3 and 512 3 . We tested the different running times for the Single projection method (S), Multiple projection method (M) and Hybrid ordering method (H).…”

“…al. [3] used this method in CUDA-based SART. This method is better when cache-miss penalty is larger than memory writing overhead.…”

“…2) are mapped to thread blocks in CUDA. This volume-to-slabs decomposition has been used in [2][3][4]. Another type of mapping is based on decomposing the volume into horizontal tiles and each tile is mapped to a CUDA thread block, as used in [5], [6].…”

“…2) to ensure better cache hit-rate. Back-projection implementations [1][2][3][4][5][6] all use texture memory for projection data storage to deal with irregular memory fetching pattern. Using texture memory has several advantages.…”

Iterative denoising algorithms for perfusion C-arm CT with a rapid scanning protocol

“…Equation 1 is solved using Algorithm 1 by alternately minimizing the data consistency term Ax − p 2 and the regularization term R (x) for a fixed number of iterations N ART . In step 3 of Algorithm 1 data consistency is enforced by applying three iterations of the GPU-based Ordered Subsets-ART (OS-ART) method presented in [9]. In step 4 prior knowledge about the reconstructed volume is incorporated by applying operator T to the current volume estimation to reduce the penalty term R(x).…”