【作者】 范文茹；

【导师】 王化祥；

【作者基本信息】 天津大学 ， 检测技术与自动化装置， 2010， 博士

【摘要】 生物电阻抗成像技术(Biological electrical impedance tomography,BEIT)是一种新型无损功能成像技术。其基本原理是根据人体内不同组织器官在不同生理或病理状态下具有不同的阻抗特性。通过配置体表电极阵列施加一定频率的低幅交变安全电流,并通过扫描电极测量电压数据,重构被测组织或器官的断层分布图像,从而提取与人体生理、病理状态相关组织或器官的电特性信息。BEIT具有安全无辐射、非侵入、低成本、便携、可视化等优点,可实现长期、动态、连续图像监测。本课题基于人体呼吸过程的BEIT图像监护技术进行系统研究:1、对图像重建算法展开研究。基于广义Tikhonov正则化算法,针对图像重建算法的分辨率、鲁棒性、实时性等指标,分别提出最大熵正则化算法有效提高了EIT成像精度,消除图像伪影;引入先验信息构建正则化矩阵,改进人体肺部呼吸EIT成像效果;组合算法克服了单一迭代中收敛慢及半收敛现象,增强了算法的鲁棒性;基于预处理技术的组合算法则在保证成像分辨率和稳定性的前提下,进一步提高算法的实时性。2、研究三维EIT电极结构与激励模式,获取三维场空间分布信息。为获取更好的三维图像重构结果,必需构建多层传感器阵列系统,以增加独立测量数据量。文中介绍两种三维电极分布模型,并对四种不同激励模式进行研究。此外,针对三维EIT成像提出了一种新的图像重建算法,在保证算法稳定性前提下,有效提高了重建分辨率和实时性。3、基于MEIT-TJU数字EIT系统对人体肺部呼吸过程进行实时监测,并分别对测量数据以及重建图像序列进行分析研究,挖掘有效信息。4、对三维人体模型进行仿真研究。根据人体胸部CT扫描图,分别构建2.5D和3D不同程度肺萎陷模型,分析比较电导率特性与复电导率特性对于不同模型EIT图像重构的影响。基于EIT引入几种肺通气量指标分析评价各种状态下肺部通气量变化,对比不同电特性对通气指标参数计算的影响。更多还原

【Abstract】 Biological electrical impedance tomography ( BEIT ) is a new non-invasive functional imaging technology. The principle of BEIT is that different tissues and organs have different electrical impedance properties in different physiological or pathological conditions. A set of electrodes are positioned around the periphery of the body with low amplitude safe currents being applied into the body through a pair of electrodes, and the resulting voltages on the other electrodes are subsequently collected. Based on the measured data, the internal conductivity distribution and its change can be estimated. Therefore, the electrical properties related to different physiological or pathological conditions can be explored. BEIT has the advantages of non-radiation, safe, non-invasive, low cost, portability, and visualization, etc. It can be used for continuous dynamic monitoring at bedside. The research work is focused on EIT technique applied for lung ventilation monitoring.1. Image reconstruction methods are studied. In order to improve the accuracy of reconstructed image, stability of the reconstruction, and the imaging speed, respectively, different methods are introduced based on the generalized Tikhononv regularization. The maximum entropy regularization method can effectively reduce the artefacts of reconstructed images and improve the accuracy. By incorporating prior information into Tikhonov regularization matrix, the differentiability and quality of the lung are improved. The hybrid method can overcome the phenomenon of slow convergence or semi-convergence in single-iteration algorithms and enhance the stability. Pre-conditioned hybrid method can further improve the algorithm in real time without losing accuracy and stability.2. To obtain the information of three-dimensional (3D) spatial distribution, 3D electrode structure and drive pattern are investigated. Two different structures of electrode array are proposed and four drive patterns are compared. Moreover, a new image reconstruction algorithm is put forward for 3D image reconstruction, in order to ensure the real-time property, accuracy and stability.3. The experiments for monitoring lung ventilation are carried out based on the digital system of MEIT_TJU. Measuremed data and corresponding image sequences of reconstruction are analysed to explore effective information.4. 3D human thorax models are established for simulation. Based on CT images of human thorax, true 3D thorax models with conductivity distribution or complex conductivity distribution under different ARDS conditions are built up in comparison with the 2.5D ones, and EIT-derived numeric indices are also employed for evaluation of the lung ventilation. The purpose of the research is to study different effects of different thorax models with either conductivity or complex conductivity on the reconstructed images and ventilation indices.更多还原