CT Data as A-Priori Information for Multimodal Magnetorelaxometry Imaging

Abstract

Magnetorelaxometry imaging (MRXI) [1] enables a quantitative and specific localization of magnetic nanoparticles (MNPs) inside living organisms [2] which is required by several novel therapeutic approaches such as magnetic hyperthermia therapy and magnetic drug targeting. MRXI involves aligning the MNPs’ magnetic moments using electromagnetic coils and subsequently recording their decaying net magnetization (relaxation). The recorded signals are connected to the MNP iron masses inside the discretized region of interest (ROI) via a system of linear equations [3]. The MNP distribution is recovered by solving the corresponding ill-posed inverse problem. However, the ROI discretization is limited to rather large voxel sizes. Finer voxel grids increase the number of unknowns and aggravate accurate reconstructions of MNP distributions. To date, finer voxel resolutions with reliable reconstruction accuracies are still necessary to obtain clinically relevant MRXI reconstructions. Standard MRXI provides only information about the MNP distributions and every voxel can contain MNPs with equal probability. Obviously, this is not realistic in clinical practice due to the different anatomical structures inside a human body. Accompanying traditional medical imaging techniques such as CT are necessary to gain anatomical information about the ROI. Here, we develop a multimodal reconstruction using CT data as prior for MRXI reconstructions, enabling a distinction between voxels with impossible, improbable and probable MNP occurrences. We test the approach on a cylindrical ROI (see Fig. 1) containing objects that produce different radiographic density (meat, wood, water, etc.) as well as MNPs. CT and MRXI measurements were conducted and the incorporation of the CT data allows us to reconstruct MNP distributions more accurately with half the voxel edge lengths than with mere MRXI (see Fig. 2). Our multimodal approach demonstrates that resolution and quantification accuracy of MRXI substantially benefits from anatomical data provided by alternative techniques such as CT or MRI.