【摘要】 本文提出一种基于径向基函数(Radial Basis Function—RBF)神经网络图像表示实现对计算机断层成像装置不完备投影数据的重建方法,并分析了方法的计算效率、重建质量和适用范围.该方法采用RBF神经网络表示断层图像,降低了问题的计算规模,并通过ART(Algebraic Reconstruction Technique)迭代算法重建出断层图像.在模拟实验中,我们将本方法与FBP、ART的重建图像算法进行了比较.实验结果表明所提出的方法其重建质量和计算效率都有明显地改进.更多还原

【Abstract】 A method based on RBF (Radial Basis Function) neural network image representation is proposed for the computerized tomography image reconstruction from a small amount of projection data,and the computation efficiency,reconstruction quantity as well as applicable range of the proposed method are analyzed.In this method,the cross-sectional image is represented by a RBF network and reconstructed by computing the network’s weight matrix with the ART (Algebraic Reconstruction Technique).In order to evaluate the proposed technique,we experimentally compared it with the ART and the FBP (Filtered Back Projection).Results show the method has obviously improved both in reconstruction quantity and computation efficiency.更多还原