The realistic breast phantom developed by the French team at Supelec provides an excellent opportunity to explore various measurement systems and associated algorithms. In this case there is a known ground truth for the geometry of the segmented regions for direct comparison of images with actual sizes, shapes and locations of the different features. It also allows for testing utilizing a range of dielectric materials for both the adipose and fibroglandular region which can be useful given the disparity of published property values. For this experiment, we are exploring our tomographic imaging algorithm with a log transformation in the context of restricting the imaging zone strictly to that space occupied by the phantom. In this case, the 2-dimensional perimeter of the breast was approximated by a simple ellipse which was extracted from the original image. All images were recovered without the assistance of a priori information and also included our 2-step imaging scheme. Simple parameter tests were performed to assess the limitations of the technique, especially when the imaging zone was not accurately determined. The final results were also analyzed by examination of histograms of the field residuals to assess whether the algorithm is robust and unbiased.