Experimental evaluation of 3D electrical impedance tomography with total variation prior

摘要

In electrical impedance tomography (EIT), two- or three-dimensional (2D or 3D) distribution of the electrical conductivity is reconstructed based on potential measurements from the surface of the object. In many industrial and medical applications of EIT, the conductivity distribution is discontinuous - due to, e.g. phase or tissue boundaries. Previous studies have shown that such features in the conductivity can be preserved by using total variation (TV) prior model in the EIT image reconstruction. Recently, both 2D and 3D TV models have been utilized in EIT simulation studies. This far, however, EIT reconstructions with TV models have been evaluated experimentally only in 2D cases or translationally invariant 3D cases. In this paper, an experimental study of EIT with a TV prior in a 3D geometry is presented. In addition, we propose the selection of the prior parameters in the TV model based on the prior information of materials in the target, and their conductivity ranges. The results demonstrate the robustness of the proposed parameter selection strategy, and verify that the use of the TV prior yields sharp reconstructions in 3D EIT.