Hybrid Approaches in Microwave Imaging Using Quantitative Time- and Frequency-Domain Algorithms

In this work we propose two hybrid time-and frequency-domain microwave imagingschemes aimed to improve time-to-solution of quantitative time-domain imaging algorithms and image resolution of quantitative frequency-domain imaging algorithms. The proposed hybrid methods combine discontinuous Galerkin method (DGM) implementations of the time-domain (TD) forward-backward time-stepping (FBTS) algorithm and the frequency-domain (FD) contrast source inversion (CSI) or Gauss Newton Inversion (GNI). Simply put, an initial inversion in one domain (time or frequency) is used as prior information for the other. These schemes, referred to as FD-TD when FD prior is used in a TD algorithm, and TD-FD when TD prior is used in a FD algorithm, are applied to experimental and synthetic data. The results of the hybrid imaging approaches manifest an appreciable improvement relative to the stand-alone of FD and TD algorithms. Specifically, this study demonstrates that low-resolution frequency-domain prior information improves TD convergence. Additionally, we show that early-iteration time-domain solutions improves FD algorithm performance. We hope that these hybridization techniques pave the way for future investigations of optimal strategies for combining TD and FD schemes.