Implementation of parallel 3-D DGM-CSI and DGM-GNI for electromagnetic imaging applications

Abstract

This thesis presents the development and implementation of the time-harmonic discontinuous Galerkin method (DGM) as a robust forward solver for the contrast source inversion (CSI) and Gauss-Newton inversion (GNI) imaging algorithms for electromagnetic imaging. Modern microwave imaging systems require versatile imaging algorithms that support electric and/or magnetic field data, and recent advances in material sciences have shown promise in the use of magnetic contrast agents which require the ability to recover permeability, both of which can be accomplished by using a DGM forward solver. DGM features also include high-order field and constitutive expansions and support for high-order representations of inhomogeneous backgrounds, which have been shown to improve imaging results. Presented herein are implementation details of a fully parallel C++ code for the DGM forward solver and DGM-CSI and DGM-GNI inversion algorithms, as well as synthetic and experimental results highlighting the capabilities of these algorithms, with a focus on breast cancer imaging.