A two-stage microwave image reconstruction procedure for improved internalfeature extraction

We have developed a two-stage Gauss-Newton reconstruction process with an a utomatic procedure for determining the regularization parameter. The combin ation is utilized by our microwave imaging system and has facilitated recov ery of quantitatively improved images. The first stage employs a Levenberg- Marquardt regularization along with a spatial filtering technique for a few iterations to produce an intermediate image. In effect, the first set of i terative image reconstruction steps synthesizes a priori information from t he measurement data versus actually requiring physical prior information on the interrogated object. Because of the interaction of the Levenberg-Marqu ardt regularization and spatial filtering at each iteration, the intermedia te image produced from the first reconstruction stage represents an improve ment in terms of the least squared error over the initial uniform guess; ho wever. it has not completely converged in a least squared sense. The second stage involves using this distribution as a priori information in an itera tively regularized Gauss-Newton reconstruction with a weighted Euclidean di stance penalty term. The penalized term restricts the final image to a vici nity (determined by the scale of the weighting parameter) about the interme diate image while allowing more flexibility in extracting internal object s tructures. The second stage makes use of an empirical Bayesian/random effec ts model that enables an optimal determination of the weighting parameter o f the penalized term. The new approach demonstrates quantifiably improved i mages in simulation, phantom and in vivo experiments with particularly stri king improvements with respect to the recovery of heterogeneities internal to large, high contrast scatterers such as encountered when imaging the hum an breast in a water-coupled configuration. (C) 2001 American Association o f Physicists in Medicine.