Biomedical microwave inversion in conducting cylinders of arbitrary shapes

We introduce a non-linear inversion algorithm for use in microwave biomedical imaging when the object of interest is surrounded by an arbitrarily shaped conducting enclosure. The algorithm utilizes the Gauss-Newton inversion method and a combined additive and multiplicative regularizer. The conducting enclosure is taken into account via a FEM-based forward solver which is able to efficiently model arbitrarily shaped boundaries. Results for the 2D scalar case are given when the enclosure is a circle, triangle, and square, and include simple and complex biological scatterers, based on synthetic data. The results show that the algorithm is capable of reconstructing objects in all cylinder types.