Field-free Line Magnetic Particle Imaging: Radon-based Artifact Reduction with Motion Models

TL;DR

This paper introduces a Radon-based artifact reduction method for field-free line magnetic particle imaging that leverages motion models to improve dynamic particle concentration reconstruction, reducing motion artifacts.

Contribution

It develops a forward model for dynamic particle concentrations in MPI and integrates motion models with total variation regularization for artifact reduction.

Findings

Successful validation on synthetic data

Joint reconstruction of particle concentration and Radon data

Reduced motion artifacts in reconstructed images

Abstract

Magnetic particle imaging is a promising medical imaging technique. Applying changing magnetic fields to tracer material injected into the object under investigation results in a change in magnetization. Measurement of related induced voltage signals enables reconstruction of the particle distribution. For the field-free line scanner the scanning geometry is similar to the one in computerized tomography. We make use of these similarities to derive a forward model for dynamic particle concentrations. We validate our theoretical findings for synthetic data. By utilizing information about the object's dynamics in terms of a diffeomorphic motion model, we are able to jointly reconstruct the particle concentration and the corresponding dynamic Radon data without or with reduced motion artifacts. Thereby, we apply total variation regularization for the concentration and an optional sparsity constraint on the Radon data.