A Deep learning informed Polynomial Fitting Approach for Electrical Properties Tomography

This work presents a neural network informed fitting approach for conductivity reconstructions in MR-Electrical Properties Tomography. First, an artificial neural network is used to predict weights from T2-weighted images. These weights are used in a weighted fitting approach to calculate polynomial coefficients that parametrize the phase map. The conductivity is finally reconstructed from these coefficients. The reconstruction approach is tested on simulated data and in-vivo data and shows more accurate results than conventional fitting methods.