Study Of Electrical Impedance Tomography Algorithms Of The Lung

The lung is a vital organ in the human body and its pathologies such as pulmonary blood clots and pulmonary embolism are characterised by a high number of patients and a high mortality rate.The new coronavirus pneumonia COVID-19 and its sequelae also posed a serious and long-term risk to the lung tissue.Electrical Impedance Tomography(EIT)is an imaging technique that reconstructs the impedance distribution of the region to be measured.It is inexpensive,non-invasive,simple to perform and easily accessible,which are its advantages.And it has a promising application in lung imaging,but still suffers from a lack of a priori information,severe pathology and many imaging artifacts.This paper focuses on the lung EIT imaging algorithm,simulates the lung blood clot imaging experiment.And designs a lung blood clot simulation model with reference to the conductivity of human lung tissues and blood through the EIDORS platform in view of the insufficient a priori information of EIT.The model was applied to the finite element solution of the EIT positive problem and the construction of the priori databases set to solve the problem of insufficient priori databases.In order to address the drawbacks of the EIT inverse problem,this paper proposes a Coefficient Weighted Generalized Regression Neural Network(CWGRNN)EIT imaging algorithm,which improves its mode layer transfer function without affecting the structure of the traditional GRNN.The CWGRNN algorithm improves its mode layer transfer function by setting different coefficient weights for each mode layer neuron through the Spearman rank difference method,increasing the contribution of useful neurons and decreasing the contribution of useless neurons through the coefficient weights,thus targeting the inverse problem the algorithm has better effectiveness and robustness by increasing the contribution of useful neurons and decreasing the contribution of useless neurons,thus targeting the drawbacks of the inverse problem and achieving the effect of suppressing or removing artifacts.In order to further optimize the EIT imaging algorithm CWGRNN,this paper uses an evolutionary algorithm to optimize it.At the same time,because the evolutionary algorithm is prone to fall into local optimum,poor optimization effect and low quality,this paper proposes an Adaptive Decomposition Mechanism Dynamic Chaotic Particle Swarm Optimization(ADCPSO)algorithm.This algorithm is based on the PSO algorithm,which proposes a particle adaptive decomposition mechanism to ensure the high quality of the particle population,and applies a dynamic Chebyshev chaotic mapping to the particles at update time to ensure that they can jump out of the local optimum,increasing the optimization capability of the algorithm and optimizing the image quality of the CWGRNN.In this paper,the ADCPSO-CWGRNN imaging algorithm is subjected to lung clot simulation experiments and water tank experiments for single,dual and triple targets.The simulation results verified the effectiveness and robustness of the algorithm.And the average ICC and SSIM for noise-free imaging results were: 0.8243 and 0.7989,the average ICC and SSIM for imaging results with noise were: 0.8180～0.8239 and 0.7908～0.7980.The reconstructed images of the water tank experiments show that ADCPSO-CWGRNN can image single and dual targets well,proving that it has some generalization capability.