Magnetic particle imaging (MPI) for NMR and MRI researchers

Abstract

Magnetic Particle Imaging (MPI) is a new tracer imaging modality that is gaining significant interest from NMR and MRI researchers. While the physics of MPI differ substantially from MRI, it employs hardware and imaging concepts that are familiar to MRI researchers, such as magnetic excitation and detection, pulse sequences, and relaxation effects. Furthermore, MPI employs the same superparamagnetic iron oxide (SPIO) contrast agents that are sometimes used for MR angiography and are often used for MRI cell tracking studies. These SPIOs are much safer for humans than iodine or gadolinium, especially for Chronic Kidney Disease (CKD) patients. The weak kidneys of CKD patients cannot safely excrete iodine or gadolinium, leading to increased morbidity and mortality after iodinated X-ray or CT angiograms, or after gadolinium-MRA studies. Iron oxides, on the other hand, are processed in the liver, and have been shown to be safe even for CKD patients. Unlike the "black blood" contrast generated by SPIOs in MRI due to increased T2* dephasing, SPIOs in MPI generate positive, "bright blood" contrast. With this ideal contrast, even prototype MPI scanners can already achieve fast, high-sensitivity, and high-contrast angiograms with millimeter-scale resolutions in phantoms and in animals. Moreover, MPI shows great potential for an exciting array of applications, including stem cell tracking in vivo, first-pass contrast studies to diagnose or stage cancer, and inflammation imaging in vivo. So far, only a handful of prototype small-animal MPI scanners have been constructed worldwide. Hence, MPI is open to great advances, especially in hardware, pulse sequence, and nanoparticle improvements, with the potential to revolutionize the biomedical imaging field.

Figure 1

X-space FFP and Projection (FFL) MPI imagers at UC Berkeley. (a-b) The FFP MPI imager’s main field gradient is constructed using two opposed NdFeB permanent magnet cylinders, which produce a Field Free Point (FFP) at the magnet iso-center. (c) An MPI image of a “Cal” phantom filled with 10× diluted Resovist SPIO tracer taken using the FFP imager shows millimeter-scale resolution. (d-e) The projection (FFL) MPI imager’s main field gradient is constructed using two opposed NdFeB permanent magnet assemblies to produce a Field Free Line (FFL) along the instrument’s y-axis. (f) An MPI image of a similar “Cal” phantom filled with 10× diluted Resovist tracer taken using the projection scanner. This image has three times lower resolution than (c) due to a weaker magnetic field gradient, but was otherwise taken using nearly identical acquisition parameters.

Figure 2

Principles of x-space MPI imaging. (a) SPIO magnetization, characterized by a Langevin function, is nonlinear with the applied magnetic field and converges to saturation above a certain threshold. This nonlinear magnetization response determines the point spread function (PSF) in MPI. The simulated PSF with a FWHM of 6 mT is for a log-normally distributed iron oxide nanoparticle with diameter 20 +/− 2.5 nm. For a 6 T/m gradient field, this FWHM corresponds to 1 mm resolution. (b) Two permanent magnets create a strong magnetic field gradient and a sensitive point, called the field free point (FFP). Only the SPIOs in the instantaneous location of the FFP create an MPI signal. (c) To cover the imaging field-of-view, the FFP is moved rapidly in a trajectory across the imaged volume. Using X-space reconstruction, we grid the MPI signal to the instantaneous position of the FFP to form a native MPI image. Figure adapted from [23].

Figure 3

(a) The signal in MPI is perfectly linear (R² = 0.99) with iron quantity, i.e., the input nanoparticle quantity. (b) Theoretical resolution of MPI improves cubically with the nanoparticle size and linearly with the gradient strength. The resolutions of our two MPI scanners in Fig. 1 are marked for a 20 nm particle size. The FFP MPI imager and the Projection (FFL) imager feature 7 T/m/ ₀ and 2.4 T/m/ ₀ gradient strengths, respectively. (c) Comparison of tracer and contrast agent sensitivities in small animal (pre-clinical) imaging modalities, and where we expect MPI to be in the next couple of years. MPI compares well to existing imaging techniques and is a promising competitor. Here"smart” U/S refers to ultrasound with micro-bubbles. Subfigure (b) adapted from [23], and subfigure (c) adapted from [82].

Figure 4

Physics of superparamagnetic iron oxide (SPIO) nanoparticles. (a) For bigger SPIO nanoparticles, the ability to rotate to align their magnetization with the applied field is slowed down by magnetoviscous drag. (b) TEM images of a commercially available contrast agent, Resovist, and tailored nanoparticles with 20 nm mean iron oxide core diameter synthesized at the University of Washington (Scale bars = 20 nm) [34]. (c) Point spread functions (PSFs) for Resovist and 20-nm UW particles, measured using x-space MPI relaxometry [24]. The tailored 20-nm UW particles are already performing at the level of Resovist in terms of their FWHM resolution, but are more efficient because they produce higher MPI signal for the same iron content [37]. Note that the PSFs for both Resovist and 20-nm UW particles demonstrate lagging due to relaxation effects from magnetoviscous drag [24]. Figure adapted from [23].

Figure 5

MPI images of “carotid artery” phantoms, mimicking a healthy (top) and an occluded (bottom) internal carotid artery. MPI image successfully captures the stenosis in the phantom, depicted with a reduced brightness and a narrowing. (a) Photos of carotid artery phantoms made of acrylic with a 3 mm thickness for the branches and a 4 mm thickness for the common artery (about 75% the size of typical carotid artery in humans). The channels were injected with 20× diluted Resovist tracer. (b) MPI images were acquired on our high-resolution FFP MPI scanner, and mildly deconvolved with a Wiener filter. FOV: 4.5 cm by 4.5 cm by 6.2 cm. Scan time: 141 seconds.

Figure 6

MPI images demonstrating imaging of ex vivo rabbit kidney on the high-resolution FFP MPI scanner at UC Berkeley. (a) Ex vivo rabbit kidney photograph and (b) rabbit kidney anatomy for reference. (c) MPI image following injection of 3X diluted Resovist tracer into ureter. Visible is the ureter and the renal pelvis. (d) Injecting DI water washes the Resovist tracer further into the kidney. The MPI image shows that the tracer has now entered the renal medulla. FOV: 5.5 cm × 4.5 cm. Scan time: 2 minutes per image, including all robot movement.

Figure 7

3D projection reconstruction MPI image showing the Resovist tracer flowing to the heart and filtering in the liver of a mouse. (a) The volume-rendered MPI image and (b) the photo of the mouse injected with 100 L undiluted Resovist via tail vein and sacrificed 30 s post-injection. Imaging was performed with the Projection (FFL) MPI scanner with 180 projections, 3 min per projection acquisition time, FOV: 12 cm × 6 cm.

Figure 8

3D MPI experiment demonstrating no depth attenuation or tissue contrast. (a-b) Projection Reconstruction 3D MPI images show no attenuation with depth for the phantom embedded in tissue, nor any tissue contrast. (c) Helical phantom filled with SPIOs and (d) the same phantom embedded in animal (Meleagris Gallopavo) tissue. Images were acquired the UC Berkeley Projection (FFL) MPI scanner with 60 projections, volume reconstructed using Projection Reconstruction, and then displayed as a maximum intensity projection. Comparing the two images, there are no qualitative differences and a negligible change in signal level. FOV: 12 cm × 6 cm, 2 averages, 21 s per projection acquisition time, 10× diluted Resovist (50 mM) injected in 0.8 mm ID tubing wrapped around a 2.6 cm OD acrylic tube.

Figure 9

Stem cell imaging with MPI. (a) Projection MPI image of two injections of hESC-derived cells (1×10⁵ cells on the left vs. 2×10⁵ cells on the right site) introduced subdermally into a postmortem mouse (injection sites marked in cyan in the photo). The ratio of signal intensities between the right and left injection regions in the MPI image was found to be 2.1. Image acquisition time was 3 minutes total with FOV of 5×10 cm and 16 averages. (b) Plot of MPI signal intensity vs. number of stem cells in scanner, demonstrating excellent linear fit (R² = 0.99). MPI images were acquired for 9 stem cell populations ranging from 1×10⁴ to 1×10⁶ cells and compared for maximum signal intensity. Our current stem cell detection threshold (i.e., the noise floor) is limited by system interference, and is approximately 1×10⁴ cells in our prototype system.

Figure 10

Safety limits for human-size MPI scanners. (a) Magnetostimulation is the dominant safety concern for the excitation field in MPI, which typically operates around 25 kHz. The SAR threshold curve was calculated for a continuous (i.e., 100% duty cycle) sinusoidal magnetic field and a 4 W/kg limit. (b) Human subject experiments results for magnetostimulation in the arm and leg. Peak magnetic field amplitudes for the median stimulation thresholds and the 30th-70th percentile are shown (N = 20 and N = 17 for the arm and leg experiments, respectively). The magnetostimulation thresholds for the human torso were extrapolated using the results in the arm and the leg, and assuming a radius of r = 20 cm (typical radius of human torso). (c) Setups for testing the magnetostimulation thresholds in the human arm and leg. Figure adapted from [79].