Time Domain System Matrix (TD-SM) for Isotropic Resolution and Artifact Removal image reconstruction in X-space MPI

Abstract

Magnetic Particle Imaging (MPI) is an emerging medical imaging modality based on the non-linear response of superparamagnetic iron oxide nanoparticles (SPIONs), and imaging the concentration of the SPIONs. At present, MPI reconstruction mainly includes x-space and system matrix methods. The x-space method has an extremely fast reconstruction speed and the final reconstructed image can be seen as the result of convolution of particle distribution with point spread function (PSF), but it causes anisotropic resolution and artifacts due to the effect of anisotropic PSF. In this work, we propose a hybrid approach that combines the x-space and model-based system matrix methods to remove the PSF blur and reconstruct the isotropic resolution and artifact removal MPI images. We first build the system matrix in the time domain transformed by the ideal MPI scan model in the time domain. Then, we convert the x-space reconstruction process into a matrix operation, and subsequently combine several matrices into a system matrix (forward model). The input of this forward model is particle distribution and the output is the x-space reconstructed image, which is affected by the PSF blurring. The PSF blur removal is then converted to an inverse problem solving, which we implement using the Kaczmarz method to obtain isotropic resolution and artifact removal images. The results of simulation and phantom experiments demonstrate that our method could achieve the best isotropic resolution and image quality comparing with the standard x-space method and other PSF removal methods.