Simulation Of Back-Projection For The Helical Cone-Beam CT 3dD Image Reconstruction And Design Of Data Channel

With the rapid development of the modern medical science and image processing technology, the spiral cone-beam CT (Computed Tomography) is playing increasingly significant role in the field of clinic due to its reconstructed image has advantages of fast scan speed, high spatial resolution and others. Image reconstruction is one of key techniques of CT scanner, because the selected algorithm and its implementation determine the execution time and the quality of the reconstructed image to a large extent, thereby affecting the application of CT machines.But the filtered back-projection algorithm for reconstruction has complex formula, intense computations, especially is a bottleneck in terms of speed and is difficult to hardware implement.Since the back-projection part is the most time-consuming step, two optimizations are made for back-projection formula in this thesis: one is combining the ray traverse algorithm to propose an improved back-projection algorithm, so that reading each filtered projection data is reduced to only once from O ( N×S) times (there are voxels on each PI line, and there are PI lines covering the source position) to update all the dedicated voxels. This structure saves a great many of memories, facilitates parallel processing, greatly reduces the time and complexity of memory operations, and tends to hardware design; the other is applying bi-linear interpolation instead of nearest neighbor interpolation, which effectively eliminates some artifacts and improves the image quality.On the other hand, the proposed SystemC modeling of the improved back-projection algorithm in this thesis, from the perspective of optimizing future hardware design, reasonably divides into sub-modules, achieves a better acceleration effect, and makes adequate preparation for hardware implementation; also, pipeline can be applied to this structure, which affords certain reference for efficiently improving the reconstruction speed by parallel processing. In addition, it is this research project--the FPGA hardware acceleration system of spiral cone-beam CT 3D reconstruction that aims to improve the reconstruction speed and quality. The CT reconstruction controller designed in this thesis is able to achieve reading and writing operations of the projections data, meet the requirements of transmission speed in the process of hardware acceleration, and supply excellent conditions for the overall realization of real-time image reconstruction.