Image Reconstruction Based on ResV-Net for Electrical Impedance Tomography
Qian Wang, Zichen Wang, Di Wang, Xiaoyan Chen · Proceedings of International Conference on Artificial Life and Robotics · 2022
Electrical impedance tomography (EIT) is a nonlinear and ill-posed inverse mathematical problem.Due to the above problem, the reconstruction image suffers from serious artifacts.To overcome shortcomings, we proposed a residual V-shaped deep convolutional neural network (ResV-Net).It consists of the feature extraction module and image reconstruction module which are optimized by ResBlock.The residual connection method can effectively increase the number of the forward information flow and reverse gradient flow in deep CNN and alleviate the problem of non-convergence caused by gradient vanishing.The simulation and experimental results show that the ResV-Net has a better visualization effect than the related imaging method.