Sparse-Representation-Based Image Reconstruction for Magnetic Particle Imaging

Magnetic particle imaging (MPI) is an emerging medical imaging technique that measures the nonlinear magnetization response of magnetic nanoparticles (MNPs). The image reconstruction of MPI is to solve the unknown spatial distribution of MNPs from the measured magnetic response signal, which plays a significant role in MPI. In this study, a sparse-representation-based image reconstruction method is proposed to improve the spatial resolution and reduce the artifacts of MPI images. In the proposed method, the spatial distribution of MNPs is sparsely represented by the Gaussian radial basis functions (GRBFs). The inverse problem in MPI is consequently transformed to obtain the optimal weight coefficient vector of the GRBFs. It helps to reduce the number of unknowns to be reconstructed and improve the robustness of the image reconstruction process. By incorporating the prior knowledge from the preliminary reconstructed images, the center points of the GRBFs are selected densely in the target area and sparsely outside to further reduce the dimension of the system matrix and the artifacts. Numerical simulations are performed to optimize the key parameters in the proposed method. Furthermore, phantom experiments are carried out using a single-harmonic-based narrowband MPI scanner to demonstrate the feasibility of the proposed method. Experimental results show that the proposed method improves the spatial resolution from 0.5 to 0.3 mm and reduces the artifacts compared with the algebraic reconstruction technique (ART) method and the Newton-Raphson method. We envisage that the proposed method is of great significance to biomedical applications for MPI.