Tomography of Materials and Structures最新文献

Mineral dissolution is an important process that occurs in both natural as well as anthropogenic processes. The kinetics of such dissolution processes are influenced not only by the characteristics of the solution but also by the characteristics of the minerals, such as crystal defects on the microscopic scale or macroscopic features such as the intersection of crystal planes to form edges and corners. Macroscopic features are known to increase the population of steps and kinks that may, in turn, affect the dissolution rate over time. Hence, this study presents a 3D empirical dissolution model aimed at examining the time-series evolution of macroscopic features together with the corresponding changes in the dissolution rate under far from equilibrium batch reactor conditions. The developed empirical model is based on the mineral geometry (surface topography and volume) derived from X-ray computed tomography (CT) measurements. The macroscopic features are identified using surface curvature which are then used to generate reactivity maps for dissolution model. As a study case, the dissolution of monomineralic galena (PbS) in ethaline and iodine as oxidizing agent is experimentally observed and then modelled. The model is then applied to seven particles of various shapes and sizes. The finding suggests that the surface reactivity increases over time as the particle shrinks and the macroscale steps and edges become dominant over the initial terraces. This implies that the persistent highly reactive surface sites defined by a particle’s geometry may play a dominant role in the overall particle dissolution in addition to the dissolution mechanisms typically studied on near atomic-flat surfaces. The model developed in this investigation offers the opportunity to be extended providing the possibility of simulating the dissolution of multi-mineral particles during batch dissolution experiments.

Dual-beam x-ray tomography systems are paving the way for new experimental procedures, such as multi-resolution and multi-energy imaging, where synchronous acquisitions are essential. However, in such systems, cross-detector scatter between the detecting devices can occur as the two beamlines operate simultaneously. This paper proposes a new affine image transformation model of each projection to correct for these cross-detector scatter issues. A toy tomography test case is presented to assess the feasibility and performance of the proposed correction method.

Cardiovascular tissues possess a complex microstructure, which remodel and adapt due to ageing and diseases. This complex and evolving microstructure is intrinsically linked to the tissue’s mechanical properties. To better understand how changes in the microstructure can impact the mechanical behavior, 4D-contrast-enhanced microCT (4D-CECT) can be used (i.e. in situ mechanical loading combined with 3D microstructural visualization). Since absorption-based CECT requires the use of contrast-enhancing staining agents (CESAs), we investigated six different CESAs for their suitability for 4D-CECT imaging of arterial tissue, considering their ability to provide good microstructural visualization and segmentation while ensuring the preservation of the mechanical properties. For this purpose, the penetration speed, contrast-enhancement, volume change, and stiffness change of porcine arterial tissue stained with the different CESA solutions were studied. Based on our results, we selected 1:2 Hafnium-substituted Wells-Dawson Polyoxometalate as the most suited CESA for 4D-CECT of arterial tissue. Phosphotungstic acid (PTA) and Lugol iodine with Sorensen’s buffer (Lugol), despite being the reference in the state-of-the-art as CESA and having excellent contrast-enhancement properties, were the only ones that significantly affected the mechanical properties of porcine arterial tissue. Additionally, for these two solutions, tissue shrinkage was observed, resulting in a volume reduction of approximately − 12 % for PTA and − 17 % for Lugol. Finally, it was observed that the penetration speed of all CESA solutions exhibited a ratio of 60–40 % from the intimal side to the adventitial side, which is likely due to the denser packing of elastic lamellae towards the adventitia. Overall, our study offers valuable new insights for selecting and comparing various CESA solutions for (4D-)CECT.

Beam tracking and edge illumination are phase contrast imaging techniques that rely on amplitude modulated x-ray beams to generate sensitivity to refraction and scattering. While each technique has its advantage (“single shot” three-contrast imaging in beam tracking; the ability to work with relatively large pixels in edge illumination), they also share a common drawback, namely that the modulator shields parts of the sample and, thus, prevents those areas from contributing to the image (under-sampling). Sample stepping, by which frames are acquired with the sample in a different position relative to the modulator (sometimes referred to as “dithering”) can produce well-sampled images. However, in computed tomography (CT), stepping must be performed at each rotation angle, enforcing step-and-shoot acquisitions and leading to long scan times. To enable faster acquisitions, fly scan compatible scanning schemes based on “roto-translating” the sample in the modulated x-ray beam were recently developed. This article reviews these schemes and provides practical guidance for their implementation.

pt4, an open-source software tool to describe time-evolving phantoms is presented. pt4 allows users to create detailed time-evolving phantoms for testing novel 4D-CT reconstruction algorithms. Ground-truth volumes and simulated X-ray projections can be produced at arbitrary time-points during time-evolution and with customisable pixel dimensions, noise models, and X-ray source trajectories. Phantoms are built up from 3D primitives whose parameters and attenuation can be made arbitrary functions of time. This feature permits both complex continuous and discontinuous time-evolution necessary for thorough testing of 4D-CT reconstruction algorithms. Various phantoms built using pt4 are also presented to demonstrate the versatility of pt4 phantom description. pt4 is written in C++ and is highly parallelised leading to a performant implementation which is feasible to use for up to thousand-of-voxel volume pixel dimensions.

Image segmentation with deep learning models has significantly improved the accuracy of the pixel-wise labeling of scientific imaging which is critical for many quantitative image analyses. This has been feasible through U-Net and related architecture convolutional neural network models. Although the adoption of these models has been widespread, their training data pool and hyperparameters have been mostly determined by educated guesses through trial and error. In this study, we present observations of how training data volume, data augmentation, and patch size affect deep learning performance within a limited data set. Here we study U-Net model training on four different samples of x-ray CT images of fiber-reinforced composites. Because the training process is not deterministic, we relied on seven-fold replication of each experimental condition to avoid under-sampling and observe model training variance. Unsurprisingly, we find greater training data volume strongly benefits individual models’ final accuracy and learning speed while depressing variance among replicates. Importantly, data augmentation has a profound benefit to model performance, especially in cases with a low abundance of ground truth, and we conclude that high coefficients of data augmentation should be used in scientific imaging semantic segmentation models. Future work to describe and measure image complexity is warranted and likely to ultimately guide researchers on the minimum required training data volume for particular scientific imaging deep learning tasks.

Splits are delaminations that may appear perpendicular to the crack plane during fracture toughness tests of certain materials, such as hot-rolled metal alloys. X-ray computed tomography (CT) was used to conduct a 3D analysis of the geometrical and morphological characteristics of the splits in SE(B) specimens machined from a DH36 steel. Tomograms and 3D reconstructions of the CT results were compared with high-resolution images obtained through optical microscopy (OM) and scanning electron microscopy (SEM). Quantitative and qualitative comparisons revealed a good agreement between the results, validating the split characterization by CT. It was discussed whether characterizing the splits just by the routinary fracture surface observation conducted in fracture mechanics specimens can hide important phenomena such as plane changes, branching, and interactions between delaminations. On the other hand, CT enables an accurate and comprehensive characterization of the morphological and geometrical attributes of splits. Contrasts between the analysis and characteristics of deformed and undeformed splits were made. Finally, the limitations and challenges of the 3D split characterization by CT were also discussed, exploring experimental and image processing issues. These findings emphasize that a more thorough understanding of the internal structure of splits can be achieved by applying CT analysis, contrasting with traditional fracture surface examination. This study highlights the relevance of CT in revealing hidden complexities within the internal structure of specimens with splits.

This paper discusses the development of a Dual Beam microfocus high-energy X-ray tomography system for laboratory experiments, aiming to enhance temporal resolution and multimodal capabilities. Initially, X-ray computed tomography (XRCT) in materials science, particularly using synchrotron sources, provided valuable insights into microstructures. Digital volume correlation (DVC) emerged as a tool for measuring displacement fields during in situ XRCT tests. High-speed XRCT became possible with synchrotrons, but laboratory devices still face limitations due to moderate X-ray flux. This paper describes the design and implementation of a new dual high-energy X-ray tomograph with two twin beamlines. The paper also covers the first in situ dual-beam experiment involving the in situ compression test of an aluminium foam sample. It discusses calculating DVC displacement fields from radiographs, comparing them to control tomographic scans, and assessing their quality. The paper explores the potential for deforming initial scans using DVC fields, both from radiographs and tomographic scans. The approach shows reasonable quantitative agreement with control scans but does not capture rotational motion along the vertical axis.

Synchrotron-based X-ray microtomography (μCT) is a valuable technique to study the internal structure of heterogeneous samples with high spatial and temporal resolution. However, synchrotron X-ray imaging, such as many microscopy methods, is solidly limited by its Field of View (FOV): a challenge when approaching large and/or highly detailed volumes at high spatial imaging resolution. In this study, we consider two techniques used to increase the FOV in µCT for studying Li-ion batteries, Local Tomography Stitching (LTS) and Sinogram Oriented Stitching (SOS), and compare in terms of scan time, processing efficiency and computing storage. We complement our study by estimating the impact of different battery geometries on the stitching performance for similar scanning parameters. Evaluation of the two presented techniques reveals that both provide equally satisfyingly stitched volumes. Nevertheless, it is demonstrated that SOS is predominantly more efficient for the considered battery geometries, requires fewer scans in total, and hence has a lower time and storage demand in comparison to LTS. However, technical constraints make the SOS technique more difficult to implement. Further discussed are differences in acquisition and reconstruction for the two techniques, addressing the processing efficiency for both SOS and LTS techniques and we shall provide indicators for selecting the most suitable stitching technique.

Our ability to visualize and quantify the internal structures of objects via computed tomography (CT) has fundamentally transformed science. As tomographic tools have become more broadly accessible, researchers across diverse disciplines have embraced the ability to investigate the 3D structure-function relationships of an enormous array of items. Whether studying organismal biology, animal models for human health, iterative manufacturing techniques, experimental medical devices, engineering structures, geological and planetary samples, prehistoric artifacts, or fossilized organisms, computed tomography has led to extensive methodological and basic sciences advances and is now a core element in science, technology, engineering, and mathematics (STEM) research and outreach toolkits. Tomorrow's scientific progress is built upon today's innovations. In our data-rich world, this requires access not only to publications but also to supporting data. Reliance on proprietary technologies, combined with the varied objectives of diverse research groups, has resulted in a fragmented tomography-imaging landscape, one that is functional at the individual lab level yet lacks the standardization needed to support efficient and equitable exchange and reuse of data. Developing standards and pipelines for the creation of new and future data, which can also be applied to existing datasets is a challenge that becomes increasingly difficult as the amount and diversity of legacy data grows. Global networks of CT users have proved an effective approach to addressing this kind of multifaceted challenge across a range of fields. Here we describe ongoing efforts to address barriers to recently proposed FAIR (Findability, Accessibility, Interoperability, Reuse) and open science principles by assembling interested parties from research and education communities, industry, publishers, and data repositories to approach these issues jointly in a focused, efficient, and practical way. By outlining the benefits of networks, generally, and drawing on examples from efforts by the Non-Clinical Tomography Users Research Network (NoCTURN), specifically, we illustrate how standardization of data and metadata for reuse can foster interdisciplinary collaborations and create new opportunities for future-looking, large-scale data initiatives.