A Lung Electrical Impedance Tomography (EIT) Reconstruction Method using Deep Learning

Due to the inherent soft-field characteristics of Electrical Impedance Tomography (EIT) and the nonlinearity of the inverse problem, the lung contour and lesion structure cannot be effectively reconstructed by conventional model-based algorithms. In this study, a direct estimation model based on, Residual Network (ResNet) is proposed to improve the reconstruction accuracy of lung contour and lesion structure. The imaging results indicate that the contour of the lung and the position and size of the lesion can be reconstructed more accurately than the Tikhonov Regularization (TR) algorithm, Conjugate Gradient (CG) algorithm and 1D-CNN.