Fast trajectory-independent model-based reconstruction algorithm for multi-dimensional magnetic particle imaging.

Area of Science:

• Biomedical Imaging
• Medical Physics
• Signal Processing

Background:

• Magnetic Particle Imaging (MPI) visualizes superparamagnetic nanoparticles for medical applications.
• Traditional MPI reconstruction methods are often time-consuming or limited to specific scanning trajectories.
• Existing Chebyshev-polynomial-based methods are restricted to sinusoidal Lissajous trajectories.

Purpose of the Study:

• To present MoBiT-2S, a novel reconstruction pipeline for 2D MPI data.
• To develop a trajectory-independent, model-based reconstruction algorithm for MPI.
• To enable general-purpose and flexible MPI reconstruction.

Main Methods:

• A two-stage model-based algorithm utilizing a reconstruction formula.
• The first stage reconstructs the MPI core response via variational formulation.
• The second stage deconvolves the core response using a zero-shot plug-and-play algorithm.

Main Results:

• MoBiT-2S successfully reconstructs 2D MPI data from Lissajous and non-Lissajous scans.
• The pipeline incorporates channel-specific thresholding and artifact reduction techniques.
• Quantitative and qualitative comparisons show competitive performance against state-of-the-art methods.

Conclusions:

• MoBiT-2S demonstrates robust reconstruction capabilities on real 2D MPI data.
• The trajectory-independent approach enhances flexibility for various scanning sequences.
• MoBiT-2S shows potential for general-purpose model-based MPI reconstruction.