【作者】 徐管鑫；

【导师】 何为；

【作者基本信息】 重庆大学 ， 电气工程， 2004， 博士

【摘要】 电阻抗成像技术(Electrical Impedance Tomography,EIT)是当今生物医学工程学重大研究课题之一，它是继形态、结构成像之后，于最近二十年才出现的新一代无损伤成像技术。电阻抗成像技术通过配置于人体体表的电极阵列，提取与人体生理、病理状态相关的组织或器官的电特性信息，不但反映了解剖学结构，更重要的是可望给出功能性图像结果，这是X-CT、核磁共振等其它成像技术欠缺的。电阻抗成像技术不使用核素或射线、对人体无害、可以多次测量、重复使用，可以成为对病人进行长期、连续监护而不给病人造成损伤或带来不适的医院监护设备，加之其成本低廉，不要求特殊的工作环境等，因而是一种理想的、具有诱人应用前景的无损伤医学成像技术和图像监护技术。 论文以电阻抗成像技术为主要目标开展基础应用研究，研究实现一种实用、快速、具有一定分辨率的电阻抗成像算法，以为临床提供一种能进行连续监护、相对廉价的图像监护设备打下技术理论基础。论文在分析、研究国内外有关电阻抗成像重建算法及其研究状况的基础上，对电阻抗成像正问题、动态、静态电阻抗成像、三维电阻抗成像以及电阻抗成像技术在医学图像监护中的应用等方面进行了大量的仿真、实验研究。 在电阻抗成像正问题研究中，研究了场域内电导率变化位置、大小对边界电位的影响。基于电阻抗成像正问题研究结果，参与了无创脑血肿、水肿动态监护仪的研制，该仪器可以对脑血肿、水肿患者进行实时监护。目前该仪器已分别在重庆医科大学附一院和第三军医大学大坪医院获得应用，并已获得中华人民共和国国家知识产权局发明专利、实用新型专利、外观设计专利等五项专利权、《医疗器械注册证》[渝药管械(试)字2002第2040059号]、《医疗器械生产企业许可证》[渝药管械生产许可第20030081号]。 在动态电阻抗成像技术研究中，从X-CT的基本原理阐述了等位线反投影算法的原理，并从灵敏度、注入电流方式、电极数、模型单元数、抗噪声等方面对等位线反投影算法进行了系统深入的研究，给出一种性能相对较优的权系数W的计算公式。在静态电阻抗成像技术研究中，基于NOSER算法思想提出一步牛顿法，并对该方法的实现过程及其图像重建性能进行了详细的研究，给出了牛顿迭代法中场域电导率初值的快速确定方法。 在颅内血肿动态监护仿真研究中，基于CT切片建立了人体头部几何模型，并基于该模型对脑血肿病例颅内出血过程进行了动态实时监测的仿真研究。 基于论文理论研究结果，研制出16电极实时电阻抗成像系统实验平台(样机、重庆大学博士学位论文成像系统软件)，并在盛有盐水的水槽上进行实验，实验表明该成像系统对目标定位准确、成像速度较快(4秒一幅图像)、具有较高的分辨率。还在人体胸腔上进行了实时电阻抗成像，重建的图像完全能够分辨出人体胸腔中的左、右肺和心脏，但图像分辨率还不够理想，有待进一步提高。在利用国外人体实测数据的图像重建研究表明，重建图像能够分辨出胸腔(心脏、左右肺)及小腿的生理结构特征，重建算法是有效的，具有一定的临床参考价值。论文还对水槽实验和人体实验的测量数据进行分析，分析结果表明从测量数据的变化情况可以反映水槽内部物体的相对大小及位置，本论文的硬件测量系统具有较好的稳定性，其精度基本能满足水槽实验，但要实现人体阻抗成像，需要提高测量系统的精度和速度; 在三维电阻抗成像技术研究中，基于三维圆柱体有限元模型，将等位线反投影法成功推广应用于三维电阻抗成像上，提出等位面反投影法。该方法重建的图像定位准确，从重建图像可以分辨出成像目标个数，而且成像速快，通过对B矩阵的预处理，成像速度可达到20幅/秒，该速度下完全可以做到实时动态成像。 最后对论文的研究工作进行了总结，给出了所完成的主要研究工作及取得的成果，指出了目前存在的问题，提出了一些不足之处及对本课题今后的研究展望。关键词:电阻抗成像，图像监护，等位线反投影，一步牛顿法，三维更多还原

【Abstract】 Electrical Impedance Tomography (EIT), which is one of the significant research .. projects in current biomedicine engineering, is an effective noninvasive imaging technique following the techniques of shape imaging and structure imaging. EIT is a relatively new medical imaging modality that produces images by computing electrical properties within the human body. Sinusoidal electrical currents are applied to a subject’s body using electrodes attached to the skin, and the voltages that are developed on the electrodes are measured. Reconstruction algorithms use the applied electrical current and the measured electrode voltages to compute the electrical conductivity (or resistivity) distributions in the body. The reconstructed images can give not only the anatomy structure, but also the functional image. The X-CT, MRI and other imaging techniques can’t compare this with EIT. EIT is a noninvasive imaging technique as it doesn’t use nuclide and radial. It can be used to image repeatedly and monitor the patient without any damnification and discomfort for a long time. The EIT device is cheap and can work in any environment. EIT is an ideal and seductive noninvasive imaging technique in biomedical engineering.This dissertation aims at EIT to study its basic application and realize a practical, fast EIT algorithm with better resolution. It provides the imaging monitoring device with the basic theory featuring real-time monitoring and cheapness. Based on the study and analysis of the domestic and overseas EIT algorithms, simulations and experiments on EIT was studied from the aspects of the forward problem, dynamic EIT, static EIT and three-dimensional (3-D) EIT.In the study on forward problem of EIT, the effect of the position and area of electrical conductivity change in the field on the boundary potential was studied. Based on the study results, the noninvasive monitoring instrument for cerebral hematoma and edema was developed. Now the instrument has been applied in the first affiliated hospital of Chongqing university of Medical and Daping hospital, third military medical university in Chongqing. The instrument has got five certificates of Chinese patents.In the study on dynamic EIT, The theory of the equi-potential back-projection algorithm was described based on the principle of X-CT. The algorithm was studied from the aspects of sensitivity, current injection modality, number of electrodes, number of elements of FEM model, anti-noise, and so on. A good formula to calculate theweight coefficient W was brought up. The static EIT has been studied detailedly by using of One Step Newton Method (OSNM). A good method to initialize the initial conductivity in Newton-Raphson algorithm was given.Two dimensional FEM model can be modeled according to patient’s CT slices. The model is more close to the real human brain dissection structure. Simulation study on dynamic imaging for cerebral hemorrhage has been done based on the model.The experimental platform of real-time EIT system with 16 electrodes has been developed. The experiment on real-time EIT has been done in a slain filled tank with different objects and also tested on the human thorax. The results show that the real-time imaging system has the feature of high precise location, fast imaging (reconstructed one image per four seconds), and high space resolution. From the reconstructed image on the human thorax, we can distinguish the left from right lung, heart of human thorax. Images are reconstructed according to the practical data of human obtained from foreign EIT. The results show that the equi-potential back-projection and OSNM algorithms can distinguish the physiological characteristics of heart and cms. These prove that the algorithms are valid.The dissertation also analyzed the experimental data. The results of analysis show that the position and relative volume of targets in the tank can be reflect from the change of measurement data. The stability of the hardware system is good. The system’s precision is enough to the experiment on slain filled tank, but not eno更多还原