Design and Optimization of a Finite Element Model for Electrical Resistance Tomography of Human Lungs

Abstract

To improve the reconstruction quality of electrical resistance tomography (ERT) images, this paper designs and optimizes the finite element model for human lungs. According to the computer tomography (CT) scan image on human chest, the entire simulation domain was divided into a region of lungs, heart, and spine, and a region of adipose tissue, and the boundary curve equations of each region were derived by the improved particle swarm optimization (PSO). Based on the prior knowledge, a structural model was established for human lungs; the ERT forward problem was solved by the finite element model based on grid reconstruction, and the calculated boundary voltage of sensitivity field was taken as the theoretical value. Next, two optimization goals were set up: improving the calculation accuracy of forward problem, and easing the ill-conditionedness of the sensitivity matrix; two variables were configured: the number of layers of the finite element model in each region, and the polar diameter ratio of the finite element nodes on each layer to the finite element nodes on the boundary of each region corresponding to the same polar angle. On this basis, the finite element model was optimized by the improved PSO to adapt to human lung ERT. Simulation results show that, under the same experimental conditions, the proposed finite element model could solve the forward problem more accurately, improve the ill-conditionedness of the sensitivity matrix and the Hessian matrix, and make the sensitivity distribution more uniformly, thereby enhancing the accuracy of image reconstruction.