Recent progress in magnetic particle imaging: from hardware to preclinical applications

Magnetic particle imaging (MPI) is a tracer-based tomographic imaging method that is capable of determining the spatial distribution of magnetic nanoparticles with excellent contrast, signal-to-noise ratio (SNR), temporal resolution and spatial resolution. To be clear, MPI scans cannot be obtained in an MRI scanner; a special-purpose MPI scanner is required. Since the seminal publication from Gleich and Weizenecker (2005), the field of MPI has seen rapid progress in scanner instrumentation, imaging methodology, reconstruction, mathematics, tracer development, and preclinical applications. Until recently, only engineering and physics groups worked on homebuilt MPI scanners. Just in the last few years, commercial vendors have released preclinical MPI scanners, bringing the unique advantages of this new imaging technology to molecular and cellular imaging experts.

What are the unique advantages of MPI? First, the MPI scanner subjects the patient to absolutely no radiation (like MRI and Ultrasound) and the MPI tracers are thought to be safe for humans use. The MPI tracers are superparamagnetic iron oxide nanoparticles (SPIOs), which were first developed more than 30 years ago as an MRI contrast agent in Nobel Laureate Paul Lauterbur's lab. One SPIO is FDA approved for human use as an iron replacement therapy in Chronic Kidney Disease (CKD) patients, who are especially susceptible to kidney damage from the most common radiologic contrast agents, Iodine (x-ray and x-ray CT) and Gadolinium (MRI). MPI has superb sensitivity, with 150 nanomolar experimentally demonstrated, and 10 nanomolar may be feasible soon after improvements in hardware, SPIOs and data acquisition techniques. Since millmolar concentrations of [Fe] are considered safe, the dose-limited sensitivity of MPI may already be complementary to nuclear medicine studies—but with zero radiation, zero attenuation with depth, and no half-life confounds.

Only a dozen labs worldwide have begun to explore MPI's potential for medical applications. Early preclinical studies have demonstrated MPI advantages, including stem cell tracking, vascular imaging, tumor imaging, CBV and CBF imaging, catheter tracking, pulmonary perfusion and ventilation, traumatic brain injury imaging, viscosity imaging, localized hyperthermia, and others. This special section in Physics in Medicine and Biology includes ten research articles that highlight cutting-edge developments in MPI. The first four articles report on recent progress on hardware development and imaging methodology. The remaining six articles focus on targeted preclinical applications and report on recent animal studies.

In the first article, Graeser et al (2017) presented the first 2D magnetic particle spectrometer (MPS). It allows for the study of 2D MPI sampling trajectories by emulating the static MPI selection field using tunable offset fields. In particular, the device allows for the direct access of the dynamic response of magnetic nanoparticles when applying complex excitation patterns in two dimensions that go beyond the most simple unidirection excitation. The device has been used by von Gladiss et al (2017) to measure high quality MPI system matrices that were used to reconstruct tomographic data measured with a preclinical MPI scanner. Using the multidimensional MPS for system calibration is a very promising approach to replace the commonly applied robot-based calibration where a small delta sample is shifted through the field of view while continuously measuring the system response on a full trajectory at any point. Gladiss et al (2017) have shown that the MPS-based approach significantly shortens the calibration time and even improves the image quality in terms of spatial resolution of reconstructed phantom data.

An experimental comparison between the Lissajous trajectory and the Cartesian trajectory was carried out by Werner et al (2017). The study was done using the same scanner hardware ensuring comparable conditions. It was found that the Lissajous trajectory performed better than the unidirectional Cartesian trajectory but similarly as the bidirectional Cartesian trajectory. In terms of spatial resolution, the bidirectional Cartesian trajectory even performed slightly better than the Lissajous trajectory.

While MPI in its most simple form only images the concentration of the applied tracer, Utkur et al (2017) studied the feasibility of additionally determining the viscosity of the medium surrounding the particles. By studying the effect of the viscosity on the relaxation behavior in spectroscopic MPI measurements, Utkur et al (2017) have shown that the viscosity can be discriminated across a biologically relevant range from 0.89 mPa s to 15.33 mPa s.

Two papers (Keselman et al (2017) and Kaul et al (2017)) studied the tracer kinetics of two of the most common MPI SPIO tracers: LS-008 and Ferucarbotran for in vivo applications. Both publications found that LS-008 has a much longer blood half life (about 4 hours) than Ferucarbotran and, hence, is ideal for vascular imaging. In Keselman et al (2017) time course MPI studies show that organs of rats can even be imaged after 70 days. In Kaul et al (2017) it was also shown that LS-008 shows a factor of 3.4 stronger MPI signal than Ferucarbotran at the same iron concentration. This sensitivity, coupled with zero background signal, allows for MPI to visualize submillimeter vessels like the aorta in mice experiments.

In vivo liver visualization in mice has been studied by Dieckhoff et al (2017). Ferucarbotran was used as a tracer with fast liver uptake. Since the mobility of the magnetic nanoparticles is limited, the authors used a reference sample for system matrix calibration that shows a similar mobility behavior. The mobility of particles in liver was determined using magnetorelaxometry measurements.

Magnetic fluid hyperthermia (MFH) began in the 1950s to nonivasively kill diseased cells and tissues where surgerical access was risky. The procedure involves first targeting SPIOs specifically to pathophysiology, followed by applying magnetic fields that heat only the SPIOs. The goal is of course to kill only pathological cells while leaving healthy background tissues undamanged. However, despite significant efforts, non-specific uptake to healthy organs (especially to liver, spleen) remains common. Hence, to avoid damaging the liver while treating a breast tumor, it would be crucial to spatially limit heating of the liver. Unfortunately, conventional 300 kHz RF hyperthermia offers only crude focusing, since the long wavelength of 300 kHz RF in the body—$\lambda /2\approx 50$ m. To ensure no damage to the liver during MFH, Hensley et al (2017) developed a platform for localized RF hyperthermia. Here, they showed one can exploit the same nonlinear Langevin physics employed in MPI to localize SPIO heating to just a few millimeters. A conventional 2.35 T m⁻¹ field free line MPI selection field gradient freezes rotation of SPIOs outside the field free region (FFR), effectively restricting the heating effects of the 300 kHz RF to the FFR. Moreover, a realtime MPI imaging and heating system could soon become a powerful platform for localized heating with realtime temperature feedback.

Traumatic brain injury is a common diagnostic imaging challenge following car accidents, bicycle accidents, and any blunt force trauma. Current vascular imaging methods (e.g. x-ray CT angiogpraphy, MRI Angiography) typically use the two contrast agents, Iodine and Gadolinium. The cerebral blood volume (CBV) in brain parenchyma ranges between 2.7% (white matter) to 5.2% (gray matter). This leaves roughly 95% background tissue, which often obscures capillary-level CBV in MRI and CT studies, despite innovations to suppress background tissue. MPI naturally has superb sensitivity and zero tissue background signal so it offers unprecedented contrast to noise ratio. Indeed, MPI can even see capillaries, which are not spatially resolvable. Here, in collaboration with neuroscientists who study disruptions in the blood brain barrier (BBB), Orendorff et al (2017) performed the first in vivo preclinical experimental demonstration that MPI can visualize bleeding to brain parenchyma following tramatic brain injury.

Pulmonary embolism (PE) is a deadly cardiovascular condition that effects roughly 500 000 patients in the USA each year, killing 60 000–100 000 of these. The preferred clinical diganosis is x-ray CT Pulmonary Angiography using Iodine. However the radiation risk is significant for pregnant women on bed rest, who are at high risk for PE. Also, the high iodine dose is risky for CKD patients, which afflicts 46.8% of Americans over the age of 70. Often a Ventilation–Perfusion (VQ) scintigraphy study is performed in these patients. During the perfusion study, physicians inject cleverly designed 40-micron Macroaggregated Albumin (MAA) labeled with the radioactive reporter Tc99m. The 40-micron sized MAA is trapped biomechanically in the tight 6-micron capillaries of the lungs. Any existing pulmonary embolism blockage is revealed by the absence of Tc99m-MAA at the distal ends of the lungs. However, the VQ study requires hours to setup the tracers, subjects the patient to radiation, has poor spatial resolution, and it is limited to 2D projection studies. Here, Zhou et al (2017) have developed a noninvasive version of the lung perfusion study by replacing the Tc99m reporter with an SPIO magnetic reporter, creating the first preclinical experimental validation of SPIO-MAA. This shows great promise because an MPI version of Ventilation Perfusion is faster (a few minutes), safer (no radiation), 3D instead of 2D, and there is zero preparation time of the SPIO-MAA.