Magnetic particle imaging (MPI) shows extraordinary promise for biomedical applications: it is highly sensitive, linearly quantitative anywhere in the body, has zero signal from biological tissue, and is safe for patients. One of the major challenges in MPI is spatial resolution, which is fundamentally governed by selection field gradient strength and nanoparticle properties. In Chapter 3 of this dissertation, I describe the design and construction of a high gradient field-free line (FFL) scanner for improved spatial resolution. This is the world's first iron return MPI scanner, and the highest resolution FFL imager in the world. Armed with MPI-tailored nanoparticles created by our collaborators at the University of Washington and Lodespin Labs, we were able to achieve 700 micron resolution in vivo. We also successfully demonstrated dynamic projection MPI as well as 3D projection reconstruction MPI with this scanner. In addition to hardware innovations, I have also demonstrated the use of MPI in two very important applications, namely cancer imaging and gastrointestinal (GI) bleed detection. In Chapter 2 of this dissertation, I demonstrated the first use of MPI for in vivo cancer imaging with systemic tracer administration. Due to the enhanced permeability and retention (EPR) effect, the tumor was highlighted with tumor-to-background ratio of up to 50. The nanoparticle dynamics in the tumor was also well appreciated, with initial wash-in on the tumor rim, peak uptake at 6 hours, and eventual clearance beyond 48 hours. In Chapter 4 of this dissertation, I demonstrated the first use of MPI along with long-circulating, PEG-stabilized SPIOs for rapid in vivo detection and quantification of GI bleed. MPI-tailored, long-circulating SPIOs were systemically administered into a disease model of GI bleed. The tracer biodistribution was then tracked over time using our custom-built high resolution FFL MPI scanner which was described in Chapter 3. Dynamic MPI projection images captured tracer accumulation in the lower GI tract with excellent contrast. The robust contrast, sensitivity, safety, ability to image anywhere in the body, along with our MPI-tailored long-circulating SPIOs, demonstrates the outstanding promise of MPI as a clinical diagnostic tool.

Magnetic Particle Imaging is a novel imaging modality with many applications in the preclinical, and soon, clinical space. In particular, MPI

shows promise for imaging of various brain related pathologies, such as stroke, hemmorhage, and traumatic brain injury. In this thesis, we

demonstrate a series of first-in-animal proof-of-concept experiments that MPI could soon be superior to our conventional imaging technologies, including X-ray CT, MRI, ultrasounds and nuclear medicine for particular neuroimaging applications. In the process, we will develop algorithmic enhancements to the image reconstruction of MPI signals in order to achieve real time interventional imaging, much like X-ray fluoroscopic imaging, but without ionizing radiation and significant risksof catheter and iodinated contrast media.

Magnetic Particle Imaging (MPI) is an emerging tracer imaging modality that utilizes safe, low-frequency magnetic fields and an existing, human-safe, superparamagnetic iron oxide (SPIO) nanoparticle tracer. MPI already shows high contrast and high sensitivity in small animal imaging. The technique exploits the nonlinear magnetization response of SPIO nanoparticles to time-varying magnetic fields at very low frequencies (VLF). Hence, for medical imaging, MPI only detects a signal from the tracers and not from the diamagnetic biological tissue. Moreover, since tissue is completely transparent to VLF magnetic fields, there is no depth attenuation of the MPI signal. Thus, the physics of MPI shows that it has the ideal contrast for tracer imaging, and is ideally suited for clinical applications such as angiography, cancer imaging, inflammation imaging, and in vivo stem cell therapy tracking.

The fundamental advantages of MPI as a tracer imaging modality are its superb tracer sensitivity, ideal image contrast, and safety of the tracer and modality. Our lab has already shown experimentally that MPI can detect a minuscule sample of 10 nanogram (100 nM) of tracer in a prototype scanner. In principle, there is another 2 orders of magnitude achievable improvement before the sensitivity reaches the physical limit of the technique. MPI is currently shown to have 200x higher net signal-to-noise ratio (SNR) than magnetic resonance imaging (MRI), and this sensitivity is approaching that of the nuclear medicine, such as positron emission tomography (PET) and single-photon emission computerized tomography (SPECT). Moreover, because MPI tracer is not radioactive, the dose-limited sensitivity can easily exceed PET and SPECT and effectively be more sensitive. MPI has ideal image contrast because the contrast is positive, quantitative, has no tissue background, and independent of depth. Lastly, MPI has ideal tracer and modality safety. MPI tracers, notably SPIO nanoparticles, have been shown to be much safer for patients with chronic kidney disease than currently available tracers (iodine and gadolinium) used in planar X-ray imaging, X-ray computed tomography (CT), and MRI. In addition, MPI uses no ionizing radiation, and thus is safer than X-ray, CT, PET and SPECT.

MPI is still a young technology in the medical imaging field. With 10 years of development since the first introduction of the technique in 2005, the current state of MPI research is very much like MRI in the 1980s. There remain many open challenges to be addressed, which makes this field very exciting. In this thesis, we will investigate and address three major challenges in MPI that are crucial for the preclinical and clinical adoption. These challenges are: 1) Restoration of MPI's linearity and shift-invariance (LSI), which are hallmarks of almost all clinically relevant imaging modalities; 2) Achieving isotropic resolution,which is an indispensable characteristic for any diagnostic and quantitative imaging technique; and 3) Understanding the source of the background image haze and eliminate it, which is essential for further improving image resolution, contrast and conspicuity.

We begin by investigating the LSI properties in MPI. In MPI, high-pass filters designed to remove unavoidable direct feedthrough interference also remove information crucial to ensuring LSI in MPI scans. We present a complete theoretical and experimental description of the image artifacts from filtering, and propose and validate a robust algorithm to completely restore the lost information for the x-space MPI method. We provide the theoretical, simulated, and experimental proof that our algorithm indeed restores the LSI properties of MPI, which is indispensable for quantification and diagnostic utility.

We then detailed an investigation into one of MPI's unique resolution challenge: MPI's point spread function (PSF) is highly dependent on the scanning parameters, and every experimental MPI scan ever created lacks desirable isotropic resolution, which leads to ambiguous and inaccurate diagnosis. In this thesis, we generalized a tensor imaging theory for multidimensional x-space MPI to explore the physical source of this anisotropy, presented a multichannel hardware and scanning trajectory to remove anisotropy, and designed and constructed two orthogonal excitation and detector coils to enable isotropic resolution. We experimentally verified the resolution improvement with the new hardware and reconstruction, and showed that isotropic resolution enabled accurate diagnosis of stenosis in small human arterial phantoms.

Lastly, we investigated on MPI's reduced image contrast due to significant background image haze. We have found that the image haze comes from the undesirable rotation of the nanoparticle magnetic moment in response to the applied field outside of the scanning region. Consequently, the native PSF contains a hazy component that falls off as 1/Gr, where G is the gradient field strength, and r is the radially symmetric spatial coordinate. This haze resembles the haze seen in CT images reconstructed with non-filtered backprojection. We propose that we can reshape the MPI PSF with k-space equalization filter that dehazes the image without any noise amplification. We demonstrate experimentally that equalization dramatically increases image contrast and enables the first quantitative measurements of lumen size in a sub-millimeter diameter blood vessel phantom.

In conclusion, this thesis work has proposed significant advancement in imaging theory, hardware and algorithm for MPI that ensures LSI properties, and improves image resolution, contrast and conspicuity. Taken together, these are major contributions to the fundamental imaging science of MPI. LSI and sharp isotropic resolution is essential for quantitative imaging, and could foster clinical adoption of MPI.

In MRI, a powerful and uniform static B0 field is responsible for image signal and contrast. However, regions of different magnetic susceptibilities in the static field give rise to field inhomogeneities that cause image artifacts. For example, air and tissue interfaces for a patient in a MRI study induce up to ±5 ppm static B0 field perturbations within the patient. Unfortunately shim coils are too large to compensate for the very steep field variations found near the breast, cervical spine, shoulders, occipital lobe, and other regions of the body. Many MRI applications are vulnerable to B0 inhomogeneity, including robust fat suppression, which requires better than ±1 ppm homogeneity.

To address this challenge, we have developed a flexible, conforming composite foam with magnetic susceptibility matched to human tissue. We surround the region with enough matching material to move the field gradients outside of the body where they cannot cause MRI artifacts. Our matching material is a pyrolytic graphite (PG) composite foam. It has many advantages over existing matching agents: it is lightweight; it adds no noise; and it is safe to embed receiver coils within the PG foam.

In this thesis, we propose theory that describes the magnetic susceptibility matching properties of composite PG foam. We show experimental proof of concept susceptibility matching and safety in phantom experiments. We also experimentally demonstrate that PG foam cushions improve the B0 field uniformity to the critical threshold of ±1 ppm in both phantoms and in the neck of 6 normal volunteers at 3T. The tissue susceptibility matched PG foams consistently mitigate signal drop out, improve image SNR, and enable far more robust frequency selective fat suppression in T1-weighted FLASH images in volunteers. PG foam has many practical applications in MR applications that require a pristine B0 field.

Magnetic particle imaging (MPI) is a new and rapidly developing imaging modality. Here we explore new ways to use MPI for molecular imaging and theranostic applications. A major focus is exploring methods to leverage the magnetic relaxation dynamics associated with the tracers used in MPI. In the context of canonical MPI implementations, these dynamics can reduce resolution and signal; however, they are also a source of molecular imaging contrast. In the first part of this thesis, we experimentally characterize relaxation in the canonical sinusoidal MPI implementation and explore novel ways to exploit this relaxation for colorized imaging and theranostic applications. In the second part, we describe a new approach to signal encoding in MPI that we call pulsed MPI (pMPI). We show that pMPI can allow one to fully prevent relaxation-induced resolution loss and can even provide ways to improve resolution over traditional continuous wave encoding techniques. Furthermore, pMPI can be leveraged to directly quantify magnetic relaxation that does occur and report this information in an imaging format as relaxation images (in contrast to the standard tracer mass/concentration images). Such approaches may pave the way for greatly improved color MPI and related molecular imaging methods. We believe a major takeaway from this work is that there remains significant untapped potential in MPI by way of unexploited physics. We believe development of these possibilities will be an exciting part of our field's future.

Cardiovascular Disease (CVD) is the leading cause of death in the United States. CVD is

caused by a buildup of plaque in the blood vessels of the heart and brain. CVD is comorbid

with Peripheral Vascular Disease (PVD), which is a buildup of plaque in the extremities

and organs below the stomach. Medical imaging is a clinically indispensable tool for diagnosing CVD and PVD, as well as guiding real-time interventional procedures. Currently,

the two main procedures for diagnosing CVD and PVD are X-Ray and CT angiography,

both of which rely on injection of iodinated contrast agents. However, about 25% of patient

undergoing these procedures suer from Chronic Kidney Disease (CKD). For this patient

subpopulation, iodinated contrast agents are risky and can lead to complete kidney failure.

Physicians risk damaging patients' kidneys in order to alleviate the more imminent threat of

death from a heart attack. Hence, an open challenge remains to develop a safer diagnostic

angiography method that could match current methods in resolution, contrast, and speed.

Here we introduce and develop two new imaging methods for imaging patients with CKD.

The rst method is MR Saline Angiography, an alternative Magnetic Resonance (MR) based

angiographic method for coronary imaging. In combination with the inherent resolution and

speed of traditional MR methods, MR Saline Angiography has the advantage of utilizing

saline as a safe contrast agent. This approach is an improvement upon the commonly used

Gadolinium-based contrast agent that is also toxic to patients with CKD. Using a novel

electromagnetically shielded catheter and a tailored pulse sequence, we have demonstrated

that MR Saline Angiography is a feasible alternative for imaging coronary arteries. The second method is Magnetic Particle Imaging (MPI), a tracer imaging modality, which images

Superparamagnetic Iron Oxide nanoparticles. MPI is an emerging imaging technique with

possible applications in angiography as well as targeted and non-targeted tumor imaging,

hyperthermia, perufsion, cell tracking and targeted drug delivery. Here we study the feasibility

of MPI as an angiographic imaging method by looking at short-term biodistribution

and long-term iron clearance for two most common MPI tracers.

Magnetic Particle Imaging (MPI) is an emergent medical imaging technology that directly images the intense magnetization of clinically safe superparamagnetic iron oxide (SPIO) nanoparticles. Because biological tissues do not produce signals detectable in MPI scanners, MPI images have extremely high image contrast and sensitivity for SPIO tracers, akin to nuclear medicine imaging techniques. The MPI signal is also linearly proportional to SPIO tracer concentrations in the imaging volume, making it a truly quantitative imaging technique. Hence, because of its high image contrast, sensitivity, quantitativeness, and tracer safety, MPI may be useful in applications ranging from coronary angiography to stem cell therapy tracking and is extremely promising for translation to the clinic.

The physical basis behind signal generation in MPI is unlike that of any other imaging modality, giving MPI the potential to be one of the most sensitive medical imaging modalities. However, several limitations have prevented existing MPI scanners from reaching the physical limits of detection sensitivity. These limitations include direct feedthrough interference from the MPI transmitter to the receiver, which can obscure the desired SPIO magnetization signal and limit SNR. In Chapter 2 of this dissertation, I aim to determine the sources of interference generation in MPI scanners and to develop engineering solutions to attenuate the feedthrough interference in the detected signal spectrum. My results indicate that feedthrough interference can arise from high-power passive components used to generate the MPI drive fields, as well as from the interaction between the MPI drive field and the magnets used to generate the MPI magnetic field gradient. To remove these interfering signals for improved detection sensitivity, I designed an actively-controlled magnetic interference cancellation system using a Cartesian feedback controller. Data from this active cancellation system has shown the ability to suppress interfering MPI signals by over 55 dB.

Another limitation in the sensitivity of existing MPI scanners is the use of non-optimized electronics in the MPI detector chain, which can add substantial noise beyond the noise mechanisms generated in the patient and in the detector coil. In Chapter 3 of this dissertation, I investigate the sources of electronic noise in MPI and describe three methods to reduce noise from the MPI detector preamplifier to below the noise generated by the MPI detector coil. Using these noise-matching techniques, I then describe the design and implementation of a custom transformer-coupled MPI preamplifier. Finally, using a 7 T/m preclinical MPI scanner, I demonstrate that the custom MPI preamplifier can achieve an 11-fold improvement in the signal-noise ratio (SNR) of MPI scanners over a commercially available low-noise preamplifier.

Building upon these improvements in the sensitivity and SNR of our preclinical MPI scanners, I then perform the first two in vivo experiments to track implanted stem cell therapies in rodent models. In Chapter 4, we show that MPI can be used to sensitively and quantitatively track stereotactically implanted neural progenitor cell xenografts over an 87- day period. In Chapter 5, we show the first use of MPI to systemically monitor intravenously implanted therapeutic cells. MPI was able to visualize the entrapment of large mesenchymal stem cells in lung vasculature during circulation and quantify the gradual clearance of these cells through the liver over a period of 12 days. Importantly, these experiments demonstrate the ability of MPI to sensitively trace small quantities of SPIOs in the body, potentially enabling new clinical approaches to metastatic cancer detection and the diagnosis of other systemic diseases.

Magnetic particle imaging (MPI) is an emerging tracer imaging modality. In this work, we develop novel scanning strategies to improve the imaging performance of MPI and explore novel methods to use MPI for theranostic applications. A major focus is inventing and exploring different scanning strategies to overcome or leverage the magnetic relaxation dynamics associated with the tracers used in MPI in order to find optimal trade-offs in imaging performance metrics such as spatial resolution, SNR and scanning time. In the first part of this thesis, we perform an experimental study to underscore the significant discrepancy between experimental performance of large core size tracers and their theoretical performance, and demonstrate that this is a major obstacle to improving MPI performance. Subsequently, we describe the hardware design and construction of a frequency-flexible, arbitrary waveform tabletop scanner to enable the investigation of novel scanning strategies to help address this issue. Conventional MPI uses a single frequency excitation wave around 20 kHz. With this new device, we show the first study of non-sinusoidal excitation waveforms such as square, trapezoidal, triangle waveforms, and based on this, we design a novel approach to scanning and signal encoding in MPI which we name pulsed MPI (pMPI). Subsequently, we show that pMPI unlocks the potential of large core size nanoparticles by suppressing deleterious relaxation-based blurring effects. This enabled the first viable use of these nanoparticles which previously have been unusable in MPI. We also exploit the frequency-flexibility to optimize parameters for continuous (sinusoidal) wave MPI and elucidate an optimal waveform that achieves both good resolution and SNR where previous work has suboptimally traded-off one for another.

In the second part, we perform extensive preclinical studies as the first in vivo proof-of-concept of theranostic MPI. We show that MPI is unique as a theranostic modality due to many key advantages such as precise localization of therapy in vivo, an ability to receive measure and predict dosage based on image-guidance, and finally elucidate an approach for future development of real-time feedback for fine control and definite quantification of dosage. Our in vivo proof-of-concept results show robust localization of the heating dose deposited as well as therapeutic outcomes in a dual tumor xenograft rodent model. Importantly, we prove that this localization method can address one key challenge in Magnetic Hyperthermia, which is to avoid collateral damage to off-target organs such as the liver which tends to accumulate magnetic nanoparticles and is prone to unintended heat damage. Furthermore, we investigate aerosolized magnetic nanoparticles as a method to image the lungs with MPI. We demonstrate that MPI lung imaging with magnetic aerosol can be a viable alternative to clinically established radioaerosol procedures. This lays important groundwork for use of magnetic aerosols and MPI for safer lung imaging and lung theranostics in combination with the abovementioned work.

Magnetic Particle Imaging (MPI) is a noninvasive imaging modality that exploits the saturation properties of superparamagnetic iron oxide particles (SPIOs). A major thrust of MPI research aims to sharpen the magnetic resolution of biocompatible SPIOs, which will be crucial for affordable and safe clinical translation. We recently reported on a new class of MPI tracers —called superferromagnetic iron oxide nanoparticles (SFMIOs) — which offer much sharper magnetic saturation curves. SFMIOs experimentally demonstrate 5-13x improvement in both resolution and sensitivity. However, superferromagnetism is a relatively unexplored branch of physics and the nanoscale physics and dynamics of SFMIOs remain a mystery. Here we show experimentally that chaining of SPIOs can explain SFMIO’s boost in SNR and resolution. We show how concentration, viscosity, transmit amplitude, and prepolarization time can all affect SPIO chain formation and SFMIO behavior. These experiments will inform strategies on SFMIO chemical synthesis as well as SFMIO data acquisition pulse sequences.

Magnetic Particle Imaging (MPI) is an emerging imaging modality with potential clinical applications in rapid angiography, cell therapy tracking, cancer imaging, and inflammation imaging. While still in its infancy, MPI already has excellent contrast, safety, depth penetration, and sensitivity without the serious health risks posed by current modalities such as ionizing radiation and iodinated contrast agents.

With any tracer imaging modality, high sensitivity is necessary to improve safety and enable new clinical applications. The first Magnetic Computed Tomography (MCT) system and experimental images were created in this work with the goal of improving sensitivity in MPI. MCT uses projection reconstruction algorithms similar to those in X-ray computed tomography to reconstruct 3D images from a projection MPI sytem. Analytical derivation and experimental evidence demonstrate that MCT has an order-of-magnitude higher SNR than previous 3D MPI methods for the same scan time as well as a 40% resolution improvement. The MCT experimental system acquisition speed was improved from initial work to acquire images twenty-fold faster at less than two minutes per 3D image.

Critical to any medical imaging technology is the reliability and accuracy of image reconstruction. Unfortunately, prior approaches to x-space MPI reconstruction suffer from image artifacts such as banding and haze. In this work, a priori knowledge of image continuity and non-negativity are introduced into a new optimization formulation to reduce these non-physical artifacts in 2D and 3D reconstructions.