Hyperspectral X-ray Research Articles

Characterisation of HEXITECMHz - A 1 MHz continuous frame rate spectroscopic X-ray imaging detector system

Spectroscopic X-ray imaging techniques including Compton X-ray Imaging, X-ray Fluorescence Imaging and Hyperspectral X-ray Tomography require energy-resolving detectors capable of operating at high incident X-ray fluxes to make time resolved measurements. HEXITECMHz, operates at a continuous 1 MHz frame rate and can make fully spectroscopic measurements at >106 ph s−1 mm−2. This is enabled by an integrating Front End, in-pixel digitisation and high-speed serialisers. A 300 μm thick p-type Si HEXITECMHz detector was characterised on the B16 Test Beamline at the Diamond Light Source and are the first measurements taken at a 1 MHz frame rate. At 10 keV and 15 keV) the device displayed average FWHM of 656 eV and 682 eV respectively, with minimal changes in spectroscopic performance over ∼8 h. Analysis of charge-sharing events show low charge loss and a linear energy-signal response. Higher-flux measurements illustrated the capability of the ASIC to operate as a photon-counting device.

Quantifying multiple stain distributions in bioimaging by hyperspectral X-ray tomography

Chemical staining of biological specimens is commonly utilised to boost contrast in soft tissue structures, but unambiguous identification of staining location and distribution is difficult without confirmation of the elemental signature, especially for chemicals of similar density contrast. Hyperspectral X-ray computed tomography (XCT) enables the non-destructive identification, segmentation and mapping of elemental composition within a sample. With the availability of hundreds of narrow, high resolution (~ 1 keV) energy channels, the technique allows the simultaneous detection of multiple contrast agents across different tissue structures. Here we describe a hyperspectral imaging routine for distinguishing multiple chemical agents, regardless of contrast similarity. Using a set of elemental calibration phantoms, we perform a first instance of direct stain concentration measurement using spectral absorption edge markers. Applied to a set of double- and triple-stained biological specimens, the study analyses the extent of stain overlap and uptake regions for commonly used contrast markers. An improved understanding of stain concentration as a function of position, and the interaction between multiple stains, would help inform future studies on multi-staining procedures, as well as enable future exploration of heavy metal uptake across medical, agricultural and ecological fields.

Data science enables X-ray vision

In their recent publication in Patterns, the authors proposed a novel workflow to derive compositional and stress maps for positive electrode materials of Li-ion batteries from hyperspectral X-ray imaging data. They describe their interdisciplinary collaboration, the elements that sustain such collaborations, and their effect on the flourishing of the domain and data science.

Multivariate hyperspectral data analytics across length scales to probe compositional, phase, and strain heterogeneities in electrode materials

The origins of performance degradation in batteries can be traced to atomistic phenomena, accumulated at mesoscale dimensions, and compounded up to the level of electrode architectures. Hyperspectral X-ray spectromicroscopy techniques allow for the mapping of compositional variations, and phase separation across length scales with high spatial and energy resolution. We demonstrate the design of workflows combining singular value decomposition, principal-component analysis, k-means clustering, and linear combination fitting, in conjunction with a curated spectral database, to develop high-accuracy quantitative compositional maps of the effective depth of discharge across individual positive electrode particles and ensembles of particles. Using curated reference spectra, accurate and quantitative mapping of inter- and intraparticle compositional heterogeneities, phase separation, and stress gradients is achieved for a canonical phase-transforming positive electrode material, α-V2O5. Phase maps from single-particle measurements are used to reconstruct directional stress profiles showcasing the distinctive insights accessible from a standards-informed application of high-dimensional chemical imaging.

Multispectral Large-Panel X-ray Imaging Enabled by Stacked Metal Halide Scintillators.

Conventional energy-integration black-white X-ray imaging lacks the spectral information of X-ray photons. Although X-ray spectra (energy) can be distinguished by the photon-counting technique typically with CdZnTe detectors, it is very challenging to be applied to large-area flat-panel X-ray imaging (FPXI). Herein, multilayer stacked scintillators of different X-ray absorption capabilities and scintillation spectra are designed; in this scenario, the X-ray energy can be discriminated by detecting the emission spectra of each scintillator; therefore, multispectral X-ray imaging can be easily obtained by color or multispectral visible-light camera in a single shot of X-rays. To verify this idea, stacked multilayer scintillators based on several emerging metal halides are fabricated in a cost-effective and scalable solution process, and proof-of-concept multispectral (or multi-energy) FPXI are experimentally demonstrated. The dual-energy X-ray image of a "bone-muscle" model clearly shows the details that are invisible in conventional energy-integration FPXI. By stacking four layers of specifically designed multilayer scintillators with appropriate thicknesses, a prototype FPXI with four energy channels is realized, proving its extendibility to multispectral or even hyperspectral X-ray imaging. This study provides a facile and effective strategy to realize multispectral large-area flat-panel X-ray imaging.

Comparing AutoEncoder Variants for Real-Time Denoising of Hyperspectral X-Ray

Hyperspectral X-ray analysis is used in many industrial pipelines, from quality control to detection of low-density contaminants in food. Unfortunately, the signal acquired by X-ray sensors is often affected by a great amount of noise. This hinders the performance of most of the applications building on top of these acquisitions (e.g., detection of food contaminants). Therefore, a good denoising pipeline is necessary. This article proposes a comparison between three different AutoEncoder variants: the Variational AutoEncoder, the Augmented AutoEncoder, and a plain vanilla AutoEncoder. All the networks are trained in an unsupervised fashion to denoise a given noisy spectrum. Focusing on the specific application of recognizing possible food contaminants, we force the latent space of the networks to have just two parameters, as suggested by the physical law of Lambert–Beer. We validate our experiments on a synthetic dataset composed of roughly 15 million spectra. Results suggest that the Augmented AutoEncoder is the best network configuration for this task, showing excellent performance without suffering from the nondeterministic behavior of the Variational AutoEncoder.

Hyperspectral X-Ray Imaging: Progress Towards Chemical Analysis in the SEM

The Hyperspectral X-ray Imaging (HXI) project will enable non-destructive chemical-state determination of nano-scale samples in the electron microscope for nuclear safeguards applications. To efficiently measure chemical state through electron-excited X-ray emission requires a combination of wide spectral bandwidth, high resolution, and high count rate capability. We are building a next-generation X-ray detector based on an array of transition edge sensors (TESs) to make these measurements possible and routine in the scanning electron microscope (SEM). Leveraging the large pixel densities afforded by microwave multiplexing readout and continuous, uninterrupted operation of a cryogen-free dilution refrigerator, this instrument will have efficiency to allow chemical species identification of nano-scale samples in hours instead of days to weeks. We describe prototype pixel designs for this HXI instrument, comprising three pixel types that will make up the hybrid TES array. Engineering design of the integrated HXI cryostat and SEM system is in progress, with full detector commissioning expected in Spring 2021 followed closely by full-scale integration with the SEM. We also report on the commissioning of a complementary TES-based X-ray emission spectroscopy platform for bulk samples to build a spectral library for HXI sample identification.

An edge-on energy-resolved X-ray semiconductor detector

The hyperspectral X-ray imaging has long been sought in various fields from material analysis to medical diagnosis. Here we propose a semiconductor detector structure to realize multi-energy imaging at potentially low cost. The device is designed based on the dramatically energy-dependent attenuation of X-ray in solids. An array or a two-/three-dimensional matrix of semiconductor cells is used to map the X-ray intensity along its trajectory. The X-ray spectrum could be extracted from a Laplace like transform or a regression via supervised machine learning. We conceptually demonstrated an energy-resolved X-ray detection with a regular silicon CMOS camera and a dummy sample of copper wires on top of an aluminum flake. The experiment provides energy-resolved X-ray image of the sample. The demonstration shows an energy resolution of less than 10 keV. We concluded that the proposed edge-on structure is potentially useful as energy-resolved high efficiency low-cost X-ray detectors.

Probelab ReImager: An Open-Source Software for Streamlining Image Processing in an Electron Microscopy Laboratory

Providing high-quality electron images and hyperspectral X-ray maps is a focus of many modern electron microscopy laboratories. Nevertheless, further image processing and annotations are often needed to prepare them for publications and reports. For multi-user facilities, accessibility to processing software can be a limitation either through license costs or availability of processing stations. Open-source software running on multiple platforms allows for post-acquisition data processing in-lab or on user-owned devices. We developed Probelab ReImager to supersede our vendor-supplied acquisition software's exportation by being efficient and highly customizable. This article describes its main features and capabilities.

The Spectral X-ray Imaging Data Acquisition (SpeXIDAQ) Framework.

Photon counting X-ray imagers have found their way into the mainstream scientific community in recent years, and have become important components in many scientific setups. These camera systems are in active development, with output data rates increasing significantly with every new generation of devices. A different class of PCD (Photon Counting Detector) devices has become generally available, where camera data output is no longer a matrix of photon counts but instead direct measurements of the deposited charge per pixel in every frame, which requires significant off-camera processing. This type of PCD, called a hyperspectral X-ray camera due to its fully spectroscopic output, yet again increases the demands put on the acquisition and processing backend. Not only are bandwidth requirements increased, but the need to do extensive data processing is also introduced with these hyperspectral PCD devices. To cope with these new developments the Spectral X-ray Imaging Data Acquisition framework (SpeXIDAQ) has been developed. All aspects of the imaging pipeline are handled by the SpeXIDAQ framework: from detector control and frame grabbing, to processing, storage and live visualisation during experiments.

Erratum: “Mouse livers machine learning identification based on hyperspectral x-ray computed tomography reconstructed x-ray absorption spectra” [AIP Advances 10, 115009 (2020)

Erratum: “Mouse livers machine learning identification based on hyperspectral x-ray computed tomography reconstructed x-ray absorption spectra” [AIP Advances 10, 115009 (2020)

Microchemical analysis of Leonardo da Vinci\u2019s lead white paints reveals knowledge and control over pigment scattering properties

Leonardo da Vinci (1452–1519) is a key artistic and scientific figure of the Renaissance. He is renowned for his science of art, taking advantage of his acute observations of nature to achieve striking pictorial results. This study describes the analysis of an exceptional sample from one of Leonardo’s final masterpieces: The Virgin and Child with St. Anne (Musée du Louvre, Paris, France). The sample was analyzed at the microscale by synchrotron-based hyperspectral photoluminescence imaging and high-angular X-ray diffraction. The results demonstrate Leonardo’s use of two subtypes of lead white pigment, thus revealing how he must have possessed a precise knowledge of his materials; carefully selecting them according to the aesthetical results he aimed at achieving in each painting. This work provides insights on how Leonardo obtained these grades of pigment and proposes new clues regarding the optical and/or working properties he may have tried to achieve.

Mouse livers machine learning identification based on hyperspectral x-ray computed tomography reconstructed x-ray absorption spectra

X-ray computed tomography (X-CT) is often used to examine organs, but the reconstructed images can only be used for structural identification. Whether the organs are healthy or not requires a professional doctor to examine the reconstructed image and judge from his or her own experience. The purpose of this paper is to identify the cirrhotic mouse liver and normal mouse liver with hyperspectral x-ray CT (HXCT) and machine learning. HXCT is proposed to reconstruct the x-ray absorption spectrum (XAS) characteristics of a single pixel in the reconstructed mouse liver images. HXCT uses a cadmium telluride photon counter as the x-ray detector, which can improve the spectral resolution and separate spectral lines. Filtered back-projection and algebra reconstruction technique reconstruction algorithms are used for image and XAS reconstruction. In the machine learning model, principal component analysis is utilized to reduce the dimensionality of XAS. Besides, the neural network algorithm Artificial Neural Network (ANN) is used to train and identify the reconstructed XAS of two different kinds of livers. These two different mouse livers can be well recognized since the accuracy goes to almost 100% based on ANN. It is feasible to employ the machine learning algorithm to identify the XAS of different mouse livers.

CdZnTe-Based X-Ray Spectrometer for Absolute Density Determination

Density is a very important property of materials because of its influence on their mechanical properties. For this reason, the possibility of measuring material density is of strategic importance for several industrial applications such as wood boards sawing, wood panels production, or ceramic tiles manufacturing. In order to answer to this need, X-ray densitometry techniques and computed tomography (CT) scanners are used nowadays. However, these traditional X-ray inspection techniques usually measure only relative density, or, in other terms, they need a calibration procedure before operation. To overcome this issue, Due2lab s.r.l. in collaboration with the Istituto dei Materiali per l’Elettronica ed il Magnetismo (IMEM)–Consiglio Nazionale delle Ricerche (CNR) Institute developed a CdZnTe (CZT)-based X-ray spectrometer capable of determining the absolute density value of a sample material in terms of g/cm3 (and additionally extracting information about its elemental composition). The system has been specifically designed for in-line real-time nondestructive inspection of industrial production; thus, it is potentially capable to return density information within a very short acquisition time. In this article, we will describe the CZT spectrometer and the hyperspectral X-ray analysis set-up in detail and we will show how to perform absolute density measurement on three different materials: 1) a plastic polyethylene sample, 2) ceramic tiles, and 3) pine and fir wood samples.

Effect of Reconstruction Algorithm on the Identification of 3D Printing Polymers Based on Hyperspectral CT Technology Combined with Artificial Neural Network.

Hyperspectral X-ray CT (HXCT) technology provides not only structural imaging but also the information of material components therein. The main purpose of this study is to investigate the effect of various reconstruction algorithms on reconstructed X-ray absorption spectra (XAS) of components shown in the CT image by means of HXCT. In this paper, taking 3D printing polymer as an example, seven kinds of commonly used polymers such as thermoplastic elastomer (TPE), carbon fiber reinforced polyamide (PA-CF), acrylonitrile butadiene styrene (ABS), polylactic acid (PLA), ultraviolet photosensitive resin (UV9400), polyethylene terephthalate glycol (PETG), and polyvinyl alcohol (PVA) were selected as samples for hyperspectral CT reconstruction experiments. Seven kinds of 3D printing polymer and two interfering samples were divided into a training set and test sets. First, structural images of specimens were reconstructed by Filtered Back-Projection (FBP), Algebra Reconstruction Technique (ART) and Maximum-Likelihood Expectation-Maximization (ML-EM). Secondly, reconstructed XAS were extracted from the pixels of region of interest (ROI) compartmentalized in the images. Thirdly, the results of principal component analysis (PCA) demonstrated that the first four principal components contain the main features of reconstructed XAS, so we adopted Artificial Neural Network (ANN) trained by the reconstructed XAS expressed by the first four principal components in the training set to identify that the XAS of corresponding polymers exist in both of test sets from the training set. The result of ANN displays that FBP has the best performance of classification, whose ten-fold cross-validation accuracy reached 99%. It suggests that hyperspectral CT reconstruction is a promising way of getting image features and material features at the same time, which can be used in medical imaging and nondestructive testing.

Deciphering the Complex Mineralogy of River Sand Deposits through Clustering and Quantification of Hyperspectral X-Ray Maps.

Alluvial mineral sands rank among the most complex subjects for mineral characterization due to the diverse range of minerals present in the sediments, which may collectively contain a daunting number of elements (>20) in major or minor concentrations (>1 wt%). To comprehensively characterize the phase abundance and chemistry of these complex mineral specimens, a method was developed using hyperspectral x-ray and cathodoluminescence mapping in an electron probe microanalyser (EPMA), coupled with automated cluster analysis and quantitative analysis of clustered x-ray spectra. This method proved successful in identifying and quantifying over 40 phases from mineral sand specimens, including unexpected phases with low modal abundance (<0.1%). The standard-based quantification method measured compositions in agreement with expected stoichiometry, with elemental detection limits in the range of <10–1,000 ppm, depending on phase abundance, and proved reliable even for challenging mineral species, such as the multi-rare earth element (REE) bearing mineral xenotime [(Y,REE)PO4] for which 24 elements were analyzed, including 12 overlapped REEs. The mineral identification procedure was also capable of characterizing mineral groups that exhibit significant compositional variability due to the substitution of multiple elements, such as garnets (Mg, Ca, Fe, Mn, Cr), pyroxenes (Mg, Ca, Fe), and amphiboles (Na, Mg, Ca, Fe, Al).

Hyperspectral X-ray Imaging with TES Detectors for Nanoscale Chemical Speciation Mapping

We are developing an imaging capability (“Hyperspectral X-ray Imaging”) for mapping chemical information (molecular formula, phase, oxidation state, hydration) that is based on ultra-high-resolution X-ray emission spectroscopy with large transition-edge sensor microcalorimeter arrays in the scanning electron microscope. By combining microcalorimeter arrays with hundreds of pixels, high-bandwidth microwave frequency-division multiplexing, and fast digital electronics for near real-time data processing, our goal is to enable measurements using laboratory-scale instrumentation rather than synchrotron beamlines. Our application focus here is on mapping the chemical form of uranium compounds on the nanoscale. We will present our approach to developing the Hyperspectral X-ray Imaging capability, progress toward a 128-pixel microwave multiplexed X-ray fluorescence instrument at LANL, and the path to high-throughput nanoscale chemical mapping.

Full-field spectroscopic measurement of the X-ray beam from a multilayer monochromator using a hyperspectral X-ray camera.

Multilayer monochromator devices are commonly used at (imaging) beamlines of synchrotron facilities to shape the X-ray beam to relatively small bandwidth and high intensity. However, stripe artefacts are often observed and can deteriorate the image quality. Although the intensity distribution of these artefacts has been described in the literature, their spectral distribution is currently unknown. To assess the spatio-spectral properties of the monochromated X-ray beam, the direct beam has been measured for the first time using a hyperspectral X-ray detector. The results show a large number of spectral features with different spatial distributions for a [Ru, B4C] strip monochromator, associated primarily with the higher-order harmonics of the undulator and monochromator. It is found that their relative contributions are sufficiently low to avoid an influence on the imaging data. The [V, B4C] strip suppresses these high-order harmonics even more than the former, yet at the cost of reduced efficiency.

Application of hyperspectral CT technology combined with machine learning in recognition of plastic components

A new method based on a hyperspectral X-ray CT (HXCT) system is proposed in this paper to identify the material composition in a region of interest (ROI) during nondestructive testing (NDT). The HXCT system (based on the principle of photon counting) was used to scan the samples containing 10 kinds of plastic components, and the Algebraic Reconstruction Technique (ART) was applied to compute the tomography images. The X-ray absorption spectra (XAS) corresponding to the 360 pixel points in the tomography images were reconstructed. After wavelet denoising and principal component analysis (PCA) preprocessing, machine learning algorithms named Artificial Neural Network (ANN) and Support Vector Machine (SVM) were used to construct the model for classification. The highest ten-fold cross-validation accuracy of material identification in ROI reached 80.00%. It is demonstrated that NDT using hyperspectral CT technology based on the photon counting principle can not only perform structural analysis and defect detection, but also identify the components in the ROI, which has important research significance in radiology research.

Implementation of a Hyperspectral X-ray Camera control and processing software chain

Hyperspectral X-ray detectors, which provide the full energy spectrum detected by each individual pixel, have become available for use in lab-based facilities. The addition of spectral information currently comes at the cost of low overall acceptable flux rates, and can introduce countrate nonlinearity at higher energies. Neither of these drawbacks are desirable for transmission imaging and tomography. In this paper a new data processing software chain is presented for the SLcam, a pnCCD-based hyperspectral camera prototype, allowing for more control over the conversion from raw frames to hyperspectral images. Complementary to the processing software, a set of live data visualisations was developed to aid in monitoring ongoing experiments and to allow for preliminary data processing on-the-fly. The combination of these software elements forms the first step towards general applicability of hyperspectral imaging at laboratory tomography setups.