A simulation study on image reconstruction in magnetic particle imaging with field-free-line encoding

TL;DR

This study evaluates image reconstruction techniques for magnetic particle imaging using field-free-line encoding, demonstrating that ML-EM improves image quality especially with limited data and weak drive fields.

Contribution

It introduces the application of the ML-EM algorithm for MPI with FFL encoding, showing its advantages over traditional FBP in simulation studies.

Findings

ML-EM improves image quality over FBP, especially with fewer projections.

Spatial resolution increases with magnetic field gradient and nanoparticle size up to 30 nm.

Weaker drive magnetic fields lead to better image resolution.

Abstract

The purpose of this study was to present image reconstruction methods for magnetic particle imaging (MPI) with a field-free-line (FFL) encoding scheme and to propose the use of the maximum likelihood-expectation maximization (ML-EM) algorithm for improving the image quality of MPI. The feasibility of these methods was investigated by computer simulations, in which the projection data were generated by summing up the Fourier harmonics obtained from the MPI signals based on the Langevin function. Images were reconstructed from the generated projection data using the filtered backprojection (FBP) method and the ML-EM algorithm. The effects of the gradient of selection magnetic field (SMF), the strength of drive magnetic field (DMF), the diameter of magnetic nanoparticles (MNPs), and the number of projection data on the image quality of the reconstructed images were investigated. The spatial resolution of the reconstructed images became better with increasing gradient of SMF and with increasing diameter of MNPs up to approximately 30 nm. For the diameter greater than approximately 30 nm, the spatial resolution was almost constant. It also became better with decreasing strength of DMF. The image quality was improved by use of the ML-EM algorithm compared to the FBP method, especially when the strength of DMF was weak and the number of projection data was small. In conclusion, we presented image reconstruction methods for MPI with an FFL encoding scheme and our preliminary results suggest that the ML-EM algorithm will be useful for improving the image quality of MPI.