Accelerating Algebraic Reconstruction Using CUDA-Enabled GPU

Yuqiang Lu, Weiming Wang, Shifu Chen, Yongming Xie, Jing Qin, Wai-Man Pang, Pheng-Ann Heng

Abstract: In this paper, we apply the compute unified device architecture (CUDA) to the 3D cone-beam CT reconstruction using simultaneous algebraic reconstruction technique (SART). With the hardware acceleration, the computationally complex SART can run at speed comparable to the commonly used filtered back-projection, and provide even better quality volume with less samples. The main contributions include two novel techniques to accelerate the reconstruction. We introduce a ray-driven projection along with hardware built-in trilinear interpolation, as well as a voxel-driven back-projection that can avoid redundant computation by combining CUDA shared memory. Significant performance boost is reported from experiments using our techniques. A real-time reconstruction is achieved within 3 seconds for a 128 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sup> volume from 80 128 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> projections, without compromising image quality. Our proposed method realizes the instantaneous presentation of CT volume to the physician once projection images are acquired.