Artifact Reduction for MPI

High-quality images are essential for any imaging modality to make a reliable diagnosis, and although MPI is highly sensitive, artifacts are common. This issue poses significant challenges for applications that operate in environments with extremely low levels of iron, such as cell tracking. As a result, our objective is to reduce the amount of image artifacts in MPI by implementing different methods in the reconstruction process that allow for these applications. Key components for artifact reduction are:

Extrapolating the system matrix beyond the drive-field field of view reduces artifacts at the patch boundaries in multi-patch imaging scenarios.

Publications

[164737]
Title: Suppression of Motion Artifacts in Multi-Patch Magnetic Particle Imaging of a Phantom with Periodic Motion.
Written by: M. Boberg, N. Gdaniec, M. Möddel, P. Szwargulski, and T. Knopp
in: <em>SIAM Conference on Imaging Science (IS22)</em>. (2022).
Volume: Number:
on pages:
Chapter:
Editor:
Publisher:
Series:
Address:
Edition:
ISBN:
how published:
Organization:
School:
Institution:
Type:
DOI:
URL:
ARXIVID:
PMID:

Note: inproceedings, multi-patch, artifact

Abstract: Magnetic particle imaging (MPI) is a tracer based imaging technique, which determines the spatial distribution of superparamagnetic iron oxide nanoparticles with a high spatial and temporal resolution. Therefore, MPI is able to image dynamic tracer distributions like cardiac or respiratory motion in in-vivo experiments. As a matter of fact, the imaging volume covers only a few cubic centimeters due to physiological constraints. To cover larger objects a multi-patch approach is used where the imaging volume is shifted relative to the object. Since this reduces the temporal resolution, motion artifacts can occur during the measurement and reconstruction of dynamic tracer distributions. For periodic motions such as the aforementioned cardiac motion, this problem can be solved by reordering the raw measurement data. In a first step, the motion frequency is calculated by analyzing the raw data without reconstruction and without an additional navigator signal. Afterwards data snippets of the raw data corresponding to a specific motion state are rearranged into a virtual frame by using multiple repetitions of the motion state. Finally, the virtual frames can be reconstructed by standard reconstruction techniques. In our experiments, we successfully reconstructed a rotating phantom with a repetition time of 0.56 s without any motion artifacts, while a single full multi-patch measurement cycle takes at least 0.69 s.