Study On The Optimization Of Discrete Error In The Numerical Calculation For Spiral Cone-beam CT Image Reconstruction Algorithm

In the field of medical imaging, with the development of medical technology,cone-beam CT (Cone-Beam Computed Tomography, CBCT) imaging is widely applied,the reason is that it can reconstruct long objects, scan most area of objects and havehigh resolution for reconstruction image. Three-dimensional reconstruction Katsevichalgorithm is a classic reconstruction algorithm, and this algorithm can theoreticallyexactly reconstruct image, but in fact all of the projection data are discrete, we shoulduse discrete reconstruction formula, therefore it can lead to discrete error, so thereconstruction image will have severe artifacts, and this will limit the application ofexact reconstruction algorithm in the clinical operation.This paper is done by improving the discrete error of each module in the threedimensional reconstruction Katsevich algorithm, so then we will improve the imageprecision of the whole algorithm.This paper mainly finishes the following work:(1) This paper analysis discrete error of each module of the threedimensional reconstruction Katsevich algorithm, and a design scheme about reducingdiscretization error is proposed.(2) With the solution of the previous analysis, the modules of the threedimensional reconstruction Katsevich algorithm can be designed and implemented touse error optimization. Especially in the design and implementation of the backprojection module optimization, in order to ensure the location of the projection data ofeach frame, we should ensure the intersection the detector with the line passing throughradiographic source and x. With the different intersection, we use different interpolation,such as nearest neighbor interpolation, linear interpolation with weight of0.5or bilinearinterpolation with weight of0.25to select projection data, and finally we will completethe image reconstruction.(3) Each optimization module and the entire algorithm are simulated: first theimage of the algorithm with one optimization module is verified by the image of theordinary algorithm; then, we use MSE and histogram of reconstruction image to verifythe each optimization module; at last, the whole optimization algorithm, with MSE andhistogram of reconstruction image, are tested. In this paper, the MSE of reconstructionimage of the optimization algorithm is lost47%by the MSE of reconstruction image ofthe ordinary algorithm, and meets the task requirements of error optimization.