The Pacific Northwest National Laboratory (PNNL) has recently developed a next-generation cylindrical millimeter-wave imaging system. This system is based on linear sparse multistatic imaging arrays. Data from this system can be focused using 3D FFT-based reconstruction algorithms, which are reasonably efficient and can be performed in near real time, or by back-projection methods that are versatile and more accurate but are computationally intensive and require lengthy post-processing. Cylindrical Fast Backprojection (CFBP) is a novel image reconstruction algorithm developed at PNNL that radically increases the efficiency of backprojection and is ideally suited to microwave and millimeter-wave imaging systems based on scanned linear arrays such as body scanners in common use for aviation security screening. This method achieves its gains in efficiency by separating a full backprojection into a sequence of three steps, range focusing, vertical focusing, and lateral focusing, with intermediate results used to avoid repetitive multidimensional computation. The method is called cylindrical fast backprojection due to the use of two-dimensional stored results, or look-up tables, that have cylindrical symmetry about the linear array. The method is well suited to cylindrically scanned linear arrays but is equally valid for linear arrays scanned to form planar or arbitrary apertures. This paper describes the CFBP algorithm and validates its performance using simulated data.

An active 3D microwave / millimeter-wave shoe scanner was previously developed at the Pacific Northwest National Laboratory (PNNL) using two linear arrays scanned over a rectilinear aperture. The radar system chirps a frequency sweep from 10-40 GHz. These frequencies allow imaging through optically opaque material such as leather, rubber, plastics, and other dielectrics. The system was designed to detect concealed items in the soles of shoes while allowing people to leave their shoes on through a security checkpoint. To shrink the footprint of the system, a new iteration of the design has been developed that scans the two linear arrays over a circular aperture. This new footprint opens the possibility of it being installed in the floor of a cylindrical millimeter-wave body scanner. The backprojection-based multilayer dielectric image reconstruction developed at PNNL can easily handle arbitrary spatial sampling, accommodating the new rotational shoe scanner design. Commonly, the fast Fourier transform (FFT) is used to efficiently compute the range response from the data collected by the system as a preprocessing step to the backprojection algorithm. It was found that converting to range using the discrete Fourier transform (DFT) directly has some advantages over the FFT. For example, nonlinear and non-uniform frequency sweeps can easily be compensated for during the computation of the DFT and only the range bins of interest need to be computed and their spacing can be chosen arbitrarily. Because the range conversion step of the image reconstruction is the fastest part of the process there is very little speed penalty for using the DFT over the FFT and it can even increase the speed of image reconstruction when the ranges of interest are fewer than the total span that is calculated in the FFT.

Prompt gamma ray activation analysis (PGAA) is a non-destructive nuclear measurement technique that quantifies isotopes present in a sample. Here, we use PGAA spectra to train different types of models to elucidate how discriminating these spectra are for various classes of materials. We trained discriminative models for closed set scenarios, where all possible material classes are known. We also trained class models to address open set conditions, where this enumeration is impossible. After appropriate pre-processing and data treatments, all such models performed nearly perfectly on our dataset, suggesting PGAA spectra may serve as powerful nuclear fingerprints for robust material classification.

Indoor geolocation of radio frequency (RF) transmitters is challenging due to site-specific multipath effects, especially when the sources are uncooperative. A source being uncooperative means that a-priori the transmit time, location, and power are unknown. It is proposed to use a site-specific ray tracing model to generate realistic indoor RF responses, perform geolocation using multiple methods, and compare their performance. This work will look at two types of geolocation algorithms: (1) time-difference of arrival (TDOA) and (2) received signal strength Indicator (RSSI) fingerprinting or pattern-matching-based methods. An indoor space will be simulated with a grid of fixed receivers and a grid of transmit locations using Remcom's Wireless InSite. Using the output of the Wireless InSite simulation the response from any given transmit location can be generated and used to evaluate geolocation performance.

This work focuses on the problem of RF geolocation in complex multipath environments. Using 3D electromagnetic propagation modeling to characterize environments of interest will enable more accurate RF geolocation. Specifically, a path-loss radio map can be generated in simulation for use in received signal strength indicators (RSSI) fingerprinting, or pattern matching. RSSI fingerprinting is an example of data-based method that takes site specific information into account which should allow for better performance than other model-based methods that use a generalized model of electromagnetic propagation. This modeling capability will also be used to evaluate the relative performance of RSSI fingerprinting, pathloss model based RSSI methods such as differential received signal strength circles (DRSS), RSSI joint gaussian estimation, and time-difference of arrival (TDOA). New methods using this simulation derived electromagnetic characterization could improve the efficacy of currently deployed and future RF spectral monitoring solutions. Wireless InSite developed by Remcom is used as the simulation tool of choice in this work. An indoor location is simulated with a grid of fixed receivers and a grid of transmit locations. Using the output of the Wireless InSite simulation the response from a given transmit location to a given receive location can be generated. During the first year of the project various geolocation methods evaluated on purely synthetic, but realistic, data. The second year focused on testing and validating the efficacy of simulation informed RF geolocation using two physical testbeds. This work has shown that data-based approaches are more accurate than model-based ones at the expensive of requiring measured or simulated site-specific training data.

Active microwave and millimeter-wave imaging is commonly used for security screening at airport checkpoints and other protected facilities. This paper explores an imaging system concept that may improve screening convenience, reduce cost, and enable alternative operational concepts by allowing a person to walk naturally through the system. Millimeter-wave imaging systems require data to be acquired over a 2D spatial aperture to form a high-resolution image. This requirement is usually met using mechanical scanners or large antenna arrays that provide a 2D aperture and provide strict control over the position of the array in relation to a motionless target. The new concept explored in this paper replaces the mechanical scan with motion of the passenger. The complex motion of the passenger is expected to be optically tracked as he or she passes by stationary linear vertical millimeter-wave arrays and can be modeled using skeletal animation. Multiple linear arrays illuminate the passenger from a wide variety of angles to provide full coverage of the body. The radar data are then correlated with the skeletal animation model by employing generalized synthetic aperture focusing or back-projection techniques. These methods accurately reconstruct the image by integrating the measured response multiplied by the conjugate of the expected response from a point scatterer anywhere within a 3D image volume. This process yields an optimally focused image and can be applied to situations involving complex target motion. This paper describes this concept in detail and provides numerous simulation-based imaging results to explore the effectiveness of the proposed methods.

The toxicity and bioavailability of arsenic is heavily dependent on its speciation. Therefore, robust and accurate methods are needed to determine arsenic speciation profiles for materials related to public health initiatives, such as food safety. Here, X-ray spectroscopies are attractive candidates as they provide in situ, nondestructive analyses of solid samples without perturbation to the arsenic species therein. This work provides a speciation analysis for three certified reference materials for the food chemistry community, whose assigned values may be used to assess the merit of the X-ray spectroscopy results. Furthermore, extracts of SRM 3232 Kelp Powder, which is value-assigned for arsenic species, are measured to provide further evidence of its efficacy. These analyses are performed on the results of As K-edge X-ray Absorption Near Edge Structure (XANES) measurements collected on each sample. Notably, such analyses have traditionally relied on linear combination fitting of a minimal subset of empirical standards selected by stepwise regression. This is known to be problematic for compounds with meaningfully collinear spectra and can yield overestimates of the accuracy of the analysis. Therefore, the least absolute shrinkage and selection operator (lasso) regression method is used to reduce the risk of overfitting and increase the interpretability of statistical inferences. As this is a biased statistical method, results and uncertainties are estimated using a bootstrap method accounting for the dominant sources of variability. Finally, this method does not separate model and data selection from regression analysis. Indeed, a survey of many spectral influences is presented including changes in the: state of methylation, state of protonation, oxidation state, coordination geometry, and sample phase. These compounds were all included in the model's training set, preventing model over-simplification and enabling high-throughput and robust analyses.

The most direct approach to determining if two aqueous solutions will phase-separate upon mixing is to exhaustively screen them in a pair-wise fashion. This is a time-consuming process that involves preparation of numerous stock solutions, precise transfer of highly concentrated and often viscous solutions, exhaustive agitation to ensure thorough mixing, and time-sensitive monitoring to observe the presence of emulsion characteristics indicative of phase separation. Here, we examined the pair-wise mixing behavior of 68 water-soluble compounds by observing the formation of microscopic phase boundaries and droplets of 2278 unique 2-component solutions. A series of machine learning classifiers (artificial neural network, random forest, k-nearest neighbors, and support vector classifier) were then trained on physicochemical property data associated with the 68 compounds and used to predict their miscibility upon mixing. Miscibility predictions were then compared to the experimental observations. The random forest classifier was the most successful classifier of those tested, displaying an average receiver operator characteristic area under the curve of 0.74. The random forest classifier was validated by removing either one or two compounds from the input data, training the classifier on the remaining data and then predicting the miscibility of solutions involving the removed compound(s) using the classifier. The accuracy, specificity, and sensitivity of the random forest classifier were 0.74, 0.80, and 0.51, respectively, when one of the two compounds to be examined was not represented in the training data. When asked to predict the miscibility of two compounds, neither of which were represented in the training data, the accuracy, specificity, and sensitivity values for the random forest classifier were 0.70, 0.82 and 0.29, respectively. Thus, there is potential for this machine learning approach to improve the design of screening experiments to accelerate the discovery of aqueous two-phase systems for numerous scientific and industrial applications.

Millimeter-wave (MMW) imaging systems require high resolution and spurious free dynamic range (SFDR) to produce images with low artifacts. In these systems, lateral resolution is determined by the center frequency and the ratio of the aperture size to the distance to the target. The downrange resolution is determined by the system bandwidth; for very fine downrange resolution of a target, or material layers/discontinuities, the bandwidth must be very broad. Additionally, the response over the bandwidth must be measured very rapidly to support high speed acquisition over large apertures. A voltage-controlled oscillator (VCO) is a standard means of generating the ultra-broadband frequency chirp. This paper investigates the use of a broadband VCO for use in MMW imaging systems, a linear-phase calibration technique, and digital-to-analog converter (DAC) parameter considerations for controlling the VCO.

Active millimeter and microwave imaging techniques can be used to create a high-resolution 3D image volume of a target's reflectivity. Millimeter-wave imaging is commonly used for personnel security screening and numerous other applications. Backprojection based image reconstruction techniques form a 3D complex-valued volume. The complex-valued volume is commonly reduced to a real-valued volume by taking the magnitude. For anomaly detection and cross image registration of an object it is beneficial to generate an accurate representation of an object's surface. Extracting a smooth and accurate surface from a magnitude only image is challenging. One difficulty is the magnitude image has limited resolution in the depth direction which normally limits precision to a moderate fraction of the depth resolution. Another difficulty is that the magnitude of the image depends heavily on the geometry and orientation of the object being imaged. The phase information in a complex-valued image volume provides a means to decouple the magnitude of the image from the geometry of an object and provide precision much finer than the depth resolution would indicate. This enables the generation of a smooth and accurate point cloud representation of the surface of an imaged object. A method to extract a point cloud from the phase information in a 3D complex-valued millimeter-wave image volume is developed and results with simulated and experimental data are presented.

Active millimeter-wave imaging is in widespread use for security screening and other applications. The Pacific Northwest National Laboratory (PNNL) has developed a variety of microwave and millimeter-wave imaging systems and technology, including the cylindrical imaging technology that forms the basis of the L3/Leidos ProVision system. Since 2016, PNNL has been actively participating in a working group that is developing a proposed American National Standards Institute (ANSI) standard (N42.59) that will be used to evaluate and verify performance of active millimeterwave imaging systems used for security screening of humans. The standard is developing image quality tools (IQTs) that will be used to assess a variety of imaging metrics, such as lateral resolution, contrast, and depth resolution. Depth resolution is vital for high-performance microwave and millimeter-wave imaging because it enables precise focusing over a full 3D volume, and allows for differentiation of reflections from multiple surfaces, such as a layer of clothing over the human body. In this paper, depth resolution is analyzed using theoretical simulations and experimental 3D imaging studies. Presented results examine depth resolution using IQTs developed from a thin partially transparent film placed in front of a metallic surface, creating reflections that are laterally aligned and at variable separation. Coherence between these reflections is investigated as it complicates the interpretation of the imaging results.

Active three-dimensional (3D) microwave and millimeter-wave imaging techniques have been extensively developed for concealed threat detection at the Pacific Northwest National Laboratory (PNNL), most notably the cylindrical millimeterwave imaging method currently in use for airport screening. Typically, a linear array is mechanically scanned over a cylindrical or planar aperture in order to form a high-resolution 3D image. A linear array mounted on a low-cost encoderdriven rail system was desired for rapid data collection and evaluation of concealed threat detection on a stationary target. A rail system to sweep out a planar aperture was quickly developed, however, due to the low-cost implementation of the rail system and encoder, resulting images were lower quality than expected. It was determined that the position information provided by the rail system encoder was not accurate enough to generate an image of the desired quality. Instead of using a traditional encoder wheel with the rail system, optical motion tracking was used to record 3D position information of the linear array synced with the radar as it was manually scanned over a nominally planar aperture. While optical motion tracking can provide position information with sub-millimeter level accuracy, it doesn’t guarantee that the scanned aperture is strictly planar or uniformly sampled. Reconstruction techniques necessary to incorporate 3D position information and compensate for an irregular imaging aperture are developed. Experimental results showing the benefit of precise optical motion tracking for a manually scanned linear array are presented.

Here we review recent work by the authors to revisit the concept of extrapolating thermodynamic properties of classical systems using statistical mechanical principles. Specifically, we discuss how the combination of these principles with biased sampling techniques enables the prediction of free energy landscapes and other detailed information, such as structural properties, of the system in question. Remarkably accurate estimates of physical properties across a broad range of conditions have been achieved using this approach, greatly reducing the number of simulations needed to explore a given system's behaviour. While approximate, these extrapolations significantly amplify the amount of reasonably accurate information that can be extracted from simulations enabling a small set of them to feed data-intensive regression algorithms such as neural networks. Thus, this extrapolation methodology represents a useful tool for performing tasks such as high-throughput screening of physical properties, optimising force field parameters, exploring equilibrium phase behaviour, and enabling theory-guided data science for these systems.

Protein therapeutics have numerous critical quality attributes (CQA) that must be evaluated to ensure safety and efficacy, including the requirement to adopt and retain the correct three-dimensional fold without forming unintended aggregates. Therefore, the ability to monitor protein higher order structure (HOS) can be valuable throughout the lifecycle of a protein therapeutic, from development to manufacture. 2D NMR has been introduced as a robust and precise tool to assess the HOS of a protein biotherapeutic. A common use case is to decide whether two groups of spectra are substantially different, as an indicator of difference in HOS. We demonstrate a quantitative use of principal component analysis (PCA) scores to perform this decision-making, and demonstrate the effect of acquisition and processing details on class separation using samples of NISTmAb monoclonal antibody Reference Material subjected to two different oxidative stress protocols. The work introduces an approach to computing similarity from PCA scores based upon the technique of histogram intersection, a method originally developed for retrieval of images from large databases. Results show that class separation can be robust with respect to random noise, reconstruction method, and analysis region selection. By contrast, details such as baseline distortion can have a pronounced effect, and so must be controlled carefully. Since the classification approach can be performed without the need to identify peaks, results suggest that it is possible to use even more efficient measurement strategies that do not produce spectra that can be analyzed visually, but nevertheless allow useful decision-making that is objective and automated.

Requirements for blends of drop-in petroleum/bio-derived fuels with specific thermophysical and thermochemical properties highlights the need for chemometric models that can predict these properties. Multivariate calibration methods were evaluated using the measured thermograms (i.e., change in temperature with time) of 11 diesel/biodiesel fuel blends (including four repeated runs for each fuel blend). Two National Institute of Standards and Technology Standard Reference Material® (SRM®) pure fuels were blended by serial dilution to produce fuels having diesel/biodiesel volumetric fractions between (0 to 100) %. The fuels were evaluated for the prepared fuel-blend volume fraction and total specific energy release (heating value), using a laser-driven calorimetry technique, termed 'laser-driven thermal reactor'. The experimental apparatus consists of a copper sphere-shaped reactor (mounted at the center of a stainless-steel chamber) that is heated by a high-power continuous wave Nd:YAG laser. Prior to heating by the laser, liquid sample is injected onto a copper pan substrate that rests near the center of the reactor and is in contact with a fine-wire thermocouple. A second thermocouple is in contact with the sphere-reactor inner surface. The thermograms are then used to evaluate for the thermochemical characteristic of interest. Partial least squares (PLS) and support vector machine (SVM) models were constructed and evaluated for SRM-fuel-blend quantification, and determination of prepared fuel-blend volume fraction and heating value. Quantification of the fuel-blend thermograms by the SVM method was found to better correlate with the experimental results than PLS. The combination of laser-driven calorimetry and multivariate calibration methods has demonstrated the potential application of using thermograms for fuels quantification and analysis of fuel-blend properties.

Protein therapeutics are vitally important clinically and commercially, with monoclonal antibody (mAb) therapeutic sales alone accounting for $115 billion in revenue for 2018.[1] In order for these therapeutics to be safe and efficacious, their protein components must maintain their high order structure (HOS), which includes retaining their three-dimensional fold and not forming aggregates. As demonstrated in the recent NISTmAb Interlaboratory nuclear magnetic resonance (NMR) Study[2], NMR spectroscopy is a robust and precise approach to address this HOS measurement need. Using the NISTmAb study data, we benchmark a procedure for automated outlier detection used to identify spectra that are not of sufficient quality for further automated analysis. When applied to a diverse collection of all 252 1H,13C gHSQC spectra from the study, a recursive version of the automated procedure performed comparably to visual analysis, and identified three outlier cases that were missed by the human analyst. In total, this method represents a distinct advance in chemometric detection of outliers due to variation in both measurement and sample.

The Pacific Northwest National Laboratory (PNNL) has recently developed an active 3D microwave/millimeter-wave shoe scanner. This system is designed to detect threats concealed within the soles of common footwear. The system was designed in response to the security incident involving Richard Reid, known as the "Shoe Bomber". The system operates over the 10-40 GHz frequency range. Waves in this band readily pass through common shoe materials, such as leather, rubber, plastics, foams, and synthetic and natural cloth materials. The shoe scanner system consists of a linear array positioned underneath a low loss dielectric window that the person is directed to stand upon. The linear array is positioned so the antenna propagation is vertical, and the array axis is horizontal across the width of the shoes. A linear mechanical scan translates the arrays along the length of the shoes. A frequency-modulated continuous wave (FM-CW) transceiver is used to collect the signal scattered from the scene. The data collected from the system is fully 3D covering two spatial and one frequency dimensions. The system presents several challenges for efficient image reconstruction, including the dielectric window, multi-row linear arrays, and focusing close to the antenna elements. The dielectric window presents a significant challenge for image reconstruction since the waves will travel through an inhomogeneous layered media. In this paper, an efficient back-projection reconstruction algorithm is presented that overcomes these challenges. Experimental imaging results are shown that demonstrate high-resolution imaging performance for this new scanner.

A detailed characterization of the chemical composition of complex substances, such as products of petroleum refining and environmental mixtures, is greatly needed in exposure assessment and manufacturing. The inherent complexity and variability in the composition of complex substances obfuscate the choices for their detailed analytical characterization. Yet, in lieu of exact chemical composition of complex substances, evaluation of the degree of similarity is a sensible path toward decision-making in environmental health regulations. Grouping of similar complex substances is a challenge that can be addressed via advanced analytical methods and streamlined data analysis and visualization techniques. Here, we propose a framework with unsupervised and supervised analyses to optimally group complex substances based on their analytical features. We test two data sets of complex oil-derived substances. The first data set is from gas chromatography-mass spectrometry (GC-MS) analysis of 20 Standard Reference Materials representing crude oils and oil refining products. The second data set consists of 15 samples of various gas oils analyzed using three analytical techniques: GC-MS, GC×GC-flame ionization detection (FID), and ion mobility spectrometry-mass spectrometry (IM-MS). We use hierarchical clustering using Pearson correlation as a similarity metric for the unsupervised analysis and build classification models using the Random Forest algorithm for the supervised analysis. We present a quantitative comparative assessment of clustering results via Fowlkes-Mallows index, and classification results via model accuracies in predicting the group of an unknown complex substance. We demonstrate the effect of (i) different grouping methodologies, (ii) data set size, and (iii) dimensionality reduction on the grouping quality, and (iv) different analytical techniques on the characterization of the complex substances. While the complexity and variability in chemical composition are an inherent feature of complex substances, we demonstrate how the choices of the data analysis and visualization methods can impact the communication of their characteristics to delineate sufficient similarity.