Dictionary-Based Background Signal Estimation For Magnetic Particle Imaging

Magnetic Particle Imaging is a highly sensitive tracer-based imaging method able to image as little as a nanogram of tracer material. In practice, background signals stemming from imperfections in the scanner instrumentation make it challenging to achieve these very high sensitivities. In case where the background signal is static over time it can be simply subtracted using an empty reference measurement. In this work we develop a method that is capable of handling dynamic background signals. Since the particle signal and the background signal are superimposed, we propose to perform a joint estimation of both quantities. The background signal space is modeled by a dictionary of background signals, build from a set of representative drifting background measurements. The proposed method is evaluated on experimentally and it is shown that it is capable of accurately estimating the background and tracer signal in a dynamic phantom experiment.