不采用灵敏度矩阵的ECT图像迭代重建

Iterative reconstruction of ECT image without using sensitivity matrix

针对ECT图像重建质量有待提高的现状,研究ECT的正问题及逆问题.目前普遍采用灵敏度矩阵建立ECT模型,最常用的两种算法是基于灵敏度矩阵模型的线性反投影算法(简称LBP法)和基于灵敏度矩阵模型的投影LandWeber迭代法(简称PLI法).给出一种不采用灵敏度矩阵的ECT图像迭代重建新算法,用多元线性回归建立ECT系统正向模型、用投影LandWeber迭代法求解逆问题.采用ECT有限元分析软件,在多种介质分布下获得投影数据,对LBP法、PLI法和新算法进行了比较研究,研究结果表明,新算法能明显提高重建图像的质量.

Both positive problem and inverse problem of ECT were investigated to improve reconstruction quality of ECT image. At present, ECT models are usually established using sensitivity matrix. The linear back-projection algorithm based on sensitivity-matrix model （LBP） and the projected Landweber algorithm based on sensitivity-matrix model （PLI） are the most widely used algorithms. A new iterative reconstruction algorithm of ECT image without using sensitivity matrix was proposed. In this algorithm, the positive model of ECT system was established using multiple linear regression and the inverse problem was solved using projected Landweber iterative method. The projected data under various medium distributions were obtained using ECT finite element analysis software. The LBP, PLI and new algorithms were compared extensively. The results show that the new algorithm can obviously improve reconstruction quality of ECT image.