Permittivity Estimation for Patient-Specific Head Model via DBIM

In this paper, we present a method for generating a patient-specific head model comprising four homogenous tissues. This model is meant to provide a convenient starting point for the initialization of quantitative microwave imaging algorithms. First, an approximate distribution of the head permittivity is rapidly estimated using a linear inverse scattering model based on the Born approximation, regularized with polynomial basis functions for the contrast. Segmentation of the obtained images is used to generate a four-domain 3D phantom comprising a thin surface layer mimicking skin and fat tissue covering the skull, the skull, the brain, and the soft tissue. The parameters of the four tissues are estimated using a properly cast distorted Born iterative inversion approach. The quality of the obtained model was assessed by comparing the scattering parameters of the original and the generated model.