high performance computing on graphics processing units: hgpu.org

In order to accelerate imaging process and lower overall system cost, many practical CT systems use single slice detectors and big-pitch helical scanning, especially in next generation Baggage Scanning system. In this paper, we present a reconstruction strategy for this kind of systems. We researched on the imaging qualities resulting from different scanning parameters, especially the thickness of the detectors. The experimental study is both on simulation data and experimental data. We found out that reconstruction results from relatively thicker detectors are better than those from thinner detectors. We also compared post-processing methods of single dimensional smoothing and multi-dimensional smoothing in improving the imaging quality for such resulting reconstructions from single-slice-detector helical CT (SSDHCT). Finally, an implementation scheme of the reconstruction algorithm and post-processing using the latest GPU is presented. We achieve a 10 times speedup and the performance meets the requirement for practical applications. At last, we produced a sample machine using this technique, and get some real baggage scanning result. Our method works effectively in reality applications.