Simulation of reconstruction based on the system matrix for magnetic particle imaging

Magnetic particle imaging (MPI) is a new tomographic imaging modality which images the distribution of the superparamagnetic iron oxide nanoparticles (SPIONs) based on its non-linear response to the applied magnetic field. MPI outperforms other imaging techniques with high resolution and sensitivity. The MPI reconstruction with high accuracy and no artifacts is of enormous practical importance for its applications. MPI reconstruction based on the system matrix is an important part of MPI reconstruction. In this work, 2D simulation of MPI based on the system matrix is conducted according to the mathematical theory of the imaging process. Before that, the basic principles and the mathematical models of MPI are first introduced. The MPI reconstruction is conducted and compared by three different methods for three different phantoms: the simulated stenosis, the heart-shaped graphics and the capital letters BIT. The MPI reconstruction is realized based on the Tikhonov regularization with different regularization parameters, the preconditioned conjugate gradient method and the least squares QR-decomposition method. The reconstruction results are compared from visualization and performance indicators, including the normalized root mean squared error, the structural similarity index measurement and the peak signal-to-noise ratio. Then, we study and explore the effect of Gaussian white noise with different levels in the reconstruction quality. We also discuss the influence of two parameters on the reconstruction performance: the size of the nanoparticles and the gradient strength of the selection field.