Interplay between iteration step size and spatial filtering in microwave tomography

Abstract

We have performed an analysis of the impact of iteration step size with respect to our spatial filtering approach in the context of microwave tomography. While reducing the step size intuitively reduces the convergence rate of the iterative reconstruction process, it has been shown to produce considerable benefits with respect to overall image stability. Especially with regards our Gauss-Newton approach that incorporates a log transformation of the minimization criterion, reducing the step size can be essential in maintaining proper phase unwrapping in more challenging imaging cases. Our early results demonstrate that the step size can be reduced considerably while still achieving relatively rapid convergence and maintaining good stability even under challenging reconstruction problem cases. However, the interplay between the spatial filtering process and our relatively new two-step reconstruction process is less obvious. In this paper we discuss the challenges and opportunities with respect to optimizing these parameters.