电阻抗成像正则化算法的优化  

Optimization of Electrical Impedance Tomography Regularization Algorithm

作　　者：李冬晔 康彬[2] 

机构地区：[1]南京邮电大学电子科学与工程学院,江苏南京210003 [2]南京邮电大学信号处理与传输研究院,江苏南京210003

出　　处：《计算机技术与发展》2016年第5期188-190,196,共4页Computer Technology and Development

基　　金：国家自然科学基金资助项目(60671065)

摘　　要：电阻抗成像技术是一种以人体内部电导率分布为成像目标的医学成像技术。该技术具有非入侵、无损伤、实时成像、系统结构简单、设备价格较低等优点。但其成像反问题的求解中存在严重的不适定性,和主流医学成像技术相比,其成像分辨率不高。针对电阻抗成像中传统Tikhonov正则化分辨率低及边界模糊的问题,文中将归一化测量电压的总变差作为正则化函数,根据电阻抗图像中非均匀的电导率具有稀疏性的特点,提出一种新颖的电阻抗成像的正则化优化算法。电导率分布的求解是通过两步迭代阈值算法(Tw IST)完成。通过对模拟域的不同部位及域内数量不等的成像目标进行仿真,测试了该算法的性能。计算机仿真结果表明,所研究的正则化优化算法适用于生物医学成像应用,它可显著改善成像质量和边界分辨率。Electrical Impedance Technology（ EIT） is a kind of medical imaging technology taking the human＇s conductivity distribution as the target. There are lots of advantages belonged to this technique such as non- invasive,no damage,real- time imaging,simple system structure and lower price. But there are still serious ill- posed in solving its inverse problem. So compared with the mainstream medical imaging technology,its imaging resolution is not good. Aiming at solving the lowresolution and blurry boundaries in electrical impedance tomography where the traditional Tikhonov regularization is used,a newoptimization regularization algorithm is proposed in this paper. It takes the total variation of the normalized measured voltage as the regularization function and takes account of the inhomogeneous conductivity ＇s sparse characteristics in electrical impedance image. The Two- step Iterative Shrinkage / Threshold（ TwIST） algorithm is considered to solve the conductivity distribution. The performance of this newalgorithm is tested through the different parts and different numbers of imaging targets in the analog domain. The simulation shows that the studied regularization algorithm for biomedical imaging application can significantly improve the image quality and boundary resolution.

关 键 词：电阻抗成像 正则化 反问题 两步迭代阈值算法 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]