Magnetic Particle Imaging (MPI) offers the potential for background- and radiation-free tomographic imaging with high temporal resolution, surpassing limitations of Computed Tomography (CT) and Magnetic Resonance Imaging (MRI) in terms of radiation exposure, patient accessibility, scan duration, and portability. Current limitations in ICU settings, including patient transport risks and difficulties in monitoring during scans, highlight the need for bedside imaging devices. Portable MRI systems with low B0 fields have shown promise, but MPI offers advantages in size and environmental compatibility. The ability to acquire MPI brain images directly at the patient's bedside within the ICU could significantly reduce workload for medical staff, minimize patient transport risks, and expedite treatment decisions. As a quantitative, tracer-based modality, MPI can visualize blood vessels and quantify tissue perfusion with high temporal resolution, making it ideal for neurovascular disease diagnosis and monitoring (ischemic stroke, intracranial hemorrhage, traumatic brain injury). Given the 17+ million annual cases worldwide and their contribution to mortality and disability, continued research and development of MPI technology is crucial. Current MPI research is focused on scaling up from preclinical (small rodent) to human-sized systems for head and extremity applications. The challenge is to minimize power consumption, particularly for the selection field (SF), while ensuring patient safety near high-power components. This paper describes a human-sized MPI system for brain applications, designed to meet these challenges using standard-socket power, an unshielded environment, and adherence to medical device safety regulations.

The authors reviewed existing literature on MPI technology, highlighting its advantages over CT and MRI for brain imaging, especially in ICU settings where portability and patient monitoring are critical. The literature review also included studies on the use of MPI in preclinical settings for various applications, demonstrating the versatility of the technology and laying the groundwork for its clinical translation. Specific mention is made of prior work on human-sized MPI scanners, including challenges and improvements in the current design.

The study involved the design, construction, and comprehensive system characterization of a human-sized MPI brain scanner. The system comprised four main parts: operational control, field generation, signal reception, and data processing. The operational control utilized open-source software for coordinating signal generation and reception, providing an adaptable and scalable framework. Field generation involved four DACs, two drive-field (DF) coils for 2D excitation in the xz-plane, and a dynamic selection field generator (SFG) using two coils mounted on an iron yoke inside a copper cabin to achieve 3D imaging via a slow shift of the FFP along the y-axis. Signal reception was achieved with a gradiometric receive coil for the x-direction and a saddle coil for the y-direction, both connected to a symmetric band-stop filter and a custom low-noise amplifier (LNA). Data processing utilized the system matrix approach and open-source MPI reconstruction framework, incorporating a Kaczmarz solver, system matrix over-gridding, L1 and L2 regularization, background subtraction, and frequency selection. For perfusion images, post-processing calculated time-to-peak (TTP), mean-transit-time (MTT), relative cerebral blood flow (rCBF), and relative cerebral blood volume (rCBV). The methodology included field analysis using spherical harmonic expansion to determine the actual imaging trajectory. A system matrix was acquired using a robot-based approach with a cubic sample of Resotran. Sensitivity and spatial resolution were evaluated using dilution series and a two-sample approach, respectively. Dynamic perfusion experiments were performed using a flow phantom simulating brain hemispheres with varying stenosis levels, and a bolus injection of Resotran. Multi-contrast imaging was demonstrated using Resotran and Synomag. The hardware implementation details include the drive-field generator (DFG), high-current resonator (HCR), inductive coupling network (ICN), transmit filter, power amplifier, and the receive chain components.

The human-sized MPI brain scanner achieved 3D single- and multi-contrast imaging at a 4 Hz frame rate. Key performance characteristics were determined: a spatial resolution of 12 mm in the x-direction, 7 mm in the y-direction, and 31 mm in the z-direction; a detection limit of 8 µg of iron; and the ability to differentiate five levels of stenosis (25% increments) in perfusion experiments. Multi-contrast imaging successfully discriminated between Resotran and Synomag. Field analysis revealed field inhomogeneities, particularly towards the edges of the FOV. System matrix analysis, interpreted as both multi-patch and single-patch datasets, showed the expected wave-like structure and provided insights into the imaging performance. The scanner exhibited a total power consumption of less than 4 kW and demonstrated robustness to electromagnetic interference in an unshielded environment. The sensitivity study demonstrated the ability to resolve spatial positions down to 8 µg of iron. The quantitative analysis showed a good match between reconstructed and applied iron content at higher iron masses, with deviations at lower masses attributed to background noise and reconstruction artifacts. The spatial resolution experiments revealed varying resolution along the x, y, and z axes. Perfusion experiments, using a flow phantom with varying levels of stenosis, provided time response graphs and perfusion maps showing TTP, MTT, rCBF, and rCBV maps which indicated that fast dynamic imaging was feasible and different levels of stenosis could be detected. Multi-contrast imaging successfully discriminated Resotran and Synomag, although minor channel leakage was observed.

The findings demonstrate the feasibility of real-time 3D MPI brain imaging with high spatial and temporal resolution, suitable for detecting neurovascular diseases. The achieved resolution is sufficient to distinguish between brain hemispheres and varying stenosis levels, making it suitable for clinical application scenarios like ischemic stroke. The use of clinically approved tracers like Resotran is a significant step towards human trials. Future improvements in tracer technology and system design could enhance sensitivity and spatial resolution, expanding the range of potential applications. The system's low power consumption and robustness to electromagnetic interference make it suitable for use in unshielded environments like the ICU. The limitations of the study, such as the use of phantoms and the need for further investigations into channel leakage in multi-contrast imaging, were also noted.

This study provides a comprehensive system characterization of a 3D human-sized MPI scanner for real-time cerebral applications. The results show promising capabilities for neurovascular disease diagnosis and monitoring. Future work should focus on in vivo studies and further optimization of system parameters. The use of tailored MPI tracers and improvements in reconstruction algorithms promise to further enhance system performance.

The study's limitations include the use of phantoms instead of in vivo experiments, the need for further investigation into channel leakage in multi-contrast imaging, and the potential influence of field imperfections on the accuracy of quantitative measurements. Additionally, while the power consumption is low, further optimization could reduce it even further. The study focused on a specific imaging sequence and FOV size. Further work is needed to assess the system's performance under diverse conditions and with larger FOV.