Examination of the sensitivity matrix during the iterative reconstruction process for microwave tomography

Paper in proceeding, 2018

We have reconstructed microwave tomographic images utilizing a Gauss-Newton scheme with a log transformation. While the recovered images are quite good, we have analyzed the behavior of the associated Jacobian, or sensitivity, matrix with respect to how it evolves during the reconstruction process. We found that in general the highest sensitivities are along the periphery of the imaging zone confirming the fact that most algorithms operate best for features nearest the outside. Interestingly, our log transformed algorithm appears to evolve during the reconstruction process to increase its internal sensitivity once the major exterior ones have been resolved. These results provide useful insight into a key driver of the reconstruction process and provide useful tools for assessing different system design criteria.