Implementation of Exact Reconstruction Algorithm of Helical Cone-Beam CT on Cell Broadband Engine Architecture

Jin Yannan He,Qing Han,Zhenxing Cao Liji

2007-01-01

Abstract:Helical cone-beam ct is an important medical imaging modality for its high longitudinal resolution and efficient use of radiation dose.The data acquisition system consists of an x-ray source and a planar detector attached to a rotating gantry while the patient is translated steadily through the gantry. Therefore, from the perspective of the patient, the trajectory of the x-ray source forms a helix. The conebeam x-rays generated by x-ray source pass through the patient and the attenuation signals are recorded by the planar detector on the other side of the patient. The task of ct reconstruction is to recover the 3d images from the projection data. Katsevich algorithm[1], proposed in 2002, was the first exact reconstruction algorithm for helical ct. It is a major breakthrough in the development of ct reconstruction algorithm. However, the intensive computation required by this algorithm prohibits its clinical use, especially when the amount of projection data becomes increasingly large. Parallel computing can be an effective approach to resolving the problem of heavy computation burden. Cell broadband engine architecture (cbea), a novel computing architecture, is a very promising mode for this purpose. The cell processor offers an immense computing power due to its high level of parallelism, its high clock rate and its enormous internal communication bandwidth. Moreover, it can be programmed using high level programming languages. In this work, we implement the exact bpf (backprojection-filtration) reconstruction algorithm[2] on a cell/ps3 andperformnumerical simulations. Weparallelize and optimize the exact bpf reconstruction algorithm to accelerate the imaging process. The simulation results demonstrate a significant reduction of reconstruction time and indicate the potential of cbea’s application in medical image reconstruction.