Model-based reconstruction in MPI accounting for field imperfections

To date a system matrix has to be obtained through a tedious calibration measurement when employing a system matrix-based reconstruction in magnetic particle imaging. This problem can be effectively addressed by model-based reconstruction, which takes into account both particle and scanner parameters. In this study, we focus on the scanner parameters and in particular on the fact that the fields of experimental systems are imperfect. For experimental Lissajous-type data we show that the modeling error can be substantially reduced by about 18 % by incorporating field imperfections in both the transmit and receive coils.