Annelies Coene, Guillaume Crevecoeur and Luc Dupré, Ghent University, Belgium
Volume: 62, Issue: 6, Page(s): 1635 – 1643
Magnetic nanoparticles (MNPs) are an important asset in many biomedical applications. Their small sizes allow them to reach virtually every part in the body, while their large magnetic saturation enables their non-invasive detection. For these applications to function properly, it is important to accurately determine the spatial MNP distribution. A promising, sensitive and non-invasive technique for visualizing this spatial MNP distribution is an adapted form of Electron Paramagnetic Resonance (EPR). In this imaging technique an MNP sample is placed in a homogenous alternating magnetic field in which the magnetic moments of the unpaired electrons are excited by an incident electromagnetic wave. This changes the orientation of the magnetic moments which enable measurement by a pickup coil. One drawback is that only the total MNP amount in the sample can be determined. In recent work an extension to 1D EPR was developed which allowed to recover the 1D MNP distribution by moving the sample through the magnetic field and by solving an inverse problem. We investigate the impact of errors on the reconstruction performance of 1D EPR by modeling EPR setup errors, measurement errors and sample positioning errors and by performing the MNP imaging for different types of MNP distributions. The error models are validated by comparison to EPR lab measurements with intentional errors. We introduce a new numerical solution method for the inverse problem which significantly improves the MNP reconstruction quality. Based on the analysis, further requirements and possible extensions (2D/3D EPR) for accurate MNP reconstruction are made. This way EPR can become a valuable MNP imaging method.