Search

Time- and memory-consuming calibration measurements are a major drawback in system-matrix-based reconstructions in magnetic particle imaging (MPI). Especially, exchanging the receive coils requires new system matrices and therefore new calibration measurements. To reduce the number of such measurements, the MPI transfer function can be used to transfer the system matrix of one receive coil set to another. The transfer function can be obtained by a direct measurement or by estimation using a measured system matrix of each setup. In this abstract, we extend the latter to incorporate coupling of the receive coils while using only a few voxels of the system matrices. In this way, we can transfer system matrices between two different receive setups, both of which may contain non-orthogonal coils., International Journal on Magnetic Particle Imaging IJMPI, Vol 9 No 1 Suppl 1 (2023)

Accurate modeling of the mean magnetic moment of an ensemble of magnetic particles in dynamic magnetic fields is a challenging task that requires sophisticated differential equation solvers. However, these methods are computationally costly and therefore not practical for long excitation sequences such as those of the Lissajous type. In this paper we propose to accelerate simulations by using a neural network mapping from the input parameter functions that are applied to the original particle simulator directly to the mean magnetic moment output function. The architecture of the neural network is based on the Fourier neural operator, which allows to train mappings between function spaces. Our results show that the particle simulation can be accelerated by a factor of about 200 while the relative error of the neural network simulator remains below 1.5%., International Journal on Magnetic Particle Imaging IJMPI, Vol 9 No 1 Suppl 1 (2023)

In iterative system-matrix-based reconstruction in Magnetic Particle Imaging, three major parameters control the amount of regularization. Finding the right choice for these parameters is commonly done by user input and requires time and experience. We propose a method that enables automatic reconstruction and achieves good results on a measured concentration series., International Journal on Magnetic Particle Imaging IJMPI, Vol 9 No 1 Suppl 1 (2023)

Magnetic particle imaging (MPI) is a pre-clinical imaging modality, whose system design is still evolving, in particular towards human studies and clinical use. Therefore, many MPI scanners are custom-made distributed systems, both on the hard- and the software side. In this work we present the open-source Julia framework MPIMeasurements.jl, which implements a composable representation of imaging systems. It also offers flexible data structures that allow the implementation of specific imaging protocols, such as online/offline measurements, repeated measurements and system matrix calibrations. %that are reusable across systems. The project is designed to be expanded to new systems through community development and component reuse. To showcase the versatility of the software package, we give an overview of four very different MPI systems, which were realized with MPIMeasurements.jl., International Journal on Magnetic Particle Imaging IJMPI, Vol 9 No 1 Suppl 1 (2023)

A recent work introduced a balloon catheter from a polymer mixed with magnetic nanoparticles that was visible in magnetic particle imaging (MPI) in initial trials. In this work, dynamic multi-patch and multi-contrast experiments were performed to validate the usability of the balloon catheter for complex application scenarios. In both cases, the balloon was successfully imaged, although there are limitations in resolution and dynamic imaging range., International Journal on Magnetic Particle Imaging, Vol 8 No 1 Suppl 1 (2022)

Non-sinusoidal excitation waveforms have the ability to improve the signal-to-noise ratio and image resolution under certain conditions. Yet, the ability to use phase information for spatial encoding is expected to diminish as sharp pulses lead to concurrent signal response due to steep slopes and therefore less phase information. This motivates investigations into alternate sampling approaches that mitigate a loss in spatial encoding. However, measurements and image reconstruction results indicate that 10 times faster slew rates compared to sine excitation lead to enough phase information to resolve basic features using system matrix reconstruction., International Journal on Magnetic Particle Imaging, Vol 8 No 1 Suppl 1 (2022)

In multi-patch magnetic particle imaging an artifact-free image can be obtained by using a joint reconstruction and measuring the system matrices not only in the field-of-view but also in a huge overscan. This leads to a long calibration time and heavy memory consumption and therefore an unsuitability of this method for large three-dimensional measurements. In this work we propose to measure the system matrices only in the field-of-view and use a diffusion based extrapolation step to extant the system matrices computationally into the overscan. In this way we massively reduce the calibration time while maintaining a nearly artifact-free image., International Journal on Magnetic Particle Imaging, Vol 8 No 1 Suppl 1 (2022)

A major issue for human-sized Magnetic Particle Imaging (MPI) scanners is the generation of sufficiently large magnetic gradient fields. By taking advantage of the field amplification properties of soft iron, a considerable amount of power can be saved. In this work, an optimized selection and focus field generator is presented, that can generate flexible and high gradient fields at comparatively low power consumption. Coil spacing and possible field-free-point positions are similar to conventional MPI scanners designed with air coils, but with significantly less demands on infrastructure and cooling design. The optimization process is discussed and first field measurements are presented., International Journal on Magnetic Particle Imaging, Vol 8 No 1 Suppl 1 (2022)

Magnetic particle imaging is capable of determining very small concentrations of particles if only a single concentration is present in the field-of-view. Meanwhile the determination of particles with widely differing concentrations is still challenging. In a recent work, we introduced a two-step reconstruction method that tackles this problem by isolating the signal of the lower concentrated tracer for a separate reconstruction. In this work, we adapt the two-step reconstruction method in order to apply a joint reconstruction to the entire signal of all particle concentrations. This is achieved by spatially adaptive Tikhonov regularization., International Journal on Magnetic Particle Imaging, Vol 8 No 1 Suppl 1 (2022)

Nanoparticles tend to agglomerate following their in vivo or in vitro application. This leads to particle interaction and, for magnetic particle imaging (MPI) tracers, to magnetic coupling phenomena. Here, we investigate these effects and their influence on magnetic particle spectroscopy (MPS) and MPI signal stability. Highly magnetic flame-made Zn-ferrites with controlled interparticle distance are suggested as a stable MPI tracer system. Due to their pre-aggregated morphology, additional agglomeration does not substantially alter their magnetic response. This is in strong contrast to frequently investigated polymer-coated iron oxide nanoparticles, which show a massive MPS signal loss in a biologically relevant dispersion medium compared to water. This effect is also shown during MPI and renders these tracers inapplicable to further applications. Our flame-made Zn-ferrites, on the other hand, show sufficient signal stability, which allows their detailed quantification via MPI., International Journal on Magnetic Particle Imaging, Vol 8 No 1 Suppl 1 (2022)

Magnetic Particle Imaging (MPI) is a tracer based medical imaging modality with great potential due to its high sensitivity, high spatio-temporal resolution, and ability to quantify the tracer concentration. Image reconstruction in MPI is an ill-posed problem, which can be addressed by the use of regularization methods. The corresponding optimization problem is most commonly solved using the Kaczmarz algorithm. Reconstruction using the Kaczmarz method for single-contrast MPI is very efficient as it produces the desired images fast with a small number of iterations. For multi-contrast MPI, however, the regular Kaczmarz algorithm fails to obtain good quality images without channel leakage when using a small number of iterations. In this work, we propose to use an accelerated Kaczmarz method in order to reduce the reconstruction time needed to achieve a good separation of the channels and a good image quality in multi-contrast MPI., International Journal on Magnetic Particle Imaging, Vol 8 No 1 Suppl 1 (2022)

Magnetic particle imaging uses time-dependent drive fields for signal generation, which are designed to be homogeneous in order to achieve a similar image quality all over the volume of interest. In practice, the fields are not exactly homogeneous and in turn a precise knowledge of the spatial field profile is necessary when using a model-based reconstruction approach. In this work, we propose an efficient method for the measurement and compact representation of the drive fields using a small 3D calibration coil and solid harmonics., International Journal on Magnetic Particle Imaging, Vol 8 No 1 Suppl 1 (2022)

It was shown that the encapsulation of SPIO-based contrast agents in the red blood cells (RBCs) increases the circulation time in blood of these nanomaterials. Not all iron oxide particles are eligible for the entrapment into RBCs, depending on several factors and synthesis protocol. We have recently identified some type of nanoparticles that can be loaded with our method into RBCs to produce biocompatible SPIO-RBCs carriers that could be used as new intravascular tracers for biomedical applications, such as Magnetic Particle Imaging (MPI). Here, we report the first in vitro results obtained by using the Synomag��-D-PEG-OMe nanoparticles with both human and murine RBCs. MPS analysis showed that human Synomag��-D-PEG-OMe-loaded RBCs produced a signal that is weaker respect to the remarkable signal obtained with ferucarbotran loaded-RBCs prepared at the same condition, but it is to be noted that the encapsulation efficiency of Synomag��-D-PEG-OMe into cells is lower compared to ferucarbotran nanoparticles., International Journal on Magnetic Particle Imaging, Vol 8 No 1 Suppl 1 (2022)

Magnetic particle imaging (MPI) uses orthogonal receive coils aligned with the drive-field coil directions for signal detection. In case of misaligned receive coils, the particle magnetization couples differently into the receive coils such that a re-calibration involving the measurement of a new system function would be necessary. In this work, we propose a method for decoupling the channels algorithmically into the drive-field coil directions using an MPI transfer function matrix. In that way, system functions for different receive coil units can be reused., International Journal on Magnetic Particle Imaging, Vol 8 No 1 Suppl 1 (2022)

Modern imaging modalities, such as magnetic particle imaging (MPI), are based on complex sequences which require synchronous multi-channel signal generation and reception. The component of an MPI scanner responsible for this functionality is the data acquisition (DAQ) system. Different scanner topologies and the nature of (digital) signal processing impose varying requirements on such a system. In this work, we introduce the RedPitayaDAQServer project, which implements a flexible scalable DAQ system. It is based on the low-cost hardware RedPitaya STEMlab 125-14 and is able to meet the requirements of most MPI scanner concepts that have been proposed to date., International Journal on Magnetic Particle Imaging, Vol 8 No 1 Suppl 1 (2022)

The achievable sensitivity in Magnetic Particle Imaging is not only limited by noise, but also depends on the stability of the system. Thermal dependencies of the current carrying components lead to drive-field distortions in amplitude and phase causing drifting background signals. In this work, an active capacitor heating system is developed that allows for thermal stabilization and trimming a resonance circuit to the desired frequency., International Journal on Magnetic Particle Imaging, Vol 8 No 1 Suppl 1 (2022)

For imaging and force experiments in Magnetic Particle Imaging several field generating coils are required to produce sufficiently high and flexible magnetic fields. To minimize power consumption, coils with iron cores are the best choice for low and medium frequency ranges. Such coils have comparatively high reactance and often are inductively coupled. The trivial approach to ensure target currents is to provide each coil with a current controlled source resulting in high system complexity and high costs. This paper presents a circuit design to distribute bipolar target currents from a single unipolar source with high accuracy, reducing unwanted coil coupling by a feedback controller. Thus, the number of current sources can be significantly reduced. With a regenerative concept, reactive power is stored and can be reused, allowing efficient and fast current switching., International Journal on Magnetic Particle Imaging, Vol 8 No 1 Suppl 1 (2022)

For the utilization of the magnetic field of rotating Halbach cylinders in MPI scanners, a rotation unit is required which houses the used permanent magnets and allows their rotation. The setup proposed here allows the concentric arrangement and individual rotation of four Halbach cylinders, whose purely mechanic agitation allows the movement of the field-free region through the field of view. Based on the requirements for such a rotation system, basic design approaches are presented and compared. Initial strength calculations and further simulations were performed on the resulting constructive concept, which, upon implementation, forms the basis for almost purely mechanical MPI scanners., International Journal on Magnetic Particle Imaging, Vol 8 No 1 Suppl 1 (2022)

Stroke is one of the leading worldwide causes of death and sustained disability. Rapid and accurate assessment of cerebral perfusion is essential to diagnose and successfully treat stroke patients. Magnetic particle imaging (MPI) is a new technology with the potential to overcome some limitations of established imaging modalities. It is an innovative and radiation-free imaging technique with high sensitivity, specificity, and superior temporal resolution. MPI enables imaging and diagnosis of stroke and other neurological pathologies such as hemorrhage, tumors, and inflammatory processes. MPI scanners also offer the potential for targeted therapies of these diseases. Due to lower field requirements, MPI scanners can be designed as resistive magnets and employed as mobile devices for bedside imaging. With these advantages, MPI could accelerate and improve the diagnosis and treatment of neurological disorders. This review provides a basic introduction to MPI, discusses its current use for stroke imaging, and addresses future applications, including the potential for clinical implementation. This article is categorized under: Diagnostic Tools > In Vivo Nanodiagnostics and Imaging Therapeutic Approaches and Drug Discovery > Nanomedicine for Neurological Disease. Forschungszentrum Medizintechnik Hamburg Deutsche Forschungsgemeinschaft (DFG) Hermann und Lily Schilling Stiftung Bundesministerium für Bildung und Forschung (BMBF)

Recently, the potential of red blood cells (RBCs) loaded with superparamagnetic iron oxide (SPIO)-based nanoparticles as new blood-pool tracer material for the Magnetic Particle Imaging (MPI) has been investigated. It was shown that the encapsulation of SPIO-based contrast agents in the RBCs increase the circulation time in blood of these nanomaterials. However, not all iron oxide nanoparticles are eligible to the encapsulation into RBCs, depending on several factors such as dispersant agent nature, nanoparticle size and synthesis protocol. Therefore, we have recently started a program to identify those nanoparticles that can be potentially loaded with our method into RBCs. The goal is to produce biocompatible SPIO-RBCs carriers that can be used as new intravascular magnetic susceptible agents in biomedical applications, such as MRI and MPI. Here, we report the in vitro results obtained by using the Synomag®-D-PEG-OMe nanoparticle suspension (micromod Partikeltechnologie GmbH) with both human and murine red blood cells. MPS analysis showed that human Synomag®-D-PEG-OMe-loaded RBCs produced a signal that is weaker respect to the remarkable signal obtained with ferucarbotran loaded-RBCs prepared at the same condition, but it is to be noted that the encapsulation efficiency of Synomag®-D-PEG-OMe into cells is lower compared to ferucarbotran nanoparticles. Int. J. Mag. Part. Imag. 8(1), 2022, Article ID: 2209001, DOI: 10.18416/IJMPI.2022.220900, International Journal on Magnetic Particle Imaging, Vol 8 No 1 (2022)

In Magnetic Particle Imaging (MPI), the system matrix plays an important role, as it encodes the relationship between particle concentration and the measured signal. Its acquisition requires a time-consuming calibration scan, which can be a limiting factor in practical applications. Calibration time can be reduced using compressed sensing, which exploits the knowledge that the MPI system matrix has a sparse representation in a suitably chosen domain. This work seeks to further enhance sparse system matrix recovery by optimizing the sampling points to the signal class at hand. For this purpose we introduce an experiment design method based on the Bayesian Fisher information matrix. Our technique uses a previously measured system matrix to tailor the sampling pattern to the signal class at hand. Our tests show that the optimized sampling patterns lead to a more accurate system matrix recovery than popular random sampling approaches. Moreover, our tests demonstrate that the optimized sampling patterns are sufficiently robust to enhance the recovery of system matrices for other types of particles or other experimental conditions. Int. J. Mag. Part. Imag. 7(2), 2021, Article ID: 2112001, DOI: 10.18416/IJMPI.2021.2112001, International Journal on Magnetic Particle Imaging, Vol 7 No 2 (2021)

Purpose: The aim of this work is to develop a high-performance, flexible and easy-to-use MRI reconstruction framework using the scientific programming language Julia. Methods: Julia is a modern, general purpose programming language with strong features in the area of signal / image processing and numerical computing. It has a high-level syntax but still generates efficient machine code that is usually as fast as comparable C/C++ applications. In addition to the language features itself, Julia has a sophisticated package management system that makes proper modularization of functionality across different packages feasible. Our developed MRI reconstruction framework MRIReco.jl can therefore reuse existing functionality from other Julia packages and concentrate on the MRI-related parts. This includes common imaging operators and support for MRI raw data formats. Results: MRIReco.jl is a simple to use framework with a high degree of accessibility. While providing a simple-to-use interface, many of its components can easily be extended and customized. The performance of MRIReco.jl is compared to the Berkeley Advanced Reconstruction Toolbox (BART) and we show that the Julia framework achieves comparable reconstruction speed as the popular C/C++ library. Conclusion: Modern programming languages can bridge the gap between high performance and accessible implementations. MRIReco.jl leverages this fact and contributes a promising environment for future algorithmic development in MRI reconstruction.

Objective: The aim is to evaluate whether smart worklist prioritization by artificial intelligence (AI) can optimize the radiology workflow and reduce report turnaround times (RTATs) for critical findings in chest radiographs (CXRs). Furthermore, we investigate a method to counteract the effect of false negative predictions by AI—resulting in an extremely and dangerously long RTAT, as CXRs are sorted to the end of the worklist. Methods: We developed a simulation framework that models the current workflow at a university hospital by incorporating hospital-specific CXR generation rates and reporting rates and pathology distribution. Using this, we simulated the standard worklist processing “first-in, first-out” (FIFO) and compared it with a worklist prioritization based on urgency. Examination prioritization was performed by the AI, classifying eight different pathological findings ranked in descending order of urgency: pneumothorax, pleural effusion, infiltrate, congestion, atelectasis, cardiomegaly, mass, and foreign object. Furthermore, we introduced an upper limit for the maximum waiting time, after which the highest urgency is assigned to the examination. Results: The average RTAT for all critical findings was significantly reduced in all prioritization simulations compared to the FIFO simulation (e.g., pneumothorax: 35.6 min vs. 80.1 min; p < 0.0001), while the maximum RTAT for most findings increased at the same time (e.g., pneumothorax: 1293 min vs 890 min; p < 0.0001). Our “upper limit” substantially reduced the maximum RTAT in all classes (e.g., pneumothorax: 979 min vs. 1293 min/1178 min; p < 0.0001). Conclusion: Our simulations demonstrate that smart worklist prioritization by AI can reduce the average RTAT for critical findings in CXRs while maintaining a small maximum RTAT as FIFO. Key Points: • Development of a realistic clinical workflow simulator based on empirical data from a hospital allowed precise assessment of smart worklist prioritization using artificial intelligence. • Employing a smart worklist prioritization without a threshold for maximum waiting time runs the risk of false negative predictions of the artificial intelligence greatly increasing the report turnaround time. • Use of a state-of-the-art convolution neural network can reduce the average report turnaround time almost to the upper limit of a perfect classification algorithm (e.g., pneumothorax: 35.6 min vs. 30.4 min).

Magnetic Particle Imaging (MPI) is used to visualize the distribution of superparamagnetic nanoparticles within 3D volumes with high sensitivity in real time. Recently, MPI is utilized to navigate micron-sized particles and micron-sized swimmers, since the magnetic field topology of the MPI scanner is well suited to apply magnetic forces. In this work, we analyze the magnetic mobility and imaging performance of nanomag/synomag-D for Magnetic Particle Imaging/Navigation (MPIN). With MPIN the focus fields are constantly switching between imaging and magnetic force mode, thus enabling quasi-simultaneous navigation and imaging of particles. In flow bifurcation experiment with a 100 % stenosis on one branch, we determine the limiting flow velocity of 1.36 mL/s, which allows all particles to flow only through one branch towards the stenosis. During this experiment, we image the accumulation of the particles within the stenosis. In combination with therapeutic substances, this approach has a high potential for targeted drug delivery. Int. J. Mag. Part. Imag. 6(2), Suppl. 1, 2020, Article ID: 2009025, DOI: 10.18416/IJMPI.2020.2009025, International Journal on Magnetic Particle Imaging, Vol 6 No 2 Suppl. 1 (2020)

In Magnetic Particle Imaging, much of the power consumed during an imaging sequence is used for the generation of the selection and focus fields. In today’s MPI scanners three different concepts are applied to generate the gradient fields: Air coils, permanent magnets and coils with soft iron. Air coils and permanent magnets have the great advantage of good calculability by the Biot-Savart Law. On the way to a clinical imaging modality, the needed power for sufficient gradient strength demand the use of soft iron. In order to make good use of the ferromagnetic amplification properties, much more complex simulations have to be done. A recently published head scanner uses a soft iron yoke for field generation. In this study, we investigated different coil geometries with soft iron with respect to this head scanner. Int. J. Mag. Part. Imag. 6(2), Suppl. 1, 2020, Article ID: 2009042, DOI: 10.18416/IJMPI.2020.2009042, International Journal on Magnetic Particle Imaging, Vol 6 No 2 Suppl. 1 (2020)

The MPI image reconstruction problem requires, particularly for 2D and 3D excitation patterns, a measured system matrix due to the lack of an accurate model that is capable of describing the nanoparticles’ magnetization behavior in the MPI setup. Here we exploit a model based on Néel rotation for large particle ensembles and we find model parameters that describe measured 2D MPI data with much higher precision than state of the art MPI models, which is also illustrated in phantom experiments. This is a short summary of the recent work [4] to which we refer to for all further details. Int. J. Mag. Part. Imag. 6(2), Suppl. 1, 2020, Article ID: 2009004, DOI: 10.18416/IJMPI.2020.2009004, International Journal on Magnetic Particle Imaging, Vol 6 No 2 Suppl. 1 (2020)

Magnetic Particle Imaging (MPI) is a tracer based medical imaging modality with great potential due to its high sensitivity, high spatial and temporal resolution, and ability to quantify the tracer concentration. Image reconstruction in MPI is an ill-posed problem that can be addressed by regularization methods that each lead to a bias. Reconstruction bias in MPI is most apparent in a mismatch between true and reconstructed tracer distribution. This is expressed globally in the spatial support of the distribution and locally in its intensity values. In this work, MPI reconstruction bias and its impact are investigated and a two-step debiasing method with significant bias reduction capabilities is introduced. Int. J. Mag. Part. Imag. 6(2), Suppl. 1, 2020, Article ID: 2009041, DOI: 10.18416/IJMPI.2020.2009041, International Journal on Magnetic Particle Imaging, Vol 6 No 2 Suppl. 1 (2020)

The reconstruction of multi-patch magnetic particle imaging data requires a compromise between image quality and calibration time. While optimal image quality is ensured by the joint reconstruction approach, a system matrix needs to be acquired for each patch. One can reuse system matrices by shifting them in space, which decreases the calibration effort but leads to distortions due to field imperfections. In this work, we introduce a method for reducing displacement artifacts in the efficient joint multi-patch reconstruction. Based on the magnetic fields we propose a mapping that warps the central system matrix to capture the spatial displacement of off-center system matrices. In this way, we can maintain the low calibration time while significantly improving the image quality. Int. J. Mag. Part. Imag. 6(2), Suppl. 1, 2020, Article ID: 2009030, DOI: 10.18416/IJMPI.2020.2009030, International Journal on Magnetic Particle Imaging, Vol 6 No 2 Suppl. 1 (2020)

In the context of system function modeling for magnetic particle imaging, computing fast and accurate approximations to the time evolution of magnetic nanoparticles' (MNPs) mean magnetic moment is a problem of interest. In a software toolbox we comprise algorithms and methods that can simulate Brownian and Néel rotation of MNPs' magnetic moments that can be used to obtain more accurate model-based system matrices than those relying on the so-called equilibrium model. We present and discuss results obtained with these implementations which are made available in the software toolbox under github.com/MagneticParticleImaging/MNPDynamics.jl to inspire further research in these directions Int. J. Mag. Part. Imag. 6(2), Suppl. 1, 2020, Article ID: 2009020, DOI: 10.18416/IJMPI.2020.2009020, International Journal on Magnetic Particle Imaging, Vol 6 No 2 Suppl. 1 (2020)

In Magnetic Particle Imaging the determination of the system matrix is a time-consuming process. As the transfer function differs depending on the electronics of the receive chain, a system matrix recorded with one chain cannot be used for reconstruction of measurement data using a different receive chain. This leads to huge amounts of data as system matrices have to be recorded for every set of drive field strength, particle system and receiver chain setting. In this paper the complexity of this is reduced by the factor of the receiver electronics. For each receive path of our MPI system the transfer function is recorded and stored on the measurement device. When loading the data, the transfer-function is corrected which allows the exchange of the system matrices data between receiver chains. To demonstrate this system matrix data is compared and a successful reconstruction of an in vivo dataset is shown. Int. J. Mag. Part. Imag. 6(2), Suppl. 1, 2020, Article ID: 2009044, DOI: 10.18416/IJMPI.2020.2009044, International Journal on Magnetic Particle Imaging, Vol 6 No 2 Suppl. 1 (2020)

In magnetic particle imaging, the time consuming measurement of a system matrix is required before image reconstruction. Reduction of measurement time can be achieved with the help of compressed sensing, which is based on the sparsity of the system matrix in a suitable transform domain. In this work, we propose regularization functions to exploit the additional correlations in multi- patch system matrices. Experiments show that the resulting recovery method allows successful matrix recovery at higher undersampling factors than a standard compressed sensing recovery. Int. J. Mag. Part. Imag. 6(2), Suppl. 1, 2020, Article ID: 2009035, DOI: 10.18416/IJMPI.2020.2009035, International Journal on Magnetic Particle Imaging, Vol 6 No 2 Suppl. 1 (2020)

Large selection-field power is required to generate a sufficient gradient strength in Magnetic Particle Imaging (MPI). Without cooling, the subsequent heat generation can limit the maximum experiment time. For commercially available MPI scanners a lot of effort was put into active cooling requiring space and infrastructure to dissipate heat. In this abstract, a promising power handling for the selection-field generation is presented. Using a pulsed instead of a continuous selection-field the gradient strength can be increased and no active cooling is required. Int. J. Mag. Part. Imag. 6(2), Suppl. 1, 2020, Article ID: 2009054, DOI: 10.18416/IJMPI.2020.2009054, International Journal on Magnetic Particle Imaging, Vol 6 No 2 Suppl. 1 (2020)

Magnetic particle imaging is a very useful tool in the detection of stroke. To study the ability of stroke in a mouse model the data acquisition is challenging as a mouse brain contains only a very small ratio of blood compared to large animals or humans. The effective concentration within the whole organ is therefore very small, especially compared to the heart or the liver. Typical MPI receiver coils however cover a sensitive region of around 30 mm to 50 mm and have a bore size of above 40 mm. This leads on the one hand to non-optimal signal coupling due to the distance to the particles and on the other hand strong signals from the heart can cause artifacts in the low signal regions. In this work we present a coil optimized for mouse brain imaging, which due to its small size, also dampens signal from regions outside of the coil. Int. J. Mag. Part. Imag. 6(2), Suppl. 1, 2020, Article ID: 2009046, DOI: 10.18416/IJMPI.2020.2009046, International Journal on Magnetic Particle Imaging, Vol 6 No 2 Suppl. 1 (2020)

One approach of image reconstruction in MPI is the system matrix based reconstruction. With this approach, in addition to the particle behavior, the sequence and the scanner properties are also calibrated and stored in a system matrix, so that a linear system of equations for the image reconstruction must be solved. However, the measurement of the system matrix is very time-consuming, depending on the desired spatial resolution. Independently of this, there are some remarkable symmetries within the system matrix that could be exploited to significantly reduce the calibration time. In the context of this work the theoretical description of a system matrix about Chebyshev polynomials is used to completely build a 3D system matrix by mirroring an octant and to successfully reconstruct an image. Int. J. Mag. Part. Imag. 6(2), Suppl. 1, 2020, Article ID: 2009051, DOI: 10.18416/IJMPI.2020.2009051, International Journal on Magnetic Particle Imaging, Vol 6 No 2 Suppl. 1 (2020)

Magnetic particle imaging is a tomographic imaging technique capable of measuring the local concentration of magnetic nanoparticles that can be used as tracers in biomedical applications. Since MPI is still at a very early stage of development, there are only a few MPI systems worldwide that are primarily operated by technical research groups that develop the systems themselves. It is therefore difficult for researchers without direct access to an MPI system to obtain experimental MPI data. The purpose of the OpenMPIData initiative is to make experimental MPI data freely accessible via a web platform. Measurements are performed with multiple phantoms and different image sequences from 1D to 3D. The datasets are stored in the magnetic particle image data format (MDF), an open document standard for storing MPI data. The open data is mainly intended for mathematicians and algorithm developers working on new reconstruction algorithms. Each dataset is designed to pose a specific challenge to image reconstruction. In addition to the measurement data, computer aided design (CAD) drawings of the phantoms are also provided so that the exact dimensions of the particle concentrations are known. Thus, the phantoms can be reproduced by other research groups using additive manufacturing. These reproduced phantoms can be used to compare different MPI systems.

Image reconstruction plays an important role for the tomographic imaging technique magnetic particle imaging (MPI) since the measured raw data cannot be directly interpreted. Instead, one needs to invert the image formation process, which involves the solution of an ill-conditioned linear system of equations. Currently, most MPI researchers have implemented custom reconstruction algorithms that cannot be directly compared since the source code is not openly available. The software package MPIReco.jl aims to change this situation by providing a reference implementation for a variety of reconstruction algorithms. With the recently proposed magnetic particle imaging data format and its reference implementation MPIFiles.jl we have taken the first steps towards standardised data exchange. With MPIReco.jl we complement these initiatives to standardise the reconstruction algorithms and to facilitate reproducible research. We chose to implement the algorithms in the programming language Julia, which provides a high level syntax making the software accessible even for non-professional software developers. On the other hand Julia code has a high run-time performance comparable to low-level C code. In the present paper, we outline some of the design principles of MPIReco.jl and give an overview of the software package. Int. J. Mag. Part. Imag. 5(1-2), 2019, Article ID: 1907001, DOI: 10.18416/IJMPI.2019.1907001, International Journal on Magnetic Particle Imaging, Vol 5 No 1-2 (2019)

Magnetic particle imaging visualizes the spatial distribution of superparamagnetic nanoparticles. Because of its key features of excellent sensitivity, high temporal and spatial resolution and biocompatibility of the tracer material it can be used in multiple medical imaging applications. The common reconstruction technique for Lissajous-type trajectories uses a system matrix that has to be previously acquired in a time-consuming calibration scan, leading to long downtimes of the scanning device. In this work, the system matrix is determined by a hybrid approach. Using the hybrid system matrix for reconstruction, the calibration downtime of the scanning device can be neglected. Furthermore, the signal to noise ratio of the hybrid system matrix is much higher, since the size of the required nanoparticle sample can be chosen independently of the desired voxel size. As the signal to noise ratio influences the reconstruction process, the resulting images have better resolution and are less affected by artefacts. Additionally, a new approach is introduced to address the background signal in image reconstruction. The common technique of subtraction of the background signal is replaced by extending the system matrix with an entry that represents the background. It is shown that this approach reduces artefacts in the reconstructed images.

Magnetic Particle Imaging (MPI) achieves a high temporal resolution, which opens up a wide range of real-time medical applications such as device tracking and navigation. These applications usually rely on automated tech- niques for finding and localizing devices and fiducial markers in medical images. In this work, we show that submillimeter-accurate automatic marker localization from low gradient MPI tomograms with a spatial resolution of several millimeters is possible. Markers are initially identified within the tomograms by a thresholding-based segmentation algorithm. Subsequently, their positions are accurately determined by calculating the center of mass of the voxel intensities inside the pre-segmented regions. A series of phantom measurements taken at full temporal resolution (46 Hz) is used to analyze statistical and systematical errors and to discuss the performance and stability of the automatic submillimeter-accurate marker localization algorithm., International Journal on Magnetic Particle Imaging, Vol.3, No.1, Article ID 1703011

Objectives Fat quantification by dual-energy computed tomography (DECT) provides contrast-independent objective results, for example, on hepatic steatosis or muscle quality as parameters of prognostic relevance. To date, fat quantification has only been developed and used for source-based DECT techniques as fast kVp-switching CT or dual-source CT, which require a prospective selection of the dual-energy imaging mode. It was the purpose of this study to develop a material decomposition algorithm for fat quantification in phantoms and validate it in vivo for patient liver and skeletal muscle using a dual-layer detector-based spectral CT (dlsCT), which automatically generates spectral information with every scan. Materials and Methods For this feasibility study, phantoms were created with 0%, 5%, 10%, 25%, and 40% fat and 0, 4.9, and 7.0 mg/mL iodine, respectively. Phantom scans were performed with the IQon spectral CT (Philips, the Netherlands) at 120 kV and 140 kV and 3 T magnetic resonance (MR) (Philips, the Netherlands) chemical-shift relaxometry (MRR) and MR spectroscopy (MRS). Based on maps of the photoelectric effect and Compton scattering, 3-material decomposition was done for fat, iodine, and phantom material in the image space. After written consent, 10 patients (mean age, 55 ± 18 years; 6 men) in need of a CT staging were prospectively included. All patients received contrast-enhanced abdominal dlsCT scans at 120 kV and MR imaging scans for MRR. As reference tissue for the liver and the skeletal muscle, retrospectively available non-contrast-enhanced spectral CT data sets were used. Agreement between dlsCT and MR was evaluated for the phantoms, 3 hepatic and 2 muscular regions of interest per patient by intraclass correlation coefficients (ICCs) and Bland-Altman analyses. Results The ICC was excellent in the phantoms for both 120 kV and 140 kV (dlsCT vs MRR 0.98 [95% confidence interval (CI), 0.94-0.99]; dlsCT vs MRS 0.96 [95% CI, 0.87-0.99]) and in the skeletal muscle (0.96 [95% CI, 0.89-0.98]). For log-transformed liver fat values, the ICC was moderate (0.75 [95% CI, 0.48-0.88]). Bland-Altman analysis yielded a mean difference of -0.7% (95% CI, -4.5 to 3.1) for the liver and of 0.5% (95% CI, -4.3 to 5.3) for the skeletal muscle. Interobserver and intraobserver agreement were excellent (>0.9). Conclusions Fat quantification was developed for dlsCT and agreement with MR techniques demonstrated for patient liver and muscle. Hepatic steatosis and myosteatosis can be detected in dlsCT scans from clinical routine, which retrospectively provide spectral information independent of the imaging mode.

Ein möglicher Anwendungsbereich der Magnetpartikelbildgebung ist die Echtzeit-Perfusionsbildgebung. Diese wird zur Beurteilung des vaskulären Systems und der Durchblutung von Organen und Geweben eingesetzt. Anhand der Konzentrations-Zeitverläufe in einem interessierenden Volumen, die durch die intravenöse Injektion eines Tracer-Bolus hervorgerufen werden, können unterschiedliche Perfusionsparameter, wie z.B. die Zeit zum Peak, die mittlere Transitzeit, das Blutvolumen und der Blutfluss bestimmt und in Form von räumlichen Perfusionskarten dargestellt werden. Bei bestimmten Erkrankungen kann eine regelmäßige Überwachung der Perfusion bestimmter Organe erforderlich sein. Da sich die injizierten Partikel mit der Zeit im Blut des Patienten verteilen, bis ein Steady State erreicht ist und die Konzentrations-Zeitverläufe kein Profil mehr zeigen, ist für jede Aufnahme die Injektion eines weiteren Bolus notwendig. Die maximale Tracer-Menge, die einem Patienten in einem bestimmten Zeitraum injiziert werden darf, ist begrenzt. Außerdem steigt die Tracerkonzentration im Blut mit jedem Bolus weiter an, sodass durch Tracer-Boli mit der Zeit möglicherweise keine erkennbaren Konzentrationsänderungen mehr verursacht werden. In dieser Arbeit wurde ein Ansatz zur Lösung dieses Problems untersucht, bei dem die Form der Injektion für die Aufnahme eines erneuten Perfusionsbildes invertiert wird, indem anstelle eines weiteren Tracer-Bolus (positiver Bolus) ein Bolus aus Wasser oder NaCl (negativer Bolus) verwendet wird. Anhand eines an eine Pumpe angeschlossenen Rattenherz-Phantoms wurde untersucht, ob negative Boli ebenso für die Perfusionsbildgebung verwendet werden können wie positive Boli und ob dieselben Ergebnisse für die Berechnung von Perfusionsparametern erzielt werden können. Voraussetzung dafür ist, dass die Konzentrations-Zeitverläufe der negativen Boli auf die der positiven Boli normiert werden können. Für diese Untersuchungen wurden optische Vorversuche mithilfe von Tinte und Wasser sowie MPI-Messungen unter der Verwendung von Tracer und Wasser für die positiven bzw. negativen Boli durchgeführt. Auf Basis dieser Versuche wurden Zeitverläufe erstellt und hinsichtlich der Normierbarkeit ausgewertet sowie Perfusionskarten berechnet und ein quantitativer Vergleich der Karten der positiven und negativen Boli vorgenommen. Anhand der Experimente konnte gezeigt werden, dass positive und negative Boli identischer Größe aufeinander normiert werden können und negative genauso wie positive Boli prinzipiell zur Perfusionsbildgebung genutzt werden können., Real-time perfusion imaging is a potential application of magnetic particle imaging. It is used to assess the vascular system and the perfusion of organs and tissues. The intravenous injection of a tracer bolus causes concentration-time curves in a volume of interest. Based on this, various perfusion parameters can be assessed and visualized in perfusion maps, e.g. time to peak, mean transit time, blood volume and blood flow. At certain diseases the consistent monitoring of the perfusion of certain organs is required. Until the reach of a steady state (when the concentration-time curves do not show a profile anymore), the injected particles spread in the patient’s blood. Thus, the injection of another bolus is necessary for every image acquisition. The maximum amount of tracer that may be injected to a patient in a certain time frame is limited. Additionally, the concentration of the tracer in the blood is increased by every bolus. Though, the change of concentration might not be detectable anymore. In this work a possible approach to this problem is considered. Therein, the form of the injection for the acquisition of a new perfusion image is inverted by the use of a water or NaCl bolus (negative bolus) instead of a tracer bolus (positive bolus). On the basis of a phantom of a rat’s heart that is connected to a pump, it is investigated if negative boluses might also be used for perfusion imaging as positive boluses to obtain the same results for the calculation of perfusion parameters. The concentration-time curves of the negative boluses necessarily need to be normalized to the positive ones. For this study, optical experiments were performed by use of ink and water as well as MPI measurements by use of tracer and water for the positive and negative boluses, respectively. Based on these experiments, time-curves were derived and assessed concerning their possibility of normalization. Besides, perfusion maps were calculated and compared quantitatively. As can be seen from the experiments, positive and negative boluses of identic size can be normalized to each other.

Diese Arbeit befasst sich mit der automatischen Analyse von Thorax-Röntgenbildern mithilfe von Deep Learning, einer sich schnell entwickelnden Methode des maschinellen Lernens. Nach einer detaillierten Evaluierung von CNN-Architekturen auf dem ChestX-ray14 -Datensatz wird gezeigt, wie Modelle angepasst werden können, um Nicht-Bild-Merkmale einzubeziehen und ihre Leistung zu verbessern. Um das Erscheinungsbild von CXRs zu vereinfachen und CNNs dabei zu helfen, mit hochdimensionalen und begrenzten Daten umzugehen, sowie um ihre Leistung zu verbessern, wurde eine Reihe von fortschrittlichen Vorverarbeitungsmethoden eingesetzt. Schließlich wird durch Simulationen, die auf der Warteschlangentheorie und Markov-Prozessen basieren, die signifikante Verbesserung durch intelligente Arbeitslistensortierung aufgezeigt., This work addresses the automatic analysis of chest radiographs using Deep Learning, a rapidly evolving machine learning method. After a detailed evaluation of state-of-the-art CNN architectures on the ChestX-ray14 dataset, it is shown how models can be adapted to incorporate non-image features and improve their performance. Further, to simplify the appearance of CXRs and help CNNs deal with high-dimensional and limited data, as well as to improve their performance, a number of advanced pre-processing methods have been employed. Lastly, through simulations based on queuing theory and Markov processes, the significant clinical impact of smart worklist ordering is shown.

Magnetic-Particle-Imaging (MPI) visualisiert die quantitative Verteilung von magnetischen Nanopartikeln mit statischen und oszillierenden Magnetfeldern. In dieser Arbeit wird eine submillimetergenaue Lokalisierungsmethode entwickelt, um Markierungen für die Registrierung von MRT- und MPI-Volumen zu identifizieren. Mit dieser Methode wird auch ein IVOCT-Katheter in 4D bimodal mit MPI verfolgt. Des Weiteren werden Magnetic-Particle-Imaging und Magnetic-Particle-Navigation verwendet, um magnetische Partikel durch eine Verzweigung zu navigieren. Zuletzt wird die easy-axis-Orientierung von Proben mit dem neuartigen easy-axis-aligned-MPI-Kontrast bestimmt., The modality Magnetic Particle Imaging (MPI) visualizes the quantitative distribution of magnetic nanoparticles by using static and oscillating magnetic fields. In this work, a submillimeter-accurate localization method is developed to identify fiducial markers more precisely for the registration of MRI and MPI volumes. With this method a IVOCT catheter is also tracked in 4D for bimodal imaging of IVOCT and MPI. Further, the method of Magnetic Particle Imaging and Magnetic Particle Navigation is used to navigate magnetic particles through a bifurcation within a vascular flow. Finally, the novel easy-axis aligned MPI contrast is used to determine the easy-axis orientation of samples.