Microwave tomographic image improvement by fitting to a Cole-Cole relaxation model

Paper in proceeding, 2017

We have reconstructed microwave tomographic images of the Supelec breast phantom using our imaging fixture in combination with a multi-channel vector network analyzer. During this study we were able to recover images without the support of a priori information over a broad frequency range - 1100-1900 MHz. We then fitted the spectral values at each pixel within the field of view to a Cole-Cole curve and extracted the coefficients at each location. While the individual images at each frequency provided reasonably representations of the target permittivity and conductivity distributions, the fibroglandular features were generally quite blurred with the surrounding adipose region. However, several of the Cole-Cole coefficient plots provided a higher level of resolution for the inclusions. While there was a noticeably high level of artifacts outside of the breast phantom perimeter, the internal structures are quite representative of the target. This is one of the first Cole-Cole image fittings using broadband measurement data on an actual phantom and sets the stage for higher specificity imaging.