Impedance image reconstruction using neural networks

Impedance imaging can be used in a variety of practical applications, such as medical diagnosis, geological exploration, multicomponent fluid flow analysis, and quality control. We consider the electrical impedance imaging where the spatial conductivity distribution within the object is reconstructed based on the voltage-current relationship measured by using a system of electrodes located on the surface of the object. The solution of the associated inverse problem requires a substantial amount of computation. In this paper, we present a new neural network architecture, with a relatively simple and inexpensive hardware, that can be employed efficiently to solve this inverse problem.