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 1,081 records · Page 60

Segmentation of tomography datasets using 3D convolutional neural networks

Dendritic microstructures are ubiquitous in nature and are the primary solidification morphologies in metallic materials. Techniques such as X-ray computed tomography (XCT) have provided new insights into dendritic phase transformation phenomena. However, manual identification of dendritic morphologies in microscopy data can be both labor intensive and potentially ambiguous. The analysis of 3D datasets is particularly challenging due to their large sizes (terabytes) and the presence of artifacts scattered within the imaged volumes. Here, in this study, we trained 3D convolutional neural networks (CNNs) to segment 3D datasets. Three CNN architectures were investigated, including a new version of FCDenseNet which we extended to 3D. We show that using hyperparameter optimization (HPO) and fine-tuning techniques, both 2D and 3D CNN architectures outperform the previous state of the art. The 3D U-Net architecture trained in this study produced the best segmentations according to quantitative metrics (intersection-over-union of 95.56% and a boundary displacement error of 0.58 pixels), while 3D FCDense produced the smoothest boundaries and best segmentations according to visual inspection. The trained 3D CNNs are able to segment entire 852 × 852 × 250 voxel 3D volumes in only ~60 s, thus hastening the progress towards a deeper understanding of phase transformation phenomena such as dendritic solidification.

36 MATERIALS SCIENCE↗

3D Imaging of Microbial Biofilms: Integration of Synchrotron Imaging and an Interactive Visualization Interface

Understanding the interactions of structured communities known as “biofilms” and other complex matrixes is possible through the X-ray micro tomography imaging of the biofilms. Feature detection and image processing for this type of data focuses on efficiently identifying and segmenting biofilms and bacteria in the datasets. The datasets are very large and often require manual interventions due to low contrast between objects and high noise levels. Thus new software is required for the effectual interpretation and analysis of the data. This work specifies the evolution and application of the ability to analyze and visualize high resolution X-ray micro tomography datasets.

Biofilm Viewer↗

3D Monte Carlo simulation of backscattered electron signal variation across pore-solid boundaries in cement-based materials

Three-dimensional (3D) Monte Carlo simulation was used to study the variation of backscattered electron (BSE) signal across pore-solid boundaries in cement-based materials in order to enhance quantitative analysis of pore structure. The effects of pore size, depth and boundary inclination angle were investigated. It is found that pores down to 1 nm can generate sufficient contrast to be detected. Visibility improves with larger pore size, smaller beam probe size and lower acceleration voltage. However, pixels in shallow pores or near pore boundaries display higher grey values (brightness) than expected due to sampling sub-surface or neighbouring solid material. Thus, cement-based materials may appear less porous or the pores appear smaller than they actually are in BSE images. Simulated BSE images were used to test the accuracy of the Overflow pore segmentation method. Results show the method is generally valid and gives low errors for pores that are 1 μm and greater.

36 MATERIALS SCIENCE↗

SU-F-J-22: Lung VolumeVariability Assessed by Bh-CBCT in 3D Surface Image Guided Deep InspirationBreath Hold (DIBH) Radiotherapy for Left-Sided Breast Cancer

Purpose: With the increasing use of DIBH techniques for left-sided breast cancer, 3D surface-image guided DIBH techniques have improved patient setup and facilitated DIBH radiation delivery. However, quantification of the daily separation between the heart and left breast still presents a challenge. One method of assuring separation is to ensure consistent left lung filling. With this in mind, the aim of this study is to retrospectively quantify left lung volume from weekly breath hold-CBCTs (bh-CBCT) of left-sided breast patients treated using a 3D surface imaging system. Methods: Ten patients (n=10) previously treated to the left breast using the C-Rad CatalystHD system (C-RAD AG, Uppsala Sweden) were evaluated. Patients were positioned with CatalystHD and with bh-CBCT. bh-CBCTs were acquired at the validation date, first day of treatment and at subsequent weekly intervals. Total treatment courses spanned from 3 to 5 weeks. bh-CBCT images were exported to VelocityAI and the left lung volume was segmented. Volumes were recorded and analyzed. Results: A total of 41 bh-CBCTs were contoured in VelocityAI for the 10 patients. The mean left lung volume for all patients was 1657±295cc based on validation bh-CBCT. With the subsequent lung volumes normalized to the validation lung volume, the mean relative ratios for all patients were 1.02±0.11, 0.97±0.14, 0.98±0.11, 1.02±0.01, and 0.96±0.02 for week 1, 2, 3, 4, and 5, respectively. Overall, the mean left lung volume change was ≤4.0% over a 5-week course; however left lung volume variations of up to 28% were noted in a select patient. Conclusion: With the use of the C-RAD CatalystHD system, the mean lung volume variability over a 5-week course of DIBH treatments was ≤4.0%. By minimizing left lung volume variability, heart to left breast separation maybe more consistently maintained. AN Gutierrez has a research grant from C-RAD AG.

60 APPLIED LIFE SCIENCES↗

GraphEM

An unsupervised segmentation method for structural analysis of Scanning Transmission Electron Microscopy (STEM) images

Pope, Jenna↗

Molecular Predictors of 3D Morphogenesis by Breast Cancer Cell Lines in 3D Culture

Correlative analysis of molecular markers with phenotypic signatures is the simplest model for hypothesis generation. In this paper, a panel of 24 breast cell lines was grown in 3D culture, their morphology was imaged through phase contrast microscopy, and computational methods were developed to segment and represent each colony at multiple dimensions. Subsequently, subpopulations from these morphological responses were identified through consensus clustering to reveal three clusters of round, grape-like, and stellate phenotypes. In some cases, cell lines with particular pathobiological phenotypes clustered together (e.g., ERBB2 amplified cell lines sharing the same morphometric properties as the grape-like phenotype). Next, associations with molecular features were realized through (i) differential analysis within each morphological cluster, and (ii) regression analysis across the entire panel of cell lines. In both cases, the dominant genes that are predictive of the morphological signatures were identified. Specifically, PPAR? has been associated with the invasive stellate morphological phenotype, which corresponds to triple-negative pathobiology. PPAR? has been validated through two supporting biological assays.

3D cell culture models, breast cancer cell lines, ↗

Single-cell correlations of mRNA and protein content in a human monocytic cell line after LPS stimulation

The heterogeneity of mRNA and protein expression at the single-cell level can reveal fundamental information about cellular response to external stimuli, including the sensitivity, timing, and regulatory interactions of genes. Here we describe a fully automated system to digitally count the intron, mRNA, and protein content of up to five genes of interest simultaneously in single-cells. Full system automation of 3D microscope scans and custom image analysis routines allows hundreds of individual cells to be automatically segmented and the mRNA-protein content to be digitally counted. Single-molecule intron and mRNA content is measured by single-molecule fluorescence in-situ hybridization (smFISH), while protein content is quantified though the use of antibody probes. To mimic immune response to bacterial infection, human monocytic leukemia cells (THP-1) were stimulated with lipopolysaccharide (LPS), and the expression of two inflammatory genes, IL1β (interleukin 1β) and TNF-α (tumor necrosis factor α), were simultaneously quantified by monitoring the intron, mRNA, and protein levels over time. The simultaneous labeling of cellular content allowed for a series of correlations at the single-cell level to be explored, both in the progressive maturation of a single gene (intron-mRNA-protein) and comparative analysis between the two immune response genes. In the absence of LPS stimulation, mRNA expression of IL1β and TNF-α were uncorrelated. Following LPS stimulation, mRNA expression of the two genes became more correlated, consistent with a model in which IL1β and TNF-α upregulation occurs in parallel through independent mechanistic pathways. This smFISH methodology can be applied to different complex biological systems to provide valuable insight into highly dynamic gene mechanisms that determine cell plasticity and heterogeneity of cellular response.

59 BASIC BIOLOGICAL SCIENCES↗

WE-AB-303-05: Breathing Motion of Liver Segments From Fiducial Tracking During Robotic Radiosurgery and Comparison with 4D-CT-Derived Fiducial Motion

Purpose: To quantify respiratory-induced motion of liver segments using the positions of implanted fiducials during robotic radiosurgery. This study also compared fiducial motion derived from four-dimensional computed tomography (4D-CT) maximum intensity projections (MIP) with motion derived from imaging during treatment. Methods: Forty-two consecutive liver patients treated with liver ablative radiotherapy were accrued to an ethics approved retrospective study. The liver segment in which each fiducial resided was identified. Fiducial positions throughout each treatment fraction were determined using orthogonal kilovoltage images. Any data due to patient repositioning or motion was removed. Mean fiducial positions were calculated. Fiducial positions beyond two standard deviations of the mean were discarded and remaining positions were fit to a line segment using least squares minimization (LSM). For eight patients, fiducial motion was derived from 4D-CT MIPs by calculating the CT number weighted mean position of the fiducial on each slice and fitting a line segment to these points using LSM. Treatment derived fiducial trajectories were corrected for patient rotation and compared to MIP derived trajectories. Results: The mean total magnitude of fiducial motion across all liver segments in left-right, anteroposterior, and superoinferior (SI) directions were 3.0 ± 0.2 mm, 9.3 ± 0.4 mm, and 20.5 ± 0.5 mm, respectively. Differences in per-segment mean fiducial motion were found with SI motion ranging from 12.6 ± 0.8 mm to 22.6 ± 0.9 mm for segments 3 and 8, respectively. Large, varied differences between treatment and MIP derived motion at simulation were found with the mean difference for SI motion being 2.6 mm (10.8 mm standard deviation). Conclusion: The magnitude of liver fiducial motion was found to differ by liver segment. MIP derived liver fiducial motion differed from motion observed during treatment, implying that 4D-CTs may not accurately capture the range of liver motion across fractions and during treatment. Author V. Nair was funded by the Cushing estate for a SABR clinical research fellowship.

60 APPLIED LIFE SCIENCES↗

A new paradigm in electron microscopy: Automated microstructure analysis utilizing a dynamic segmentation convolutional neutral network

Over the past half century, the transmission electron microscope enabled insight into the fundamental arrangements and structures of materials. State-of-the-art electron microscopes can acquire large image datasets across multiple imaging modalities. However, the manual annotation process for feature or defect quantification may not be feasible with the modern microscope. Convolutional neural networks emerged to characterize individual microstructural features from an image in a cost-effective, consistent manner. However, many of these neural network approaches rely on thousands to hundreds of thousands of manual annotations of each feature type across hundreds of images to train the network for adequate performance. This work focused on the development and application of a pixel-wise defect detection machine-learning dynamic segmentation convolutional neural network with associated automated acquisition and postprocessing to identify microstructural features rapidly and quantitatively from a small initial dataset incorporating multiple imaging modes. The approach was demonstrated for characterization of superalloy 718 from both single image acquisition on multiple detectors to in-situ evolution captured with a single detector on a standard desktop computer to demonstrate the low barrier to entry required for widespread adoption. Pixel-by-pixel class identification was excellent with strong identification of chemically distinct phases, structurally distinct phases, and defect structures, thus demonstrating the new paradigm of machine learning-assisted characterization.

36 MATERIALS SCIENCE↗

Annular beam with segmented phase gradients

An annular beam with a single uniform-intensity ring and multiple segments of phase gradients is proposed in this paper. Different from the conventional superposed vortices, such as the modulated optical vortices and the collinear superposition of multiple orbital angular momentum modes, the designed annular beam has a doughnut intensity distribution whose radius is independent of the phase distribution of the beam in the imaging plane. The phase distribution along the circumference of the doughnut beam can be segmented with different phase gradients. Similar to a vortex beam, the annular beam can also exert torques and rotate a trapped particle owing to the orbital angular momentum of the beam. As the beam possesses different phase gradients, the rotation velocity of the trapped particle can be varied along the circumference. The simulation and experimental results show that an annular beam with three segments of different phase gradients can rotate particles with controlled velocities. The beam has potential applications in optical trapping and optical information processing.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

2D-to-3D image translation of complex nanoporous volumes using generative networks

Image-based characterization offers a powerful approach to studying geological porous media at the nanoscale and images are critical to understanding reactive transport mechanisms in reservoirs relevant to energy and sustainability technologies such as carbon sequestration, subsurface hydrogen storage, and natural gas recovery. Nanoimaging presents a trade off, however, between higher-contrast sample-destructive and lower-contrast sample-preserving imaging modalities. Furthermore, high-contrast imaging modalities often acquire only 2D images, while 3D volumes are needed to characterize fully a source rock sample. In this work, we present deep learning image translation models to predict high-contrast focused ion beam-scanning electron microscopy (FIB-SEM) image volumes from transmission X-ray microscopy (TXM) images when only 2D paired training data is available. We introduce a regularization method for improving 3D volume generation from 2D-to-2D deep learning image models and apply this approach to translate 3D TXM volumes to FIB-SEM fidelity. We then segment a predicted FIB-SEM volume into a flow simulation domain and calculate the sample apparent permeability using a lattice Boltzmann method (LBM) technique. Results show that our image translation approach produces simulation domains suitable for flow visualization and allows for accurate characterization of petrophysical properties from non-destructive imaging data.

58 GEOSCIENCES↗

Automated detection of vessel lumen and stent struts in intravascular optical coherence tomography to evaluate stent apposition and neointimal coverage

Purpose: Intravascular optical coherence tomography (IV-OCT) is a high-resolution imaging method used to visualize the microstructure of arterial walls in vivo. IV-OCT enables the clinician to clearly observe and accurately measure stent apposition and neointimal coverage of coronary stents, which are associated with side effects such as in-stent thrombosis. In this study, the authors present an algorithm for quantifying stent apposition and neointimal coverage by automatically detecting lumen contours and stent struts in IV-OCT images. Methods: The algorithm utilizes OCT intensity images and their first and second gradient images along the axial direction to detect lumen contours and stent strut candidates. These stent strut candidates are classified into true and false stent struts based on their features, using an artificial neural network with one hidden layer and ten nodes. After segmentation, either the protrusion distance (PD) or neointimal thickness (NT) for each strut is measured automatically. In randomly selected image sets covering a large variety of clinical scenarios, the results of the algorithm were compared to those of manual segmentation by IV-OCT readers. Results: Stent strut detection showed a 96.5% positive predictive value and a 92.9% true positive rate. In addition, case-by-case validation also showed comparable accuracy for most cases. High correlation coefficients (R > 0.99) were observed for PD and NT between the algorithmic and the manual results, showing little bias (0.20 and 0.46 μm, respectively) and a narrow range of limits of agreement (36 and 54 μm, respectively). In addition, the algorithm worked well in various clinical scenarios and even in cases with a low level of stent malapposition and neointimal coverage. Conclusions: The presented automatic algorithm enables robust and fast detection of lumen contours and stent struts and provides quantitative measurements of PD and NT. In addition, the algorithm was validated using various clinical cases to demonstrate its reliability. Therefore, this technique can be effectively utilized for clinical trials on stent-related side effects, including in-stent thrombosis and in-stent restenosis.

60 APPLIED LIFE SCIENCES↗

SU-F-J-211: Scatter Correction for Clinical Cone-Beam CT System Using An Optimized Stationary Beam Blocker with a Single Scan

Purpose: X-ray scatter photons result in significant image quality degradation of cone-beam CT (CBCT). Measurement based algorithms using beam blocker directly acquire the scatter samples and achieve significant improvement on the quality of CBCT image. Within existing algorithms, single-scan and stationary beam blocker proposed previously is promising due to its simplicity and practicability. Although demonstrated effectively on tabletop system, the blocker fails to estimate the scatter distribution on clinical CBCT system mainly due to the gantry wobble. In addition, the uniform distributed blocker strips in our previous design results in primary data loss in the CBCT system and leads to the image artifacts due to data insufficiency. Methods: We investigate the motion behavior of the beam blocker in each projection and design an optimized non-uniform blocker strip distribution which accounts for the data insufficiency issue. An accurate scatter estimation is then achieved from the wobble modeling. Blocker wobble curve is estimated using threshold-based segmentation algorithms in each projection. In the blocker design optimization, the quality of final image is quantified using the number of the primary data loss voxels and the mesh adaptive direct search algorithm is applied to minimize the objective function. Scatter-corrected CT images are obtained using the optimized blocker. Results: The proposed method is evaluated using Catphan@504 phantom and a head patient. On the Catphan©504, our approach reduces the average CT number error from 115 Hounsfield unit (HU) to 11 HU in the selected regions of interest, and improves the image contrast by a factor of 1.45 in the high-contrast regions. On the head patient, the CT number error is reduced from 97 HU to 6 HU in the soft tissue region and image spatial non-uniformity is decreased from 27% to 5% after correction. Conclusion: The proposed optimized blocker design is practical and attractive for CBCT guided radiation therapy. This work is supported by grants from Guangdong Innovative Research Team Program of China (Grant No. 2011S013), National 863 Programs of China (Grant Nos. 2012AA02A604 and 2015AA043203), the National High-tech R&D Program for Young Scientists by the Ministry of Science and Technology of China (Grant No. 2015AA020917)

60 APPLIED LIFE SCIENCES↗

SU-E-I-87: Automated Liver Segmentation Method for CBCT Dataset by Combining Sparse Shape Composition and Probabilistic Atlas Construction

Purpose: The aiming of this study was to extract liver structures for daily Cone beam CT (CBCT) images automatically. Methods: Datasets were collected from 50 intravenous contrast planning CT images, which were regarded as training dataset for probabilistic atlas and shape prior model construction. Firstly, probabilistic atlas and shape prior model based on sparse shape composition (SSC) were constructed by iterative deformable registration. Secondly, the artifacts and noise were removed from the daily CBCT image by an edge-preserving filtering using total variation with L1 norm (TV-L1). Furthermore, the initial liver region was obtained by registering the incoming CBCT image with the atlas utilizing edge-preserving deformable registration with multi-scale strategy, and then the initial liver region was converted to surface meshing which was registered with the shape model where the major variation of specific patient was modeled by sparse vectors. At the last stage, the shape and intensity information were incorporated into joint probabilistic model, and finally the liver structure was extracted by maximum a posteriori segmentation.Regarding the construction process, firstly the manually segmented contours were converted into meshes, and then arbitrary patient data was chosen as reference image to register with the rest of training datasets by deformable registration algorithm for constructing probabilistic atlas and prior shape model. To improve the efficiency of proposed method, the initial probabilistic atlas was used as reference image to register with other patient data for iterative construction for removing bias caused by arbitrary selection. Results: The experiment validated the accuracy of the segmentation results quantitatively by comparing with the manually ones. The volumetric overlap percentage between the automatically generated liver contours and the ground truth were on an average 88%–95% for CBCT images. Conclusion: The experiment demonstrated that liver structures of CBCT with artifacts can be extracted accurately for following adaptive radiation therapy. This work is supported by National Natural Science Foundation of China (No. 61201441), Research Fund for Excellent Young and Middle-aged Scientists of Shandong Province (No. BS2012DX038), Project of Shandong Province Higher Educational Science and Technology Program (No. J12LN23), Jinan youth science and technology star (No.20120109)

60 APPLIED LIFE SCIENCES↗

THE DYNAMICALLY DISRUPTED GAP IN HD 142527

The vestiges of planet formation have been observed in debris disks harboring young and massive gaseous giants. The process of giant planet formation is terminated by the dissipation of gas in the protoplanetary disk. The gas-rich disk around HD 142527 features a small inner disk, a large gap from {approx}10 to {approx}140 AU, and a massive outer disk extending out to {approx}300 AU. The gap could have been carved out by a giant planet. We have imaged the outer regions of this gap using the adaptive optics camera NICI on Gemini South. Our images reveal that the disk is dynamically perturbed. The outer boundary of the roughly elliptical gap appears to be composed of several segments of spiral arms. The stellar position is offset by 0.''17 {+-} 0.''02 from the centroid of the cavity, consistent with earlier imaging at coarser resolutions. These transient morphological features are expected in the context of disk evolution in the presence of a perturbing body located inside the cavity. We perform hydrodynamical simulations of the dynamical clearing of a gap in a disk. A 10 M{sub jup} body in a circular orbit at r = 90 AU perturbs the whole disks, even after thousands of orbits. By then the model disk has an eccentric and irregular cavity, flanked by tightly wound spiral arms, but it is still evolving far from steady state. A particular transient configuration that is a qualitative match to HD 142527 is seen at 1.7 Myr.

79 ASTRONOMY AND ASTROPHYSICS↗

RadioGalaxyNET: Dataset and novel computer vision algorithms for the detection of extended radio galaxies and infrared hosts

Abstract Creating radio galaxy catalogues from next-generation deep surveys requires automated identification of associated components of extended sources and their corresponding infrared hosts. In this paper, we introduce RadioGalaxyNET, a multimodal dataset, and a suite of novel computer vision algorithms designed to automate the detection and localization of multi-component extended radio galaxies and their corresponding infrared hosts. The dataset comprises 4 155 instances of galaxies in 2 800 images with both radio and infrared channels. Each instance provides information about the extended radio galaxy class, its corresponding bounding box encompassing all components, the pixel-level segmentation mask, and the keypoint position of its corresponding infrared host galaxy. RadioGalaxyNET is the first dataset to include images from the highly sensitive Australian Square Kilometre Array Pathfinder (ASKAP) radio telescope, corresponding infrared images, and instance-level annotations for galaxy detection. We benchmark several object detection algorithms on the dataset and propose a novel multimodal approach to simultaneously detect radio galaxies and the positions of infrared hosts.

Astronomy & Astrophysics↗

Multimodal Nanoscale Tomographic Imaging for Battery Electrodes

Accurate representations of the 3D structure within a lithium-ion battery are key to understanding performance limitations. However, obtaining exact reconstructions of electrodes, where the active particles, the carbon black and polymeric binder domain, and the pore space are visualized is challenging. In this work, it is shown that multimodal imaging can be used to overcome this challenge. High-resolution ptychographic X-ray computed tomography are combined with lower resolution but higher contrast transmission X-ray tomographic microscopy to obtain 3D reconstructions of pristine and cycled graphite-silicon composite electrodes. This cross-correlation enables quantitative analysis of the surface of active particles, including the heterogeneity of carbon-black and binder domain and solid-electrolyte interphase coverage. Capturing the active particles as well as the carbon black-binder domain allows using these segmented structures for electrochemical simulations to highlight the influence of the particle embedding on local state of charge heterogeneities.

25 ENERGY STORAGE↗