Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “image segmentation”

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 235 records · Page 13
All sourcesNASA NTRSDOE OSTI
NASA NTRS1991

Supervised pixel classification using a feature space derived from an artificial visual system

Image segmentation involves labelling pixels according to their membership in image regions. This requires the understanding of what a region is. Using supervised pixel classification, the paper investigates how groups of pixels labelled manually according to perceived image semantics map onto the feature space created by an Artificial Visual System. Multiscale structure of regions are investigated and it is shown that pixels form clusters based on their geometric roles in the image intensity function, not by image semantics. A tentative abstract definition of a 'region' is proposed based on this behavior.

Baxter, Lisa C.↗
DOE OSTI2023

Lithium Metal Battery Quality Control via Transformer–CNN Segmentation

Lithium metal battery (LMB) has the potential to be the next-generation battery system because of its high theoretical energy density. However, defects known as dendrites are formed by heterogeneous lithium (Li) plating, which hinders the development and utilization of LMBs. Non-destructive techniques to observe the dendrite morphology often use X-ray computed tomography (XCT) to provide cross-sectional views. To retrieve three-dimensional structures inside a battery, image segmentation becomes essential to quantitatively analyze XCT images. This work proposes a new semantic segmentation approach using a transformer-based neural network called TransforCNN that is capable of segmenting out dendrites from XCT data. In addition, we compare the performance of the proposed TransforCNN with three other algorithms, U-Net, Y-Net, and E-Net, consisting of an ensemble network model for XCT analysis. Our results show the advantages of using TransforCNN when evaluating over-segmentation metrics, such as mean intersection over union (mIoU) and mean Dice similarity coefficient (mDSC), as well as through several qualitatively comparative visualizations.

Quenum, Jerome (ORCID:0000000271265853)↗
DOE OSTI2022

Design of a graphical user interface for few-shot machine learning classification of electron microscopy data

The recent growth in data generation by modern electron microscopes requires rapid, scalable, and flexible approaches to image segmentation and analysis. Few-shot machine learning, which can richly classify images from a handful of user-provided examples, is a promising route to high-throughput analysis. However, current command-line implementations of such approaches can be slow and unintuitive to use, lacking the real-time feedback necessary to perform effective classification. Here we report on the development of a Python-based graphical user interface that enables end users to easily conduct and visualize the output of few-shot learning models. This interface is portable and can be hosted locally or on the web, providing the opportunity to reproducibly conduct, share, and crowd-source few-shot analyses.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2015

Pore-Scale and Multiscale Numerical Simulation of Flow and Transport in a Laboratory-Scale Column

Pore-scale models are useful for studying relationships between fundamental processes and phenomena at larger (i.e., Darcy) scales. However, the size of domains that can be simulated with explicit pore-scale resolution is limited by computational and observational constraints. Direct numerical simulation of pore-scale flow and transport is typically performed on millimeter-scale volumes at which X-ray computed tomography (XCT), often used to characterize pore geometry, can achieve micrometer resolution. In contrast, the scale at which a continuum approximation of a porous medium is valid is usually larger, on the order of centimeters to decimeters. Furthermore, laboratory experiments that measure continuum properties are typically performed on decimeter-scale columns. At this scale, XCT resolution is coarse (tens to hundreds of micrometers) and prohibits characterization of small pores and grains. We performed simulations of pore-scale processes over a decimeter-scale volume of natural porous media with a wide range of grain sizes, and compared to results of column experiments using the same sample. Simulations were conducted using high-performance codes executed on a supercomputer. Two approaches to XCT image segmentation were evaluated, a binary (pores and solids) segmentation and a ternary segmentation that resolved a third category (porous solids with pores smaller than the imaged resolution). We used a mixed Stokes-Darcy simulation method to simulate the combination of Stokes flow in large open pores and Darcy-like flow in porous solid regions. Simulations based on the ternary segmentation provided results that were consistent with experimental observations, demonstrating our ability to successfully model pore-scale flow over a column-scale domain.

CFD↗
DOE OSTI2018

Progress Update on Segmentation of Neutron Tomographic Images, 2018

The present report documents the techniques developed to segment fast-neutron tomographic images of objects consisting of assemblies of machined parts with distinct boundaries. This report satisfies the fiscal year 2018 technical deliverable for project OR16-3DTomography-PD3Jb, “3D Tomography and Image Processing Using Fast Neutrons,” to report on the extension of geometric shape-finding algorithms to three dimensions. The project has two overall goals. The first of these goals is to extend associated-particle fast-neutron transmission and, particularly, induced-reaction tomographic imaging algorithms to three dimensions. This aspect of the project is beyond the scope of this report. The second goal is to automatically segment the resultant tomographic images into constituent parts and then extract information about the parts, such as the class of shape and potentially the shape parameters. This report describes progress for the image segmentation part of the project. Imaging techniques have been developed as high-confidence methods for confirming the presence and configuration of special nuclear materials (SNM). This high confidence is achieved at the cost of revealing considerable information that may be undesirable to share with the operator of the equipment. One potential way to make use of the high-confidence of imaging methods without revealing imaging data is to employ automated analysis that can extract meaningful attributes of the SNM without showing imaging data to the operator. An essential step of this automated analysis is the segmentation of the image into its constituent parts. For example, a three-dimensional (3D) fast-neutron tomographic image of an assembly containing uranium could be reconstructed and then segmented into discrete parts. The properties of a constituent part identified as uranium would be of interest and could be extracted. For instance, the volume and density of the part would together determine its mass. Moreover, its shape might be expected to fall into a particular geometric class, such as a cube, cylinder, or sphere, and the parameters of the shape (such as side length, height, or diameter) would be expected to fall in a given range. In such an analysis, automated identification of the boundaries and properties of the uranium part is an essential step. A key goal of the present work is to infer the boundaries of objects in fast-neutron tomographic images via the general method of image segmentation. Development of this capability can be broken into more manageable steps, including 1. operator-guided segmentation of two-dimensional (2D) fast-neutron tomographic images, 2. operator-guided segmentation of 3D fast-neutron tomographic images, 3. development of automated segmentation algorithms, and 4. extraction of shape parameters from constituent volumes in 3D. In the first year, the project team reported progress on the first step. Last year, progress on the second step, the extension into 3D, was reported with application to simulated data. The present report describes additional progress on the second step, and progress toward the third step, automated segmentation algorithms. In particular, a multiphase level set approach for image segmentation in three dimensions has been applied to measured 3D data for the first time. Activities related to the fourth step will also be reported, but it will focus on estimation of 3D volumes rather than shape parameters. In this report, the steps before image segmentation will be described, and the multiphase level sets approach to image segmentation will be briefly reviewed. This will be followed by a discussion of the issues encountered when applying the level set technique to measured 3D transmission images, and then a description of the inclusion of induced-reaction data will be provided. Finally, segmentation results for various measured target assemblies will be presented and the image segmentation portion of this project will be summarized and future directions will be discussed.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2024

Uncertainty quantification and propagation in lithium-ion battery electrodes using bayesian convolutional neural networks

The complex nature of manufacturing processes stipulates electrodes to possess high variability with increased heterogeneity during production. X-ray computed tomography imaging has proved to be critical in visualizing the complicated stochastic particle distribution of as-manufactured electrodes in lithium-ion batteries. However, accurate prediction of their electrochemical performance necessitates precise evaluation of kinetic and transport properties from real electrodes. Image segmentation that characterizes voxels to particle/pore phase is often meticulous and fraught with subjectivity owing to a myriad of unconstrained choices and filter algorithms. Here we utilize a Bayesian convolutional neural network to tackle segmentation subjectivity and quantify its pertinent uncertainties. Otsu inter-variance and Blind/Referenceless Imaging Spatial Quality Evaluator are used to assess the relative image quality of grayscale tomograms, thus evaluating the uncertainty in the derived microstructural attributes. We analyze how image uncertainty is correlated with the uncertainties and magnitude of kinetic and transport properties of an electrode, further identifying pathways of uncertainty propagation within microstructural attributes. The coupled effect of spatial heterogeneity and microstructural anisotropy on the uncertainty quantification of transport parameters is also understood. This work demonstrates a novel methodology to extract microstructural descriptors from real electrode images through quantification of associated uncertainties and discerning the relative strength of their propagation, thus facilitating feedback to manufacturing processes from accurate image based electrochemical simulations.

25 ENERGY STORAGE↗
NASA NTRS1990

Real-time image processing II; Proceedings of the Meeting, Orlando, FL, Apr. 16-18, 1990

The present conference discusses topics in the fields of feature extraction and implementation, filter and correlation algorithms, optical correlators, high-level algorithms, and digital image processing for ranging and remote driving. Attention is given to a nonlinear filter derived from topological image features, IR image segmentation through iterative thresholding, orthogonal subspaces for correlation masking, composite filter trees and image recognition via binary search, and features of matrix-coherent optical image processing. Also discussed are multitarget tracking via hybrid joint transform correlator, binary joint Fourier transform correlator considerations, global image processing operations on parallel architectures, real-time implementation of a differential range finder, and real-time binocular stereo range and motion detection.

Juday, Richard D.↗
DOE OSTI2019

Juxtatumoral perinephric fat analysis in clear cell renal cell carcinoma

Purpose: The purpose of the study was to evaluate the feasibility of using contrast-enhanced computed tomography (CECT)-based texture analysis (CTTA) metrics to differentiate between juxtatumoral perinephric fat (JPF) surrounding low-grade (ISUP 1–2) versus high-grade (ISUP 3–4) clear cell renal cell carcinoma (ccRCC). Methods: In this IRB-approved study, we retrospectively queried the surgical database between June 2009 and April 2016 and identified 83 patients with pathologically confirmed ccRCC (low grade: n = 54, mean age = 61.5 years, 18F/35M; high grade n = 30, mean age = 61.7 years, 8F/22M) who also had pre-operative multiphase CT acquisitions. CT images were transferred to a 3D workstation, and nephrographic phase JPF regions were manually segmented. Using an in-house developed Matlab program, a CTTA panel comprising of texture metrics extracted using six different methods, histogram, 2D- and 3D-Gray-level co-occurrence matrix (GLCM) and Gray-level difference matrix (GLDM), and 2D-Fast Fourier Transform (FFT) analyses, was applied to the segmented images to assess JPF textural heterogeneity in low- versus high-grade ccRCC. Univariate analysis and receiver-operator characteristics (ROC) analysis were used to assess interclass differences in texture metrics and their prediction accuracy, respectively. Results: All methods except GLCM consistently revealed increased heterogeneity in the JPF surrounding high- versus low-grade ccRCC. FFT showed increased complexity index (p < 0.01). Histogram analysis showed increased kurtosis and positive skewness in (p < 0.03), and GLDM analysis showed decreased measure of correlation coefficient (MCC) (p < 0.04). Several of the GLCM metrics showed statistically significant (p < 0.04) textural differences between the two groups, but with no consistent trend. ROC analysis showed that MCC in GLCM analysis had an area under the curve of 0.75. Conclusions: Our study suggests that CTTA of ccRCC shows statistically significant textural differences in JPF surrounding high- versus low-grade ccRCC.

62 RADIOLOGY AND NUCLEAR MEDICINE↗
DOE OSTI2023

Deep Deception: Exemplars of Adversarial Machine Learning and Countermeasures Applicable to International Safeguards

As a follow-up to our more comprehensive report on Adversarial Machine Learning (AML), here we provide demonstrations of AML attacks against the Limbo image database of UF6 cylinders in a variety of orientations and amongst a variety of distractor images. We demonstrate the Carlini & Wagner AML attack against a subset of Limbo images, with 100% attack success rate; meaning all attacked images were misclassified by a highly accurate trained model, yet the image changes were imperceptible to the human eye. We also demonstrate successful attacks against segmented images (images with more than one targeted object). Finally, we demonstrated the Fast Fourier Transform countermeasure that can be used to detect AML attacks on images. The intent of this and our previous report is to inform the IAEA and stakeholders of both the promise of machine learning, which could greatly improve the efficiency of surveillance monitoring, but also of the real threat of AML and potential defenses.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS1995

Image Processing for Planetary Limb/Terminator Extraction

A novel image segmentation technique for extracting limb and terminator of planetary bodies is proposed. Conventional edge- based histogramming approaches are used to trace object boundaries. The limb and terminator bifurcation is achieved by locating the harmonized segment in the two equations representing the 2-D parameterized boundary curve. Real planetary images from Voyager 1 and 2 served as representative test cases to verify the proposed methodology.

image processing planetary limb planetary terminat↗
NASA NTRS2011

Marker-Based Hierarchical Segmentation and Classification Approach for Hyperspectral Imagery

The Hierarchical SEGmentation (HSEG) algorithm, which is a combination of hierarchical step-wise optimization and spectral clustering, has given good performances for hyperspectral image analysis. This technique produces at its output a hierarchical set of image segmentations. The automated selection of a single segmentation level is often necessary. We propose and investigate the use of automatically selected markers for this purpose. In this paper, a novel Marker-based HSEG (M-HSEG) method for spectral-spatial classification of hyperspectral images is proposed. First, pixelwise classification is performed and the most reliably classified pixels are selected as markers, with the corresponding class labels. Then, a novel constrained marker-based HSEG algorithm is applied, resulting in a spectral-spatial classification map. The experimental results show that the proposed approach yields accurate segmentation and classification maps, and thus is attractive for hyperspectral image analysis.

Tarabalka, Yuliya↗
DOE OSTI2014

TH-A-BRF-11: Image Intensity Non-Uniformities Between MRI Simulation and Diagnostic MRI

Purpose: MRI simulation for MRI-based radiotherapy demands that patients be setup in treatment position, which frequently involves use of alternative radiofrequency (RF) coil configurations to accommodate immobilized patients. However, alternative RF coil geometries may exacerbate image intensity non-uniformities (IINU) beyond those observed in diagnostic MRI, which may challenge image segmentation and registration accuracy as well as confound studies assessing radiotherapy response when MR simulation images are used as baselines for evaluation. The goal of this work was to determine whether differences in IINU exist between MR simulation and diagnostic MR images. Methods: ACR-MRI phantom images were acquired at 3T using a spin-echo sequence (TE/TR:20/500ms, rBW:62.5kHz, TH/skip:5/5mm). MR simulation images were obtained by wrapping two flexible phased-array RF coils around the phantom. Diagnostic MR images were obtained by placing the phantom into a commercial phased-array head coil. Pre-scan normalization was enabled in both cases. Images were transferred offline and corrected for IINU using the MNI N3 algorithm. Coefficients of variation (CV=σ/μ) were calculated for each slice. Wilcoxon matched-pairs and Mann-Whitney tests compared CV values between original and N3 images and between MR simulation and diagnostic MR images. Results: Significant differences in CV were detected between original and N3 images in both MRI simulation and diagnostic MRI groups (p=0.010, p=0.010). In addition, significant differences in CV were detected between original MR simulation and original and N3 diagnostic MR images (p=0.0256, p=0.0016). However, no significant differences in CV were detected between N3 MR simulation images and original or N3 diagnostic MR images, demonstrating the importance of correcting MR simulation images beyond pre-scan normalization prior to use in radiotherapy. Conclusions: Alternative RF coil configurations used in MRI simulation can Result in significant IINU differences compared to diagnostic MR images. The MNI N3 algorithm reduced MR simulation IINU to levels observed in diagnostic MR images. Funding provided by Advancing a Healthier Wisconsin.

60 APPLIED LIFE SCIENCES↗
NASA NTRS2020

Medics: Medical Decision Support System for Long-Duration Space Exploration

The Autonomous Medical Operations (AMO) group at NASA Ames is developing a “medical decision support system” to enable astronauts on long-duration exploration missions to operate autonomously. The system will support clinical actions by providing medical interpretation advice and procedural recommendations during emergent care and clinical work performed by crew. The current state of development of the system, called MedICS (Medical Interpretation Classification and Segmentation) includes two separate aspects: a set of machine learning diagnostic models trained to analyze organ images and patient health records, and an interface to ultrasound diagnostic hardware and to medical repositories. Three sets of images of different organs and medical records were utilized for training machine learning models for various analyses, as follows: 1. Pneumothorax condition (collapsed lung). The trained model provides a positive or negative diagnosis of the condition. 2. Carotid artery occlusion. The trained model produces a diagnosis of 5 different occlusion levels (including “normal”). 3. Ocular retinal images. The model extracts optic disc pixels (image segmentation). This is a precursor step for advanced autonomous fundus clinical evaluation algorithms to be implemented in FY20. 4. Medical health records. The model produces a differential diagnosis for any particular individual, based on symptoms and other health and demographic information. A probability is calculated for each of 25 most common conditions. The same model provides the likelihood of survival. All results are provided with a confidence level. Item 1 images were provided by the US Army and were part of a data set for the clinical treatment of injured battlefield soldiers. This condition is relevant to possible space mishaps, due to pressure management issues. Item 2 images were provided by Houston Methodist Hospital, and item 3 health records were acquired from the MIT laboratory of computational physiology. The machine learning technology utilized is deep multilayer networks (Deep Learning), and new models will continue to be produced, as relevant data is made available and specific health needs of astronaut crews are identified. The interfacing aspects of the system include a GUI for running the different models, and retrieving and storing data, as well as support for integration with an augmented reality (AR) system deployed at JSC by Tietronix Software Inc. (HoloLens). The AR system provides guidance for the placement of an ultrasound transducer that captures images to be sent to the MedICS system for diagnosis. The image captured and the associated diagnosis appear in the technician’s AR visual display.

Colombano, Silvano↗
DOE OSTI2014

MINER - A Mobile Imager of Neutrons for Emergency Responders

We have developed a mobile fast neutron imaging platform to enhance the capabilities of emergency responders in the localization and characterization of special nuclear material. This mobile imager of neutrons for emergency responders (MINER) is based on the Neutron Scatter Camera, a large segmented imaging system that was optimized for large-area search applications. Due to the reduced size and power requirements of a man-portable system, MINER has been engineered to fit a much smaller form factor, and to be operated from either a battery or AC power. We chose a design that enabled omnidirectional (4π) imaging, with only a ~twofold decrease in sensitivity compared to the much larger neutron scatter cameras. The system was designed to optimize its performance for neutron imaging and spectroscopy, but it does also function as a Compton camera for gamma imaging. This document outlines the project activities, broadly characterized as system development, laboratory measurements, and deployments, and presents sample results in these areas. Additional information can be found in the documents that reside in WebPMIS.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2024

Pore-scale visualization of natural hydrate-bearing sediments

Accurate modeling of gas hydrate reservoir productivity and geomechanical risks associated with subsurface dissociation of natural gas hydrates (NGH) requires the determination of model parameters through physical testing on natural hydrate-bearing sediments (HBS). This involves investigating the hydro-mechanical behavior of undisturbed hydrate samples from nature under in situ conditions using pressure core characterization and analysis, which provides a unique opportunity for research. By employing state-of-the-art micro computed tomography imagery on cryogenically preserved, hydrate-bearing sediment samples, we can determine hydrate saturation as well as permeability with and without the presence of hydrates in the sediment. Furthermore, utilizing a machine learning based image segmentation technique, it is possible to extract pore space and grain information. Subsections of the entire image volume were used to determine anisotropic permeabilities using a finite-difference method Stokes solver (FDMSS). Additionally, permeability measurements on whole pressure and temperature preserved hydrate-bearing core were analyzed by utilizing the National Energy Technology Laboratory’s (NETL) Pressure Core Characterization and X-ray CT Visualization Tool (PCXT) to manipulate, cut, and analyze pressure preserved sediment. Permeabilities were measured under a broad range of vertical stress states to simulate expected pressure changes during production scenarios, and the results show that permeabilities derived from images are in agreement with those from traditional core derived experiments. The collected stress-dependent permeability, permeability anisotropy, and corresponding gas hydrate saturations provide valuable input into numerical simulations of reservoir productivity. These properties have been proven to be key parameters determining a long-term reservoir response under depressurization.

Liu, Mengwei [Oak Ridge Institute for Science and ↗
NASA NTRS2017

Vascular Patterning Analysis by VESGEN 2D/3D with Bioinformatics: Updates for Rodent Tissues

Fractally branching vascular systems are a complex physiological requirement shared by humans with all higher terrestrial life forms, including other vertebrates, insects, and higher land plants. Vascular trees, networks, and tree-network composites are therefore mapped and quantified by the VESsel GENeration Analysis (VESGEN) software according to weighted physiological vascular rules that include vessel connectivity, tapering and bifurcational branching. According to fluid dynamics, successful vascular transport depends upon a complex distributed system of highly regulated laminar flow. VESGEN has elucidated changes in vascular patterning resulting from inflammatory, developmental and other signaling pathways within numerous tissues of major model organisms important for Space Biology, especially for rodents. Important early stage regenerative opportunities have been identified by VESGEN vascular analysis for visual impairments in the human retina, and is currently being used for research into astronaut visual and ocular disorders associated with long duration missions. The VESGEN 2D software is a mature, automated, widely published capability for which beta testing and public release by NASA is planned for the upcoming year. Early-stage capabilities for VESGEN 3D analysis are under development for the rodent retina and intestine as prototype tissues. A prototype VESGEN 2D Bioinformatics software capability has also been developed to associate phenotypic changes in molecular expression with vascular structure and function. By new VESGEN bioinformatic innovations, expression patterns of the genetic, transcriptional, protein and other markers for regulatory molecules such as vascular endothelial growth factor (VEGF) and their receptors, often indicators of tissue oxygenation status, are co-localized with alterations in vascular pattern. Biomarkers are therefore mapped and quantified as information dimensions directly correlated with the spatial dimensions of a vascular pattern. Further important technology innovations by NASA include substantial image segmentation advances for more automated binary extraction of the grayscale vascular patterns, together with informative associated image quality assessments. Vascular mapping and quantification capabilities for the rodent retina and intestine are illustrated for VESGEN 2D, along with technology status reports on VESGEN 3D and Bioinformatic capabilities. Research partially supported by Ames Center Innovation Awards.

Parsons-Wingerter, P.↗
DOE OSTI2014

WE-G-18C-05: Characterization of Cross-Vendor, Cross-Field Strength MR Image Intensity Variations

Purpose: Variations in MR image intensity and image intensity nonuniformity (IINU) can challenge the accuracy of intensity-based image segmentation and registration algorithms commonly applied in radiotherapy. The goal of this work was to characterize MR image intensity variations across scanner vendors and field strengths commonly used in radiotherapy. Methods: ACR-MRI phantom images were acquired at 1.5T and 3.0T on GE (450w and 750, 23.1), Siemens (Espree and Verio, VB17B), and Philips (Ingenia, 4.1.3) scanners using commercial spin-echo sequences with matched parameters (TE/TR: 20/500 ms, rBW: 62.5 kHz, TH/skip: 5/5mm). Two radiofrequency (RF) coil combinations were used for each scanner: body coil alone, and combined body and phased-array head coils. Vendorspecific B1- corrections (PURE/Pre-Scan Normalize/CLEAR) were applied in all head coil cases. Images were transferred offline, corrected for IINU using the MNI N3 algorithm, and normalized. Coefficients of variation (CV=σ/μ) and peak image uniformity (PIU = 1−(Smax−Smin)/(Smax+Smin)) estimates were calculated for one homogeneous phantom slice. Kruskal-Wallis and Wilcoxon matched-pairs tests compared mean MR signal intensities and differences between original and N3 image CV and PIU. Results: Wide variations in both MR image intensity and IINU were observed across scanner vendors, field strengths, and RF coil configurations. Applying the MNI N3 correction for IINU resulted in significant improvements in both CV and PIU (p=0.0115, p=0.0235). However, wide variations in overall image intensity persisted, requiring image normalization to improve consistency across vendors, field strengths, and RF coils. These results indicate that B1- correction routines alone may be insufficient in compensating for IINU and image scaling, warranting additional corrections prior to use of MR images in radiotherapy. Conclusions: MR image intensities and IINU vary as a function of scanner vendor, field strength, and RF coil configuration. A two-step strategy consisting of MNI N3 correction followed by normalization was required to improve MR image consistency. Funding provided by Advancing a Healthier Wisconsin.

60 APPLIED LIFE SCIENCES↗