Novel inversion tools to improve performance of the DBIM algorithm for microwave medical imaging

Abstract

This paper presents recent results from the application of new linear inversion methods to the Distorted Born Iterative Method (DBIM) for quantitative microwave breast imaging. We first review a recently proposed method based on a regularization by projection scheme, which exploits wavelets to accommodate the trade off between reducing the number of unknowns and representing highly heterogeneous breast compositions. Second, we review a compressed sensing recovery method that can be used to improve microwave imaging resolution when sparsity can be argued in the reconstruction process. Finally, we present the application of a two-step approach for the solution of the linear problem at each DBIM, which results in fast and accurate reconstructions. We note that the presented results from all these methods cannot be obtained from traditional DBIM approaches.