Image Reconstruction of Electrical Impedance Tomography Based on Deep Learning

Abstract

Image reconstruction is a key step in electrical impedance tomography (EIT). The inverse problem of EIT is a nonlinear, highly morbid and under-qualitative problem, the resolution of EIT image reconstructed by traditional methods is relatively low. In order to apply electrical impedance tomography to practice successfully, it is urgent to find an effective image reconstruction algorithm. After the advantage of deep learning in self-learning nonlinear mapping between input and output was found, it was used in electrical impedance tomography image reconstruction. In this paper, convolution neural network (CNN) is used to solve the problem of image reconstruction. The data set and the 16-electrode EIT model are obtained through the joint simulation of MATLAB and COMSOL. The data set is divided into training set and test set, and the expected result is obtained after training the network.