This PDF file contains the front matter associated with SPIE Proceedings Volume 11113, including the Title Page, Copyright information, Table of Contents, Author and Conference Committee lists

Hard X rays are the probe of choice to three-dimensionally visualize optically opaque hard and soft objects as well as their combinations without physical cutting. The currently available methods allow for the quantification of the real and the imaginary parts of local refractive indices. The related phase- and absorption-contrast data are often complementary. The datasets are of GB or TB size due to the object's size and the selected spatial resolution. Therefore the related quantification is challenging. Nevertheless, the current developments pave the way towards imaging larger and larger objects with true micro- and nanometer resolution. Since 1997, the SPIE conference on Developments in X-ray Tomography, which was established by U. Bonse, has set the benchmark in the dissemination of knowledge between experts in instrumentation, software development, and a variety of applications. This paper summarizes key aspects of the contributions in the current volume.

X-ray cone-beam holo-tomography of unstained tissue from the human central nervous system reveals details down to sub-cellular length scales.¹ This visualization of variations in the electron density of the sample is based on phase contrast techniques using intensities formed by self-interference of the beam between object and detector. Phase retrieval inverts diffraction and overcomes the phase problem by constraints such as several measurements at different Fresnel numbers for a single projection. Therefore, the object-to-detector distance (defocus) can be varied. However, for cone beam geometry, changing defocus changes magnification, which can be problematic in view of image processing and resolution. Alternatively, the photon energy can be altered (multi-E). Far from absorption edges, multi-E data yield the wavelength independent electron density. In this contribution we present multi-E holo-tomography at the GINIX setup of the P10 beamline at DESY. The instrument is based on a combined optics of elliptical mirrors and an x-ray waveguide positioned in the focal plane for further coherence, spatial Filtering and high numerical aperture.² Previous results showed the suitability of this instrument for nanoscale tomography of unstained brain tissue.¹ We demonstrate that upon energy variation, the focal spot is stable enough for imaging. To this end, a double crystal monochromator and automated alignment routines are required. Three tomograms of human brain tissue were recorded and jointly analyzed using phase retrieval based on the contrast transfer function formalism generalized to multiple photon energies. Variations of the electron density of the sample are successfully reconstructed.

Composite materials, both biominerals such as bone or shells but also increasingly synthetic materials, often have a complex hierarchical 3D structure ranging over several length scales. To fully understand the structure of such materials, a method for probing the nanoscale structure in 3D is needed. Materials such as bone are particularly challenging due to their complex composition and hierarchical structure. Recent developments in synchrotron x-ray focusing optics have paved the way for smaller x-ray beams with high brilliance. Herein, we present how we probe the 3D elemental distribution and crystallographic properties of human bone using combined fluorescence and diffraction tomography (F-CT and XRD-CT) with a 50 nm pencil X-ray beam. The 2.6×3.1 µm2 cross section sample was a FIB-cut rod of human iliac crest bone. We recorded 2D diffraction patterns and fluorescence spectra for each point in a 50 nm raster scan grid pattern from 92 projection angles covering 0-182°. This allowed us to reconstruct tomographically both x-ray diffraction patterns and the elemental composition in a ∼5×5×3 µm3 volume encompassing the sample. We show that tomographic reconstruction of x-ray diffraction and fluorescence information is possible at <140 nm spatial resolution estimated from features in reconstructed images. Thereby it is possible to probe crystalline structure and elemental composition in 3D at length scales an order of magnitude smaller than hitherto available. This allows studying a very broad range of materials from biominerals to energy materials in more detail than ever before.

The three-dimensional (3D) architecture of organs, tissues and cells together with the spatial distribution of specific molecules, both enables and drives the close interactions between hosts and microbes. To image the 3D anatomy and molecular composition of an animal-microbe system at a single cell scale we developed a correlative workflow combining phase-contrast synchrotron radiation based micro-computed tomography (PC-SRμCT) and mass spectrometry imaging (MSI). The key challenge in combining both techniques was the synchronization of sample preparation and imaging conditions. We used PC-SRμCT for high-resolution 3D imaging, which is the most suitable x-ray tomographic technique to image soft tissue samples without contrast enhancing agents at a single-cell level. For method development, we used small invertebrates that live in symbiosis with bacteria. For PC-SRμCT samples were paraformaldehyde-fixed and submersed in a buffer-filled capillary. We tested sample preparation for Matrix-assisted laser desorption ionization (MALDI)-MSI after PC-SRμCT and achieved the best results for PFA-fixed samples, processed under cryogenic conditions. The resulting molecular composition permitted imaging of hundreds of different molecule distributions from single tissue sections on a micrometer resolution. Image analysis revealed organ- and cell-specific metabolite distributions, which we could match to corresponding structures in the anatomical 3D PC-SRμCT model.
Our PC-SRμCT-MSI sample preparation- and imaging workflow results in a detailed 3D scenario, which provides an atlas to guide other microscopy or omics approaches towards specific organs and cells. Particularly, for animal-microbe systems this approach presents a powerful tool to visualize anatomical and chemical processes at a microscale, linking structure and function in the symbiosis.

EMBL Hamburg operates the P14 beamline at the high-brilliance PETRA III storage ring at DESY. Delivering hard X-rays (6 – 30 keV), P14 is very versatile: shape, size and intensity of the X-ray beam are easily tunable by using both reflective and refractive optical elements. Sample stage and detecting systems are state-of-the-art including a high-precision vertical goniometer, a robotized sample changer and a DECTRIS EIGER 16M detector. These features enable crystallography of large macromolecular complexes, serial crystallography, time-resolved and in-situ diffraction data collection. Exploiting the flexibility of the beamline optics, we are investigating new methods and techniques to look into samples of interest via full-field phase-contrast X-ray imaging [1] and microscopy. Unlike conventional absorption X-ray imaging, phase contrast makes interfaces and density gradients clearly visible in a biological sample. To achieve higher resolution and to see fine structure, we also implement X-ray refractive lenses [2] as an objective in the microscopy mode. This approach allows us to magnify details of the sample up to 15 times prior to the optical detection of X-rays that carry the phase-contrast information from the interaction with sample. With the current setup, a sub-100nm resolution can be achieved. High-precision rotation of the sample stage at P14 also allows to perform X-ray tomography. Due to the use of high-energy X-rays for the imaging process, materials and tissues with thicknesses in the mm-range can be penetrated. The described setup apply phase-contrast imaging techniques for full-field visualization and fast X-ray tomography of crystals in optically opaque mesophase with sub-micron resolution at low dose (Polikarpov et al., 2019, in press). With the help of the method, it is possible to visualize crystal boundaries and their local deformations in situ, thus supporting the most challenging MX applications. In addition to crystallographic applications, the imaging assembly at P14 is used for investigation of multi-cellular organisms. The short time-scale for completing measurements and the existing automation of P14 allow us to apply high-throughput X-ray tomography, providing a way for large-field-of-view screening of biological samples prior to further high-resolution electron microscopy studies. [1] A. Snigirev, et al., Rev. Sci. Instrum., 66 (1995), 5486. [2] A. Snigirev, et al., Nature, 384 (1996) 49. [3] Fischer et al., Frontiers in Zoology 7:31 (2010).

X-ray grating interferometry (XGI) is a phase-contrast imaging technique that allows for a quantitative measurement of the refractive index with high density resolution in a model-independent manner—i.e. without a priori knowledge of the specimen composition. However, the retrieval of the X-ray wavefront phase shift relies on the accurate measurement of the interference pattern phase shift, making XGI vulnerable to phase wrapping when the interference pattern phase shift, related to the derivative of the wavefront phase shift, is large. Standard procedure for avoiding phase wrapping involves submerging the specimen in a water bath to reduce the mismatch of the index of refraction at the boundaries, but this requires a top-down rotation stage and is susceptible to gas bubble formation inside the water bath. Our team has presented an algorithm to remove phase wrapping artifacts for cylindrically shaped specimens that is applied to the phase-retrieved sinogram. This algorithm models and replaces phase-wrapped data to prevent the spread of “cupping” artifacts due to the integration of the differential phase during reconstruction. We give a criterion for selecting the modeling parameters so that the resulting measurement of the index of refraction matches the results of measurements without phase wrapping. We also apply this technique to cases where phase wrapping occurs at multiple interfaces. This algorithm allows for XGI measurements without a water bath and top-down rotation stage at synchrotron and laboratory facilities, especially as sensitivity increases.

In x-ray computed tomography (CT) scanning, the presence of metal objects gives rise to artifacts. Although dual-energy CT scanning and decomposition can reduce metal artifacts, in practice, the decomposition is unstable in the presence of noise and yields residual and new artifacts. A common practice in metal artifact reduction (MAR) algorithms is to use a prior-image as a guide to estimating the underlying data that are corrupted by the metal. We have developed a method in which one prior-image can be used to correct the various sinograms generated by dual-energy decomposition. We applied our method to data acquired on a commercial CT scanner. Compared to the uncorrected images, the MAR images have superior uniformity in known uniform regions while preserving edges, and better visual definition of structures.

We propose a new image reconstruction algorithm for CT, which is able to reduce the so-called metal artifact well. The most existing reconstruction algorithms for the metal artifact reduction consist of detecting metallic parts in the sinogram followed by image reconstruction after excluding or interpolating projection data corresponding to the identified metallic parts. However, the proposed algorithm consists of only a single computational step, leading to unifying the two steps into a single step. The proposed algorithm can be considered a particular application of Fault-Tolerant image reconstruction discovered by Kudo et al. [1]. The main idea is to use the L₁ norm error Ax −b ¹₁ between Ax and b (x denotes image and b denotes projection data), or the error defined by using the Huber loss function Huber(Ax−b), instead of the ordinary L₂ norm. The use of these robust error functions leads to excluding abnormal projection data passing through the metallic parts implicitly from the data fitting. The simulation result using a clinical dental CT image demonstrates that the proposed algorithm is able to reduce the metal artifact well by accurately identifying the location of metallic parts in the sinogram.

Features such as particles, pores, or cracks are challenging to measure accurately in CT data when they are small relative to the data resolution, characterized as a point-spread function (PSF). These challenges are particularly acute when paired with segmentation, as the PSF distributes some of the signal from a voxel among neighboring ones; effectively dispersing some of the signal from a given object to a region outside of it. Any feature of interest with one or more dimensions on the order of the PSF will be impacted by this effect, and measurements based on global thresholds necessarily fail. Measurements of the same features should be consistent across different instruments and data resolutions. The PVB (partial volume and blurring) method successfully compensates by quantifying features that are small in all three dimensions based on their attenuation anomaly. By calibrating the CT number of the phase of interest (in this case, gold) it is possible to accurately measure particles down to <6 voxels in data acquired on two instruments, 14 years apart, despite severe artifacts. Altogether, the PVB method is accurate, reproducible, resolution-invariant, and objective; it is also notable for its favorable error structure. The principal challenge is the need for representative effective CT numbers, which reflect not only the features of interest themselves, but also the X-ray spectrum, the size, shape and composition of the enclosing sample, and processing details such as beam-hardening correction. Empirical calibration is the most effective approach.

Images from MARS spectral CT scanners show that there is much diagnostic value from using small pixels and good energy data. MARS scanners use energy-resolving photon-counting CZT Medipix3RX detectors that measure the energy of photons on a five-point scale and with a spatial resolution of 110 microns. The energy information gives good material discrimination and quantification. The 3D reconstruction gives a voxel size of 70 microns. We present images of pre-clinical specimens, including excised atheroma, bone and joint samples, and nanoparticle contrast agents along with images from living humans. Images of excised human plaque tissue show the location and extent of lipid and calcium deposition within the artery wall. The presence of intraplaque haemorrhage, where the blood leaks into the artery wall following a rupture, has also been visualised through the detection of iron. Several clinically important bone and joint problems have been investigated including: site-specific bone mineral density, bone-metal interfaces (spectral CT reduces metal artefacts), cartilage health using ionic contrast media, gout and pseudogout crystals, and microfracture assessment using nanoparticles. Metallic nanoparticles have been investigated as a cellular marker visible in MARS images. Cell lines of different cancer types (Raji and SK-BR3) were incubated with monoclonal antibody-functionalised AuNPs (Herceptin and Rituximab). We identified and quantified the labelled AuNPs demonstrating that Herceptin-functionalised AuNPs bound to SK-BR3 breast cancer cells but not to the Raji lymphoma cells. In vivo human images show the bone microstructure. Fat, water, and calcium concentrations are quantifiable.

X-ray imaging is well-suited for deep tissue analysis, and is routinely used in medical diagnosis, but is normally blind to local biochemical signals. Here we describe simple passive implantable sensors that can report local chemical concentrations (e.g. pH), which are important for detecting and studying infection and other diseases or conditions. The sensors contain hydrogels with chemically-responsive swelling; the chemical concentration is determined using an X-ray to measure the position of radio-opaque markers in embedded in the hydrogel. For example, to measure local pH near the surface of an orthopedic plate, a sensor containing a polyacrylic acid hydrogel was attached to the plate. The hydrogel displayed a pH-dependent swelling, expanding approximately and moving an embedded radio-opaque tungsten marker. The sensor was calibrated in standard pH buffers and tested during bacterial growth in culture. Its response was negligibly affected by changes in temperature and sodium chloride concentration within the normal physiological range. Radiographic measurements were also performed in cadaveric tissue with the sensor attached to an implanted orthopedic plate fixed to a tibia. Pin position readings varied by 100 µm between observers surveying the same radiographs, corresponding to 0.065 pH unit precision in the range pH 4-8 (1.5 mm/pH unit). We are expanding the approach to other analytes (infection markers and mechanical strain sensors to track bone heaing), miniaturized injectable sensors, and imaging at multiple locations. Acknowledgement: This research was supported by the following grants: NSF CHE1255535, NIH NIGMS 5P20GM103444-07, NIH 1R21EB019709-01A1, and NIH NIAMS R01 AR070305-01.

The past decade has witnessed tremendous growth in both interest and available techniques for laboratory X-ray analysis. From the progression of commercially-available micro- and nano-CT scanners to the resolution and sensitivity enhancements of x-ray fluorescence spectrometers, the scientific community is benefiting from a rapid expansion of laboratory-based x-ray techniques. In our work, we have developed a suite of advanced x-ray instrumentation providing a wide range of enhanced capabilities for specimen characterization. The key enabling technology lies in the X-ray source, which features a microstructured target capable of providing 5-10x higher brightness than conventional sealed-tube x-ray sources and offering power flux densities that rival rotating anode sources. The target array can be custom-designed to incorporate a variety of materials, facilitating fast & easy switching between characteristic emission lines and radiation spectra. This source has been subsequently integrated with state-of-the-art X-ray focusing optics, such as ellipsoidal/paraboloidal capillary lenses and finely-structured Fresnel zone plate imaging objective lenses, and sensitive scintillator-coupled CCD detection systems, opening up new opportunities for advancing laboratory x-ray inspection equipment. Here, we will describe the system geometries in detail and demonstrate how these new advancements have led us to the development of laboratory micro-XRF, nano-XRM, and XAS instrumentation. We will also briefly introduce the image-centric software workspace, which facilitates novice users to collect data quickly and reliably with minimal training overhead.

The NBS-1 photo-neutron source is the primary standard for neutron calibrations in the United States of America. The source is constructed from 1 g of radium in a RaBr2 salt contained in a Pt-Ir capsule located in the geometric center of a 4 cm beryllium sphere. Since the source was fabricated in 1949, it is desirable to determine the integrity of the capsule and current geometry of the radium salt. Several challenges are presented in inspecting the source such as the high-Z materials used for the capsule which make X-ray tomography challenging with lower keV generators and the radioactive material which creates an intense neutron and gamma radiation fields. The radiation emitted from the source makes it impractical to move the source to a capable imaging facility and makes it difficult to work with as it is not possible to handle directly. To overcome these issues and allow for a feasibility study, a dummy source has been constructed using PbBr2 powder. This talk will provide background on the dummy source and the results of the feasibility study to determine if it possible to relocate a current X-ray system to the neutron calibration facility to inspect the NBS-1 source.

Bright synchrotron x-ray sources enable imaging with short exposure times, and hence in a high-speed image sequence. These x-ray movies can capture not only sample structure, but also how the sample changes with time, how it functions. The use of a synchrotron x-ray source also provides high spatial coherence, which facilitates the capture of not only a conventional attenuation-based x-ray image, but also phase-contrast and dark-field signals. These signals are strongest from air/tissue interfaces, which means that they are particularly useful for examining the respiratory system. We have performed a range of x-ray imaging studies that look at lung function, airway surface function, inhaled and instilled treatment delivery, and treatment effect in live small animal models [Morgan, 2019]. These have utilized a range of optical set-ups and phase-contrast imaging methods in order to be sensitive to the relevant sample features, and be compatible with high-speed imaging. For example, we have used a grating interferometer to measure how the airsacs in the lung inflate during inhalation, via changes in the dark-field signal [Gradl, 2018], a single-exposure, single-grid set-up to capture changes in the liquid lining of the airways [Morgan, 2015] and propagation-based phase contrast to image clearance of inhaled debris [Donnelley, 2019]. Studies have also utilized a range of analysis methods to extract how the sample features change within a time-sequence of two-dimensional projections or three-dimensional volumes. While these imaging studies began in large-scale synchrotron facilities, we have recently performed these kinds of studies at an inverse-Compton-based compact synchrotron, the Munich Compact Light Source (MuCLS) [Gradl, 2018b]. 1. Morgan, Kaye, et al., “Methods for dynamic synchrotron X-ray imaging of live animals.”, under review 01/2019. 2. Gradl, R., et al. "Dynamic in vivo chest x-ray dark-field imaging in mice." IEEE Transactions on Medical Imaging (2018). 3. Morgan, Kaye S., et al. "In vivo X-ray imaging reveals improved airway surface hydration after a therapy designed for cystic fibrosis." American Journal of Respiratory and Critical Care Medicine 190.4 (2014): 469-472. 4. Donnelley, Martin, et al. "Live-pig-airway surface imaging and whole-pig CT at the Australian Synchrotron Imaging and Medical Beamline." Journal of Synchrotron Radiation 26.1 (2019). 5. Gradl, Regine, et al. "In vivo Dynamic Phase-Contrast X-ray Imaging using a Compact Light Source." Scientific Reports 8.1 (2018b): 6788.

Talbot-Lau interferometry (TLI) enables X-ray imaging in multiple contrasts, especially for low-Z materials, complementary to conventional absorption radiography. A novel microstructured array anode target (MAAT) source offers major advantages over the combination of an extended X-ray source coupled with an absorption grating (also known as the source grating), including more efficient use of source X-rays and a larger field of view, especially so in the case of Xray energies greater than 30 keV. Therefore, there is an increasing interest in optimizing performance of the MAAT X-ray source to achieve high energy TLI. With the Monte Carlo simulation, here we systematically evaluate the MAAT source parameters for TLI and report the optimal values for high X-ray energies. In particular, the spatial dependence of the Xray generating capability of the MAAT is studied with respect to the incidence angle of the electron beam to the normal of the MAAT target surface.

X-ray phase tomography by Talbot interferometry functions with X-rays of a broad energy band, and a pink- beam extracted by a multilayer mirror from white synchrotron radiation is used for four-dimensional X-ray phase tomography, which was developed at BL28B2, SPring-8, Japan. Polymer samples under infrared laser irradiation were observed as model experiments of laser processing. In this paper, laser ablation for a carbon-fiber reinforced polymer (CFRP) sample is reported. Thanks to the fact that the fibrous structure generates visibility (or dark-field) contrast, four-dimensional tomogram from the visibility reduction is presented. A technical effort to improve the temporal resolution from 1 s to 160 ms is also described.

Quantitative measurements of the phase shift of X-rays passing through a matter allow us to perform X-ray phase tomography for visualization of soft materials. Combination of an X-ray microscope and a grating interferometer is a promising approach to realize quantitative phase measurements with a microscopic spatial resolution. A Lau interferometer consisting of a source grating and a phase grating is available for this purpose with an incoherent laboratory X-ray source. We installed a Lau interferometer into a laboratory-based X-ray microscope adopting a copper rotating anode source and Fresnel zone plates (ZEISS Xradia 800 Ultra). A “twin-phase image”, which consists of positive and negative phase images overlaid with a certain separation, is generated through a fringe-scanning measurement with this microscope. A step for generating a quantitative phase image from the twin phase image should be developed to perform phase tomography. However, conventional deconvolution operations are not suitable because of artifacts and noise remained in resultant phase images. To reduce the artifacts and noise, an iterative calculation algorithm has been developed. The evaluation of the algorithm shows that the artifacts and noise are suppressed and quantitative phase images are obtained. Finally, results of phase tomography obtained for soft materials are demonstrated.

Laboratory-based X-ray tomography systems with conventional electron-impact sources can provide resolutions from hundreds of microns down to below 50 nm. They are commercially available and widely used for a broad range of applications. However, the limited brightness, monochromaticity, tunability and coherence of such sources require relatively long imaging times (especially at high resolution) and limit their applicability for quantitative, spectroscopic or phase contrast imaging applications. Such applications are served well by synchrotron radiation sources with their extremely high brightness, tunability and monochromaticity. However, synchrotrons are very large and expensive, and typically operated as national user facilities with limited access. An Inverse Compton Scattering (ICS) X-ray source can bridge this gap by providing a narrow-band, high flux and tunable X-ray source that fits into a laboratory at a cost of a few percent of a large synchrotron facility. It works by colliding a high-power laser beam with a relativistic electron beam, in which case the back-scattered photons have an energy in the X-ray regime. This presentation describes the working principle of the Lyncean Compact Light Source, a storage-ring based ICS source, and its applicability to X-ray tomography applications. We provide recent application examples from the field of biomedical phase-contrast imaging that demonstrate synchrotron-like capabilities in a laboratory setting. Furthermore, we explore how such an X-ray source can advance laboratory-based micro- and nano-CT by enabling high throughput or spectroscopic imaging applications.

Although the mortality rate of breast cancer has been reduced with the introduction of screening mammography, many women undergo unnecessary subsequent examinations due to inconclusive diagnoses. Therefore, spatial resolution and soft tissue contrast must meet very high standards. The former is necessary in order to be able to detect and distinguish closely spaced microcalcifications from each other. Thus, a high spatial resolution is demanded. Superposition of anatomical structures especially within dense breasts in conjunction with the inherently low soft tissue contrast of absorption images compromises image quality. This can be overcome by phase-contrast imaging. In this study, we analyze the spatial resolution of grating-based multimodal mammography dose-dependently using a mammographic phantom and one freshly dissected mastectomy specimen at an inverse Compton X-ray source. Here, the main focus was on estimating the spatial resolution with the sample in the beam path and discussing benefits and drawbacks of the method used as well as the estimation of the mean glandular dose. Finally, the possibility of improving the spatial resolution is investigated by comparison of the monochromatic grating-based mammography with the standard one. In almost all cases, the spatial resolution is higher in our proposed approach while the dose can be significantly reduced. Additionally, phase-contrast imaging helps to improve the detection of tumor lesions

In the present work, we evaluate and compare the contrast and resolution obtained on different neuronal tissues with propagation-based x-ray phase contrast computed tomography (PB-CT). At our laboratory-based liquid metal-jet setup, we obtain overview datasets at sub-micron resolution of mm³ -sized volumes. In order to evaluate these parameters down to the sub-cellular level, we utilize the synchrotron endstation GINIX at P10, DESY. At this dedicated endstation1 developed and operated by our group, we utilize x-ray waveguide optics for highresolution cone-beam scans at strong geometrical magnification M. Exploiting this multi-scale approach, we investigate the image quality of cerebellum tissue treated by different heavy-metal stains. In addition, we study the electron density contrast in unstained tissues. Different embedding media are utilized depending on the stain, which also significantly affects contrast and image quality. With this work, we want to contribute to an optimized sample preparation to study the neuronal architecture of the brain tissue in greater detail in three dimensions (3d).

The application of 3D X-ray imaging for biological samples (e.g. biopsies) to gain a deeper understanding of microscopic structures on a (sub)cellular level is restricted by the weak attenuation contrast of soft tissue. The development of novel staining tools for X-ray soft-tissue imaging will overcome these challenges. Here, we present the application of a recently developed method combining a laboratory-based nanoscopic X-ray CT setup enabling resolutions down to 100 nm with a target-specific X-ray staining protocol. The results clearly show that the X-ray attenuation contrast in the samples is remarkably improved by our staining method and detailed tissue (sub)structures are apparent, which cannot be visualized without the staining. The nanoscopic CT data reproduce the tissue morphology with a similar level of detail as the corresponding histological light microscopy images in 2D and enable pathological characterization of the crucial structures. Beyond that, the applied method allows for visualization of the 3D tissue architecture, offering deeper insights into the 3D microscopic structure of soft-tissue. Moreover, we demonstrate the compatibility of the X-ray stain with standard histological staining methods. Beside medical research, the methodology has the potential to contribute to advances in zoology and developmental biology.

The DIAMOND beamline I13L is dedicated to multi-scale and multi-modal imaging in real and reciprocal space. The beamline consists of two independently operating experimental stations, located at a distance of more than 200 m from the source. The Imaging Branch performs micro-tomography with in-line phase contrast in the 6-30 keV energy range. In addition, a grating interferometry setup and a full-field microscope for nano-tomography are currently implemented. Other techniques providing high-resolution three-dimensional information, in particular coherent X-ray diffraction, are hosted on the Coherence Branch. All imaging methods are tested to operate with large energy bandwidths and therefore shorter exposure times. To this end, two options are currently used: the so-called ‘pink-beam’ mode using a reflecting mirror and X-ray filters and monochromatic mode using a multilayer monochromator. The operation mode enables science for in-situ and operando studies across a wide range of scientific areas.

Systems capable of performing phase-contrast X-ray micro-CT currently represent only a fraction of all installed micro- CT devices. Institutions without dedicated phase-contrast research therefore often lack the necessary equipment and expertise to apply these techniques. Until this gap is closed, X-ray contrast agent-enhanced absorption-contrast micro-CT remains the primary means for soft tissue imaging where enhanced sensitivity, but not quantitative imaging is required. Commercial contrast agents for in vivo applications are, however, typically small molecular compounds that can easily diffuse across the blood vessel walls, which makes them unsuitable for ex vivo high resolution X-ray micro-CT imaging. We evaluated the suitability for kidney capillary imaging of the nanoparticle-based blood pool contrast agents Aurovist 15 nm and ExiTron nano 12000, along with the vascular casting resin μAngiofil, by post mortem imaging of whole mouse kidneys with 4-5 μm voxel size using the laboratory source micro-CT system Nanotom m. Based on the results, we identified specific obstacles to reliable capillary filling in the kidney, which will aid in the development of next generation X-ray contrast agents and protocols.

In clinical X-ray imaging, the quantitative information in a CT scan has recently been extended by the possibility of using dual-energy information. Dual-energy CT has found its way into clinical imaging during the last few years and has been proven to add additional diagnostic information in different pathologies. It is based on a dual measurement at different photon energies, such that the energy dependence of the linear attenuation coefficient can be used for improved material discrimination. Here, we demonstrate how the dual information accessed with grating-based phase-contrast CT can be used to provide the same quantitative information. Different from dual energy, the phase-contrast measurement directly yields the electron-density and the total attenuation coefficient in a single measurement. With algebraic basis transformation this can be used for quantitative material decomposition, allowing the visualization of quantitative material maps. Further, a simple interaction parametrization has been used for the generation of effective atomic number maps and virtual monochromatic images. The approach has been demonstrated with an experimental angiography simulation with a chicken heart. The results have been compared with iodine staining, which is a current approach for ex-vivo soft-tissue contrast enhancement. The measurements have been performed at a compact laser-undulator synchrotron X-ray source with a tunable quasi-monochromatic X-ray energy. The simultaneous image acquisition guarantees an inherent registration of the two original data-sets. In total, the method provides a range of novel quantitative image representations which can be helpful for specific material discrimination tasks in medical imaging in the future.

Here we present a phase contrast X-ray tomography study of of wild type C57BL/6 mouse hearts, as nondestructive approach to the microanatomy on the scale of the entire excised organ. Based on the partial coherence at a home-built phase-contrast μ-CT setup installed at a liquid metal jet (LMJ) source, we exploit phase retrieval and hence achieve superior image quality for heart tissue, almost comparable to previous synchrotron data on the whole organ scale. The fact that this is possible with compact laboratory instrumentation opens up new opportunities for screening samples and to optimize sample preparation, also prior to synchrotron beamtimes. In this work, different embedding methods and also heavy metal-based stains have been explored. From the tomographic reconstructions, quantitative structural parameters describing the 3D architecture have been derived, by two different fiber tracking algorithms. The first algorithm is based on the local gradient of the reconstructed electron density. By performing a principal component analysis (PCA) on the local structure-tensor of small sub-volumes the dominant direction inside the volume can be determined. In addition to this approach which is already well established for heart tissue, we have implemented and tested an algorithm which is based on a local 3D Fourier transform. In addition to the determination of the fiber orientation, the degree of filament alignment and local thickness of single muscle fiber bundles can be obtained.

Over the last two decades, phase-contrast X-ray imaging has been demonstrated to provide superior soft-tissue contrast in comparison to conventional attenuation-based X-ray imaging. However, visualizing biomedical soft tissue at high spatial resolution and high image quality still has been limited to brilliant synchrotron radiation sources with its very limited access. Thus, the possibility to perform high-sensitivity X-ray phase-contrast computed tomography with laboratory setups is of great interest for a broad range of high-resolution biomedical applications. We designed and built a grating-based setup using a standard rotating-anode X-ray source and reaching comparable angular sensitivity as synchrotron-based setups. Here, we will demonstrate the performance of our instrument by highlighting the recent results on quantitative biomedical soft-tissue analysis of different human diseases.

In 2009, Yu and Wang discovered that an exact reconstruction in interior tomography is possible when the object is piecewise constant over the whole ROI based on Compressed Sensing (CS). However, the assumption that the object is piecewise constant over the whole region of interest (ROI) is not realistic for most medical CT images having smooth intensity changes and textures inside each organ. Consequently, there is a strong tendency that the CS approach leads to losing the smooth image components. To overcome this drawback, we propose an advanced CS approach, in which the piecewise constancy is imposed only over a peripheral region of ROI such as frames, left+right boundaries, or four corners. To theoretically support the proposed method, we prove that the proposed method produces an exact reconstruction of the ROI when the object is actually piecewise constant only over the peripheral prior knowledge region. We demonstrate that the proposed method works well by simulation studies.

X-ray computed tomography (CT) reconstructs cross-sectional images from projection data. However, ionizing X-ray radiation associated with CT scanning might induce cancer and genetic damage and raises public concerns. Therefore, the reduction of radiation dose has attracted major attention. Few-view CT image reconstruction is an important topic to reduce the radiation dose. Recently, data-driven algorithms have shown great potential to solve the few-view CT problem. In this paper, we develop a dual network architecture (DNA) for reconstructing images directly from sinograms. In the proposed DNA method, a point-wise fully-connected layer learns the backprojection process requesting significantly less memory than the prior art and with O(C×N×N_C) parameters where N and N_c denote the dimension of reconstructed images and number of projections respectively. C is an adjustable parameter that can be set as low as 1. Our experimental results demonstrate that DNA produces a competitive performance over the other state-of-the-art methods.Interestingly, natural images can be used to pre-train DNA to avoid overfitting when the amount of real patient images is limited.

Recently, deep learning has become the mainstream method in multiple fields of artificial intelligence / machine learning (AI/ML) applications, including medical imaging. Encouraged by the neural diversity in the human body, our group proposed to replace the inner product in the current artificial neuron with a quadratic operation on inputs (called quadratic neuron) for deep learning. Since the representation capability at the cellular level is enhanced by the quadratic neuron, we are motivated to build network architectures and evaluate the potential of quadratic neurons towards “quadratic deep learning”. Along this direction, our previous theoretical studies have shown advantages of quadratic neurons and quadratic networks in terms of efficiency and representation. In this paper, we prototype a quadratic residual neural network (Q-ResNet) by incorporating quadratic neurons into a convolutional residual structure, and then deploy it for CT metal artifact reduction. Also, we report our experiments on a simulated dataset to show that Q-ResNet performs better than the classic NMAR algorithm.

Magnetic Resonance Imaging (MRI) and Computed Tomography (CT) are widely used for screening, diagnosis and imageguided therapeutics. Due to physical, technical and economical limitations, it is impossible for MRI and CT scanners to target ideal image resolution. Given the nominal imaging performance, how to improve image resolution has been a hot topic, and referred to as super-resolution research. As a promising method for super-resolution, over recent years deep learning has shown a great potential especially in deblurring natural images. In this paper, based on the neural network model termed as GAN-CIRCLE (Constrained by the Identical, Residual, Cycle Learning Ensemble), we adapt this neural network for achieving super-resolution for both MRI and CT. In this study, we demonstrate two-fold resolution enhancement for MRI and CT with the same network architecture.

X-ray computed tomography (CT) is a promising technique for three-dimensional imaging of batteries, electrolyzers and fuel cells. Operando techniques rely significantly on reactor design, where the operando cell should approximate the actual device. Furthermore, nano- and micro-scale hierarchical materials are used to convert or store electrochemical energy. Imaging with the single scale or synchrotron beamline provides only sliver of information needed. Carbon and ionomer materials that are used as bulk materials in the electrochemical devices are soft and phase-contrast imaging is required. We present our micro- and nano-X-ray CT reactors designs for polymer electrolyte fuel cells (PEFCs), electrolyzers, and micro X-ray CT operando pouch cell for battery. With the PEFC hardware we control temperature, relative humidity, pressure of gasses, as well as current density. By imaging the PEFCs at various current densities we are able to observe physical phenomena of membrane swelling with relative humidity increase and formation and removal of liquid water from the cell. With nano X-ray CT we are able to combine morphology with x-ray absorption near edge structure (XANES) to elucidate the oxidation state of the catalyst within the active porous layer. Using phase-contrast we are able to distinguish between liquid water and carbon phases. For electrolyzers, understanding formation and removal of oxygen bubbles, which is sub-second transport phenomena is critical. We combine X-ray CT with sub-second X-ray radiography to understand this transient process. For the batteries, we have developed a high throughput pouch cell design for x-ray CT imaging featuring 1 cm diameter active area.

X-ray phase/dark-field tomography by using an X-ray grating interferometer and white synchrotron radiation has been demonstrated to observe dynamics in materials consisting of light elements, because it is available with X-rays of a broad energy bandwidth. In this work, we combined X-ray phase/dark-field tomography with a stroboscopic technique, which synchronizes image acquisitions with repetitive tension applied to a sample. Snapshot images with a 200 μs temporal resolution are measured to reconstruct phase/dark-field tomograms. A result of stroboscopic X-ray dark-field tomography is described, which was obtained for a rubber sample under a 24 Hz repetitive compression-stretch motion of a 10 mm amplitude.

The multi-phase flow within a steel diesel injector is imaged using synchrotron-produced X-rays from the Advanced Photon Source at Argonne National Laboratory. Projections gathered from several repeated injection events at up to 10,000 frames per second and with 2.1 micrometer resolution at various viewing angles resulted in a 4D data set. Photon statistics were improved by averaging the data over 200 injection events. Owing to significant attenuation caused by the injector body material, and the short exposure time, the images are obscured by various types of noise making tomographic reconstruction challenging. Attempts at denoising this data are discussed. At each time step, translational and rotational image registration was performed to align projections obtained from different lines-of-sight. This is followed by a Fourier Transform method for computed tomography to reconstruct the 3D flow-field, from the start to end of fuel injection, which is a 2 millisecond long event. The X-ray phase contrast in the data was exploited by applying a low-pass filter. Segmentation is performed to track the location of the liquid-gas interface, thus distinctly revealing a highly asymmetric flow-separation layer affected by micron-scale features in the nozzle geometry. This complete data processing pipeline converted the images acquired at a signal-to-noise ratio of 1 into a unique tomographic dataset of internal fluid flow through approximately 2 millimeters of steel. The data shows excellent validation with computational fluid dynamics simulations of the flow profile previously obtained for this nozzle.

Contrast in Computed Tomography (CT) most often relies on absorption differences, and reconstructions give absorptivity of each volume element (voxel) but not the material’s identity. This paper describes the first use of in situ position-resolved x-ray diffraction to identify different contents (see below) within an intact mummy. Hawara Portrait Mummy No. 4 (HPM4) was first imaged with a clinical CT scanner, and this 3D “roadmap” guided diffraction mapping at beamline 1-ID of the Advanced Photon Source. Only 24 hrs were available, and the CT data allowed mapping of only the volumes of interest. The long beam path through the mummy was a major complication solved by collecting each diffraction pattern at two detector-mummy separations. Accuracy in diffracting volume positions was 1-2 mm, and precision sufficed for diffraction identification of the materials. Diffraction patterns were collected from the mineralized tissues within the five year old child’s mummy (HPM4), specifically the skull, cervical vertebrae, femora and teeth. Patterns from the femorum lateral and medial sides (and the skull’s left and right sides) could be analyzed separately. Lattice parameters and crystallite sizes of the bones’ mineral phase were comparable with those from modern bone. Within the wrappings, wires were located by diffraction and identified as a modern dual phase steel, probably introduced during earlier conservation. Highly absorbing, millimeter-sized inclusions were identified as calcite. The position of these and other features were found at the positions indicated by the CT scans, and the limitations of the diffraction approach compared to absorption CT are discussed.

The bony labyrinth is the bony capsule of the inner ear, which is the center of hearing and balance in the skull of vertebrates. In mammals this osseous structure ossifies early in the development being fully formed largely before birth. This means that it does not grow after full ossification while the animal continues to grow until full adult size well after birth. At the level of an individual, the bony labyrinth thus shows a negative ontogenetic allometric relationship with the surrounding skull and with the animal’s body mass. At the evolutionary level, between species, such a negative allometry has already been evidenced for the middle ear, a component of the ear region made of three tiny bones, the ossicles, in contact with the inner ear. Herein, we test the allometric relationship between the bony labyrinth and skull length as well as body mass on a large sample in the ruminant clade (Mammalia, Ruminantia) using micro computed tomography as the imaging method of choice. We find a strong negative allometry paralleling the ontogentic allometry described earlier. This evolutionary relationship related to the timing of ossification of the bony labyrinth is probably critical in explaining the large hearing frequency range of mammals as well as their particular ecological adaptations.

Hard X-rays tomography enables us to determine the shape and inner microstructure of a wide variety of solid states and their combinations. Biological samples, and in particular soft tissues from animals and humans, can be encapsulated in suitable containers to avoid deformations during the rotation and radiograph acquisition. The quantitative characterization of soft silicone objects, however, is challenging, because their shape depends on the orientation with respect to the gravitation force. Solid supports may help, but have some impact on the more or less complex-shaped silicone part. In medicine, silicone implants and devices play a major role, since the mechanical properties can be easily tailored. Using the nanotom® m (phoenix|x-ray, GE Sensing & Inspection Technologies GmbH, Wunstorf, Germany), we measured the X-ray absorption of silicone with pre-defined softness. The density resolution of the CT-system was insufficient to establish a density-stiffness relationship. In addition, the shape of a silicone part of a soft intraoral device was measured for three angles with respect to the gravitation force, for which the bending was obvious. These successful tomographic measurements will become the basis of finite element modeling to extract the mechanical properties including their local variations.

When using X-ray microtomography (XMT) to look for small mineral changes between successive scans in, for example, extracted human teeth, it is necessary to use a carefully designed protocol to ensure maximum accuracy. Our protocol is divided into four main elements: Mounting: The sample, in this case a tooth, must be mounted in such a way that it can be precisely relocated in the same place in the scanner (height and tilt must be precise; rotation can be changed post-scanning). Calibration: Over time (possibly weeks or months), the X-ray spectrum may change (with anode potential and current constant). This is due primarily to gradual anode pitting, which causes varying amounts of self-absorption. A calibration carousel is used to characterize the X-ray spectrum after every scan. Alignment: To ensure that the same regions are measured for every scan, it is necessary that all reconstructed images are perfectly aligned. We use a 7 degree-of-freedom alignment algorithm for this. To avoid loss of resolution when realigning data sets, second and subsequent scans are reconstructed with a 2× scale factor and “shrunk” during the realignment. Equalization: In spite of best efforts for calibration, repeatability for linear attenuation coefficient (LAC) measurement is of the order of 1%. To improve on this, regions are defined in which no change in mineral concentration should occur. In every scan, the mean LAC for these regions is measured and an appropriate LAC scaling factor is applied to the reconstructed volumes for the second and subsequent scans.

Bone has a complex hierarchical structure, which is essential for its performance. Bone is typically replete with cells called osteocytes that are embedded in the mineralized bone matrix in osteocyte lacunae, which are interconnected by canaliculi only a few hundred nanometer wide to form a vast cellular network. Our understanding of the osteocyte lacuno-canalicular network has been limited because of difficulties to image the cellular network within the opaque bone matrix in 3D. Synchrotron X-ray computed tomography is ideally suited to study the lacuno-canalicular network in bone because it combines the high penetration power of X rays with sub-micron resolution while retaining a fast acquisition time and thus high throughput. We discuss how synchrotron radiation-based tomography techniques have given insights into the osteocyte network in bone both in the form of regular tomography and, for higher resolution studies, in the form of nanotomography such as holotomography. These studies have provided quantitative measures of osteocyte lacunar properties and their relation to location within bones and bone challenges such as immobilization or lactation. Nanotomography revealed new features of the canalicular network that we term canalicular junctions, which are likely to play an important but hitherto hidden role in fluid flow dynamics within the bone cellular network. The examples illustrate how tomography provides information on complex biological materials like bone and we foresee that these capabilities will continue to improve with future/upgraded synchrotron X-ray sources.

Damage evolution in ductile metals is characterized by the nucleation, growth and coalescence of small internal voids. Many laws and studies have been developed to understand and predict damage. Numerical modeling, based on full field microstructure acquisition have proven also to be quite efficient. Real microstructure input are of utmost importance as some studies have shown that the deviation from the average behavior for voids growth can be huge and therefore allows a single cavity to drive the damage behavior and lead to failure. We develop the dedicated tool to analyze individually voids during in situ test and use results as input for numerical simulations.

A load frame for in situ mechanical testing is developed for the microtomography end stations at the imaging beamline P05 and the high-energy material science beamline P07 of PETRA III at DESY, both operated by the Helmholtz- Zentrum Geesthacht. The load frame is fully integrated into the beamline control system and can be controlled via a feedback loop. All relevant parameters (load, displacement, temperature, etc.) are continuously logged. It can be operated in compression or tensile mode applying forces of up to 1 kN and is compatible with all contrast modalities available at IBL and HEMS i.e. conventional attenuation contrast, propagation based phase contrast and differential phase contrast using a grating interferometer. The modularity and flexibility of the load frame allows conducting a wide range of experiments. E.g. compression tests to understand the failure mechanisms in biodegradable implants in rat bone or to investigate the mechanics and kinematics of the tessellated cartilage skeleton of sharks and rays, or tensile tests to illuminate the structure-property relationship in poplar tension wood or to visualize the 3D deformation of the tendonbone insertion. We present recent results from the experiments described including machine-learning driven volume segmentation and digital volume correlation of load tomography sequences.

X-ray Grating Talbot Interferometer (XGTI) is already routinely used for quantitative phase contrast imaging of soft tissue samples. XGTI can be realized using various measurement techniques, depending on the X-ray source used, the required spatial resolution and the speed of acquisition. The phase-stepping measurement technique, which is commonly used for XGTI data acquisition, needs multiple acquisitions for a single projection. For fast imaging the Moiré technique, a single-shot technique, is often preferred. However, it requires two gratings which increases the dose on the sample. We have therefore examined the Spatial Harmonic Imaging (SHI) technique which is low-dose and single-shot, using the I13-2 Diamond-Manchester Beamline at Diamond Light Source (DLS). The DLS I13-2 beamline is equipped with a Double Crystal Monochromator (DCM) and a Multi-Layer Monochromator (MLM) to deliver monochromatic beam, which work at the energy bandwidths (ΔE/E) of 10^-4 and 10^-2, respectively. However, the disadvantage of using these monochromators, especially for fast imaging, is loss of X-ray flux. It has already been shown that XGTI can work with an energy bandwidth (ΔE/E) of 10^-1. Our aim is to develop a single phase grating interferometer with pink beam from an undulator source, with X-ray mirror optics and multiple absorption filters, to obtain maximum possible flux with sufficient coherence and monochromaticity. We demonstrate performance with optimized beamline parameters for a photon energy of 15 keV with some demonstrative image reconstructions.

Microtomography using synchrotron radiation has proven to be a valuable tool for characterizing samples within biology, medical science, paleontology and materials science. A variety of attenuation contrast and phase-contrast imaging techniques were established to focus on spatial resolution, density resolution, and temporal resolution. In this paper, we focus on the recent developments at beamline P07 at PETRA III, Hamburg, Germany, for high-energy microtomography to provide high-density resolution for the study of a fossil tooth sample. Due to the integration of two independent CCDs for alternating use within the X-ray detector the imaging throughput was doubled. The concept, the alignment process, and first results are presented.

X-ray luminescence computed tomography (XLCT) is an emerging hybrid molecular imaging modality and has shown great promises in overcoming the strong optical scattering in deep tissues. Though the narrow x-ray beam based XLCT imaging has been demonstrated to obtain high spatial resolution at depth, it suffers from a relatively long measurement time, hindering its practical applications. Recently, we have designed a focused x-ray beam based XLCT imaging system and have successfully performed imaging in about 7.5 seconds per section for a mouse sized object. However, its high spatial resolution capacity has not been fully implemented yet. In this paper, with a superfine focused x-ray beam we design a focused-x-ray luminescence tomography (FXLT) system for spatial resolution up to 94 μm. First, we have described our design in details. Then, we estimate the performance of the designed FXLT imaging system. Lastly, we have found that the spatial resolution of FXLT can be further improved by reducing the scan step size, which has been demonstrated by numerical simulations.

The NIST Neutron and X-ray Tomography (NeXT) system provides simultaneous, complementary, multimodal information for the characterization of materials. Neutrons and X-rays yield complementary, non-destructive views of a complex system due to the contrast differences that arise from the differences in interaction with matter for the two modes. The NeXT system orients a 90 keV microfocus X-ray tube orthogonally to the thermal neutron beam. This enables truly simultaneous capture of the two modalities, thus it is possible to perform multimodal tomography of dynamic or stochastic samples while penetrating through sample environment equipment such as pressure and flow vessels. NIST has developed and continues to update distributable software for reconstructing and analyzing the multimodal datasets. Tomographic reconstruction offers multiple CPU and GPU based algorithms for parallel and cone beam geometries. Additional analysis routines perform volume registration and region of interest segmentation of the bivariate histogram that leverages the contrast differences between the two modalities. This talk will give an overview of the NeXT system, the algorithm developments underway and provide multiple research applications using the system.

The aim is to perform qualitative and quantitative assessment of metal induced artefacts of small titanium biomaterials using photon counting spectral CT. The energy binning feature of some photon counting detectors enables the measured spectrum to be segmented into low, mid and high energy bins in a single exposure. In this study, solid and porous titanium implants submerged in different concentrations of calcium solution were scanned using the small animal MARS photon counting spectral scanner equipped with a polyenergetic X-ray source operated at 118 kVp. Five narrow energy bins (7-45 keV, 45-55 keV, 55-65 keV, 65-75 keV and 75-118 keV) in charge summing mode were utilised. Images were evaluated in the energy domain (spectroscopic images) as well as material domain (material segmentation and quantification). Results show that calcium solution outside titanium implants can be accurately quantified. However, there was an overestimation of calcium within the pores of the scaffold. This information is critical as it can severely limit the assessment of bone ingrowth within metal structures. The energy binning feature of the spectral scanner was exploited and a correction factor, based on calcium concentrations adjacent to and within metal structures, was used to minimise the variation. Qualitative and quantitative evaluation of bone density and morphology with and without titanium screw shows that photon counting spectral CT can assess bone-metal interface with less pronounced artefacts. Quantification of bone growth in and around the implants would help in orthopaedic applications to determine the effectiveness of implant treatment and assessment of fracture healing.

The liquid-metal-jet technology has developed from prototypes into fully operational and stable X-ray tubes running in many labs over the world. A conventional X-ray tube generates X-rays when highly energetic electrons are stopped in a solid metal anode. The fundamental limit for the X-ray power generated from a given spot size is when the electron beam power is so high that it locally melts the anode. The liquid-metal-jet anode (MetalJet) technology solves this thermal limit by replacing the traditional anode by a thin high-speed jet of liquid metal. Melting of the anode is therefore no longer a problem as it is already molten, and significantly (currently about 10x) higher e-beam power densities can therefore by used. The assembly of a MetalJet source consists of an upper part that is the source head with the electron gun and the vacuum pump. This is like an open-type X-ray tube apart from the metal-jet anode that is formed by ejecting a ~200 µm diameter metal jet through a nozzle. To achieve 24/7 operation the 100 % of the liquid-metal-alloy can be recirculated. This pump pressurizes the metal alloy to approximately 200 bar to allow the jet to reach a speed of ~75 m/s. The applications include X-ray diffraction and scattering, but recently several publications have also shown very impressive imaging results using liquid-metal-jet anode technology, especially in 2-D or 3-D phase-contrast imaging and X-ray microscopy. Phase-contrast imaging achieves a significant improvement on the contrast and resolution of soft-issue with hard X-rays, however, the imaging quality, has been compromised by the low flux and brilliance using traditional microfocus tubes or adding optical elements. Therefore, the high brilliance liquid-metal-jet technology paves the way for the development of laboratory-scale phase-contrast imaging, especially its biomedical applications by enabling shorter exposure time, higher imaging resolution and contrast. Besides, the high stability of the source at its top performance perfectly matches the requirement of the associated phase-contrast imaging techniques. Sharp characteristic line of Gallium, one of the main components in the liquid metal alloy, MetalJet fits well in the optics-based, i.e. X-ray zone plate, X-ray microscopy, while at the same time, reducing the exposure time and maintaining the ultimate resolution. Several application examples will be given during the conference, illustrating the capability of MetalJet in commercial or in-house built X-ray microscopy system. The Kα line of gallium, which is just above the absorption edge of copper, makes MetalJet beneficial for imaging copper with high contrast. Therefore, its advantage in electronic CT.

This paper introduces a generative adversarial network (GAN) for low-dose CT (LDCT) simulation, which is an inverse process for network-based low-dose CT denoising. Within our GAN framework, the generator is an encoder-decoder network with a shortcut connection to produce realistic noisy LDCT images. To ensure satisfactory results, a conditional batch normalization layer is incorporated into the bottleneck between the encoder and the decoder. After the model is trained, a Gaussian noise generator serves as the latent variable controlling the noise in generated CT images. With the Mayo Low-dose CT Challenge dataset, the proposed network was trained on image patches, and then produced full-size low-dose CT images of different noise distributions at various noise levels. The network-generated low-dose CT images can be used to test the robustness of the current low-dose CT denoising models and also help perform other imaging tasks such as optimization of radiation dose to patients and evaluation of model observers.

Segmentation of large images can be one of the most time-consuming steps in the analysis of materials science datasets. Convolutional neural networks (NNs) have been shown to reduce segmentation time compared to manual techniques, but training a new NN is often required for each dataset. We show that simply combining NN training datasets does not necessarily lead to a NN capable of segmenting multiple types of images. In the present study, we first show that SegNet-based neural networks (NNs) can be trained to accurately segment Al-Zn x-ray computed tomography and Pb-Sn serial sectioning images. Applying the Al-Zn NN to the Pb-Sn test image led to misclassified smudges as dendrites, and misclassified speckles as background. Applying the Pb-Sn NN to the Al-Zn test image was unsuccessful, likely because the Al-Zn dendrites had a higher luminance than the Pb-Sn dendrites. The Mix NN (trained using the combined Al-Zn and Pb-Sn datasets) was better at segmenting the Pb-Sn test image than the Al-Zn test image. This is likely because the Pb-Sn training dataset contained ~4.5 times as many dendrite pixels as the Al-Zn training dataset, thus the Mix NN was over-tuned to identify Pb-Sn dendrites. Simply combining the training datasets was overall detrimental to NN performance, but assigning different classes to the Al-Zn and Pb-Sn dendrites may lead to enhanced performance in the future. These findings serve as guidelines in the quest to develop a universal NN for segmentation of large materials science datasets.

This manuscript explores a new approach for spectral X-ray tomography that uses K-edge filtering structures to attain spectral and spatially coded illumination which enables the acquisition of compressive measurements for the reconstruction of energy-binned images. The system is coined compressive spectral X-ray imaging (CSXI). A multi-stage algorithm is proposed to solve the non-linear ill-posed problem using sparse and low-rank regularization constraints to exploit the structure of the spectral data cube. The proposed algorithm can reconstruct both the energy binned images as well as the material decomposition of the object given a set of basis materials.

X-Ray parallel-beam micro-tomography systems at synchrotron facilities are mainly bespoke designs. Technical problems from a detector system can strongly affect the quality of reconstruction results. Radial lens distortion in the visible light optics causes streak artifacts which get stronger towards the edge of the image. The irregular response of the detector gives rise to a variety of stripe artifacts in the sonogram; full stripes, partial stripes, fluctuating stripes, and unresponsive stripes. These give rise to ring artifacts of different kinds in the reconstructed image. The scattering of the scintillation photons can cause artifacts which are similar to beam hardening artifacts and reduce the resolution of the image. Here we present our practical approaches to tackle each such problem. These approaches are easy to implement and have low computational cost. The algorithms are freely available as open-source software.

This paper proposes a new image reconstruction algorithm in sparse-view CT using the so-called nonlocal Total Variation (nonlocal TV) regularization. Compared to the previous work using the nonlocal TV, the proposed algorithm possesses the following three features. First, we introduce the newly developed modified nonlocal TV regularization term to preserve smooth intensity changes. Second, we utilize Passty’s proximal splitting framework to construct an accelerated iterative algorithm to minimize the cost function. Third, we introduce a novel technique called Selective Artifact Reduction (SAR) for further reduction of streak artifacts during the iteration. We demonstrate that the proposed algorithm can achieve significant image quality from 50-100 projection data with less than 20 iterations, through simulation studies using a clinical abdominal CT image.

Cone-beam breast computed tomography (CT) provides true 3D breast images with isotropic resolution and highcontrast information, detecting calcifications as small as a few hundred microns and revealing subtle tissue differences. However, breast is highly sensitive to x-ray radiation. It is critically important for healthcare to reduce radiation dose. Few-view cone-beam CT only uses a fraction of x-ray projection data acquired by standard cone-beam breast CT, enabling significant reduction of the radiation dose. However, insufficient sampling data would cause severe streak artifacts in images reconstructed using conventional methods. We propose a deep-learning-based method for the image reconstruction to establish a residual neural network model, which is applied for few-view breast CT to produce high quality breast CT images. In this study, we respectively evaluate the breast image reconstruction from one third and one quarter of x-ray projection views of the standard cone-beam breast CT. Based on clinical breast imaging dataset, we perform a supervised learning to train the neural network from few-view CT images to corresponding full-view CT images. Experimental results show that the deep learning-based image reconstruction method allows few-view breast CT to achieve a radiation dose <6mGy per cone-beam CT scan which is a threshold set by FDA for mammographic screening.

Total variation based on the L₀− and L₁−norm regularizers is a popular and powerful method for computed tomography (CT) reconstruction from noisy data. However their solutions are difficult to compute because of non-differentiability and even discontinuity. In this paper, we propose smoothing L₀- and L₁-norm regularizers to approximate the original norms. Also, we provide two expectation maximization (EM) like iterative algorithms, which have properties similar to that of the EM algorithm, including the monotonous decrement of the cost function and the self-constraint within a non-negative space. Finally, we build the relationship between the non-local means (NLM) methods and the above methods. The simulated CT projections are used to evaluate the performance of the proposed algorithms. The results show the superiority of the new methods over the existing methods in terms of noise suppression and computational cost.

Jaws can be affected by numerous lesions. This work aimed at demonstrating the interest of registration and fusion of Cone Beam Computed Tomography (CBCT) / Magnetic Resonance Imaging (MRI) examinations within the framework of the exploration of benign tumoral lesions of the jaws. These two modalities based on different physical principles could be complementary in terms of tissues contrast. Currently, separate analysis of CBCT and Jaws can be affected by numerous lesions that require exploration by imaging methods. This work is aimed at demonstrating the benefits of registration and fusion of Cone-Beam Computed Tomography (CBCT) and Magnetic Resonance Imaging (MRI) examinations within the framework of the exploration of benign tumoral lesions of the jaws. These two modalities based on different physical principles are generally complementary in terms of tissue contrast. Currently, the analysis of CBCT and MRI based on two distinct acquisitions is performed separately, but can be timeconsuming and challenging, especially when looking at multiple areas of tissue interfaces and even more so for nonspecialist medical imaging practitioners. Therefore, co-registration should be done by the operator based on reorientation of the axes to find equivalent cutting planes in both CBCT and MRI data. Such a process appears tedious. This work was done by the analysis of imaging records of benign lesions of the jaw from seven patients and based on the use of Horos*, a common medical image processing software using a non-rigid fiducial point-based registration. Our results show that interfaces between bone and lesional content benefit from this combined view of data: it could improve diagnosis and therapeutic approach. Such an approach was not communicated in dento-maxillary imaging for jawbone lesions. The preliminary results demonstrate the clinical potential of this approach and the need to perform further investigations to optimize CBCT-MRI registration techniques in dento-maxillary imaging, or even to motivate development of combined imaging instruments specific to this field.

Machine Learning is a rapidly growing field that has the power to change the way many tasks are performed. This tutorial serves as a basic introduction to Machine Learning for those who are unfamiliar with the topic. To fully understand the neural networks that make up the main stream of Machine Learning, the neural functions in biology are first discussed, followed by the translation of these concepts to artificial neurons as an example of biomimicry. Furthermore, basic knowledge and skills are given to analyze different neural networks. Finally, applications of this technology to medical imaging are mentioned, along with important improvements on the generic networks, including ResNet, GAN, LTSM, and quadratic networks.

Intraoral scanners have been gaining importance in dental offices. The technology has become a valuable and economically reasonable alternative to conventional silicone impressions and the conventional plaster casts because of the reduced treatment time and the sufficient precision achieved. The required precision of the dental prostheses depends on the clinical needs. For the production of working models, the tolerable inaccuracy is relatively large. Especially for crowns, bridges and larger dental prostheses, which include several teeth over the dental arch, however, extra work is often performed, that presumably caused by the limited accuracy of intraoral scanners. Therefore, the paper deals with the detailed evaluation of selected, commercially available intraoral scanner systems. For this purpose, we have designed and realized a model of a full arch upper jaw on the basis of clinically relevant imaging data. As well-defined references, we have incorporated cylinders with a diameter of 4 mm. This standard to be used as reference, was quantitatively characterized by four independent measurements using the advanced CT-system phoenix nanotom® m (GE Sensing & Inspection Technologies GmbH, Wunstorf, Germany) with a pixel size of 35 μm. The standard was also scanned using five commercially available intraoral scanners. In order to compare the accuracy of the ten measurements per scanner, the data were matched to the standard. Their displacements were made visible with GOM Inspect (GOM GmbH Braunschweig, Germany). Applying the same approach, we have analyzed the accuracy of two three-dimensionally printed dental models by the stereolithography printers Form 2, Formlabs Inc., Somerville MA, USA). The results demonstrate the currently possible level of precision by the selected intraoral scanners. They differ, however, not only in scanning time necessary and the ease of handling, but also in reachable accuracy. The precision ranged from (35 ± 5) μm (TRIOS® 3) to (97 ± 22) μm (Emerald^TM). The largest deviations to the standard were found in the transversal plane (412 ± 247) μm (Medit i500). The combination of advanced conventional microCT and adequate software for quantitative analysis permits a detailed evaluation of the performance of the currently available intraoral scanners.

Phase-contrast computed tomography (CT) have advantages of analyzing low Z objects such as polymer and soft tissue. Especially, X-ray grating interferometer CT is a practical method to obtain phase-contrast CT, but it has limited object size because of the limitation of the grating size. So, if the object is larger, the interior problem is occurred. It is known that there is no exact solution to solve this problem. In this study, we used machine learning to reduce the artifacts due to data truncation. We prepared the first input as a filtered backprojection (FBP) output, which is a classical image reconstruction method that has severe artifacts when data is truncated. And we also prepared the second input as geometrical information to clarify the region of interest (ROI). These networks were compared in two cases; a single input, two inputs. Visual results and quantitative results were used to compare image quality about various methods. Simulation results showed the better results than other methods. Our results show that machine learning is a promising technique to solve the CT challenges, may have many applications to all imaging fields.

Mice are used for models of almost all human diseases and are routinely scanned by micro-CT scanners. Mouse phantoms are often used for image-quality assessment. With recent developments in deep-learning-based preclinical imaging, there is a major need for large micro-CT datasets in which ground truth is known. In this study, we investigate the feasibility of making cost-effective deformable and reconfigurable mouse phantoms to generate real micro-CT datasets that reflect realistic underlying physical characteristics. Such datasets are highly desirable; for example, complicated photon-counting micro-CT datasets are needed for deep-learning-based material decomposition. In our scheme, mouse body parts are 3Dprinted with high precision using rigid or flexible materials. Liquid tissue surrogates (LTSs) or bioinks/cell lines could be used to emulate mouse organs and physiological fluid in the animals. LTSs provide realistic x-ray properties of their biological counterparts. The LTS organs could be contained in not only 3D-printed chambers, but also dialysis tubing, which emulates the cell membrane. Furthermore, through bioprinting and tissue engineering, organs and tissues can be made even more realistic for micro-CT and other types of tomographic scanning.

Tomographically measuring the temperature distribution inside a human body has important and immediate clinical applications - including thermal ablative and hyperthermic treatment of cancers, and will enable novel solutions such as opening the blood-brain barrier for therapeutics and robotic surgery in the future. A high intensity focused ultrasound (HIFU) device can heat tumor tissues to 50-90 °C locally within seconds. Thus, accurate, real-time control of the thermal dose is critical to eliminate tumors while minimizing damage to surrounding healthy tissue. This study investigates the feasibility of using deep learning to improve the accuracy of low-dose CT (LDCT) thermometry. CT thermometry relies on thermal expansion coefficients, which is prone to inter-patient variability, and is also affected by image noise and artifacts. To demonstrate that deep neural networks can compensate for these factors, 1,000 computer-generated CT phantoms with simulated heating spots were used in training both a “divide and conquer” and “end to end” approach. In the first strategy, the first encoder-decoder network corrected for beam hardening and Poisson noise in the image domain, while a second fine-tuned differences between predicted and ground truth heat maps. The second strategy is identical to the first, except only a single convolutional autoencoder was used as the CT images were not pre-cleaned. Ultimately, the two-part divide and conquer network increased thermal accuracy substantially, demonstrating exciting future potential for the use of deep learning in this field.

To investigate the effects of x ray tube setting on image quality in industrial computed tomography, an experimental characterization with constant tube powers has been reported in this paper. A series of CT scans for a QRM Medium-Contrast-Phantom were performed with a constant tube power of 40W and other scanning parameters, varying tube voltages from 80kV to 125kV and tube currents from 320μA to 500μA. The signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) measured on the reconstructed images indicated that increasing the tube voltage can improve the SNR, as well as the CNR in high density areas. While in low density regions in the phantom, higher CNR resulted from lower voltage or higher tube current. Furthermore, a custom-made aluminum cylinder is scanned several times for the assessment of the CT spatial resolution, similarly keeping a constant tube power and variable tube voltages and currents. According to the obtained modulation transfer function (MTF)_1/10 values, defined as the spatial frequency corresponding to a contrast loss of 10 %, it is found that using the same tube power, the tube voltage has a greater impact on improving the CT spatial resolution.

A major challenge in computed tomography (CT) is how to minimize patient radiation exposure without compromising image quality and diagnostic performance. The use of deep convolutional (Conv) neural networks for noise reduction in Low-Dose CT (LDCT) images has recently shown a great potential in this important application. In this paper, we present a highly efficient and effective neural network model for LDCT image noise reduction. Specifically, to capture local anatomical features we integrate Deep Convolutional Neural Networks (CNNs) and Skip connection layers for feature extraction. Also, we introduce parallelized 1 × 1 CNN, called Network in Network, to lower the dimensionality of the output from the previous layer, achieving faster computational speed at less feature loss. To optimize the performance of the network, we adopt a Wasserstein generative adversarial network (WGAN) framework. Quantitative and qualitative comparisons demonstrate that our proposed network model can produce images with lower noise and more structural details than state-of-the-art noise-reduction methods.

The X-ray Fuel Spray research at Argonne National Laboratory is aimed at utilizing synchrotron X-ray diagnostics for providing insights into automotive fuel injection. One important task is to generate high-fidelity geometries or iso-surfaces of steel fuel injector nozzles from X-ray Computed Tomography measurements, to be used as inputs to realistic CFD simulations of fuel injector flow. These fuel nozzles contain 3D features between 5 - 500 micron and are imaged at a pixel resolution of 1 micron. The main bottleneck to automated generation of an STL geometry from X-ray CT data is the segmentation or surface determination process – conversion of the CT volume into a binary map that classifies each voxel as belonging to either injector metal or flow domain, accurately locating the metal surface at the transitions between these domains. Here, we describe our recent success in automating the segmentation process itself, which is challenging because various artifacts that arise from X-ray imaging and CT reconstruction confound the identification of threshold values needed for traditional segmentation algorithms. A convolutional neural network (CNN) coupled with a tailored loss function is implemented to achieve state-of-the-art accuracy in surface localization with limited manual intervention. Through data augmentation, the model can be trained on less than 30% of the slices drawn from two CT scans of different automotive injectors that were manually segmented and is tested on a third. Our architecture achieves state-of-the-art accuracy at lower computation time and GPU memory requirement compared to U-net, one of the most popular architectures for image segmentation.