Trigo Flávio Celso, Gonzalez-Lima Raul, Amato Marcelo Brito Passos
Department of Mechanical Engineering, Polytechnic School, University of São Paulo-SP, R. da Consolação 3064-171A, 01416-000 São Paulo, Brazil.
IEEE Trans Biomed Eng. 2004 Jan;51(1):72-81. doi: 10.1109/tbme.2003.820389.
In this paper, we propose an algorithm that, using the extended Kalman filter, solves the inverse problem of estimating the conductivity/resistivity distribution in electrical impedance tomography (EIT). The algorithm estimates conductivity/resistivity in a wide range. The purpose of this investigation is to provide information for setting and controlling air volume and pressure delivered to patients under artificial ventilation. We show that, when the standard deviation of the measurement noise level raises up to 5% of the maximal measured voltage, the conductivity estimates converge to the expected vector within 7% accuracy of the maximal conductivity value, under numerical simulations, with spatial a priori information. A two-phase identification procedure is proposed. A cylindrical phantom with saline solution is used for experimental evaluation. An abrupt modification on the resistivity distribution of this solution is caused by the immersion of a glass object. Estimates of electrode contact impedances and images of the glass object are presented.
在本文中，我们提出了一种算法，该算法利用扩展卡尔曼滤波器解决电阻抗断层成像（EIT）中电导率/电阻率分布估计的逆问题。该算法能在很宽的范围内估计电导率/电阻率。本研究的目的是为人工通气时输送给患者的空气量和压力的设置与控制提供信息。我们表明，在有空间先验信息的数值模拟中，当测量噪声水平的标准差上升到最大测量电压的5%时，电导率估计值在最大电导率值7%的精度范围内收敛到预期向量。提出了一种两相识别程序。使用装有盐溶液的圆柱形体模进行实验评估。玻璃物体的浸入会导致该溶液电阻率分布的突然变化。给出了电极接触阻抗的估计值和玻璃物体的图像。
