Sensor fusion of electron paramagnetic resonance and magnetorelaxometry data for quantitative magnetic nanoparticle imaging

Abstract

Magnetorelaxometry (MRX) imaging and electron paramagnetic resonance (EPR) are two non-invasive techniques capable of recovering the magnetic nanoparticle (MNP) distribution. Both techniques solve an ill-posed inverse problem in order to find the spatial MNP distribution. A lot of research has been done on increasing the stability of these inverse problems with the main objective to improve the quality of MNP imaging. In this paper a proof of concept is presented in which the sensor data of both techniques is fused into EPR–MRX, with the intention to stabilize the inverse problem. First, both techniques are compared by reconstructing several phantoms with different sizes for various noise levels and calculating stability, sensitivity and reconstruction quality parameters for these cases. This study reveals that both techniques are sensitive to different information from the MNP distributions and generate complementary measurement data. As such, their merging might stabilize the inverse problem. In a next step we investigated how both techniques need to be combined to reduce their respective drawbacks, such as a high number of required measurements and reduced stability, and to improve MNP reconstructions. We were able to stabilize both techniques, increase reconstruction quality by an average of 5% and reduce measurement times by 88%. These improvements could make EPR–MRX a valuable and accurate technique in a clinical environment.