Mesh error influence determination on electrical impedance tomography image reconstruction

Abstract

The Electrical Impedance Tomography (EIT) reconstructs lung images by mapping the conductivity space of the human thoracic region. This technique consists of applying patterns of electrical current to electrodes placed on the body surface and measuring the values of electrical potentials between them. The resolution of this inverse problem can be approached as a problem of optimization that seeks to minimize the difference between the potential calculated by the Finite Element Method (FEM) and the potentials measured. At each iteration, a mesh is resolved, which reflects the high computational cost of the whole process. This work aims to study mesh discretization, coarser meshes need less computational time to solve the linear system, and more refined meshes need more computational time to solve the linear system. The aim of this work is the error distribution determination related to the difference in mesh resolution; this error distribution allows us to solve coarser meshes with less computational time, but with result quality closer to refined meshes. (AU)