This PDF file contains the front matter associated with SPIE Proceedings Volume 12242, including the Title Page, Copyright information, table of Contents, and Conference Committee Page.

Several thousand X-ray computed tomography systems are currently operated in clinics, industry, and academia. While conventional attenuation-based imaging is still dominant, alternative and complementary contrast mechanisms are under development and already in use. The generated big datasets require state-of-the art methods for image reconstruction, artefact removal, and analysis. Several technical advancements are enabling or pushing applications of tomography in pathology, tissue engineering, anthropology, etc. It is common to produce impressive imagery of unique objects and derive relevant features of the underlying structures and dynamics. Multi-modal imaging, which includes synergistic and reciprocal information, has played an increasing role. This interdisciplinary field comprises researchers and users from medicine/dentistry, biology, earth and materials science, crystallography, solid-state and soft-matter physics, chemistry, computational sciences, engineering, and applied mathematics to fruitfully combine results on system and component developments, algorithmic design and optimization, performance evaluation, and tomography experiments. This paper summarizes key aspects of the 55 contributions in the 14th volume on Developments in X-ray Tomography, arranged in sections on instrumentation, algorithms, and applications.

Phase contrast micro-CT is a powerful technique allowing imaging of soft tissue at synchrotrons or using laboratory sources. The use of contrast agents is a useful approach when imaging vascular structures. However, common x-ray contrast agents typically rely on heavy metals to increase absorption, which may affect the phase contrast and cause artifacts in the reconstructed volumes. Thus, utilizing an agent with lower attenuation similar to soft tissue is clearly advantageous. Here, we evaluated different colored radiopaque solutions (tissue marking dyes) which had been injected into the vascular system of bovine lung samples, prior to embedding in paraffin. Scans were performed using a micro-focus x-tube calibrated to 10 µm spot size at 70 kV and a photon counting detector with a silicon sensor and 75 µm pixels. The resulting volumes have a voxel size of (25 µm)³ , limited by the size of the samples, but sufficient to resolve the vascular system. Experiments confirmed that sufficient perfusion of the vessels with the dyes could be achieved, and the different dyes could be clearly discerned in the reconstructed volumes without causing artifacts allowing to clearly identify features in the soft tissue. Further, the findings were confirmed by histology.

X-ray phase-contrast tomography (XPCT) is a non-destructive, three-dimensional imaging modality which provides higher contrast in soft tissue than absorption-based CT, and allows one to cover the cytoarchitecture from the centi- and millimeter scale down to the nanoscale. To further increase contrast and resolution of XPCT, for example in view of addressing connectivity issues in the central nervous system (CNS), metal staining is indispensable. However, currently used protocols, for example based on osmium and/or uranyl are less suited for XPCT, due to an excessive β/δ-ratio. Neodymium(III) acetate (NdAc) has recently been proposed as a nontoxic, non-radioactive easy to use alternative contrast agent for uranium acetate (UAc) in electron microscopy. In this work we investigate NdAc as a contrast agent for tissues of the CNS by X-ray phase-contrast tomography. Due to its vertical proximity to UAc in the periodic table, similar chemical but better suited optical properties for phase contrast can be expected. The results suggest NdAc to be an effective contrast agent by binding to negatively charged biomolecules with a uniform infiltration, high binding specificity and good reproducibility of results.

Imaging of microvasculature is primarily performed with X-ray contrast agents, owing to the wide availability of absorption-contrast laboratory source µCT compared to phase contrast capable devices. Standard commercial contrast agents used in angiography are not suitable for high-resolution imaging ex vivo, however, as they are small molecular compounds capable of diffusing through blood vessel walls within minutes. Large nanoparticle-based blood pool contrast agents on the other hand exhibit problems with aggregation, resulting in clogging in the smallest blood vessels. Injection with solidifying plastic resins has, therefore, remained the gold standard for microvascular imaging, despite the considerable amount of training and optimization needed to properly perfuse the viscous compounds. Even with optimization, frequent gas and water inclusions commonly result in interrupted vessel segments. This lack of suitable compounds has led us to develop the polymeric, cross-linkable X-ray contrast agent XlinCA. As a water-soluble organic molecule, aggregation and inclusions are inherently avoided. High molecular weight allows it to be retained even in the highly fenestrated vasculature of the kidney filtration system. It can be covalently crosslinked using the same aldehydes used in tissue fixation protocols, leading to stable and permanent contrast. These properties allowed us to image whole mice and individual organs in 6 to 12-month-old C57BL/6J mice without requiring lengthy optimizations of injection rates and pressures, while at the same time achieving greatly improved filling of the vasculature compared to resin-based vascular casting. This work aims at illuminating the rationales, processes and challenges involved in creating this recently developed contrast agent.

A novel approach for the localization of X-ray stains in individual cells with ptychographic computed tomography was recently demonstrated. With a basis material decomposition in the image domain the quantitative stain concentration can be retrieved with nanoscopic resolution. So far, the approach was demonstrated for two different staining elements, but it has the potential to be easily applied to other contrast agents. The noise in the resulting basis materials depends on the one hand on the noise in the original data, and is furthermore given by the basis materials and specifically their attenuation properties and electron density. In this work, we theoretically predict the standard deviation in basis material images for several heavy elements. This is done by applying the material decomposition algorithm to an experimental data set. It was found that for more electron dense materials the decomposition should have lower noise in the basis material image of the stain. Furthermore, the energy used for the measurement impacts the standard deviation in the basis material images especially for materials with low atomic number, for which therefore lower X-ray energy should be used for data acquisition.

Small-angle X-ray scattering (SAXS) probes nanoscale features in reciprocal space. Apart from statistical information of size and shape in the probed volume, the orientation of anisotropic nanostructures can be directly obtained from the anisotropy of the scattering patterns. By using a focused X-ray beam and raster-scanning a sample through the beam, images can be created with a real-space resolution defined by beam- and stepsize while each pixel contains information from the nanostructure in reciprocal space. For an extension of the technique to computed tomography, new reconstruction algorithms as well as experimental schemes probing additional tilt angles of the tomography axis have been developed in order to retrieve 3D reciprocal space maps in each voxel, extending the method to 6 dimensions. The technique is in particular powerful for hierarchical systems where relevant length scales are spanning over many orders of magnitudes. For the example of bone, the arrangement of mineralized collagen fibrils in the nanoscale can be probed over extended macroscopic samples millimeters or even centimeters in size.

X-ray ptychography in 3D is ideal for quantitative structural analysis with highest spatial resolution. The Ptychographic Nano-Analytical Microscope (PtyNAMi) installed at beamline P06 at PETRA III (DESY, Hamburg) is optimized for high-resolution scanning hard X-ray microscopy. The current data evaluation pipeline for ptychographic tomography at PtyNAMi is outlined and the performance of the microscope demonstrated at the example of an inverse opal Ni sample.

X-ray dark-field imaging is used to visualize the ultra-small angle x-ray scattering signal that originates from sub-resolution density fluctuations within the sample microstructure. Dark-field tomography using the edge-illumination x-ray imaging system is presented as a tool for measuring this scattering signal in a sample in three dimensions. Its applicability to different fields is shown through example images of a multi-material phantom, a tissue-engineered esophagus, a pouch cell battery and a short-fiber reinforced composite material. The multichannel contrast available in edge-illumination helps with material identification, with high contrast at boundaries enhancing dark-field reconstructions.

This works illustrates the capability of scanning Small-Angle X-ray Scattering (sSAXS) and Ptychographic X-ray nano Tomography (PXCT) to characterize nanoporous composite materials over multiple length scales on a fuel cell catalyst. Results comprise direct images of the nanostructure in a 30µm catalyst pillar at 20 nm resolution in combination with statistically averaged information of the nanostructure between 1-200nm from sSAXS with a spatial resolution of few tens of µm over macroscopic areas. The ex-situ material study is complemented by in-situ dynamic imaging of vapor condensation in 3D at 20nm spatial and temporal resolution of 10minutes.

Tomographic experiments at modern synchrotrons require software and hardware solutions for automatic selecting representative features for high-resolution scanning, real-time feedback on the sample motion and changes of the internal structure, as well as on-demand data saving mechanisms to reduce projection volumes from fast detectors. Here we present a new real-time imaging monitoring instrument allowing for automatic 3D zooming to features of interest, optimized organization of data flow significantly reducing acquired data sizes, and real-time sample reconstruction and visualization. Efficacy of the instrument was demonstrated during several tomographic experiments at beamline 2-BM of the Advanced Photon Source at Argonne National Laboratory.

The Helmholtz-Zentrum Hereon, 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. Within the recent years we built an imaging pipeline to optimize the deliverd radiation dose onto the sample with respect to the applied imaging technique. This became possible by the modernisation of the experiment control and the integration of different imaging detectors. In combination with an new concept for a high-speed X-ray shutter the applied radiation dose can be adjusted from high, for optimization of the statistic within the tomogram, to low, for avoiding any radiation based artefact. Within this talk the recent hardware and software developments integrated to the microtomography imaging system at the beamlinew P05/PETRA III and P07/PETRA III will be presented. Furthermore, the optimization of dose will be demonstrated on selected samples.

X-ray phase-contrast tomography (X-PCT) techniques are capable of imaging samples with small differences in densities. They enable scientists to study biological or medical samples using high energy X-rays, which means less X-ray absorption and less sample damage, with high contrast quality. One branch of these techniques known as speckle-based methods have been well developed and demonstrated on real applications by different groups of developers using their own codes. However, there is lack of collective effort to package these methods into an open-source software which is easy-to-install, easy-to-use, well-documented, and optimized for speed. Such software is crucial to make the X-PCT techniques accessible to generic users and become regular tools. This report demonstrates the effort which implements speckle-based phase-retrieval methods in Python and GPU.

The Helmholtz-Zentrum Hereon is operating several tomography end stations at the beamlines P05 and P07 of the synchrotron radiation facility PETRA III at DESY in Hamburg, Germany. Attenuation and phase contrast imaging techniques are provided as well as sample environments for in situ/operando/vivo experiments for applications in biology, medicine, materials science, etc. Very large and diverse data sets with varying spatiotemporal resolution, noise levels and artifacts are acquired which are challenging to process and analyze. Here we report on an active learning approach for the semantic segmentation of tomography data using a guided and interactive framework, and evaluate different acquisition functions for the selection of images to be annotated in the iterative process.

The DIAD beamline at Diamond Light Source is a new dual-beam instrument for quasi-simultaneous full-field imaging and powder diffraction. DIAD enables its users to visualise internal structures (in 2- and 3D) as well as identify and measure compositional and strain changes. The aim is to enable in-situ and in-operando experiments that require spatially correlated results from both X-ray Computed Tomography and X-ray Powder Diffraction. DIAD has two independent beams in the medium energy range (8-38keV) that are generated from a single source 10-pole wiggler. Both beams are combined at one sample position. Here, radiography and/or tomography can be performed using either a monochromatic or pink beam, with a 1.4 mm x 1.2mm field of view and a pixel size of 0.5 µm. Micro-diffraction is possible with a variable beam size between 13µm x 5 µm up to 50 µm x 50µm. The talk will start with a description of the multi-modal capabilities of DIAD and the unique data collection pipeline that has been developed to reduce post-processing workloads. This will be followed by recent results from the fields of material science and bioengineering. Finally, the talk will conclude by giving the audience tips on how to gain access to DIAD.

Radiation dose reduction is one of the most important topics in the field of computed tomography (CT). Over past years, deep learning based denoising methods have been demonstrated effective in reducing radiation dose and improving the image quality. Since the paired low-dose and normal-dose CT are usually not available in clinical scenarios, various learning paradigms are studied, including the fully-supervised learning based on simulation data, the weakly-supervised learning based on unpaired noise-clean or paired noise-noise data, and the self-supervised learning based on noisy data only. Under neither clean nor noisy reference data, unsupervised/self-supervised low-dose CT (LDCT) denoising methods are promising to processing real data and/or images. In this study, we propose the first-of-its-kind Self-Supervised Dual-Domain Network (SSDDNet) for LDCT denoising. SSDDNet consists of three modules including a projection-domain network, a reconstruction layer, and an image-domain network. During training, a projection-domain loss, a reconstruction loss, and an image-domain loss are simultaneously used to optimize the denoising model end-to-end using the single LDCT scan. Our experimental results show that the dual-domain network is effective and superior over single-domain networks in the self-supervised learning setting.

Convolutional neural networks (CNNs) have been widely used for low-dose CT (LDCT) image denoising. To alleviate the over-smoothing effect caused by conventional mean squared error, researchers often resort to adversarial training to achieve faithful structural and texture recovery. On one hand, such adversarial training is typically difficult to train and may lead to a potential CT value shift. On the other hand, these CNNs-based denoising models usually generalize poorly to new unseen dose levels. Recently, diffusion models have exhibited higher image quality and stable trainability compared to other generative models. Therefore, we present a Contextual Conditional Diffusion model (CoCoDiff) for low-dose CT denoising, which aims to address the issues of existing denoising models. More specifically, during the training stage, we train a noise estimation network to gradually convert a residual image to a Gaussian distribution based on a Markov chain with a low-dose image as the condition. During the inference stage, the Markov chain is reversed to generate the residual image from random Gaussian noise. In addition, the contextual information of adjacent slices is also utilized for noise estimation to suppress potential structural distortion. Experimental results on Mayo LDCT datasets show that the proposed model can restore faithful structural details and generalize well to other unseen noise levels.

Low performing pixels (LPP)/missing/bad channels in CT detectors, if left uncorrected cause ring and streak artifacts, structured non-uniformities, and make the reconstructed image unusable for diagnostic purposes. Many image processing methods are proposed to correct the ring and streak artifacts in reconstructed images, but it is more appropriate to correct the LPPs in sinogram domain as the errors are localized. Although Generative Adversarial Networks based sinogram inpainting methods have shown promise in interpolating the missing sinogram information, it is often observed that the reconstructed images lack diagnostic value especially in visualizing soft tissues with certain window width and level. In this work, we propose a deep-learning based solution that operates on the sinogram data to remove the distortions cause by LPPs. This method leverages the CT system geometry (including conjugate ray information) to learn the anatomy aware interpolation in the sinogram domain. We demonstrated the efficacy of the proposed method using data acquired on GE RevACT multi-slice CT system with flat-panel detector. We have considered 46 axial head scans out of them 42 sets are used for training and the remaining 4 sets for validation/testing. We have simulated isolated LPPs accounting for 10% of total channels in the central panel of the detector and corrected them using the proposed approach. Detailed statistical analysis has revealed that, approximately 5dB improvement in SNR is observed in both sinogram and reconstruction domain as compared to classical bicubic and Lagrange interpolation methods. Also, with reduction in ring and streak artifacts, the perceptual image quality is improved across all the test images.

Hard X-ray nanotomography is a commonly used tool in many research areas such as material science, biology and medicine. The nanotomography station at the P05 imaging beamline at PETRA III at DESY is operated by the Helmholtz-Zentrum Hereon and optimized for full-field X-ray nanoimaging techniques. It offers spatial resolutions down to below 50 nm, as well as a high temporal resolution. The technical design allows for a high flexibility and is optimized for in situ experiments. The two major full-field techniques offered to the user community are transmission X-ray microscopy with optionally Zernike phase contrast and near-field holography. Here, the different full-field nanoimaging techniques, as well as the latest technical developments are presented.

A novel automated workflow for the recovery of image resolution using deep convolutional neural networks (CNNs) trained using spatially registered multiscale data is presented. Spatial priors, coupled with high order voxel-based image registration, are used to correct for uncertainties in image magnification and position. A network is then trained to remove the effects of point spread from the low-resolution data, improving resolution while reducing image noise and artefact levels. While benchmarking on real materials, including biological, materials science and electronics samples, we find that resolution recovery improves quantitative and qualitative measurements, even if certain image details cannot be easily identified from the original low-resolution data.

Scanning coherent X-ray microscopy (ptychography) has gained considerable interest during the last decade since the performance of this indirect imaging technique does not necessarily rely on the quality of the X-ray optics and, in principle, can achieve highest spatial resolution in X-ray imaging. The method can be easily extended to 3D by adding standard tomographic reconstruction schemes. However, the tomographic reconstruction is often applied in a subsequent step using a sequence of aligned ptychographic 2D projections. In this contribution, we outline current developments of a GPU-accelerated framework for direct 3D ptychography, coupling 2D ptychography and tomography. The program utilizes a custom GPU-accelerated framework for ptychography that offers three distinct ptychographic reconstruction algorithms. The tomographic reconstruction runs simultaneously and uses numerical routines of the ASTRA Toolbox. This parallel-computing approach results in a high performance increase considerably reducing the reconstruction time of 3D ptychographic datasets.

Edge illumination is a phase sensitive X-ray imaging technique, compatible with lab-based X-ray sources. Phase information can be retrieved by displacing two masks relative to each other with the object positioned in between. Phase retrieval then allows the three contrasts, absorption, refraction and dark field, to be individually reconstructed with FBP. In this work, a novel joint reconstruction method is proposed using a combined forward model, allowing the three contrasts to be reconstructed simultaneously, without a retrieval step. This allows for more freedom in the acquisition scheme. The proposed objective function is minimized using a split Barzilai-Borwein gradient method. Improved convergence speed and reconstruction quality on an experimental dataset, compared to existing methods, is shown.

Dental restorations should match the color of the surrounding enamel. Carefully selecting the appropriate shade for the filling material is a challenge for dentists. Moreover, tooth color can change over time due to habits such as smoking or drinking coffee. In the last few years, single-shade dental composites have come to the market. They rely on a chameleon effect to provide acceptable to good color matching regardless of the tooth color. The chameleon effect refers to a dental filling’s ability to guide light in such a way that its color blends in with that of the tooth. Structural color is a contributing factor to the chameleon effect and an active area of research where structures at the submicron scale play a critical role. We investigated the size, shape, and three-dimensional spatial arrangement of filler particles in single-shade dental resin composites. Cylindrical samples of dental composites were prepared and imaged with the transmission X-ray microscope at the ANATOMIX beamline, Synchrotron SOLEIL, France. The centers of the filler particles were determined from the tomography data. Combined with shape information from scanning electron microscopy, a Monte Carlo approach was used to model the transmittance for light at wavelengths from the visible to the ultraviolet. The results were compared to optical transmission measurements. The combination of nanotomography and simulation can thus help to understand the influence of the size and distribution of filler particles on the chameleon effect.

Cementum deposits on mammalian teeth contain layered microstructures associated with the chronological age of an animal and other details of their life history. Hard X-ray tomography data captured this record contained within the cementum deposits from whole teeth without sectioning. We investigated three teeth of African bovids, namely gemsbok (Oryx gazella), eland (Taurotragus oryx), and African or Cape buffalo (Syncerus caffer) using the laboratorybased system nanotom m for measuring each complete tooth to identify relevant regions, which were scanned at the ANATOMIX beamline of Synchrotron SOLEIL, France. Using microtomography in archaeological materials such as teeth, eliminates the need for tooth sectioning, making it a desirable alternative for archaeologists and museum curators. Synchrotron measurements enabled the application of pixel sizes as low as 0.65µm, which generated around 40 TB of data. The three adult bovids investigated here, have a known day of death and season of death, and come from regions with distinct seasonal patterns in temperature and/or rainfall. They also have an estimated age at death based on occlusal wear. The known information serves as a control to determine the applicability of microtomography on whole teeth of large bovids. Preliminary results show that microtomography can successfully replace the need of sectioning in cementum dental analysis. Our future goal is to develop a protocol to standardize procedures of tooth cementum analysis in bovids using microtomography.

Dogfish (Squalus spp.) possess one or two dorsal-fin spines located at the dorsal midline over the vertebral column. These spines are heavily mineralized, and surface bands of light and dark contrast are used in age determinations. The interior of the spine also contains growth bands visible via optical microscopy of thin sections, but the three-dimensional pattern of growth bands does not appear to have been quantitatively mapped nor has the microstructural origin of the band contrast been established. This paper reports synchrotron microComputed Tomography (microCT) results on spines of Squalus suckleyi. MicroCT was performed at beamline 2-BM of the Advanced Photon Source (APS). There are numerous parallel bands, and their contrast consists of varying (higher and lower) values of linear attenuation coefficient, similar to growth bands observed in other mineralized tissues including mammalian cementum and dentin. The microCT data are supplemented by x-ray excited x-ray fluorescence maps of a sectioned Pacific dogfish spine recorded at beamlines 8-BM and 2-ID-E, APS; of particular note are bands of high Zn content, something which has been found in growth bands in other mineralized tissues.

The quantification of appearance and geometrical changes between structures captured across several images relies on the determination of correspondences between these image structures. Automatic methods for establishing sparse and dense spatial correspondences are feature tracking and image registration, respectively. Main registration challenges are weak image features, complex appearance change, large shape changes, and huge images. These challenges can make the problem intractable. We present approaches to overcome these challenges for registration tasks including microtomography images. Applications include assessing tissue changes due to embedding material, comparison of microtomography with optical microscopy images, and optimization of process parameters for materials.

Applications of X-ray computed tomography (CT) for porosity characterization of engineering materials often involve an extended data analysis workflow that includes CT reconstruction of raw projection data, binarization, labeling and mesh extraction. It is often desirable to map the porosity in larger samples but the computational challenge of reducing gigabytes of raw data to porosity information poses a critical bottleneck. In this work, we describe algorithms and implementation of an end-to-end porosity mapping code "Tomo2Mesh" that processes raw projection data from a synchrotron CT instrument into a porosity measurement and visualization within minutes on a single high-performance workstation equipped with GPU.

Determining fiber orientations in a fiber reinforced material with X-ray CT is challenging because the width of the individual fibers is often much smaller than the spatial resolution of the X-ray images. However, the anisotropic X-ray scattering signal of the fibers yields information about the fiber direction without the need to individually resolve them. In this work, edge illumination phase contrast imaging is used to sample the anisotropic X-ray scattering in the object along different directions. The measured scatter is represented on a sphere by a set of spherical harmonics. Next, the orientation of the fibers is extracted from the scatter function by constrained spherical deconvolution. This approach is experimentally validated, for both parallel and crossing fiber distributions in a single voxel, using Monte Carlo simulations in GATE. The retrieved fiber orientations for parallel and crossing carbon fibers are presented as result.

X-ray based inspection often relies on triangular meshes, for example to inspect objects that were manufactured from CAD models. In this work, we present three complementary implementations of X-ray mesh projectors, obtained by adapting state-of-the-art rendering techniques to the simulation of X-ray imaging. The first technique is rasterization, where the interaction of each triangle with the X-ray beam is simulated in parallel using the NVIDIA CUDA toolkit. The second approach is ray tracing, where the interaction of each ray with the mesh is simulated in parallel using the NVIDIA OptiX framework. Both recursive and non-recursive versions of ray tracing are described. The simulated XCT setup is described in terms of a cone beam projection geometry that is compatible with the corresponding geometry in the ASTRA toolbox. All three projectors were benchmarked on a series of tests with varying resolution of both the mesh and the detector. The rasterizer exhibited the best computation time in most benchmark scenarios, coupled with the best scalability w.r.t. both the mesh size and the detector size. However, the recursive ray tracing approach offers more capabilities towards implementing additional optical effects such as refraction.

X-ray thermography is a novel non-destructive method for visualizing the thermal distribution inside an object on the basis of the slight density changes caused by thermal expansion. The high transparency of X-rays and high sensitivity of crystal-based X-ray interferometry (CXI) provides a means to visualize the three-dimensional distribution of temperature changes with a few-degree resolution. We developed an X-ray thermography system that uses a combination of CXI and monochromated synchrotron radiation. We conducted experiments to visualize the time-resolved dynamical thermal flow in heated water and food with our system and successfully obtained a three-dimensional image of the thermal distribution in heated water.

Ultrahigh-resolution CT is highly desirable in many clinical applications, such as cochlear implantation and coronary stenosis assessment. Despite the wide accessibility to micro-CT for preclinical research, there are no such micro-CT scanners for clinical use, due to several issues including high radiation dose, patient movement, limited X-ray source power and long image time. To meet the challenge, here we design a robotic-arm-based clinical photon-counting micro-CT system in the interior tomography mode for temporal bone imaging at a 50 µm resolution. We adopt twin collaborative robotic arms. These robotic arms can be operated near humans without danger of injury due to collision. One robot steers the source, the other the detector. The six-axis robot pair defines a volume of interest (VOI) more flexibly than the traditional rotation gantries and C-arm systems. The robotic system allows arbitrary imaging geometry. The required spatial resolution is on the same order of the robot mechanical precision, therefore the geometric errors due to robotic coordination together with patient motion, and system misalignment must be addressed. In this paper, we propose a locally linear embedding-based motion correction method to solve this problem to correct all these geometric errors in an unified framework with a nine-degree-of-freedom model. Different from the conventional motion correction methods that rely on gradient-based parametric optimization, our method utilizes a parametric searching mechanism through iterative reconstruction and without assuming any smoothness of the errors as a function of the view angle. The effectiveness of our method is first demonstrated with numerical simulation and further verified on a set of limited-angle cone-beam scans of a sacrificed mouse using our robotic-arm-based imaging system. Sharper images with clearer anatomical details in the absence of misalignment artifacts are obtained after our data-driven geometric calibration, suggesting a great potential of our method in biomedical and industrial applications.

Edge illumination X-ray phase contrast imaging is a method which relies on two gratings to obtain attenuation, phase and dark field contrast. Current gratings consist of periodic apertures in a high-absorbing flat plate. While edge illumination was originally designed for parallel-beam imaging, it also works with cone-beam sources when the opening angle is sufficiently low. With higher opening angles, however, current gratings cause a shadow which results in a decreased intensity. In this paper, three alternative grating geometries are studied with Monte-Carlo simulations and their performance is evaluated in terms of shadowing. Results show that the alternative grating geometries significantly reduce shadowing in cone-beam based edge illumination.

X-ray phase imaging with grating interferometers, such as the Talbot interferometer, is widely performed even with a laboratory X-ray source. However, the achievable spatial resolution is normally limited by the period of gratings. In this work, two laboratory-based apparatuses are developed to overcome the constraint of the spatial resolution. One is the combination of a commercially available FZP-based X-ray imaging microscope and Lau interferometer optics. The two-step deconvolution approach is explained to attain phase tomography. The other is a sub-period super-resolution X-ray phase imaging, which is based on the sample-scanning scheme across the beamlet array formed by a triangular phase grating. A proof-of-concept result of the super-resolution approach is presented.

X-ray phase-contrast imaging by grating interferometry has in the past decades been established both at synchrotron radiation sources and at conventional x-ray tubes. Due to the sensitivity to x-ray phase shifting, contrast in soft tissue has been improved significantly and the possibilities to do virtual histology of tissue biopsies are being pioneered. Constraints in the design of a grating interferometer are dominated by the phase-shifting structures of the gratings. Typically, binary gratings are used, with alternating regions of shift and no shift. However, other structures are also possible, leading to different patterns of intensity modulations. Here we simulate the possibility of using multiple binary structures in the phase grating, to utilize Moiré effects and avoid the use of attenuating grating structures. In particular the parameters for grating-based phase-contrast imaging using a single optical element are discussed, with focus on laboratory-based x-ray imaging using microfocus sources.

Classical histology is the gold standard for soft-tissue analysis, and it is performed by imaging thin slices of a sample with a microscope. While the resolution is high in two dimensions, the sample preparation to be able to record images is time consuming, destructive and effectively limiting the recorded data to two dimensions. We propose virtual histology by x-ray phase-contrast computed tomography (CT) as a non-destructive imaging method to be used as a complement to classical histology. With our laboratory system, we focus on detailed imaging of soft-tissue tumours with the aim of distinguishing the tumour from healthy tissue.

Previously, we designed a new quantitative phase-contrast setup for the microtomography experiments at the hereon beamlines P05 and P07 at PETRA III, DESY. This setup is based on Talbot array illuminators (TAI) as high visibility wavefront markers and can reach a spatial resolution level comparable to propagation-based imaging. In this work, we focus on the progress of bringing this setup into user operation. We collaborated with INI-Research to investigate the vascular system of mouse lymph nodes, exploiting the spatial resolution capabilities and using the quantitative aspect of the data to compare different sample preparation methods. We could successfully visualize and trace the blood supply of the lymph nodes, even with fine capillaries, showing the stability and performance of the setup in user application.

Assessing the complex three-dimensional (3D) structure of the cochlea is crucial to understand fundamental aspects of signal transduction in the inner ear and is a prerequisite for the development of novel cochlear implants (CIs). Phase-contrast x-ray tomography (XPCT) offers destruction-free 3D imaging with little sample preparation, thus preserving the delicate structure of the cochlea. In this work, we present XPCT data of small-animal and human cochleae obtained at dedicated synchrotron radiation (SR) and inhouse µ-CT setups. For µ-CT we explore heavy metal stains as radiocontrast agents and show optimized phase retrieval. With the brilliant SR we achieve high contrast for unstained whole cochleae at the cellular level.

Joint tissues consist of trabecular and cortical bone as well as calcified and hyaline cartilage, which presents a challenge for hard X-ray-based visualization on the sub-cellular level due to the wide range of local X-ray absorption values. The density of the calcified tissues requires rather high photon energy, which often leads to insufficient contrast within the cartilage and impedes the visualization of individual biological cells. Decalcification of the tissues reduces the total and local X-ray absorption values and allows for selecting a lower photon energy. Further contrast enhancement can be achieved by ethanol fixation and paraffin tissue embedding. In this study, we (i) searched for an appropriate visualization method to investigate lesions generated by a laser osteotome and (ii) visualized a decalcified porcine joint after ethanol fixation and subsequent paraffin embedding using laboratory- and synchrotron radiation-based microtomography. The experiments at the ANATOMIX beamline of Synchrotron SOLEIL were performed in off-axis scan mode with a pixel size of 1.3 µm. Individual cells in all layers of the joint could be made visible and the effect of ethanol fixation and paraffin embedding demonstrated.

Although radiographs are the most frequently used medical imaging modality worldwide due to their cost-effectiveness and widespread accessibility, the structural superposition along the x-ray paths often renders lung nodules difficult to detect. In this study, we apply “X-ray dissectography” to dissect the lungs digitally in a few radiographic projections, suppress the interference of irrelevant structures, and improve lung nodule detectability. Then, we design a novel collaborative detection network to localize lung nodules in both the dissected 2D projections and the 3D physical space. Our experimental results show that our approach can significantly improve the average precision by 20+% in comparison with detecting lung nodules from the original projections using a popular detection network. Our proposed approach and results suggest a potential in re-designing the current X-ray imaging protocols and workflows and improving the diagnostic performance of chest radiographs in lung diseases.

Edge illumination is an emerging X-ray phase contrast imaging technique, which has been successfully transferred from synchrotron facilities to lab-based X-ray systems. Rather than installing a dedicated phase contrast system, our goal is to enable phase contrast imaging in a highly flexible X-ray CT system, FleXCT, a novel scanning system consisting of ten motorized axes. These axes allow movements of not only the sample stage, but also the source and detector. To enable phase contrast imaging, two gratings have to be incorporated in the FleXCT system. In this work, we report on the procedure to determine the optimal grating parameters, relying on numerical simulations. Optimal parameter values are presented for both FleXCT gratings.

A fast-moving economy requires X-ray technology that inspects container traffic promptly and with high accuracy to identify nonconforming cargo. Unfortunately, conventional X-ray CT with a rotating gantry cannot be used because of the limited throughput. Instead, X-ray radiographs are acquired, but these do not provide depth information of the objects to be inspected. A novel solution is to use rotation-free CT geometries, which aim to preserve the image quality by using multiple static X-ray sources, albeit their optimal position and orientation have not been extensively studied to date. In this work, we investigate a rectangular, fixed-gantry CT geometry, for which we implemented a comprehensive simulation framework using the ASTRA Toolbox.¹ In addition, the orientation of the X-ray sources was optimized to perform ROI imaging with enhanced image quality, in terms of the mean squared reconstruction error. Lastly, we challenged the equidistant distribution of X-ray sources as the preferred configuration and compared the reconstructions with those of a projected circular distribution. Although spatially bound improvements were observed, future research is required.

The spatial resolution of an x-ray computed tomography (CT) system is determined by its components, including the x-ray source, detector, and image reconstruction process. It is necessary to analyze the contributions of each component to understand the total spatial resolution of the system. Detector blur in flat panels is usually estimated with an edge function and data fitting, but the contribution of cross-talk (inter-pixel signal leakage) to detector blur is difficult to measure directly. Here, we report a simple method to isolate the cross-talk. We analytically modeled the spatial resolution of a CT scanner that uses a Varex flat-panel detector with 0.2 mm pixels and a 450 kVp, dual-focus x-ray tube. Our model examines the contributions of the x-ray focal spot size, detector pixel size, cross-talk, filtering, and backprojection. For the cross-talk model, we measured covariance in air, and from the covariance, we generated a digital filter that approximates the blur due to cross-talk. We validated the model by experimentally measuring spatial resolution with a 0.05 mm-diameter tungsten wire. The wire serves as an impulse function. The image is therefore an impulse response, and its Fourier transform yields the modulation transfer function (MTF). The model and measurement show correspondence within 5% at the 50% point. Our main finding was that detector cross-talk dominates the blurring, dropping the total MTF by over 30 % at the 50% point. Our methods are easily reproducible and do not require sophisticated test-objects, difficult alignment procedures, or image-processing as is common with other measurement methods.

X-ray computed tomography (CT) has been a mainstay of clinical neuroimaging since the 1970s. Its quick scans, high spatial resolution, ease of availability, and relatively low cost make it essential for the diagnosis and monitoring of injury and disease. However, low contrast resolution relative to magnetic resonance imaging (MRI) has limited its use in resolving subtle variations between different soft tissue types and between inflamed and healthy tissues. While clinical CT exploits the attenuation of X-rays through the body, phase contrast X-ray imaging (PCXI) additionally uses refraction and diffraction, increasing sensitivity to subtle inhomogeneities in tissue composition. Most PCXI-CT research has focused on lung and breast tissue, where the sparsely-distributed surrounding bone does not significantly obscure soft tissues. However, neuroimaging with X-rays poses unique challenges, since the brain is fully encased within the highly-attenuating skull, making soft-tissue segmentation difficult and creating artifacts that can obscure the underlying anatomy. We have previously demonstrated that PCXI-CT can resolve anatomical structures within the brains of small animals in situ at micron-scale resolution (Croton et al., 2018) and have developed new correction methods to address the key artifacts that appear when capturing these soft-tissue images (Croton et al., 2019). Here, we present recent progress in the detection of brain injury with PCXI-CT, including direct comparisons to MRI of the same injured brains. Our project aims to spatially resolve injuries that elude MRI, including both traumatic and diffuse injury, in order to better understand the progression of cerebral palsy near birth.

X-ray Phase-Contrast Computed Tomography (PC-CT) increases contrast in weakly attenuating samples, such as soft tissues. In Edge-Illumination (EI) PC-CT, phase effects are accessed from amplitude modulation of the x-ray beam using alternating transmitting and attenuating masks placed prior to the sample and detector. A large field of view PC-CT scanner using this technique was applied to two areas of cancer assessment, namely excised breast and esophageal tissue. For the breast tissue, Wide Local Excisions (WLEs) were studied intra-operatively using PC-CT for the evaluation of tumor removal in breast conservation surgery. Images were acquired in 10 minutes without compromising on image quality, showing this can be used in a clinical setting. Longer, higher resolution PC-CT images were also taken, with analysis showing previously undetected thinning of tumor strands. This would allow a second use of the system for “virtual histopathology”, outside of surgery. For the esophagus samples, tissues were taken from esophagectomy surgery, where the lower part of the esophagus is removed, and the stomach relocated. For the assessment of ongoing therapy, accurate staging of tumors in the removed esophagus is essential, with the current gold standard provided by histopathology. PCCT images were acquired on several samples and compare well with histopathology, with both modalities showing similar features. Examples are shown where staging of tumor penetration is possible with PC-CT images alone, which is hoped will be an important step in performing the imaging and staging intra-operatively.

While conventional computed tomography (CT) is widely used in nondestructive analysis applications, the scanning geometry poses limitations on objects with high aspect ratios, such as printed circuit boards (PCBs), which significantly reduce image quality. Alternatively, computed laminography (CL) techniques are being developed to overcome these challenges by granting additional flexibility of the scan geometry. For this study, the capabilities of a laboratory cabinet x-ray system are extended to perform CL scan acquisitions. Key issues regarding the system geometry are evaluated and calibration techniques are implemented to ensure CL reconstruction accuracy, which is particularly sensitive for high-magnification applications. Results using the CL method are evaluated against the traditional cone-beam CT method which shows improved image quality for the nondestructive inspection of stacked microvias buried within large PCBs. This study also uncovers the particular issues with our system geometry calibration that influence the quality of the reconstructed CL images. Application of these results suggest that cone-beam CL is an effective alternative to conventional CT when inspecting fine features within high aspect ratio objects.

Spondyloarthritis is an inflammatory rheumatic disease characterized by inflammation of the bone and soft tissues, such as enthesis and synovial membrane, in the spine and peripheral joint. Early diagnosis of this disease plays an important role in deciding the possible treatments but has remained a challenge for decades. Several imaging modalities exist, such as X-ray radiography, Magnetic Resonance Imaging (MRI), and Computed Tomography (CT). However, they are not suitable for the early diagnostics of Spondyloarthritis because of their inability to depict active inflammations. This paper demonstrates the application of a novel X-ray phase contrast imaging (PCI) technique that allows imaging of weakly absorbing tissues with better sensitivities than other conventional imaging modalities. A PCI imaging procedure involving 6 affected (SKG induced) and 6 control mice ankles was carried out at the European Synchrotron (ESRF). Results from the high-resolution X-ray phase contrast tomography performed on the mice’s ankles are presented. It was observed that the improved contrast of the soft tissues could allow us to visualize the inflammations in the form of swelling of the synovial membrane. Deep learning techniques are used to segment and evaluate inflammations. The results obtained in this work highlight the ability of the PCI technique to detect inflammations more efficiently than the other conventionally existing modalities.

This work presents an efficient technique for simultaneously reconstructing an image of a moving object from X-ray projections along with estimating the motion parameters. Current state-of-the-art iterative methods rely on objective functions that contain the solution of an iterative procedure as one of its terms. This complicates the reconstruction process as it leads to nested iterations and makes analytic differentiation impractical. The presented technique relies on an objective function that simultaneously depends on the image and the motion parameters. The derivatives of this objective function towards the image and the motion parameters are known exactly and implemented matrix free and in parallel. Moreover, the stepsizes of both the iterative reconstruction and motion parameter estimation schemes are chosen following a mathematical formulation that guarantees a fast convergence speed. The result is a method that not only needs less iterations, but also removes the need to tune the finite differences stepsize and the number of inner iterations, which allows for efficient, scalable reconstruction using modern optimizers, without nested iterations.

In this work we consider a well-known empirical beam-hardening correction algorithm applied to metal artifact reduction in cone beam micro-computed tomography (CT). In its basic form, this algorithm consists of a few simple steps: segmentation of metal components out of uncorrected image data, forward projection to create metal only projection data, and the creation of several correction basis images. The set of basis images can then be combined to create a corrected image with mitigated metal artifacts. The combination weights can be determined either manually or automatically by solving an optimization equation, be performed globally or locally (allowing weights to vary spatially). The latter approach may be more attractive as it may account for larger local variances in scatter artifacts, however, not practical for manual optimization, requiring an automated approach. We apply both global and spatially variant version of the algorithm to datasets from a cone-beam micro-CT (using a 3D X-ray microscope) and study the performance.

Nano-CT enables 3D imaging of micro/nano-structures and is becoming an indispensable tool. At such a high resolution, specimens must have a small diameter small enough to fit into the instrument’s field of view, typically a few tens of micrometers to a few hundred micrometers. As a result, samples are commonly glued to the tip of a steel pin for alignment before imaging. Ideally, data are collected from the part of a specimen above the pin and the x-ray opaque pin will not interfere with imaging. However, the tiny sample size makes precise mounting very tricky, and many times a region adjacent to the pin is found of to be of interest post mounting. Sometimes the sample is too fragile to remount and other times removing the specimen and repeating the tedious remounting steps is impractical due to time constraints, we find that the information occluded by the metal pin can be almost fully recovered via iterative reconstruction with simple metal-trace masked from a regular scan of an imperfectly mounted specimen. Specifically, combining the metal artifact reduction and interior tomography techniques, a metal trace mask in the sinogram is first extracted from a low-resolution global reconstruction which covers the whole cross-section of the pin, then the desired high-resolution reconstruction of a region of interest is iteratively reconstructed excluding any contribution from the metal trace. Our method is demonstrated with a 42.35 nm reconstruction of a portion of a sea urchin tooth, which is scanned on a synchrotron with the pin moving across the field of view during sample rotation, showing that streaking artifacts caused by pin occlusion can be greatly suppressed to achieve an image quality close to that without occlusion. These results suggest that our method has a great potential in simplifying the specimen preparation and relaxing the proficiency requirements, which significantly facilitates nano-CT imaging applications.

Cardiac CT is the first line imaging modality for diagnosis of cardiovascular diseases. A major challenge of cardiac CT remains motion artifacts due to fast and/or irregular cardiac dynamics. The existing motion artifact suppression algorithms can be improved based on distribution shifts due to anatomical and pathological variations in patients, protocol and technical changes of scanners, and other factors. In this paper, we construct a diversified dataset consisting of over 1,000 cardiac CT images of diverse features. Also, we provide a pipeline for source-agnostic vessel segmentation and motion artifact scoring. Our results demonstrate the merits of the approach and suggest a guideline for ensuring source-agnostic representativeness of anatomical and pathological imaging biomarkers in cardiac CT applications and beyond.

Additive manufacturing (AM) is increasingly gaining interest as a low-waste production technique, capable of producing objects using a computer-aided design file. It is particularly interesting for rapid prototyping of parts and manufacturing objects that have complex shapes. However, as in the case of AM through selective laser melting (SLM), manufactured objects may contain defects that can seriously alter their properties. These defects may appear as pores, which can be detected by X-ray computed tomography (X-CT) in a non-destructive manner. CT images can simply be segmented by thresholding or through more advanced techniques such as discrete X-CT reconstruction or machine learning techniques. Nevertheless, these techniques are vulnerable to image reconstruction artefacts. In this work, we evaluate the performance of state-of-the-art, deeply supervised 3D deep learning networks (UNet++, UNet 3+ and UNet-MSS) in terms of segmentation performance of pores from X-ray CT images. The networks have been trained on a real CT dataset, with (noisy) labels produced from both conventional thresholding of the CT images as well as more advanced discrete polychromatic reconstructions. Furthermore, the performance of the networks was evaluated on a test dataset with severe CT artefacts. Pore segmentation from real CT images, which include noise and reconstruction artefacts, revealed that the best performing network was UNet++ with an average Sørensen–Dice score of 0.869 ± 0.006.

Mammalian brains are extremely complex: a mouse brain contains one hundred million neurons. Mapping an entire brain’s three-dimensional cytoarchitecture from the nano- to centimeter-scale is a monumental challenge. For standard microtomography with sub-µm pixel size, reconstructed volume is limited to about a few mm3 . For full brain mapping, even for the mouse brain with volume of 450 mm3 , the field-of-view must be significantly increased in all three dimensions. We demonstrate mosaic tiling to extend imaged volume by 400× and a dedicated pipeline to process these tera-voxel sized datasets. Here, an entire mouse brain was imaged with 0.65 µm-wide voxels. The datasets, which are 6 TB in size at 16-bit depth, contain a wealth of microanatomical information but present challenges for registration and segmentation.

In 1936 in Basel, Johannes Hürzeler isolated the bony labyrinth out of a fossil petrosal bone of the small artiodactyl Cainotherium from the 24 million years old French locality of Branssat (Hürzeler, 1936). To remove it from the bone, he filled the hollow chambers of the bony labyrinth with celluloid and then dissolved the petrosal bone itself to get its endocast. To the best of our knowledge, this is the first cast of an extinct animal inner ear ever produced and published. Here we compare this analog model still preserved in the collection of the Natural History Museum Basel with a 3D virtual model generated by a microtomography scan of another petrosal bone of the same taxon from the same locality. A previous study published the first digital data on the ear region of a cainotheriid but without segmenting the bony labyrinth and another one visualized an inner ear without describing it. Our 3D digital model expands our knowledge of the inner ear anatomy of Cainotherium by showing the so far unclear extent and course of the vestibular and cochlear aqueducts as well as by enabling volumetric data of the different labyrinth parts to be measured.

The famous Swiss mummy, the "Lady of the Franciscan Church", is located in Basel. Subsequent to discovery in 1975, many investigations on her origin were performed without success. Four decades later, however, the mummy’s origin has been identified: Anna Catharina Gernler-Bischoff (1719-1787), who was a direct ancestor of Boris Johnson, the former Prime Minister of the United Kingdom. Further details about her life have been collected. For this purpose, teeth were removed and physically sliced for imaging using optical microscopy. One single tooth remained and became available for this cementochronology study. We have applied microtomography to visualize the annual deposits in tooth cementum, termed incremental layers, without the need of physical slicing. The applicability of synchrotron radiation-based tomography to cementochronology has previously been investigated and showed promise. It is, however, unclear how far tomography will work for the mummy tooth. Laboratory-based microtomography enabled us to discriminate between enamel, dentin, and cementum, but the annual layers remained invisible. The improved density and spatial resolution of the tomography setup at the beamline ANATOMIX, Synchrotron SOLEIL, France, however, brought the incremental layers to light. We have counted 30 layers on average. Their thickness corresponds to (4.6 ± 1.4) µm. The available tomography data of the entire tooth should be further analyzed to correlate the life history of Anna Catharina Gernler-Bischoff with thickness and contrast of the incremental layers. The pipeline developed and the gained knowledge will be beneficial for cementochronology of humans dead and alive.

Caries is detected visually, often supported by palpation and conventional radiographs, and is characterized by reduced Xray absorption. The fringe regions of carious lesions, however, are hardly distinguished. Therefore, the treatment is usually suboptimal, because either a substantial part of healthy crown is removed or the carious lesion is still partially present. This in vitro study applied conventional micro computed tomography to quantify remaining carious tissue after treatment with respect to the volume of tissue mechanically removed. For this purpose, 16 teeth with 23 lesions were treated by an experienced dentist. The teeth were imaged before and after caries removal using a SKYSCAN 1275 system, Bruker, Belgium. Pre- and post-treatment datasets were rigidly registered. Manual segmentation gave rise to the volume of the remaining carious tissue. Twenty of the 23 lesions had a residual carious lesion that amounted to less than 3% of the mechanically erased tissue. Still, residual caries with volumes between 0.8 and 3.4 mm3 was found in three cases. Thus, more detailed microtomography studies are to be performed to give the desired feedback for caries therapy to clinicians.

X-Ray Computed Tomography (CT) has revolutionised modern medical imaging. However, X-Ray CT imaging requires patients to be exposed to radiation, which can increase the risk of cancer. Therefore there exists an aim to reduce radiation doses for CT imaging without sacrificing image accuracy. This research combines phase retrieval with the ShallowU-Net CNN method to achieve the aim. This paper shows that a significant change in existing machine learning neural network algorithms could improve the X-ray phase retrieval in propagationbased phase-contrast imaging. This paper applies deep learning methods, through a variant of the existing U-Net architecture, named ShallowU-Net, to show that it is possible to perform two distance X-ray phase retrieval on composite materials by predicting a portion of the required data. ShallowU-Net is faster in training and in deployment. This method also performs data stretching and pre-processing, to reduce the numerical instability of the U-Net algorithm thereby improving the phase retrieval images.