Search NASA⌕ Search

SEARCH · Search NASA

Results for “Truth functions”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 253 records · Page 14

Evaluation and Windspeed Dependence of MODIS Aerosol Retrievals Over Open Ocean

The Maritime Aerosol Network (MAN) data set provides high quality ground-truth to validate the MODIS aerosol product over open ocean. Prior validation of the ocean aerosol product has been limited to coastal and island sites. Comparing MODIS Collection 5 ocean aerosol retrieval products with collocated MAN measurements from ships shows that MODIS is meeting the pre-launch uncertainty estimates for aerosol optical depth (AOD) with 64% and 67% of retrievals at 550 nm, and 74% and 78% of retrievals at 870 nm, falling within expected uncertainty for Terra and Aqua, respectively. Angstrom Exponent comparisons show a high correlation between MODIS retrievals and shipboard measurements (R= 0.85 Terra, 0.83 Aqua), although the MODIS aerosol algorithm tends to underestimate particle size for large particles and overestimate size for small particles, as seen in earlier Collections. Prior analysis noted an offset between Terra and Aqua ocean AOD, without concluding which sensor was more accurate. The simple linear regression reported here, is consistent with other anecdotal evidence that Aqua agreement with AERONET is marginally better. However we cannot claim based on the current study that the better Aqua comparison is statistically significant. Systematic increase of error as a function of wind speed is noted in both Terra and Aqua retrievals. This wind speed dependency enters the retrieval when winds deviate from the 6 m/s value assumed in the rough ocean surface and white cap parameterizations. Wind speed dependency in the results can be mitigated by using auxiliary NCEP wind speed information in the retrieval process.

Kleidman, Richard G.↗

Development of a dynamic quality assurance testing protocol for multisite clinical trial DCE-CT accreditation

Purpose: Credentialing can have an impact on whether or not a clinical trial produces useful quality data that is comparable between various institutions and scanners. With the recent increase of dynamic contrast enhanced-computed tomography (DCE-CT) usage as a companion biomarker in clinical trials, effective quality assurance, and control methods are required to ensure there is minimal deviation in the results between different scanners and protocols at various institutions. This paper attempts to address this problem by utilizing a dynamic flow imaging phantom to develop and evaluate a DCE-CT quality assurance (QA) protocol.Methods: A previously designed flow phantom, capable of producing predictable and reproducible time concentration curves from contrast injection was fully validated and then utilized to design a DCE-CT QA protocol. The QA protocol involved a set of quantitative metrics including injected and total mass error, as well as goodness of fit comparison to the known truth concentration curves. An additional region of interest (ROI) sensitivity analysis was also developed to provide additional details on intrascanner variability and determine appropriate ROI sizes for quantitative analysis. Both the QA protocol and ROI sensitivity analysis were utilized to test variations in DCE-CT results using different imaging parameters (tube voltage and current) as well as alternate reconstruction methods and imaging techniques. The developed QA protocol and ROI sensitivity analysis was then applied at three institutions that were part of clinical trial involving DCE-CT and results were compared.Results: The inherent specificity of robustness of the phantom was determined through calculation of the total intraday variability and determined to be less than 2.2 ± 1.1% (total calculated output contrast mass error) with a goodness of fit (R{sup 2}) of greater than 0.99 ± 0.0035 (n= 10). The DCE-CT QA protocol was capable of detecting significant deviations from the expected phantom result when scanning at low mAs and low kVp in terms of quantitative metrics (Injected Mass Error 15.4%), goodness of fit (R{sup 2}) of 0.91, and ROI sensitivity (increase in minimum input function ROI radius by 146 ± 86%). These tests also confirmed that the ASIR reconstruction process was beneficial in reducing noise without substantially increasing partial volume effects and that vendor specific modes (e.g., axial shuttle) did not significantly affect the phantom results. The phantom and QA protocol were finally able to quickly (<90 min) and successfully validate the DCE-CT imaging protocol utilized at the three separate institutions of a multicenter clinical trial; thereby enhancing the confidence in the patient data collected.Conclusions: A DCE QA protocol was developed that, in combination with a dynamic multimodality flow phantom, allows the intrascanner variability to be separated from other sources of variability such as the impact of injection protocol and ROI selection. This provides a valuable resource that can be utilized at various clinical trial institutions to test conformance with imaging protocols and accuracy requirements as well as ensure that the scanners are performing as expected for dynamic scans.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

In-process monitoring and prediction of droplet quality in droplet-on-demand liquid metal jetting additive manufacturing using machine learning

Abstract In droplet-on-demand liquid metal jetting (DoD-LMJ) additive manufacturing, complex physical interactions govern the droplet characteristics, such as size, velocity, and shape. These droplet characteristics, in turn, determine the functional quality of the printed parts. Hence, to ensure repeatable and reliable part quality it is necessary to monitor and control the droplet characteristics. Existing approaches for in-situ monitoring of droplet behavior in DoD-LMJ rely on high-speed imaging sensors. The resulting high volume of droplet images acquired is computationally demanding to analyze and hinders real-time control of the process. To overcome this challenge, the objective of this work is to use time series data acquired from an in-process millimeter-wave sensor for predicting the size, velocity, and shape characteristics of droplets in DoD-LMJ process. As opposed to high-speed imaging, this sensor produces data-efficient time series signatures that allows rapid, real-time process monitoring. We devise machine learning models that use the millimeter-wave sensor data to predict the droplet characteristics. Specifically, we developed multilayer perceptron-based non-linear autoregressive models to predict the size and velocity of droplets. Likewise, a supervised machine learning model was trained to classify the droplet shape using the frequency spectrum information contained in the millimeter-wave sensor signatures. High-speed imaging data served as ground truth for model training and validation. These models captured the droplet characteristics with a statistical fidelity exceeding 90%, and vastly outperformed conventional statistical modeling approaches. Thus, this work achieves a practically viable sensing approach for real-time quality monitoring of the DoD-LMJ process, in lieu of the existing data-intensive image-based techniques.

Gaikwad, Aniruddha (ORCID:0000000285642621)↗

Evaluation of intrinsic respiratory signal determination methods for 4D CBCT adapted for mice

Purpose: 4D CT imaging in mice is important in a variety of areas including studies of lung function and tumor motion. A necessary step in 4D imaging is obtaining a respiratory signal, which can be done through an external system or intrinsically through the projection images. A number of methods have been developed that can successfully determine the respiratory signal from cone-beam projection images of humans, however only a few have been utilized in a preclinical setting and most of these rely on step-and-shoot style imaging. The purpose of this work is to assess and make adaptions of several successful methods developed for humans for an image-guided preclinical radiation therapy system. Methods: Respiratory signals were determined from the projection images of free-breathing mice scanned on the X-RAD system using four methods: the so-called Amsterdam shroud method, a method based on the phase of the Fourier transform, a pixel intensity method, and a center of mass method. The Amsterdam shroud method was modified so the sharp inspiration peaks associated with anesthetized mouse breathing could be detected. Respiratory signals were used to sort projections into phase bins and 4D images were reconstructed. Error and standard deviation in the assignment of phase bins for the four methods compared to a manual method considered to be ground truth were calculated for a range of region of interest (ROI) sizes. Qualitative comparisons were additionally made between the 4D images obtained using each of the methods and the manual method. Results: 4D images were successfully created for all mice with each of the respiratory signal extraction methods. Only minimal qualitative differences were noted between each of the methods and the manual method. The average error (and standard deviation) in phase bin assignment was 0.24 ± 0.08 (0.49 ± 0.11) phase bins for the Fourier transform method, 0.09 ± 0.03 (0.31 ± 0.08) phase bins for the modified Amsterdam shroud method, 0.09 ± 0.02 (0.33 ± 0.07) phase bins for the intensity method, and 0.37 ± 0.10 (0.57 ± 0.08) phase bins for the center of mass method. Little dependence on ROI size was noted for the modified Amsterdam shroud and intensity methods while the Fourier transform and center of mass methods showed a noticeable dependence on the ROI size. Conclusions: The modified Amsterdam shroud, Fourier transform, and intensity respiratory signal methods are sufficiently accurate to be used for 4D imaging on the X-RAD system and show improvement over the existing center of mass method. The intensity and modified Amsterdam shroud methods are recommended due to their high accuracy and low dependence on ROI size.

60 APPLIED LIFE SCIENCES↗

Uncertainty-Aware Machine Learning for Small-Angle X-ray Scattering Analysis in Autonomous Experimentation

Small-angle X-ray scattering (SAXS) is a powerful high-throughput characterization tool for probing nanoscale structure in native sample environments, providing real-time morphological information such as nanoparticle size and shape during synthesis. However, automated SAXS data analysis for extracting meaningful structural parameters is non-trivial and remains a bottleneck in closed-loop experimentation towards autonomous materials discovery, which demands fast, reliable, and uncertainty-aware data analysis. Here, we develop a machine-learning approach for automated SAXS analysis tailored to closed-loop nanoparticle synthesis. A Random Forest (RF) regression model is trained on 100,000 synthetic SAXS curves generated from polydisperse spherical nanoparticles with realistic background contributions. Using normalized one-dimensional SAXS intensity profiles as input, the RF model directly predicts nanoparticle radius, size polydispersity, and background parameters, while the ensemble standard deviation across trees provides built-in uncertainty quantification (UQ). On synthetic data, we show that combining fit-quality metrics (R 2 , MAE) with thresholds on prediction uncertainty reliably identifies accurate parameter estimates without access to ground truth. We then apply the trained model to 365 experimental SAXS profiles of citrate-reduced gold nanoparticles synthesized using an automated droplet-flow microreactor with in situ SAXS at a synchrotron beamline, classifying the results into high- and low-confidence subsets based on UQ metrics. Finally, we integrate RF-based SAXS analysis into a simulated closed-loop optimization campaign using Gaussian process Bayesian optimization to minimize nanoparticle polydispersity, benchmarking against conventional automated Levenberg–Marquardt fitting. The RF-guided campaign exhibits substantially faster convergence and lower relative opportunity cost (∼0.07 vs ∼0.3), demonstrating that uncertainty-aware machine-learning SAXS analysis significantly enhances the efficiency and robustness of autonomous nanomaterials synthesis workflows.

Bayesian optimization↗

Ionospheric calibration for single frequency altimeter measurements

This report investigates the potential of using Global Positioning System (GPS) data and a model of the ionosphere to supply a measure of the sub-satellite Total Electron Current (TEC) of the required accuracy (10 TECU rms) for the purpose of calibrating single frequency radar altimeter measurements. Since climatological (monthly mean) models are known to be in error by as much as 50 percent, this work focused on the Parameterized Real-Time Ionospheric Specification Model (PRISM) which has the capability to improve model accuracy by ingesting (adjusting to) in situ ionospheric measurements. A set of globally distributed TEC measurements were generated using GPS data and were used as input to improve the accuracy of the PRISM model. The adjusted PRISM TEC values were compared to TOPEX dual frequency TEC measurements (which are considered truth) for a number of TOPEX sub-satellite tracks. The adjusted PRISM values generally compared to the TOPEX measurements within the 10 TECU accuracy requirements when the sub-satellite track passed within 300 to 400 km of the GPS TEC data or when the track passed through a night time ionosphere. However, when the sub-satellite points were greater than 300 to 400 km away from the GPS TEC data or when a local noon ionosphere was sampled, the adjusted PRISM values generally differed by greater than 10 TECU rms with data excursions from the TOPEX TEC measurements of as much as 40 TECU (an 8 cm path delay error at K band). Therefore, it can be concluded from this analysis that an unrealistically large number of GPS stations would be needed to predict sub-satellite TEC at the 10 TECU level in the day time ionosphere using a model such as PRISM. However, a technique currently being studied at the Jet Propulsion Laboratory (JPL) may provide a means of supplying adequate TEC data to meet the 10 TECU ionospheric correction accuracy when using a realistic number of ionospheric stations. This method involves using global GPS TEC data to estimate a global grid of vertical ionospheric TEC as a function of time (i.e. every one half hour) in a sun-fixed longitude frame. Working in a sun-fixed longitude frame, one is not limited by the spatial decorrelation distance of the ionosphere, but instead is limited more by the temporal correlations of the ionosphere in the sun-fixed frame which are a smaller effect. It is the opinion of the authors that using the global sun-fixed TEC grid data, in particular, ingesting it into PRISM, offers the best possibility of meeting the 10 TECU ionospheric correction accuracy requirement, and should be the subject of further study.

Schreiner, William S.↗

ETO cloud studies for FIRE 2, part 1

This research program in support of Project FIRE (First ISCCP Regional Experiment) involved two efforts. The results of the first effort, which were in direct support of the Extended Time Observations (ETO) component, are described here. Over the period from June 1990 through May 1991, our remote sensing systems were applied to providing ground-truth cirrus cloud observations for a total of 71 NOAA polar orbiting satellite overpasses. (Chronological tables of this effort are provided.) The primary remote sensor was a dual-polarization ruby (0.694 microns wavelength) lidar, although mid-way through the program we added a number of radiometers to assess the surface radiation budget and cirrus cloud infrared emittance, and some supplemental observations from a Ka-band (8.6 mm) radar were also collected. These studies were conducted from the Facility for Atmospheric Remote Sensing (FARS) at 40 degrees 46 minutes 00 seconds north latitude and 111 degrees 49 minutes 38 seconds east longitude. We also investigated the unusual characteristics of a subset of ETO case studies involving cirrus that generated solar and lunar corona displays. As we reported recently (reprint attached), these cirrus were atypically high and cold in relation to our total midlatitude cloud sample, and were comprised of unexpectedly small ice crystals from 10 to 30 microns in dimension. This finding lends some credence to the so-called cirrus small particle radiative anomaly, but only for very cold (less than -60 C) cirrus clouds. In a supplement, we will describe the design and testing of a prototype cirrus cloud polar nephelometer, which we constructed as part of our second research effort, to allow scattering phase functions to be obtained in future in situ cirrus research.

Sassen, Kenneth↗

On Orbit Testing of the Video Guidance Sensor

The Video Guidance Sensor (VGS), part of NASA's Automated Rendezvous and Capture program, was flown on Shuttle mission STS-95 in October of 1998 to orbitally test the functional characteristics of the VGS. This was the second flight of the VGS (the first flight was in 1997 on STS-87), and this time long-range tracking data was gathered during the experiment. The flight experiment sensor was designed to operate from 1.5 meter range out to 110 meter range, with a field-of-view of 16 X 21 degrees. The VGS tracked its target at a 5 Hz rate and returned 6-degree-of-freedom information on the target's position and attitude relative to the sensor. The VGS was mounted in the Shuttle cargo bay, and its target was mounted on the Spartan spacecraft being carried on this mission. The orbital testing of the VGS included operations with the target on the Shuttle's Remote Manipulator System (RMS) at the start of the 10-day mission, long-range data during the Shuttle rendezvous with the Spartan two days later, and some more RMS operations later in the mission. The data returned from the orbital testing included VGS diagnostics, acquisition, and tracking data, RMS positions, hand-held laser range data, tapes of the data from the VGS video camera, and orbital positioning data from the Spartan and the Shuttle to allow correlation of the VGS data with orbital best-estimate-of-truth data. The Video Guidance Sensor performed well in all phases of the testing, and the VGS is being incorporated into the ground testing of a complete automated rendezvous and docking system. Work on the development of the next generation VGS is continuing

Howard, Richard T.↗

nn-PINNs: Non-Newtonian physics-informed neural networks for complex fluid modeling

Time- and rate-dependent material functions in non-Newtonian fluids in response to different deformation fields pose a challenge in integrating different constitutive models into conventional computational fluid dynamic platforms. Considering their relevance in many industrial and natural settings alike, robust data-driven frameworks that enable accurate modeling of these complex fluids are of great interest. The main goal is to solve the coupled Partial Differential Equations (PDEs) consisting of the constitutive equations that relate the shear stress to the deformation and fully capture the behavior of the fluid under various flow protocols with different boundary conditions. Here, in this work, we present non-Newtonian physics-informed neural networks (nn-PINNs) for solving systems of coupled PDEs adopted for complex fluid flow modeling. The proposed nn-PINN method is employed to solve the constitutive models in conjunction with conservation of mass and momentum by benefiting from Automatic Differentiation (AD) in neural networks, hence avoiding the mesh generation step. nn-PINNs are tested for a number of different complex fluids with different constitutive models and for several flow protocols. These include a range of Generalized Newtonian Fluid (GNF) empirical constitutive models, as well as some phenomenological models with memory effects and thixotropic timescales. nn-PINNs are found to obtain the correct solution of complex fluids in spatiotemporal domains with good accuracy compared to the ground truth solution. We also present applications of nn-PINNs for complex fluid modeling problems with unknown boundary conditions on the surface, and show that our approach can successfully recover the velocity and stress fields across the domain, including the boundaries, given some sparse velocity measurements.

42 ENGINEERING↗

Remote sensing of earth terrain

In remote sensing, the encountered geophysical media such as agricultural canopy, forest, snow, or ice are inhomogeneous and contain scatters in a random manner. Furthermore, weather conditions such as fog, mist, or snow cover can intervene the electromagnetic observation of the remotely sensed media. In the modelling of such media accounting for the weather effects, a multi-layer random medium model has been developed. The scattering effects of the random media are described by three-dimensional correlation functions with variances and correlation lengths corresponding to the fluctuation strengths and the physical geometry of the inhomogeneities, respectively. With proper consideration of the dyadic Green's function and its singularities, the strong fluctuation theory is used to calculate the effective permittivities which account for the modification of the wave speed and attenuation in the presence of the scatters. The distorted Born approximation is then applied to obtain the correlations of the scattered fields. From the correlation of the scattered field, calculated is the complete set of scattering coefficients for polarimetric radar observation or brightness temperature in passive radiometer applications. In the remote sensing of terrestrial ecosystems, the development of microwave remote sensing technology and the potential of SAR to measure vegetation structure and biomass have increased effort to conduct experimental and theoretical researches on the interactions between microwave and vegetation canopies. The overall objective is to develop inversion algorithms to retrieve biophysical parameters from radar data. In this perspective, theoretical models and experimental data are methodically interconnected in the following manner: Due to the complexity of the interactions involved, all theoretical models have limited domains of validity; the proposed solution is to use theoretical models, which is validated by experiments, to establish the region in which the radar response is most sensitive to the parameters of interest; theoretically simulated data will be used to generate simple invertible models over the region. For applications to the remote sensing of sea ice, the developed theoretical models need to be tested with experimental measurements. With measured ground truth such as ice thickness, temperature, salinity, and structure, input parameters to the theoretical models can be obtained to calculate the polarimetric scattering coefficients for radars or brightness temperature for radiometers and then compare theoretical results with experimental data. Validated models will play an important role in the interpretation and classification of ice in monitoring global ice cover from space borne remote sensors in the future. We present an inversion algorithm based on a recently developed inversion method referred to as the Renormalized Source-Type Integral Equation approach. The objective of this method is to overcome some of the limitations and difficulties of the iterative Born technique. It recasts the inversion, which is nonlinear in nature, in terms of the solution of a set of linear equations; however, the final inversion equation is still nonlinear. The derived inversion equation is an exact equation which sums up the iterative Neuman (or Born) series in a closed form and, thus, is a valid representation even in the case when the Born series diverges; hence, the name Renormalized Source-Type Integral Equation Approach.

Yueh, Herng-Aung↗

The Synchronic Web

The Synchronic Web is a distributed network for securing data provenance on the World Wide Web. By enabling clients around the world to freely commit digital information into a single shared view of history, it provides a foundational basis of truth on which to build decentralized and scalable trust across the Internet. Its core cryptographical capability allows mutually distrusting parties to create and verify statements of the following form: “I commit to this information—and only this information—at this moment in time.” The backbone of the Synchronic Web infrastructure is a simple, small, and semantic-free blockchain that is accessible to any Internet-enabled entity. The infrastructure is maintained by a permissioned network of well-known servers, called notaries, and accessed by a permissionless group of clients, called journals. Through an evolving stack of flexible and composable semantic specifications, the parties cooperate to generate synchronic commitments over arbitrary data. When integrated with existing infrastructures, adapted to diverse domains, and scaled across the breadth of cyberspace, the Synchronic Web provides a ubiquitous mechanism to lock the world’s data into unique points in discrete time and digital space. This document provides a technical description of the core Synchronic Web system. The distinguishing innovation in our design—and the enabling mechanism behind the model—is the novel use of verifiable maps to place authenticated content into canonically defined locations off-chain. While concrete specifications and software implementations of the Synchronic Web continue to evolve, the information covered in the body of this document should remain stable. We aim to present this information clearly and concisely for technical non-experts to understand the essential functionality and value proposition of the network. In the interest of promoting discourse, we take some liberty in projecting the potential implications of the new model.

97 MATHEMATICS AND COMPUTING↗

Stimulation of isoprene emissions and electron transport rates are a key mechanism of thermal tolerance in the tropical species Vismia guianensis

Tropical forests absorb large amounts of atmospheric CO 2 through photosynthesis, but high surface temperatures suppress this absorption while promoting isoprene emissions. While mechanistic isoprene emission models predict a tight coupling to photosynthetic electron transport (ETR) as a function of temperature, direct field observations of these phenomenon are lacking in the tropics and are necessary to assess the impact of a warming climate on global isoprene emissions. In this paper, we demonstrate that in the early successional species Vismia guianensis in the central Amazon, ETR rates increased with temperature in concert with isoprene emissions, even as stomatal conductance ($g _s$) and net photosynthetic carbon fixation ($P_n$) declined. We observed the highest temperatures of continually increasing isoprene emissions yet reported (50°C). While $P_n$ showed an optimum value of 32.6 ± 0.4°C, isoprene emissions, ETR, and the oxidation state of PSII reaction centers (q L ) increased with leaf temperature with strong linear correlations for ETR (ρ = 0.98) and q L (ρ= 0.99) with leaf isoprene emissions. In contrast, other photoprotective mechanisms, such as non-photochemical quenching (NPQ), were not activated at elevated temperatures. Inhibition of isoprenoid biosynthesis repressed $P_n$ at high temperatures through a mechanism that was independent of stomatal closure. While extreme warming will decrease $g_s$ and $P_n$ in tropical species, our observations support a thermal tolerance mechanism where the maintenance of high photosynthetic capacity under extreme warming is assisted by the simultaneous stimulation of ETR and metabolic pathways that consume the direct products of ETR including photorespiration and the biosynthesis of thermoprotective isoprenoids. Our results confirm that models which link isoprene emissions to the rate of ETR hold true in tropical species and provide necessary "ground-truthing" for simulations of the large predicted increases in tropical isoprene emissions with climate warming.

54 ENVIRONMENTAL SCIENCES↗

A Multi-Scale Sampling Strategy for Detecting Physiologically Significant Signals in AVIRIS Imagery

Models of photosynthetic production at ecosystem and global scales require multiple input parameters specifying physical and physiological surface features. While certain physical parameters (e.g., absorbed photosynthetically active radiation) can be derived from current satellite sensors, other physiologically relevant measures (e.g., vegetation type, water status, carboxylation capacity, or photosynthetic light-use efficiency), are not generally directly available from current satellite sensors at the appropriate geographic scale. Consequently, many model parameters must be assumed or derived from independent sources, often at an inappropriate scale. An abundance of ecophysiological studies at the leaf and canopy scales suggests strong physiological control of vegetation-atmosphere CO2 and water vapor fluxes, particularly in evergreen vegetation subjected to diurnal or seasonal stresses. For example hot, dry conditions can lead to stomatal closure, and associated "downregulation" of photosynthetic biochemical processes, a phenomenon often manifested as a "midday photosynthetic depression". A recent study with the revised simple biosphere (SiB2) model demonstrated that photosynthetic downregulation can significantly impact global climate. However, at the global scale, the exact significance of downregulation remains unclear, largely because appropriate physiological measures are generally unavailable at this scale. Clearly, there is a need to develop reliable ways of extracting physiologically relevant information from remote sensing. Narrow-band spectrometers offer many opportunities for deriving physiological parameters needed for ecosystem and global scale photosynthetic models. Experimental studies on the ground at the leaf- to stand-scale have indicated that several narrow-band features can be used to detect plant physiological status. One physiological signal is caused by xanthophyll cycle pigment activity, and is often expressed as the Photochemical Reflectance Index (PRI). Because the xanthophyll cycle pigments are photoregulatory pigments closely linked to photosynthetic function, this index can be used to derive relative photosynthetic rates. An additional signal with physiological significance is the 970 nm water absorption band, which provides a measure of liquid water content. This feature has been quantified both using a simple 2-band ratio (900/970 nm, here referred to as the "Water Band Index" or WBI;), and using the "continuum removal" method. Current atmospheric correction methods for AVIRIS imagery also obtain quantitative expressions of surface liquid water absorption based on the 970 nm water band and may be comparable to ground-based estimates of water content using this feature. However, physiological interpretations of both the PRI and the WBI are best understood at the leaf and canopy scales, where complications of atmospheric interference and complex stand and landscape features can be minimized, and where experimental manipulations can be readily applied. Currently it is not known whether these physiological indices can be used to derive meaningful physiological information from AVIRIS imagery. In addition to the problem of atmospheric interference, another challenge is that any simple physiological index can be confounded by multiple factors unrelated to physiology, and this problem can become more severe at progressively larger spatial scales. For example, previous work has suggested that both the PRI and the WBI, are strongly correlated with other optical measures of canopy structure (e.g., the Normalized Difference Vegetation Index or green vegetation fraction), indicating a confounding effect of structure on physiological signals at the larger, landscape scale. Furthermore, the normal operating mode of most imaging spectrometers does not allow simultaneous, ground truthing at a level of detail needed for physiological sampling. Additionally, manipulative experiments of physiology are difficult to apply at a geographic scale suitable for comparison with remote imagery, which often works at spatial scales that are several orders of magnitude larger than those typically used for physiological studies. These limitations require the consideration of alternative approaches to validating physiological information derived from AVIRIS data. In this report, we present a multi-scale sampling approach to detecting physiologically significant signals in narrow-band spectra. This approach explores the multi-dimensional data space provided by narrow-band spectrometry, and combines AVIRIS imagery at a large scale, with ground spectral sampling at an intermediate scale, and detailed ecophysiological measurements at a fine scale, to examine seasonally and spatially changing relationships between multiple structural and physiological variables. Examples of this approach are provided by simultaneous sampling of the Normalized Difference Vegetation Index (NDVI), an index of fractional PAR interception and green vegetation cover, the Water Band Index (WBI, an index of liquid water absorption, and the Photochemical Reflectance Index (PRI, an index of xanthophyll cycle pigment activity and photosynthetic light-use efficiency. By directly linking changing optical properties sampled on the ground with measurable physiological states, we hope to develop a basis for interpreting similar signals in AVIRIS imagery.

Gamon, John A.↗

Alpha image reconstruction (AIR): A new iterative CT image reconstruction approach using voxel-wise alpha blending

Purpose: Iterative image reconstruction gains more and more interest in clinical routine, as it promises to reduce image noise (and thereby patient dose), to reduce artifacts, or to improve spatial resolution. Among vendors and researchers, however, there is no consensus of how to best achieve these aims. The general approach is to incorporatea priori knowledge into iterative image reconstruction, for example, by adding additional constraints to the cost function, which penalize variations between neighboring voxels. However, this approach to regularization in general poses a resolution noise trade-off because the stronger the regularization, and thus the noise reduction, the stronger the loss of spatial resolution and thus loss of anatomical detail. The authors propose a method which tries to improve this trade-off. The proposed reconstruction algorithm is called alpha image reconstruction (AIR). One starts with generating basis images, which emphasize certain desired image properties, like high resolution or low noise. The AIR algorithm reconstructs voxel-specific weighting coefficients that are applied to combine the basis images. By combining the desired properties of each basis image, one can generate an image with lower noise and maintained high contrast resolution thus improving the resolution noise trade-off. Methods: All simulations and reconstructions are performed in native fan-beam geometry. A water phantom with resolution bar patterns and low contrast disks is simulated. A filtered backprojection (FBP) reconstruction with a Ram-Lak kernel is used as a reference reconstruction. The results of AIR are compared against the FBP results and against a penalized weighted least squares reconstruction which uses total variation as regularization. The simulations are based on the geometry of the Siemens Somatom Definition Flash scanner. To quantitatively assess image quality, the authors analyze line profiles through resolution patterns to define a contrast factor for contrast-resolution plots. Furthermore, the authors calculate the contrast-to-noise ratio with the low contrast disks and the authors compare the agreement of the reconstructions with the ground truth by calculating the normalized cross-correlation and the root-mean-square deviation. To evaluate the clinical performance of the proposed method, the authors reconstruct patient data acquired with a Somatom Definition Flash dual source CT scanner (Siemens Healthcare, Forchheim, Germany). Results: The results of the simulation study show that among the compared algorithms AIR achieves the highest resolution and the highest agreement with the ground truth. Compared to the reference FBP reconstruction AIR is able to reduce the relative pixel noise by up to 50% and at the same time achieve a higher resolution by maintaining the edge information from the basis images. These results can be confirmed with the patient data. Conclusions: To evaluate the AIR algorithm simulated and measured patient data of a state-of-the-art clinical CT system were processed. It is shown, that generating CT images through the reconstruction of weighting coefficients has the potential to improve the resolution noise trade-off and thus to improve the dose usage in clinical CT.

60 APPLIED LIFE SCIENCES↗

The Video Guidance Sensor- A Flight Proven Technology

The Video Guidance Sensor (VGS) flew on Shuttle mission STS-95 in October of 1998 to test the VGS functional characteristics on orbit. This was the second flight of the VGS, and during these two flights, both long range and short range data were gathered under a variety of lighting conditions, orbital exposure times, and temperatures. The flight experiment sensor was designed to operate from 1.5 meter range out to 110 meter range, with a field-of-view of 16 X 21 degrees. The VGS tracked its target at a 5 Hz rate and returned 6-degree-of-freedom information on the target's position and attitude relative to the sensor. The VGS was mounted in the Shuttle cargo bay, and its target was mounted on the Spartan spacecraft being carried on this mission. The VGS is a sensor designed to allow an automated vehicle to dock with a spacecraft equipped with a passive target. The VGS is a part of an Automated Rendezvous and Capture (AR&C) system being developed and tested by NASA. The orbital testing included operations with the target on the Shuttle's Remote Manipulator System (RMS) near the start of each flight of the VGS, long-range data (on the STS-95 flight) during the Shuttle rendezvous with the Spartan two days later, and some more RMS operations later in the mission. The data returned from the orbital testing included VGS diagnostics, acquisition, and tracking data, RMS positions, hand-held laser range data, tapes of the data from the VGS video camera, and orbital positioning data from the Spartan and the Shuttle to allow correlation of the VGS data with orbital best-estimate-of-truth data. The Video Guidance Sensor performed well in all phases of the testing, and the VGS is being incorporated into the ground testing of a complete automated rendezvous and docking system. Work on the development of the next generation VGS is continuing.

Howard, Richard T.↗

An algorithm for the estimation of bounds on the emissivity and temperatures from thermal multispectral airborne remotely sensed data

The effective flux incident upon the detectors of a thermal sensor, after it has been corrected for atmospheric effects, is a function of a non-linear combination of the emissivity of the target for that channel and the temperature of the target. The sensor system cannot separate the contribution from the emissivity and the temperature that constitute the flux value. A method that estimates the bounds on these temperatures and emissivities from thermal data is described. This method is then tested with remotely sensed data obtained from NASA's Thermal Infrared Multispectral Scanner (TIMS) - a 6 channel thermal sensor. Since this is an under-determined set of equations i.e. there are 7 unknowns (6 emissivities and 1 temperature) and 6 equations (corresponding to the 6 channel fluxes), there exist theoretically an infinite combination of values of emissivities and temperature that can satisfy these equations. Using some realistic bounds on the emissivities, bounds on the temperature are calculated. These bounds on the temperature are refined to estimate a tighter bound on the emissivity of the source. An error analysis is also carried out to quantitatively determine the extent of uncertainty introduced in the estimate of these parameters. This method is useful only when a realistic set of bounds can be obtained for the emissivities of the data. In the case of water the lower and upper bounds were set at 0.97 and 1.00 respectively. Five flights were flown in succession at altitudes of 2 km (low), 6 km (mid), 12 km (high), and then back again at 6 km and 2 km. The area selected with the Ross Barnett reservoir near Jackson, Mississippi. The mission was flown during the predawn hours of 1 Feb. 1992. Radiosonde data was collected for that duration to profile the characteristics of the atmosphere. Ground truth temperatures using thermometers and radiometers were also obtained over an area of the reservoir. The results of two independent runs of the radiometer data averaged 7.03 plus or minus .70 for the first run and 7.31 plus or minus .88 for the second run. The results of the algorithm yield a temperature of 7.68 for the low altitude data to 8.73 for the high altitude data.

Jaggi, S.↗

MABAL: a Novel Deep-Learning Architecture for Machine-Assisted Bone Age Labeling

Bone age assessment (BAA) is a commonly performed diagnostic study in pediatric radiology to assess skeletal maturity. The most commonly utilized method for assessment of BAA is the Greulich and Pyle method (Pediatr Radiol 46.9:1269–1274, 2016; Arch Dis Child 81.2:172–173, 1999) atlas. The evaluation of BAA can be a tedious and time-consuming process for the radiologist. As such, several computer-assisted detection/diagnosis (CAD) methods have been proposed for automation of BAA. Classical CAD tools have traditionally relied on hard-coded algorithmic features for BAA which suffer from a variety of drawbacks. Recently, the advent and proliferation of convolutional neural networks (CNNs) has shown promise in a variety of medical imaging applications. There have been at least two published applications of using deep learning for evaluation of bone age (Med Image Anal 36:41–51, 2017; JDI 1–5, 2017). However, current implementations are limited by a combination of both architecture design and relatively small datasets. The purpose of this study is to demonstrate the benefits of a customized neural network algorithm carefully calibrated to the evaluation of bone age utilizing a relatively large institutional dataset. In doing so, this study will aim to show that advanced architectures can be successfully trained from scratch in the medical imaging domain and can generate results that outperform any existing proposed algorithm. The training data consisted of 10,289 images of different skeletal age examinations, 8909 from the hospital Picture Archiving and Communication System at our institution and 1383 from the public Digital Hand Atlas Database. The data was separated into four cohorts, one each for male and female children above the age of 8, and one each for male and female children below the age of 10. The testing set consisted of 20 radiographs of each 1-year-age cohort from 0 to 1 years to 14–15+ years, half male and half female. The testing set included left-hand radiographs done for bone age assessment, trauma evaluation without significant findings, and skeletal surveys. A 14 hidden layer-customized neural network was designed for this study. The network included several state of the art techniques including residual-style connections, inception layers, and spatial transformer layers. Data augmentation was applied to the network inputs to prevent overfitting. A linear regression output was utilized. Mean square error was used as the network loss function and mean absolute error (MAE) was utilized as the primary performance metric. MAE accuracies on the validation and test sets for young females were 0.654 and 0.561 respectively. For older females, validation and test accuracies were 0.662 and 0.497 respectively. For young males, validation and test accuracies were 0.649 and 0.585 respectively. Finally, for older males, validation and test set accuracies were 0.581 and 0.501 respectively. The female cohorts were trained for 900 epochs each and the male cohorts were trained for 600 epochs. An eightfold cross-validation set was employed for hyperparameter tuning. Test error was obtained after training on a full data set with the selected hyperparameters. Using our proposed customized neural network architecture on our large available data, we achieved an aggregate validation and test set mean absolute errors of 0.637 and 0.536 respectively. To date, this is the best published performance on utilizing deep learning for bone age assessment. Our results support our initial hypothesis that customized, purpose-built neural networks provide improved performance over networks derived from pre-trained imaging data sets. We build on that initial work by showing that the addition of state-of-the-art techniques such as residual connections and inception architecture further improves prediction accuracy. This is important because the current assumption for use of residual and/or inception architectures is that a large pre-trained network is required for successful implementation given the relatively small datasets in medical imaging. Instead we show that a small, customized architecture incorporating advanced CNN strategies can indeed be trained from scratch, yielding significant improvements in algorithm accuracy. It should be noted that for all four cohorts, testing error outperformed validation error. One reason for this is that our ground truth for our test set was obtained by averaging two pediatric radiologist reads compared to our training data for which only a single read was used. This suggests that despite relatively noisy training data, the algorithm could successfully model the variation between observers and generate estimates that are close to the expected ground truth.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Plant Water Profiler: A Tool for Understanding Water Use, Cost and Savings Potential for Manufacturing Plants

In the 1746 Poor Richard’s Almanack, Benjamin Franklin published “When the well’s dry, we know the worth of water” (Franklin, 1746); industrial water users have begun to see the truth in these words as they examine the potential and reality of water scarcity becoming a significant business risk. Through the authors’ interactions with manufacturers via the U.S. Department of Energy’s Better Plants program and Strategic Analysis projects, they realized that manufacturing facilities often do not fully understand the risks to the water supply and the consequential risks to their business. The authors’ also noticed that many facilities do not know how their water is used (baselining), how they compare with their peers (benchmarking), the actual cost impacts of water use (true cost of water), and potential areas for water savings. This results in a lack of motivation and ability to identify and implement effective water-efficiency measures.The Plant Water Profiler Tool (PWP) was created to help manufacturing plant management address these issues by identifying and accounting for water use in manufacturing operations, helping determine the true cost of water, and providing some high-level estimates for peer comparison. The PWP uses a water mass balance and common engineering equations to estimate water use by each of the facility’s systems (e.g., differentiating process water use from cooling and restroom water use), even when no submetering data is available. This new tool also helps users identify possible areas for water savings and associated water and energy cost savings, making it a potentially useful resource for corporations conducting water assessments. Additionally, governments and private organizations can use the methodology behind this tool to design a survey to help build a database of industrial water use. This paper and corresponding presentation provide an overview of current industrial water use practices and tools, as well as the value of determining the true cost of water. It also briefly describes the methodology behind the tool, then focuses on its functionality through an industrial case study and lessons learned from implementing the case study.

Armstrong, Kristina O.↗