This PDF file contains the front matter associated with SPIE Proceedings Volume 11840 including the Title Page, Copyright information, and Table of Contents.

Deep learning, the mainstream of artificial intelligence (AI), has made progresses in computer vision, exacting information of multi-scale features from images. Since 2016, deep learning methods are being actively developed for tomography, reconstructing images of internal structures from their integrative features such as line integrals. There are both excitements and challenges in the Wild West of AI at large, and AI-based imaging in particular, involving accuracy, robustness, generalizability, interpretability, among others. Based on the author’s plenary speech at SPIE Optics + Photonics, August 2, 2021, here we provide a background where x-ray imaging meets deep learning, describe representative results on low-dose CT, sparse-data CT, and deep radiomics, and discuss opportunities to combine datadriven and model-based methods for x-ray CT, other imaging modalities, and their combinations so that imaging service can be significantly improved for precision medicine.

In 2016, the Block Museum, Northwestern University, was planning an exhibit “Paint the Eyes Softer” on portraits attached to mummies from Roman-era Egypt. These portraits are especially important because few other paintings have survived from the Greco-Roman tradition. Perhaps only 100 portraits remain attached to their mummies, serendipity provided an example amidst Northwestern’s campus, at Garrett Evangelical Theological Seminary and this mummy became the exhibit’s centerpiece. The mummy required conservation, and Computed Tomography (CT) was used to investigate subsurface deterioration and to reveal details of the encased person and items within the wrappings. CT provides a 3D map of how absorbing each reconstructed volume element is but not the identity of the material(s) producing the contrast, and in situ position-resolved x-ray diffraction at the Advanced Photon Source identified different objects within the mummy. This talk covers the provenance of Hawara Portrait Mummy Four, what CT tells us about the person and object enclosed and what secrets x-ray diffraction reveals.

This year, we celebrate the 125th anniversary of medical imaging based on Röntgen’s discoveries. Hands with a metallic ring have been replicated numerous times. The early images show impressive details on the bony structure. Thanks to tomographic imaging, the advanced X-ray instrumentation, the sophisticated algorithms of reconstruction and data evaluation as well as the preparation beyond the well-established staining we can make visible the micro- and nanostructures of the bone with its surrounding tissues with unequalled details, just to mention one example. This paper summarizes selected key aspects of the contributions in the 13th volume on Developments in X-ray Tomography.

The mammalian brain shows a complex and hierarchical architecture, whose assessment at all functionally relevant scales requires the establishment of multiomics approaches. In this work, we propose a correlative workflow, which is based on large-scale overview PC-CT scans using the extended beams offered by laboratory μCT sources and parallel beam synchrotron radiation (SR), with subsequent zooms into specific regions-of-interest using cone-beam recordings with nanofocused laboratory sources or SR, and finally SEM in controlled and wellidentified sub-volumes obtained before. We demonstrate the workflow at the example of rOTO-stained murine corpus callosum tissue, a brain region rich in myelinated nerve fibers. Based on two different and complementary techniques, PC-CT and scanning electron microscopy (SEM), we approach the establishment of a correlative imaging workflow. As we show here, the workflow can be applied (i) in a correlative study, in order to add further quantitative value, for instance, or (ii) in a multiscale approach, to which PC-CT can contribute volume throughput, while SEM can contribute resolution. The findings from this work demonstrate the complementary strength of each modality in terms of resolution (FIB-SEM) and FOV or volume throughput (PC-CT).

Microtomography (μCT) has become indispensable for imaging unique objects. We investigated a historic case of a ‘beaten-copper’ cranium from around 5,000 BCE discovered at the site of Els Trocs in the Spanish pyrenees. The skull of an approximately five-year-old child (ET1/CET1) shows distinct impressions of the cerebral convolutions (gyri cerebri) on the inner surface (tabula interna) of the frontal, parietal and occipital bone. These changes are visually reminiscent of beaten-copper and are associated with chronically increased intracranial pressure, which may have various causes. The investigation was performed with a nanotom m (Baker Hughes, Wunstorf, Germany). The 60 cm distance between source and detector and a horizontal shift of the detector (3,072 pixels × 2,400 pixels) allows visualizing specimens up to 35 cm in diameter. The 2,000 radiographs of the μCT-study were acquired with an acceleration voltage of 150 kV and a beam current of 60 mA using a pixel length of 68 μm.

New developments in x-ray technology have quickly been applied to the scientific study of mummies. A mummified cat was radiographed within a year of Roentgen’s discovery of x-rays and radiographs of the leg and foot of a human mummy followed within another year. X-ray computed tomography, in its initial incarnation of a head scanner was used to scan a mummy’s brain in 1977, and a mummy was put into a full body scanner shortly thereafter. Now mummies are being subject to micro-CT, synchrotron, MRI and other kinds of specialized imaging modalities. One of the most important aspects of radiological examinations of mummies is that they are non-destructive. In addition, they allow us to examine the biological characteristics (age, sex, health) of the individual and the cultural process of preparing them for eternity. This information sheds important light on cultures of the past. The radiological study of mummies is necessarily interdisciplinary, as it requires the expertise of the radiologist, the bioarchaeologist and the Egyptologist, to name only a few key team members. Unfortunately, the application of modern imaging modalities to the study of ancient mummies is generally characterized by a case study approach, with very little coordination or standardization. This paper will explore how the radiological study of mummies has evolved and where it needs to go.

Physical and mechanical properties of aligners determine the clinical success of orthodontic treatments. A main element of the successful orthodontic tooth movements is the fitting of the aligner’s surface to the backside of the related teeth. The complex human tooth anatomy and the aligner’s production make gaps inevitable. The aim of the tomography study is the morphological assessment of the recently introduced NaturAligner (Bottmedical AG, Basel, Switzerland). Using the advanced micro-CT system (nanotom m, phoenix|xray, Waygate Technologies, Wunstorf, Germany), a series of eight different aligners, placed on the 3D-printed model of the upper jaw, were visualized. Based on these 3D datasets, the gaps between model and aligner were automatically segmented and the thickness distribution of the aligners automatically determined. This quantification, validated by manual inspection, clearly indicated that aligners fitted better the model, when higher process temperatures were applied.

Defects in tooth enamel are associated with a multitude of health conditions. An ongoing push to improve our understanding of enamel formation is generating a large number of mutant mouse lines to map protein function and create an Enamel Atlas. Reproducible analysis of a large amount of micro-CT data to compare these mouse lines necessitates an automated, high-throughput method of segmenting enamel, bone, and dentin. Neither simple binary segmentation nor region growing algorithms are effective for enamel in continuously growing mouse incisors due to the gradient in mineral content. To overcome these limitations, we have trained and validated a 3D convolutional neural network (CNN) to semantically segment mouse jaws. The network adopted a UNet architecture and incorporated training data from synchrotron- and laboratory-based sources. We evaluated the performance of the 3D CNN for the wildtype by comparing segmented outputs to ground truth labels and to outputs from a similarly trained 2D network. Next we tested the adaptability of the network by segmenting mutant tissues displaying phenotypes ranging in severity. Finally, we will demonstrate the use of CNN-segmented datasets to calculate metrics for quantitative comparison of the 3D mineral distribution between wildtype and mutant genotypes. We will discuss segmentation of the incisor, which allows us to track changes in the mineral during each developmental stage of enamel production. Our results show that the CNN-based segmentation and quantification pipeline is a versatile tool that will empower enamel researchers, help delineate mechanisms of disease, and enable the development of new approaches of intervention.

X-ray computed tomography (XCT) is increasingly utilized industrially at material- and process development as well as in non-destructive quality control; XCT is important to many emerging manufacturing technologies, for example metal additive manufacturing. These trends lead to increased needs of safe automatic or semi-automatic data interpretation, considered an open research question for many critical high value industrial products such as within the aerospace industry. By safe, we mean that the interpretation is not allowed to unawarely or unexpectedly fail; specifically the algorithms must react sensibly to inputs dissimilar to the training data, so called out-of-distribution (OOD) inputs. In this work we explore data interpretation with deep neural networks to address: robust safe data interpretation which includes a confidence estimate with respect to OOD data, an OOD detector; generation of realistic synthetic material flaw indications for the material science and nondestructive evaluation community. We have focused on industrial XCT related challenges, addressing difficulties with spatially correlated X-ray quantum noise. Results are reported on training auto-encoders (AE) and generative adversarial networks (GAN), on a publicly available XCT dataset of additively manufactured metal. We demonstrate that adding modeled X-ray noise during training reduces artefacts in the generated imperfection indications as well as improves the OOD detector performance. In addition, we show that the OOD detector can detect real and synthetic OOD data and still model the accepted in-distribution data down to the X-ray noise levels.

Propagation-based imaging is an X-ray phase-contrast method that does not require any additional optical elements in the imaging system. Instead, phase-contrast effects are observed as measurable intensity variations after an incident X-ray wave field has propagated some distance between the object and plane of detection. Subsequent phase-retrieval processing of the recorded images is a necessary step in order to obtain quantitative information and can greatly increase the image signal-to-noise ratio, in particular in computed tomography. Most commonly this is done by approximating the object as a single material, which cannot always be justified. An example of this is the presence of bones and soft-tissues in biomedical imaging, which motivated the work presented here. We developed the combination of spectral X-ray imaging and propagation-based X-ray phasecontrast imaging to overcome the limitations of a single-material approximation. The addition of spectral data enables phase-retrieval on multi-material objects, as well as the use of processing methods well-known from conventional dual-energy imaging, such as material decomposition, virtual monochromatic images and effective atomic-number mapping. Here, we present first computed tomography results of spectral propagation-based phase-contrast. We discuss how the phase retrieval process can be interpreted as low- and high-pass filters. Lastly, we take a look at current and future technological challenges, and the potential application areas of spectral propagation-based phase-contrast imaging.

Creating a viable reconstruction method for Compton scatter tomography remains challenging. Accounting for scatter attenuation when the underlying attenuation map is not known is particularly difficult, and current mathematical approaches to this vary widely. This work explores a novel approach to joint scatter and attenuation image reconstruction, which leverages the underlying structural similarity between the two images and incorporates a deep learning model in an alternating iterative reconstruction scheme. A single-view CT imaging procedure for recording Compton scatter is first described. A joint reconstruction model, which iterates between algebraically reconstructing scatter images and estimating the attenuation via deep learning is then proposed. This model is tested on both a generated dataset of 2D phantom images designed to mimic human tissues as well as a realistically simulated dataset based on real CT images. Testing results yield convergence of the model and decent reconstruction quality, demonstrating the potential principled utilities of this configuration and deep learning approach. The model achieved a structural similarity index measure of at least 0.84 for scatter and 0.89 for attenuation reconstructions with the realistically simulated dataset. The iterative, deep learning approach outlined in this work shows potential for future efficient medical imaging procedures, reconstructing images with limited scatter information.

We introduce a method to extract density information from an x-ray computed tomography (XCT) volume that is more accurate than simply assuming density is proportional to CT number. XCT is a versatile tool for analysis, however, for lab-based XCT machines that employ polychromatic x-rays, it is difficult to extract anything more than the crudest quantitative data from the sample. Reconstructed tomograms values are, in theory, the x-ray attenuation coefficients of the material. However, due to the polychromatic nature of the beam, and effects such as beam hardening, such an interpretation of real data is rarely feasible. The Alvarez-Macovski (AM) equation, which is used in quantitative XCT reconstruction algorithms, provides a model of x-ray attenuation. We use the AM equation to extract quantitative information from conventionally reconstructed tomograms, provided it is not too severely affected by beam-hardening artefacts. In essence, we assume that the tomogram values are proportional to the attenuation coefficients of the AM equation at a mean x-ray energy. Then, given a calibration scan which contains enough materials, we can solve the the AM equation for the unknown coefficients and exponents. We then apply it to tomograms of objects with similar shape and material composition. The quantitative data extracted thus provides a more accurate estimate of both per-material density and bulk density.

The phase imaging has a higher sensitivity for low-Z materials than conventional absorption imaging. We have developed a high-resolution X-ray phase microscope in combination with a Lau interferometer and used it for phase tomography. However, an existing method cannot avoid artifacts originating from the assumption of a two-beam interference model. In this study, we take a three-wave interference model into account to reduce the artifacts and propose a new method to attain phase tomography. In the presentation, we will demonstrate the reduction of the artifacts with the results of phase tomography.

Spatial resolution in standard phase-contrast X-ray imaging is limited by the finite number and size of detector pixels. As a result, this limits the size of features that can be seen directly in projection images or tomographic reconstructions. Dark-field imaging allows information regarding such features to be obtained, as the reconstructed image is a measure of the position-dependent small-angle X-ray scattering of incident rays from the unresolved microstructure. In this paper we utilize an intrinsic speckle-tracking-based X-ray imaging technique to obtain the effective dark-field signal from a wood sample. This effective dark-field signal is extracted using a Fokker-Planck type formalism, which models the deformations of illuminating reference-beam speckles due to both coherent and diffusive scatter from the sample. We here assume that (a) small-angle scattering fans at the exit surface of the sample are rotationally symmetric, and (b) the object has both attenuating and refractive properties. The associated inverse problem, of extracting the effective dark-field signal, is numerically stabilised using a “weighted determinants” approach. Effective dark-field projection images are presented, as well as the dark-field tomographic reconstructions obtained using Fokker-Planck implicit speckle-tracking.

The goal of this work is to develop an x-ray computed tomography (CT) capability that delivers improved imaging resolutions while reliably identifying the material composition of the interrogated object. Through the use of a hyperspectral x-ray detector along with a multi-metal patterned anode, one can simultaneously enhance achievable spatial resolution and improve the spectral signal through the use of energy intervals that capture the k-lines of each material present in the anode. This paper will present preliminary Monte-Carlo results of the anode design and simulated CT datasets along with the applied machine learning techniques to identify materials and their concentrations.

The progress of scientific research and technology development greatly depends upon effective imaging solutions for characterizing the properties and behaviors of materials. Revealing details of microstructure, ideally in 3D, is a critical part of this understanding, whether the goal is to develop and confirm models that describe material properties and behaviors or simply to visualize structural details. ZEISS offers 3D X-ray microscopes (XRM): advanced imaging solutions that have removed major hurdles for three-dimensional imaging by achieving high contrast and submicron resolution imaging even for relatively large samples. These groundbreaking advances in non-destructive, three-dimensional (3D) imaging empower a broad range of technical disciplines.

There is a large performance gap between X-ray tomography instruments with conventional laboratory X-ray sources and synchrotron radiation sources. An Inverse Compton Scattering (ICS) source can provide 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. Here we present an ICS source design that is more than two orders of magnitude brighter than sources currently in operation, well-suited for high resolution, micro-CT imaging of millimeter-sized samples at micron resolution, with a flux density similar to some high-energy synchrotron beamlines.

Talbot-Lau interferometers are widely used for X-ray phase imaging/tomography. For thick objects or materials including metals, high-energy X-rays should be used. However, absorption gratings with extremely high aspect-ratios (AR) are necessary but are hard to be fabricated by current microfabrication techniques. As an approach without using high-AR absorption gratings, we developed the high-energy X-ray phase imaging device with a structured anode X-ray source, which has tungsten target array embedded in a diamond substrate. X-rays are emitted from the tungsten region, and therefore G0 can be omitted. We will show the results of high-energy X-ray phase imaging and tomography performed with a design energy of 82 keV.

We present the most recent advances in fast X-ray grating interferometer and their applications. A dedicated setup for rapid scanning with a single grating and using filtered broadband illumination of an undulator source has been implemented. With this setup tomographic scans can be achieved within minutes. Larger number of samples can be measured and chemical processes can be studied. Improvements in the data processing technique will be presented. We will also describe the new capabilities and applications.

Magnesium-based alloys are suitable materials for biodegradable bone implants due to their high biocompatibility and mechanical properties similar to bone. Diffraction tomography (DCT) is an imaging technique based on acquiring positionresolved diffraction patterns at multiple angular orientations. DCT is able to provide information about the crystal structure of hydroxyapatite (HAp) in bone, which is necessary for understanding the influence of magnesium degradation on the bone tissue. This article presents a study of a sheep bone explant containing a biodegradable magnesium-zinc-calcium (BRI.Mag, Mg-Zn-Ca) alloy implant. The DCT data were reconstructed and analyzed using the MATLAB computing environment and ASTRA Toolbox. Thus, the DCT images with 200 μm pixel size were reconstructed and, therefore, a DCT study workflow was developed. That allowed to retrieve the crystal lattice parameters of HAp such as full width at half maxima (FWHM) and reflection position around the implant surface of a selected (002) diffraction peak. Based on these parameters, the corresponding d-spacing and crystal size of HAp were determined. It was shown that the degradation of the implant does not notably affect the lattice spacing of the bone within a distance of 3 mm from the implant surface. At the same time, the crystal size decreases closer to the implant (1.5 mm). The developed workflow for the reconstruction of a volume-resolved diffraction experiment of a bone explant containing metal implants is a convenient approach to investigate the crystal structure of the sample on exact locations in a non-invasive way. The workflow will be further applied for a broader range of samples containing Mg-based biodegradable implants.

Tomography using diffracted x-rays produces reconstructions mapping quantities including crystal lattice parameter(s), crystallite size and crystallographic texture; this information is quite different from that obtained with absorption contrast or phase contrast. Diffraction tomography can be performed using energy dispersive diffraction (EDD) and polychromatic synchrotron x-radiation. In EDD, different, properly-oriented Bragg planes diffract different x-ray energies; these intensities are measured by an energy sensitive detector. A pencil beam defines the irradiated volume, and a collimator before the energy sensitive detector selects which portion of the irradiated column is sampled at any one time. A 3D map is assembled by translating the specimen along X, Y and Z axes. This paper reports results of 3D mapping of the integrated intensity of several reflections from the bioapatite in the mineralized cartilage centrum of a blue shark. The multiple detector EDD system at 6-BM-B, the Advanced Photon Source was used to map an entire blue shark centrum. The shark centrum consists of a double cone structure (corpora calcerea) supported by the intermedialia consisting of four wedges. The integrated intensities of the c-axis reflection and of a reflection with no c-axis component reveals the bioapatite within the cone wall is oriented with its c-axes lateral, i.e., perpendicular to the axis of the backbone, whereas the bioapatite within the wedges is oriented with its c-axes axial. Results of absorption microCT (laboratory and synchrotron) and x-ray excited x-ray fluorescence mapping are included to provide higher resolution data of the structures underlying the EDD maps. Application of EDD tomography to 3D mapping of large specimens promises to add to the understanding of other mineralized tissue samples which cannot be sectioned.

The visual systems of pollinating insects are studied to understand the effects of a changing environment, and are to date imaged using microscopy or micro-CT. Microscopy only allows two-dimensional imaging and conventional micro-CT requires heavy-metal staining of the samples. Here we present virtual histology of compound eyes of bumblebee (Bombus terrestris) using a laboratory nano-CT system. Propagation-based phase-contrast CT allows three-dimensional imaging of samples without using any staining. Micrometre-resolution images of the microanatomy of the eyes were reconstructed, and the features identified in CT (cornea, crystalline cones, pigment, photoreceptor cells, basement membrane) were confirmed with electron microscopy.

Formalin fixation and paraffin embedding of post mortem tissue specimens is widely used for high-resolution neuroimaging with both conventional and X-ray virtual histology. Exchange of embedding solutions generates non-uniform brain shrinkage and changes relative tissue densities. We used synchrotron radiation-based X-ray micro computed tomography to visualize the embedding process for a single mouse brain. Non-rigid registration was employed to determine the volumetric strain fields and to track the X-ray absorption changes of corresponding features. This allows for a correction of the observed microanatomy to improve the anatomical context. Through embedding, the entire brain shrinks to around 40% of its volume in formalin. Shrinkage is non-uniform and varies over anatomical regions and the distance to external surfaces.

Surgery is an essential part of the curative plan for most patients affected with solid tumors. The outcome of such surgery, e.g., recurrence rates and ultimately patient survival, depends on several factors where the resection margin is of key importance. Presently the resection margin is assessed by classical histology, which is time-consuming (several days), destructive, and basically only gives two-dimensional information. Clearly it would be advantageous if immediate feedback on tumor extension in all three dimensions were available to the surgeon intra-operatively. In the present paper we investigate a laboratory propagation-based phase-contrast xray computed tomography (CT) system that provides the resolution, contrast, and, potentially, the speed for this purpose. The system relies on a liquid-metal jet micro-focus source and a scintillator-coated CMOS detector. The study is performed on paraffin-embedded non-stained samples of human pancreatic neuroendocrine tumors, liver intrahepatic cholangiocarcinoma, and pancreatic serous cystic neoplasm (benign). We observe tumors with distinct and sharp edges having cellular resolution (∼10 μm) as well as many assisting histological landmarks, allowing for resection margin assessment. All x-ray data is compared with classical histology. The agreement is excellent, and we conclude that the method has potential for intra-operative three-dimensional virtual histology.

A preliminary investigation into the use of cycloidal computed tomography for intraoperative specimen imaging is presented. Intraoperative imaging is applied in time-sensitive clinical settings, where obtaining a high-resolution, highquality image within minutes is paramount in evaluating the success of operations and/or the need for additional surgery. As a flexible imaging method that is compatible with x-ray phase contrast imaging, cycloidal computed tomography can provide both high spatial resolution and high image contrast, whilst keeping scan times short thanks to an effective under-sampling approach. To gather early evidence, the method was tested on resected breast and oesophageal tissue. The results, although preliminary, indicate that cycloidal imaging may indeed be beneficial for intraoperative specimen imaging, although further studies are required to confirm this potential.

X-ray speckle-based imaging (SBI), one of them most recent phase-contrast imaging methods, has received growing interest in the last years. Its simplicity, cost-effectiveness and robustness combined with the high phase sensitivity and compatibility with laboratory X-ray sources make it an attractive method for visualising even minute density differences in samples. Since its first demonstration, SBI has seen rapid development and a range of applications have been identified. Among the various ways to perform SBI, the unified modulated pattern analysis (UMPA) offers a number of advantages. Here, we present an overview of the state of the art of SBI, including some of our work using UMPA in the recent years. We demonstrate the potential of UMPA for applications such as optics characterisation, biomedical and geological imaging and discuss its translation from large-scale synchrotron facilities to the laboratory.

Inline phase tomography using synchrotron radiation with sub-micrometer voxel sizes is nowadays the gold standard for investigation of soft and hard tissues with micron resolution. Recent developments on detectors and X-ray sources allow the transfer of the technique into laboratory environment. For the comparison of three manufacturers, we performed microtomography with advanced laboratory microtomography devices with micron resolution on a porcine nerve, a zebrafish embryo and a historic human tooth. These data sets were also compared with data acquired at the ANATOMIX beamline at Synchrotron Soleil and the TOMCAT beamline at SLS. For the lab-based experiments following scanners were chosen: Skyscan 2214 (Bruker-microCT, Kontich, Belgium), Xradia 620 Versa (Zeiss, Oberkochen, Germany) and a prototype with a MetalJet X-ray source from Exillum from the company Exciscope (Stockholm, Sweden). All devices contained detectors including X-ray optics.

The Helmholtz-Zentrum Geesthacht, Germany, is operating the user experiments for microtomography at the beamlines P05 and P07 using synchrotron radiation produced in the storage ring PETRA III at DESY, Hamburg, Germany. Attenuation-contrast and phase-contrast techniques were established to provide an imaging tool for applications in biology, medical science and materials science. In the recent years we rebuilt the integration of imaging detectors. This allows the user to choose from a set of cameras based on different CMOS and CCD sensors. Here we will present the features of the different camera system together with the advantage for different applications. Furthermore, we rebuilt the data preprocessing before reconstruction to provide different scanning techniques to investigate samples larger than the field of view of the X-ray beam. Multi-scale tomography is realized by using different setups or to integrate a low-resolution together with a high-resolution region-of-interest invest

GEMS is the user platform of Helmholtz-Zentrum Geesthacht with a unique infrastructure for complementary research with photons and neutrons. The instruments using synchrotron radiation are operated at DESY in Hamburg, the instruments using neutrons are located at the research reactor FRM II of the Heinz Maier-Leibnitz Center (MLZ) in Garching near Munich. GEMS provides methods for applied materials research to user groups from both industry and science. At the PETRA III synchrotron storage ring of the German Electron Synchrotron (DESY), GEMS operates instruments for tomographic analyses from the micro- to the nanoscale. The high-energy tomography station of the beamline HEMS/P07 uses photon energies from 50 to 200 keV. A robot sample changer enables automated high-throughput measurements. The beamline is therefore of special interest for industrial users, especially from the metal industry sector. In this contribution we introduce the tomography set-up at HEMS and the possibilities for industrial access, together with examples for tomography of challenging highly-absorbing samples.

Phase-contrast imaging is one of the standard X-ray imaging methods at synchrotron beamlines and has already proven to be beneficial for soft-tissue visualization. However, most implementations use single-distance inline phase-contrast techniques, and are thus not able to provide quantitative information. To access these, grating-based imaging (GBI) setups or, rather recently, speckle-based imaging (SBI) methods can be used. We built a new grating-based setup at the beamline P05 operated by HZG at the storage ring PETRA III / DESY. This new setup overcomes the previously reported limitations in spatial resolution compared to inline phase-contrast imaging. Furthermore, it allows for accurate quantitative phase contrast micro computed tomography of biological soft tissue. We replaced the typically used sandpaper by a 2D phase-grating as a wavefront marker, which increased the visibility and allowed for using fewer phase steps. Combined with an existing SBI phase-retrieval algorithm, the so-called Unified Modulated Pattern Analysis (UMPA) and an optimized scan protocol, we reached a resolution below 4 microns in scan times less than two hours. We investigated stained and unstained tissue samples, to quantify the staining process of different tissue types and were able to observe an increase in electron density, dependent on the stain and tissue type. By this, we could show the successful operation of our setup to quantitatively investigate samples on a micro meter scale at the beamline P05.

Laboratory x-ray fluorescence (XRF) computed tomography (XFCT) with nanoparticles (NPs) as contrast agents now allows for in vivo preclinical imaging and longitudinal studies at low radiation dose. We present on developments of our XFCT arrangement capable of low-dose (<25 mGy) imaging with high signal-to-background resulting in high-spatial-resolution (200-400 μm) in vivo imaging of Ru, Rh and Mo NPs injected and accumulated locally in mice. We further demonstrate multiplexing capabilities by cross-talk-free separation of Ru, Rh and Mo XRF signal as well as envisioning the future of preclinical XFCT for active targeting and imaging of molecular markers (e.g., cancer cells).

We show exact correlation between light microscopy of historical ground section microscope slides and XMT and OCT scans of the prepared slides. Samples were selected to represent a wide range of sizes and skeletal and dental tissue types, including human femoral bone, human permanent teeth, dog carnassial tooth, narwhal mandible, black rhinoceros molars, sperm whale cementum and dentine, African elephant ivory, and prairie marmot molars. XMT was conducted using the QMUL Mucat2 system, nominal voxel size 25um, 90kV, 24 hours. Analysis used TomView, ImageJ and Drishti software. In each case we were able to match XMT and LM. We can now report mineralisation densities for all the calcified tissues in the context of LM imagery.

One of the imaging modalities offered by the Imaging and Medical Beamline (IMBL) at the Australian Synchrotron is Xray phase-contrast propagation-based computed tomography (PB-CT). The unique combination of high coherence and high brightness of radiation produced by synchrotron X-ray sources enables phase contrast imaging with excellent sensitivity to small density differences in soft tissues and tumors. The PB-CT images using spatially coherent radiation show high signal-to-noise ratio (SNR) without reducing the spatial resolution. This is due to the combined effect of forward free-space propagation and the advanced step of phase retrieval in the reconstruction processes that allows to accommodate noisier recorded images. This gives an advantage of potentially reducing the radiation dose delivered to the sample whilst preserving the reconstructed image quality. It is expected that the PB-CT technique will be well suited for diagnostic breast imaging in the near future with the advantage that it could provide better tumor detection and characterization/grading than mammography and other breast imaging modalities/techniques in general. The PB-CT technique is expected to reduce false negative and false positive cancer diagnoses that result from overlapping regions of tissue in 2D mammography and avoid patient pain and discomfort that results from breast compression. The present paper demonstrates that PB-CT produces superior results for imaging low-density materials such as breast mastectomy samples, when compared to the conventional absorption-based CT collected at the same radiation dose. The performance was quantified in terms of both the measured objective image characteristics and the subjective scores from radiological assessments. This work is part of the ongoing research project aimed at the introduction of 3D X-ray medical imaging at the IMBL as innovative tomographic methods to improve the detection and diagnosis of breast cancer. Major progress of this project includes the characterization of a large number of mastectomy samples, both normal and cancerous.

Deep neural network architectures based on unrolling optimization algorithms have been widely adopted in deep-learning based image reconstruction applications in recent years. However, these architectures only mimic the iterative schemes of the corresponding algorithms, but lack rigorous convergence guarantee; and the learned network layers are difficult to interpret. These issues have hindered their applications in clinical use. In this paper, we develop an efficient Learned Descent Algorithm with a Line Search strategy (LDA-LS) and apply it to the nonconvex nonsmooth optimization problem of low-dose CT (LDCT) reconstruction. We show that LDA-LS yields a highly interpretable neural network architecture, where the regularization parameterized as multilayer perception is explicitly integrated into the iterative scheme and learned during the training process. We demonstrate that LDA-LS retains convergence guarantee as classical optimization algorithms, while achieving improved efficiency and accuracy in LDCT image reconstruction problems.

Statistical reconstruction methods in X-ray Computed Tomography (XCT) are well-regarded for their ability to produce more accurate and artefact-free reconstructed volumes, in the presence of measurement noise. Maximum-likelihood methods are particularly salient and have been shown to result in superior reconstruction quality, compared with methods that minimise the ℓ2 residual between measured and projected line attenuations. Least-squares more generally may refer to the minimisation of quadratic forms of the projected attenuation residuals. Early maximum-likelihood methods showed promising reconstruction capabilities but were not practical to implement due to very slow convergence, especially compared with least-squares methods. More recently, leastsquares methods have been adapted to minimise quadratic approximations to (negative) log-likelihood, thereby attaining the speed of least-squares minimisation in service of likelihood maximisation for superior reconstruction fidelity. Quadratic approximation to the log-likelihood under Poisson measurement statistics has been demonstrated several times in the literature. In this publication we describe an approach to quadratically expanding loglikelihood under an arbitrary noise model, and demonstrate via simulation that this can be implemented practically to maximise likelihood under mixed Poisson-Gaussian models that describe a broad range of transmission XCT imaging systems.

Deep learning based methods have achieved promising results for CT metal artifact reduction (MAR) by learning to map an artifact-affected image or projection data to the artifact-free image in the data-driven manner. Basically, the existing methods simply select a single window in the Hounsfield unit (HU) followed by a normalization operation to preprocess all training and testing images, based on which a neural network is trained to reduce metal artifacts. However, if the selected widow contains the whole range of HU values, the model is challenged to predict the dedicated narrow windows accurately since the contribution of small HU values to the training loss may not be sufficiently weighted relative to that for large HU values. On the other hand, if a selected window is small, the opportunity will be lost to train the network effectively on features of large HU values. In practice, various tissues and organs in CT images are inspected with different window settings. Therefore, here we propose a multiple-window learning method for CT MAR. The basic idea of multiple-window learning is that the content of large HU values may help improve features of small HU values, and vice versa. Our method can precisely process multiple specified windows through simultaneously and interactively learning to remove metal artifacts within multiple windows. Experimental results on both simulated and clinical datasets have demonstrated the effectiveness of the proposed method. Due to its simplicity, the proposed multiple-window network can be easily incorporated into other deep learning frameworks for CT MAR.

Compressed sensing (CS) computed tomography has been proven to be important for several clinical applications, such as sparse-view computed tomography (CT), digital tomosynthesis and interior tomography. Traditional compressed sensing focuses on the design of handcrafted prior regularizers, which are usually image-dependent and time-consuming. Inspired by recently proposed deep learning-based CT reconstruction models, we extend the state-of-the-art LEARN model to a dual-domain version, dubbed LEARN++. Different from existing iteration unrolling methods, which only involve projection data in the data consistency layer, the proposed LEARN++ model integrates two parallel and interactive subnetworks to perform image restoration and sinogram inpainting operations on both the image and projection domains simultaneously, which can fully explore the latent relations between projection data and reconstructed images. The experimental results demonstrate that the proposed LEARN++ model achieves competitive qualitative and quantitative results compared to several state-of-the-art methods in terms of both artifact reduction and detail preservation.

Micro-/nano-CT has been widely used in practice to offer noninvasive 3D high-resolution (HR) imaging. However, increased resolution is often at a cost of a reduced field of view. Although data truncation does not corrupt high-contrast structural information in the filtered back-projection (FBP) reconstruction, the quantitative interpretation of image values is seriously compromised due to the induced shifting and cupping artifacts. State-of-the-art deep-learning-based methods promise fast and stable solutions to the interior reconstruction problem compared to analytic and iterative algorithms. Nevertheless, given the huge effort required to obtain HR global scans as the ground truth for network training, deep networks cannot be developed in a typical supervised training mode. To overcome this issue, here we propose to train the network with a low-resolution (LR) dataset generated from LR global scans which are relatively easily obtainable and obtained excellent results.

Tooth cementum annulation (TCA) is used for determining age-at-death and stress periods based on yearly deposited lines in the root cementum of human teeth. Traditionally, TCA analysis employs optical microscopy, which requires cutting sections of the root and provides only sparse sampling in the third dimension. Ancient teeth are unique specimens that should not be sliced. In this imaging study, we show that extended field of view synchrotron radiation-based tomography provides true micrometer resolution and coverage for non-destructively surveying for incremental lines. To rapidly review the root cementum layer of four teeth from early 19th century cemetery with historical records of life events, we employed machine learning for semi-automatic detection and analysis of incremental lines. Surveying large regions of the root cementum enables detection of incremental lines and hence improves TCA analysis as an alternative to slicing of the unique teeth.

It is well known that beam-hardening gives rise to errors in the reconstructed linear attenuation coefficient that vary with position, thus invalidating post-reconstruction calibration methods. Prior to reconstruction, projections can be linearized, but this works only for samples composed of a single phase. Two-dimensional beam-hardening correction has been proposed for the two-phase case. A method for forward projection from polyhedral phantoms has also been proposed in the past. Here, this method has been extended to simulate polychromatic radiation and a hierarchical representation of phantom materials is proposed. This is used to create a liquid-immersed tooth phantom. The phantom projection data was then used to validate and improve the two-dimensional beam-hardening correction algorithm.

With the appearance of the field-emission cold cathode x-ray source, the fast response and small in size innovate the traditional imaging system. Although the field-emission x-ray source array generates multiple novel imaging modalities, it still faces a long stand-off distance between source and object. To realize a portable, smart, extremely low dose imaging modality, researchers proposed a concept to design a two-dimensional array field-emission source, namely the flat-panel source. In this paper, a real imaging system based on the ZnO field-emission flat-panel source is designed. Currently, the real flat-panel source faces an extremely low dose and non-addressable situation. Hence, the measurement based on the flat-panel source is overlapped and without an application potential. We first try to realize the imaging ability of the flat-panel source by designing a rebinning algorithm. With the analysis on the overlapped measurement, a high-order prior is introduced into the rebinning algorithm to improve the performance. Simulation and real data experiments verified our proposed method. Compared to the no high-order prior, the proposed algorithm can recover a more distinct measurement.

Micro-computed tomography (micro-CT or μCT) is a standard tool for materials science research. This method allows high-resolution non-destructive analysis of the internal structure of an object. In this work, the internal structure of B₄C-C composites has been examined using µCT for quantitative and qualitative evaluation. Boron carbide is a promising material for X-ray refractive lenses due to its properties - quite high refractive ability with high transmission of x-ray radiation. Experimental samples were fabricated by the spark plasma sintering method. It has been shown that μCT can be used to determine the parameters of the internal structure of B₄C-C composite for further optimization of x-ray refractive optics manufacturing.

We report about multiscale tomography with high throughput at the Diamond beamline I13L. The beamline has the purpose of multi-scale and operando imaging and consists of two independent branchlines operating in real and reciprocal space. The imaging branch -called Diamond-Manchester branchline- hosts micro-tomography, grating interferometry and a full-field microscope. For rapid recording a broad spectrum of the undulator radiation is used either with band-passing the light with a combination of a filter and a deflecting mirror or using a multilayer monochromator. For all the methods similar recording times can be achieved, with typical scanning times of some minutes and covering the resolution range from microns to the 100nm range. Most recently a robot arm has been installed to increase the throughput to 300 samples per day. The system is now implemented for user operation in remote operation mode for the micro-tomography setup and can be expanded to the two other experiments. The instrumental capabilities are applied on various topics such as the study of biodiversity of insects or the structural variations of electrode materials in batteries. Fast recording with dedicated sample environments (not using the sample changing robot) enables operando studies in many areas, the charging/discharging cycles on batteries, the degradation of teeth enamel under various conditions or loading brine sandstone mixtures with CO2, to name some examples. For imaging with highest spatial resolution we managed to improve significantly the recording speed of ptycho-tomography, which is now in the order of hours and will be reduced further. We demonstrated in the past 2-D recording with 10kHz and expand the instrumental capability with specific hardware dependent triggering and scanning schemes. We expand the research program for multi-scale imaging across both branchlines (imaging and coherence branchlines) with first studies such as batteries, brain research, concrete.