Model-based Calibration and Image Reconstruction with Immobilized Nanoparticles

Abstract

The model-based reconstruction problem is still one of the key challenges in magnetic particle imaging (MPI) when using multi-dimensional Lissajous-type excitations. One aspect, which is often highlighted in the literature, is the magnetization behavior of the magnetic nanoparticles in fluids, which is typically modeled by a coupling of Brown and N\'{e}el rotation mechanisms. Another aspect, which is at least as important as the particle model itself but sometimes treated less prominently, is a very careful calibration of the model input, respectively the scanner parameters such as analog filter and applied magnetic fields. The careful consideration of both aspects is the essential requirement for a proper solution to the model-based problem. In the present work we combine calibrated scanner components with polydisperse particle models for immobilized nanoparticles to derive a model-based system function and an efficient calibration routine. It is experimentally validated on the Bruker preclinical MPI system using 2D Lissjous trajectories.