Magnet Resonance Electrical Impedance Tomography (MREIT): Convergence and Reduced Basis Approach

TL;DR

This paper develops a convergence theory for MREIT inverse problems, introduces a faster algorithm combining reduced basis methods, and demonstrates improved image reconstruction efficiency through numerical experiments.

Contribution

It extends the convergence theory of the Harmonic Bz Algorithm and introduces a reduced basis approach to accelerate MREIT image reconstruction.

Findings

The new algorithm converges with approximate PDE solutions.

Reduced basis method speeds up the reconstruction process.

Numerical experiments show improved image quality and efficiency.

Abstract

This article considers the inverse problem of Magnet resonance electrical impedance tomography (MREIT) in two dimensions. A rigorous mathematical framework for this inverse problem as well as the existing Harmonic $B_z$ Algorithm as a solution algorithm are presented. The convergence theory of this algorithm is extended, such that the usage an approximative forward solution of the underlying partial differential equation (PDE) in the algorithm is sufficient for convergence. Motivated by this result, a novel algorithm is developed where it is the aim to speed-up the existing Harmonic $B_z$ Algorithm. This is achieved by combining it with an adaptive variant of the reduced basis method, a model order reduction technique. In a numerical experiment a high-resolution image of the shepp-logan phantom is reconstructed and both algorithms are compared.