【作者】 徐桂芝；

【导师】 颜威利；

【作者基本信息】 河北工业大学 ， 电工理论与新技术， 2002， 博士

【摘要】 脑科学是21世纪生命科学研究的重要课题，脑内组织电阻抗特性是进行各种脑内电磁活动研究和病理分析的基础，是脑科学研究的重点和难点问题。电阻抗断层成像(EIT)技术是近年来发展起来的基于生物组织电学特性的、无损的、价格低廉的新一代功能成像技术。根据国家自然科学基金重点资助项目“脑内电活动三维动态成像”和河北省自然科学基金资助项目“脑内组织电阻抗无损检测与功能成像”的要求，本文由磁共振成像(MRI)技术获得的图像抽取解剖结构信息建立几何头模型，将脑内电磁逆问题的求解研究与EIT技术相结合，对获得人体脑内不同组织结构的电阻抗特性的计算方法、功能成像方法及软件实现和硬件设计进行了系统的研究。主要内容如下： 1．开发了基于EIT技术的正、逆问题有限元分析软件 对EIT正、逆问题求解场域的数学模型的建立及求解方法做了较详细的分析，在此基础上编制了系统的仿真软件。该软件对任意形状、结构的场域可实现自动剖分，可建立16电极、32电极求解模型；正问题部分可给出各种激励情况下的等位线分布和边界电位变化情况；逆问题部分用反投影法实现了图像重构，并成功应用于所研制的测试系统。 2．进行了基于EIT技术的二维(2D)脑电四层同心圆头模型和真实头模型的电特性分析计算 对2D四层同心圆头模型和山MRI技术获得的头的断层图片构造的2D轴位真实头模型及2D矢位真实头模型的电特性进行分析，给出了各种激励情况下的等电位线分布。结果表明，由于颅骨电导率极低，采用相对激励模式比相邻激励模式对场域、特别是中心场域的敏感性强。分析了头部各层电导率参数变化对求解区域场分布的影响，结果显示颅骨电导率变化对场域特性的分布影响最强。特别是研究了脑内组织电导率变化对头皮表面电位分布的影响，这是一项很有意义的工作，为脑内电特性的深入研究奠定了基础。 3．提出并实现了用小波神经网络求解脑内电参数 提出应用小波神经网络求解脑内组织的电导率参数，构造了用于脑电EIT问题的小波框架，并用于脑电参数的求解，较好地解决于阻抗成像的非线性映射问题，避免了其他重构算法中迭代优化收敛差、耗时长的缺点。并且该方法在脑电测量中无需做任何提前假设，这对进一步反演脑内电特性奠定了基础。 4．研制了实时EIT硬件系统并对物理模型进行了实际测试和阻抗图像重构 研制了EIT硬件系统，建立了16电极的物理模型，并利用计算机进行实时数据采 沁 狂十EIT技术的脑内电特性与功能成像研究集、处理与图像重构。分析了各种激励模式对求解场域敏感性的影响，并对测量通道性能的一致性做了测试，结果表明所设计的系统各测量通道性能基本一致。对不同结构形状、不同导电性能的模型进行实验，均可获得较好的阻抗图像重构效果。该系统具有工作稳定、实时、使用方便、扩展性强等特点。更多还原

【Abstract】 Brain research is very important subject to life science in the 21st century. The electrical characteristic of brain tissue is a basis for analyzing brain electrical and magnetic active and clinical diagnosis. Electrical impedance tomography (EIT) is a new type of technique with the feature of harmless, non-invasive and convenience, developed in recent years. In this paper, combining the technique with electromagnetic forward problem and inverse problem, the author has studied the numerical computation methods, hardware and software design of EIT system, reconstruction algorithm and brain electrical characteristic. The main research work is following:1. Development of the software package of EIT systemThe forward and inverse problems of EIT system have been analyzed and the software package has been developed based on Matlab and Fortran. This software can be used to construct automatically 16 electrodes and 32 electrical electrodes finite element method (FEM) division model. The equi-potential lines in any drive pattern can be plotted and the boundary potential changes in its forward problem can be obtained. Using back projection method, the impedance imaging in simulation and real measurement can be reconstructed.2. Study on brain electrical characteristics for two dimentional (2D) real head modelsThe brain electrical characteristics have been analyzed for 2D four layers concentric sphere head model, axial real and sagittal real head models from MRI pictures. The equi-potential lines can be gotten in different drive patterns and it shows that the sensitivity for the resolution region, especially the center region in opposite drive pattern is more than others because the head bone has very low conductivity. The potential distributions change with conductivity increase and reduce within head. And the change is the biggest when the bone conductivity changes.3. Computing brain resistivity parameters by using the wavelet neural network methodThe EIT inverse resolution is a non-linear mapping problem. Here it is presented that the brain resistivity parameters is computed by using the wavelet neural network method. Not only can it resolve non-linear mapping problem, but also overcome the shortage of non-convergence and expending long CPU time in other reconstruction algorisms.4. Research of the testing system for BIT and the reconstruction of impedance imagingA hardware system for EIT technique was made in this paper and the physical model with 16 electrodes was obtained. Real time collection, process for data and imaging reconstruction are done by a computer. The affection on sensitivity by differential drive patterns is analyzed. And the test for consistency of testing channels shows that the characteristics of every testing channel of the system are identical generally. Good impedance imaging reconstruction can be gotten for models with various shapes and differential conductivity. The system has stable, real time, convenient and extended features.更多还原