基于神经网络的少数投影CT图像重建算法

An Image Reconstruction Algorithm of Computer Tomography from Fewer Views Based on a Neural Network

基于连续型 Hopfield神经网络 (HNN)提出了一种新的少数投影 CT图像重建算法。首先将图像重建问题转化为最大熵准则下的优化问题 ,然后构造相应的连续型 HNN模型 ,网络的稳定解就是重建问题的优化解。数值模拟结果表明 ,与联合代数重建技术 (SART)相比 ,利用从投影角间距为 4 5°的 4个不同的投影视角方向获取的投影数据 ,该算法具有较高的重建精度和良好的抗噪声性能。

A new image reconstruction algorithm of computer tomography from fewer views based on the continuous Hopfield neural network (HNN) is presented.The problem of image reconstruction is formulated as optimization under the criterion of maximum entropy,and a continuous HNN is then constructed to solve the optimization problem.Compared with the simultaneous algebraic reconstruction technique (SART),the results of numerical simulation show that this technique using the projection data obtained from 4 views with the projection angles 45° apart has fairly high accuracy and good stability against noise.