Neural network for 3D inertial confinement fusion shell reconstruction from single radiographs.

In inertial confinement fusion (ICF), x-ray radiography is a critical diagnostic for measuring implosion dynamics, which contain rich three-dimensional (3D) information. Traditional methods for reconstructing 3D volumes from 2D radiographs, such as filtered backprojection, require radiographs from at least two different angles or lines of sight (LOS). In ICF experiments, the space for diagnostics is limited, and cameras that can operate on fast timescales are expensive to implement, limiting the number of projections that can be acquired. To improve the imaging quality as a result of this limitation, convolutional neural networks (CNNs) have recently been shown to be capable of producing 3D models from visible light images or medical x-ray images rendered by volumetric computed tomography. We propose a CNN to reconstruct 3D ICF spherical shells from single radiographs. We also examine the sensitivity of the 3D reconstruction to different illumination models using preprocessing techniques such as pseudo-flatfielding. To resolve the issue of the lack of 3D supervision, we show that training the CNN utilizing synthetic radiographs produced by known simulation methods allows for reconstruction of experimental data as long as the experimental data are similar to the synthetic data. We also show that the CNN allows for 3D reconstruction of shells that possess low mode asymmetries. Further comparisons of the 3D reconstructions with direct multiple LOS measurements are justified.

Microcomputed X-Ray Tomographic Imaging and Image Processing for Microstructural Characterization of Explosives.

Microstructural characterization of composite high explosives (HEs) has become increasingly important over the last several decades in association with the development of high fidelity mesoscale modeling and an improved understanding of ignition and detonation processes. HE microstructure influences not only typical material properties (e.g., thermal, mechanical) but also reactive behavior (e.g., shock sensitivity, detonation wave shape). A detailed nondestructive 3D examination of the microstructure has generally been limited to custom-engineered samples or surrogates due to poor contrast between the composite constituents. Highly loaded (>90 wt%) HE composites such as plastic-bonded explosives (PBX) are especially difficult. Here, we present efforts to improve measurement quality by using single and dual-energy microcomputed X-ray tomography and state-of-the-art image processing techniques to study a broad set of HE materials. Some materials, such as PBX 9502, exhibit suitable contrast and resolution for an automatic segmentation of the HE from the polymer binder and the voids. Other composite HEs had varying levels of success in segmentation. Post-processing techniques that used commercially available algorithms to improve the segmentation quality of PBX 9501 as well as zero-density defects such as cracks and voids could be easily segmented for all samples. Aspects of the materials that lend themselves well to this type of measurement are discussed.

Effect of a dual enteric and respiratory pathogen challenge on swine growth, efficiency, carcass composition, and pork quality1.

The objective of this study was to determine the influence of a dual respiratory and enteric pathogen challenge on growth performance, carcass composition, and pork quality of high and low feed efficient pigs. Pigs divergently selected for low and high residual feed intake (RFI, ~68 kg) from the 11th generation of Iowa State University RFI project were used to represent high and low feed efficiency. To elicit a dual pathogen challenge, half of the pigs (n = 12/line) were inoculated with Mycoplasma hyopneumoniae (Mh) and Lawsonia intracellularis (MhLI) on days post-inoculation (dpi) 0. Pigs in a separate room of the barn were not inoculated and used as controls (n = 12/RFI line). Pigs were weighed and feed intake was recorded to calculate ADG, ADFI, and G:F for the acclimation period (period 1: dpi -21 to 0), during peak infection (period 2: dpi 0 to 42), and during the remaining growth period to reach market weight (period 3: dpi 42 to harvest). At ~125 kg, pigs were harvested using standard commercial procedures. Carcasses were evaluated for composition (weight, fat free lean, loin eye area, 10th rib fat depth) and meat quality (pH decline, temperature decline, Hunter L, a, and b, subjective color and marbling, star probe, drip loss, cook loss, proximate composition, and desmin degradation). Challenged pigs had lesser ADFI than controls during period 2 (P < 0.05), but had greater ADG and G:F during period 3 (P < 0.05). Selection for feed efficiency did not result in a differential response to MhLI (P > 0.05). Loin chops from the less feed efficient, high RFI pigs, had greater drip loss, greater cook loss, lesser moisture content, greater Hunter L values, and greater Hunter b values (P < 0.05) than loin chops from low RFI pigs. Infection status did not significantly affect carcass composition or pork quality traits (P > 0.05). These results indicate that a MhLI challenge early in growth did not significantly affect ultimate carcass composition or meat quality traits. Selection for greater feed efficiency in pigs did not affect their response to pathogenic challenge.

In Situ Imaging during Compression of Plastic Bonded Explosives for Damage Modeling.

The microstructure of plastic bonded explosives (PBXs) is known to influence behavior during mechanical deformation, but characterizing the microstructure can be challenging. For example, the explosive crystals and binder in formulations such as PBX 9501 do not have sufficient X-ray contrast to obtain three-dimensional data by in situ, absorption contrast imaging. To address this difficulty, we have formulated a series of PBXs using octahydro-1,3,5,7-tetranitro-1,3,5,7-tetrazocine (HMX) crystals and low-density binder systems. The binders were hydroxyl-terminated polybutadiene (HTPB) or glycidyl azide polymer (GAP) cured with a commercial blend of acrylic monomers/oligomers. The binder density is approximately half of the HMX, allowing for excellent contrast using in situ X-ray computed tomography (CT) imaging. The samples were imaged during unaxial compression using micro-scale CT in an interrupted in situ modality. The rigidity of the binder was observed to significantly influence fracture, crystal-binder delamination, and flow. Additionally, 2D slices from the segmented 3D images were meshed for finite element simulation of the mesoscale response. At low stiffness, the binder and crystal do not delaminate and the crystals move with the material flow; at high stiffness, marked delamination is noted between the crystals and the binder, leading to very different mechanical properties. Initial model results exhibit qualitatively similar delamination.

Morphological and performance measures of polyurethane foams using X-ray CT and mechanical testing.

Meso-scale structure in polymeric foams determines the mechanical properties of the material. Density variations, even more than variations in the anisotropic void structure, can greatly vary the compressive and tensile response of the material. With their diverse use as both a structural material and space filler, polyurethane (PU) foams are widely studied. In this manuscript, quantitative measures of the density and anisotropic structure are provided by using micro X-ray computed tomography (microCT) to better understand the results of mechanical testing. MicroCT illustrates the variation in the density, cell morphology, size, shape, and orientation in different regions in blown foam due to the velocity profile near the casting surface. "Interrupted" in situ imaging of the material during compression of these sub-regions indicates the pathways of the structural response to the mechanical load and the changes in cell morphology as a result. It is found that molded PU foam has a 6 mm thick "skin" of higher density and highly eccentric morphological structure that leads to wide variations in mechanical performance depending upon sampling location. This comparison is necessary to understand the mechanical performance of the anisotropic structure.

Dimensional quantification of embedded voids or objects in three dimensions using X-ray tomography.

Scientific digital imaging in three dimensions such as when using X-ray computed tomography offers a variety of ways to obtain, filter, and quantify data that can produce vastly different results. These opportunities, performed during image acquisition or during the data processing, can include filtering, cropping, and setting thresholds. Quantifying features in these images can be greatly affected by how the above operations are performed. For example, during binarization, setting the threshold too low or too high can change the number of objects as well as their measured diameter. Here, two facets of three-dimensional quantification are explored. The first will focus on investigating the question of how many voxels are needed within an object to have accurate geometric statistics that are due to the properties of the object and not an artifact of too few voxels. These statistics include but are not limited to percent of total volume, volume of the individual object, Feret shape, and surface area. Using simple cylinders as a starting point, various techniques for smoothing, filtering, and other processing steps can be investigated to aid in determining if they are appropriate for a specific desired statistic for a real dataset. The second area of investigation is the influence of post-processing, particularly segmentation, on measuring the damage statistics in high purity Cu. The most important parts of the pathways of processing are highlighted.

Dimensional standard for micro X-ray computed tomography.

The decrease in the cost of high end computing and the availability of high quality X-ray sources in the laboratory environment has led to an increased use of three-dimensional (3D) X-ray micro computed tomography (µCT). In the medical community, the primary concern for CT is calibrating for X-ray absorption and ascertaining the difference between healthy tissue and cancerous tissue or examining fractures. Absorption calibration is also important in the materials community, however confirming dimensional accuracy of voids, defects, machined parts, cracks, or the distribution of dispersed particles is typically more important. One key aspect of µCT that is often overlooked in the literature is the number of radiographs required for dimensional accuracy of the 3D reconstruction and minimization of image noise. In µCT, a number of radiographs are collected in theta increments as the sample is rotated at least 180°. They are typically collected in 1° increments (or 181 radiographs), 0.25° increments (721 radiographs), or some other multiple. The question that arises, especially in a laboratory based instrument, where the required exposure times are longer to get high-quality signal-to-noise compared to synchrotron sources, is what is the optimal number of images required to reach the volumetric statistics of the sample, and minimize the noise while not overly scanning the sample at a cost in time? A dimensional standard based upon NIST certified glass microspheres dispersed in a low density poly(styrene) matrix to answer this question is proposed. Experiments are shown that describe the microsphere size statistics as a function of number of radiographs calculated using a commercial software package, AvizoFire. These results are important in understanding the distribution of voids in a foam and confirming the accuracy of the 3D measurements obtained.

Infrared microspectroscopic imaging using a large radius germanium internal reflection element and a focal plane array detector.

Previously, we established the ability to collect infrared microspectroscopic images of large areas using a large radius hemisphere internal reflection element (IRE) with both a single point and a linear array detector. In this paper, preliminary work in applying this same method to a focal plane array (FPA) infrared imaging system is demonstrated. Mosaic tile imaging using a large radius germanium hemispherical IRE on a FPA Fourier transform infrared microscope imaging system can be used to image samples nearly 1.5 mm x 2 mm in size. A polymer film with a metal mask is imaged using this method for comparison to previous work. Images of hair and skin samples are presented, highlighting the complexity of this method. Comparisons are made between the linear array and FPA methods.

Attenuated total internal reflection infrared microspectroscopic imaging using a large-radius germanium internal reflection element and a linear array detector.

The number of techniques and instruments available for Fourier transform infrared (FT-IR) microspectroscopic imaging has grown significantly over the past few years. Attenuated total internal reflectance (ATR) FT-IR microspectroscopy reduces sample preparation time and has simplified the analysis of many difficult samples. FT-IR imaging has become a powerful analytical tool using either a focal plane array or a linear array detector, especially when coupled with a chemometric analysis package. The field of view of the ATR-IR microspectroscopic imaging area can be greatly increased from 300 x 300 microm to 2500 x 2500 microm using a larger internal reflection element of 12.5 mm radius instead of the typical 1.5 mm radius. This gives an area increase of 70x before aberrant effects become too great. Parameters evaluated include the change in penetration depth as a function of beam displacement, measurements of the active area, magnification factor, and change in spatial resolution over the imaging area. Drawbacks such as large file size will also be discussed. This technique has been successfully applied to the FT-IR imaging of polydimethylsiloxane foam cross-sections, latent human fingerprints, and a model inorganic mixture, which demonstrates the usefulness of the method for pharmaceuticals.

Elemental and molecular characterization of aged polydimethylsiloxane foams.

The application and integration of micro X-ray fluorescence (MXRF) and Fourier transform infrared (FT-IR) imaging to polydimethylsiloxane (PDMS) foam aging issues have been applied to cross-sectional images. Previous work has shown the tin in the stannous 2-ethylhexanoate catalyst to be highly mobile and it typically migrates to the PDMS foam upper surface. The current paper discusses a method for the integration of full spectral MXRF and FT-IR imaging of aged foams. Solvent extractions have also been performed on both fresh and aged foams to further examine aged foam properties. Combining elemental and molecular imaging techniques and applying them to PDMS aging provides synergistic information that aids in understanding the sample composition and distribution of components. Application of chemometric analysis to the full spectral elemental and molecular maps demonstrates correlations within the foams of the residual tin, organo-tin functional group moieties, and the presence of nitroplasticizer from an exogenous source.

Integrating X-ray fluorescence and infrared imaging microspectroscopies for comprehensive characterization of an acetaminophen model pharmaceutical.

The integration of full spectral images using the complementary microspectroscopic imaging techniques X-ray fluorescence and Fourier transform infrared is demonstrated. This effort surpasses previous work in that a single chemometric software package is used to elicit chemical information from the integrated spectroscopic images. Integrating these two complementary spectroscopic methods provides both elemental and molecular spatial distribution within a specimen. The critical aspect in this work is using full spectral maps from each pixel within the image and subsequent processing with chemometric tools to provide integrated chemical information. This integration enables a powerful approach to more comprehensive materials characterization. Issues addressed include sample registration and beam penetration depth and how each affects post-processing. An inorganic salt and an acetaminophen pharmaceutical model mixture demonstrate the power of integrating these techniques with chemometric software.

Attenuated total internal reflectance infrared microspectroscopy as a detection technique for capillary electrophoresis.

A novel detector for capillary electrophoresis (CE) using single-bounce attenuated total internal reflectance (ATR) Fourier transform infrared (FT-IR) microspectroscopy is presented. The terminus of the CE capillary is placed approximately 1 microm from the internal reflectance crystal at the focus of an ATR infrared microscope. Using pressure driven flow injection, concentration and volume detection limits have been determined for 25- and 10-microm-i.d. silica capillaries. Upon injection of 820 pL of succinylcholine chloride in a 10-microm capillary, a concentration detection limit of approximately 0.5 parts per thousand (ppt), or 410 pg, is found. The injection volume detection limit using a 108 ppt solution is 2.0 pL (216 pg). Sample separations using a programmed series of pressure, voltage, and again pressure on 25-, 50-, and 75-microm-i.d. capillaries are shown. CE separations of citrate and nitrate, as well as succinylcholine chloride with sodium salicylate using acetone as a neutral marker, are demonstrated. Several advantages of this CE-FT-IR technique include: (1) minimization of postcolumn broadening as a result of a small detector volume; (2) the ability to signal average spectra of the same aliquot, thereby improving the signal-to-noise in a stopped-flow environment; and (3) simplicity of design.

Attenuated total internal reflectance infrared microspectroscopy as a detection technique for high-performance liquid chromatography.

The use of single reflection attenuated total internal reflection infrared microspectroscopy as a detector for high-performance liquid chromatography (HPLC) is demonstrated. The terminus of the HPLC column is placed at the focus of an ATR infrared microscope, allowing several advantages over other detection techniques. These include the following: (1) the reduction or elimination, or both, of detection cell dead volume, (2) the ability to interrogate a nearly pure aliquot of analyte, and (3) the ability to signal average spectra of the same aliquot (depending on its size) thereby improving the signal-to-noise ratio of the measurement and, concomitantly, the analytical characteristics of the method. Static measurements of succinylcholine chloride in water have shown a detection limit of 0.7 parts per thousand (ppt). Two-microliter injections of succinylcholine chloride in a 5-cm x 1-mm C-18 column with a flow rate of 60 microL/min have shown a detection limit of 1.9 ppt. This analytical concentration corresponds to a mass of 3.8 microg in the injection loop and a mass of 350 fg in the sampling volume. The potential of this detector for HPLC is demonstrated, and future improvements are also discussed.