Search NASA⌕ Search

SEARCH · Search NASA

Results for “image based modeling”

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 703 records · Page 39
All sourcesNASA NTRSDOE OSTI
DOE OSTI2023

Autonomous Molecular Structure Imaging with High-Resolution Atomic Force Microscopy for Molecular Mixture Discovery

Due to its single-molecule sensitivity, high-resolution atomic force microscopy (HR-AFM) has proved to be a valuable and uniquely advantageous tool to study complex molecular mixtures, which hold promise for developing clean energy and achieving environmental sustainability. However, significant challenges remain to achieve the full potential of the sophisticated and time-consuming experiments. Automation combined with machine learning (ML) and artificial intelligence (AI) is key to overcoming these challenges. Here we present Auto-HR-AFM, an AI tool to automatically collect HR-AFM images of petroleum-based mixtures. In this study, we trained an instance segmentation model to teach Auto-HR-AFM how to recognize features in HR-AFM images. Auto-HR-AFM then uses that information to optimize the imaging by adjusting the probe-molecule distance for each molecule in the run. Auto-HR-AFM is the initial tool that will lead to fully automated scanning probe microscopy (SPM) experiments, from start to finish. This automation will allow SPM to become a mainstream characterization technique for complex mixtures, an otherwise unattainable target.

36 MATERIALS SCIENCE↗
DOE OSTI2022

Multisource Mobile Transfer Learning Algorithm Based on Dynamic Model Compression

With the development of the Internet of Things, the application of computer vision on mobile phones is becoming more and more extensive and people have higher and higher requirements for the timeliness of the recognition results returned and the processing capabilities of the mobile phone for image recognition. However, the processing capability and storage capability of the user terminal equipment cannot meet the needs of identifying and storing a large number of pictures, and the data transmission process will cause high energy consumption of the terminal equipment. At the same time, multisource deep transfer learning has outstanding performance in computer vision and image classification. However, due to the huge amount of calculation of the deep network model, it is impossible to use the existing excellent network model to realize image recognition and classification on the mobile terminal. In order to solve the abovementioned problems, we propose a multisource mobile transfer learning algorithm based on dynamic model compression, this algorithm considers the realization of multisource transfer learning computing in the case of multiple mobile device computing source domains, and the method also guarantees data privacy and security for each device (origin domain). Meanwhile, extensive experiments show that our method can achieve remarkable results in popular image classification datasets.

Gao, Peng↗
DOE OSTI2023

Laser ultrasonic imaging of subsurface defects with the linear sampling method

Laser ultrasonics is a remote nondestructive evaluation technique suitable for real-time monitoring of fabrication processes in semiconductor metrology, advanced manufacturing, and other applications where non-contact, high fidelity measurements are required. Here we investigate laser ultrasonic data processing approaches to reconstruct images of subsurface side drilled holes in aluminum alloy specimens. We demonstrate through simulation that the model-based linear sampling method (LSM) can perform accurate shape reconstruction of single and multiple holes and produce images with well-defined boundaries. We experimentally confirm that LSM produces images that represent the internal geometric features of an object, some of which may be missed by conventional imaging.

42 ENGINEERING↗
DOE OSTI2013

Objective models of compressed breast shapes undergoing mammography

Purpose: To develop models of compressed breasts undergoing mammography based on objective analysis, that are capable of accurately representing breast shapes in acquired clinical images and generating new, clinically realistic shapes. Methods: An automated edge detection algorithm was used to catalogue the breast shapes of clinically acquired cranio-caudal (CC) and medio-lateral oblique (MLO) view mammograms from a large database of digital mammography images. Principal component analysis (PCA) was performed on these shapes to reduce the information contained within the shapes to a small number of linearly independent variables. The breast shape models, one of each view, were developed from the identified principal components, and their ability to reproduce the shape of breasts from an independent set of mammograms not used in the PCA, was assessed both visually and quantitatively by calculating the average distance error (ADE). Results: The PCA breast shape models of the CC and MLO mammographic views based on six principal components, in which 99.2% and 98.0%, respectively, of the total variance of the dataset is contained, were found to be able to reproduce breast shapes with strong fidelity (CC view mean ADE = 0.90 mm, MLO view mean ADE = 1.43 mm) and to generate new clinically realistic shapes. The PCA models based on fewer principal components were also successful, but to a lesser degree, as the two-component model exhibited a mean ADE = 2.99 mm for the CC view, and a mean ADE = 4.63 mm for the MLO view. The four-component models exhibited a mean ADE = 1.47 mm for the CC view and a mean ADE = 2.14 mm for the MLO view. Paired t-tests of the ADE values of each image between models showed that these differences were statistically significant (max p-value = 0.0247). Visual examination of modeled breast shapes confirmed these results. Histograms of the PCA parameters associated with the six principal components were fitted with Gaussian distributions. The six-component model was also used to generate CC and MLO view mammogram breast shapes, using the mean PCA parameter values of these distributions and randomly generated values based on the fitted Gaussian distributions, which resemble clinically encountered breasts. A spreadsheet with the data necessary to apply this model is provided as the supplementary material. Conclusions: Our PCA models of breast shapes in both mammographic views successfully reproduce analyzed breast shapes and generate new clinically relevant shapes. This work can aid in research applications which incorporate breast shape modeling, such as x-ray scatter correction, dosimetry, and image registration.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2018

Using Rule-Based Models for Weak Supervised Learning: A ChemNet for Transferable Chemical Property Prediction

With access to large datasets, deep neural networks (DNN) have achieved human-level accuracy in image and speech recognition tasks. However, in chemistry, data is inherently small and fragmented. In this work, we develop an approach of integrating rule-based knowledge into the Chemception CNN model via transfer learning techniques. The resulting model, ChemNet, is a transferable and generalizable deep neural network for chemical property prediction that learns in a semi-supervised manner from large unlabeled chemical databases. When ChemNet is further fine-tuned on 3 smaller datasets to predict chemical properties that it was not originally trained on, we show that ChemNet exceeds the accuracy of existing Chemception CNN models and other contemporary DNN models that were trained using conventional supervised learning approaches. These results indicate that pre-training ChemNet on a large diverse chemical database while incorporating chemistry domain knowledge, enables the development of more generalizable deep neural networks for the prediction of novel chemical properties.

Artificial Intelligence, Cheminformatics, Computat↗
DOE OSTI2015

Technical Note: MRI only prostate radiotherapy planning using the statistical decomposition algorithm

Purpose: In order to enable a magnetic resonance imaging (MRI) only workflow in radiotherapy treatment planning, methods are required for generating Hounsfield unit (HU) maps (i.e., synthetic computed tomography, sCT) for dose calculations, directly from MRI. The Statistical Decomposition Algorithm (SDA) is a method for automatically generating sCT images from a single MR image volume, based on automatic tissue classification in combination with a model trained using a multimodal template material. This study compares dose calculations between sCT generated by the SDA and conventional CT in the male pelvic region. Methods: The study comprised ten prostate cancer patients, for whom a 3D T2 weighted MRI and a conventional planning CT were acquired. For each patient, sCT images were generated from the acquired MRI using the SDA. In order to decouple the effect of variations in patient geometry between imaging modalities from the effect of uncertainties in the SDA, the conventional CT was nonrigidly registered to the MRI to assure that their geometries were well aligned. For each patient, a volumetric modulated arc therapy plan was created for the registered CT (rCT) and recalculated for both the sCT and the conventional CT. The results were evaluated using several methods, including mean average error (MAE), a set of dose-volume histogram parameters, and a restrictive gamma criterion (2% local dose/1 mm). Results: The MAE within the body contour was 36.5 ± 4.1 (1 s.d.) HU between sCT and rCT. Average mean absorbed dose difference to target was 0.0% ± 0.2% (1 s.d.) between sCT and rCT, whereas it was −0.3% ± 0.3% (1 s.d.) between CT and rCT. The average gamma pass rate was 99.9% for sCT vs rCT, whereas it was 90.3% for CT vs rCT. Conclusions: The SDA enables a highly accurate MRI only workflow in prostate radiotherapy planning. The dosimetric uncertainties originating from the SDA appear negligible and are notably lower than the uncertainties introduced by variations in patient geometry between imaging sessions.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2016

WE-AB-BRA-06: 4DCT-Ventilation: A Novel Imaging Modality for Thoracic Surgical Evaluation

Purpose: The current standard-of-care imaging used to evaluate lung cancer patients for surgical resection is nuclear-medicine ventilation. Surgeons use nuclear-medicine images along with pulmonary function tests (PFT) to calculate percent predicted postoperative (%PPO) PFT values by estimating the amount of functioning lung that would be lost with surgery. 4DCT-ventilation is an emerging imaging modality developed in radiation oncology that uses 4DCT data to calculate lung ventilation maps. We perform the first retrospective study to assess the use of 4DCT-ventilation for pre-operative surgical evaluation. The purpose of this work was to compare %PPO-PFT values calculated with 4DCT-ventilation and nuclear-medicine imaging. Methods: 16 lung cancer patients retrospectively reviewed had undergone 4DCTs, nuclear-medicine imaging, and had Forced Expiratory Volume in 1 second (FEV1) acquired as part of a standard PFT. For each patient, 4DCT data sets, spatial registration, and a density-change based model were used to compute 4DCT-ventilation maps. Both 4DCT and nuclear-medicine images were used to calculate %PPO-FEV1 using %PPO-FEV1=pre-operative FEV1*(1-fraction of total ventilation of resected lung). Fraction of ventilation resected was calculated assuming lobectomy and pneumonectomy. The %PPO-FEV1 values were compared between the 4DCT-ventilation-based calculations and the nuclear-medicine-based calculations using correlation coefficients and average differences. Results: The correlation between %PPO-FEV1 values calculated with 4DCT-ventilation and nuclear-medicine were 0.81 (p<0.01) and 0.99 (p<0.01) for pneumonectomy and lobectomy respectively. The average difference between the 4DCT-ventilation based and the nuclear-medicine-based %PPO-FEV1 values were small, 4.1±8.5% and 2.9±3.0% for pneumonectomy and lobectomy respectively. Conclusion: The high correlation results provide a strong rationale for a clinical trial translating 4DCT-ventilation to the surgical domain. Compared to nuclear-medicine, 4DCT-ventilation is cheaper, does not require a radioactive contrast agent, provides a faster imaging procedure, and has improved spatial resolution. 4DCT-ventilation can reduce the cost and imaging time for patients while providing improved spatial accuracy and quantitative results for surgeons. YV discloses grant from State of Colorado.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2018

Investigating the relationship between noise transfer inside the x-ray framing cameras and their imaging ability

We apply a cascaded linear model analysis to a micro-channel plate x-ray framing camera. We establish a theoretical expression of the Noise Power Spectrum (NPS) at the detector’s output and assess its accuracy by comparing it to the NPS of Monte Carlo simulations of the detector’s response to a uniform illumination. We also reflect that fitting the NPS of experimental data against a parametric model based on this expression can yield valuable information on the imaging ability of framing cameras, offering an alternative method to the usual avenue employed to measure their modulation transfer functions.

25 ENERGY STORAGE↗
NASA NTRS2018

Remote Sensing Based Yield Monitoring: Application to Winter Wheat in United States and Ukraine

Accurate and timely crop yield forecasts are critical for making informed agricultural policies and investments, as well as increasing market efficiency and stability. Earth observation data from space can contribute to agricultural monitoring, including crop yield assessment and forecasting. In this study, we present a new crop yield model based on the Difference Vegetation Index (DVI) extracted from Moderate Resolution Imaging Spectroradiometer (MODIS) data at 1 km resolution and the un-mixing of DVI at coarse resolution to a pure wheat signal (100% of wheat within the pixel). The model was applied to estimate the national and subnational winter wheat yield in the United States and Ukraine from 2001 to 2017. The model at the subnational level shows very good performance for both countries with a coefficient of determination higher than 0.7 and a root mean square error (RMSE) of lower than 0.6 t/ha (15–18%). At the national level for the United States (US) and Ukraine the model provides a strong coefficient of determination of 0.81 and 0.86, respectively, which demonstrates good performance at this scale. The model was also able to capture low winter wheat yields during years with extreme weather events, for example 2002 in US and 2003 in Ukraine. The RMSE of the model for the US at the national scale is 0.11 t/ha (3.7%) while for Ukraine it is 0.27 t/ha (8.4%).

remote sensing↗
DOE OSTI2023

Understanding structure-processing relationships in metal additive manufacturing via featurization of microstructural images

Understanding and predicting accurate property-structure-processing relationships for additively manufactured components is important for both forward and inverse design of robust, reliable parts and assemblies. While direct mapping of process parameters to properties is sometimes plausible, it is often rendered difficult due to poor microstructural control. Exploring the direct relationship between processing conditions and microstructural features can thus provide significant physical insights and aid the overall design process. Here, in this study, we develop an automated high-throughput framework to simulate an uncertainty-aware additive manufacturing (AM) process, characterize microstructural images, and extract meaningful features/descriptors. A kinetic Monte Carlo (KMC) based model of the AM process is used to simulate microstructural evolution for a diverse set of experimentally relevant processing conditions. We perform a parametric study to explore the relationship between microstructural features and processing conditions. Our results indicate that a many-to-one mapping can exist between processing conditions and typical descriptors; therefore, multiple descriptors are thus necessary to unambiguously represent microstructural images. Our work provides crucial quantitative and qualitative in-formation that would aid in the selection of features for microstructural images. Featurized microstructures could then be utilized to build data-driven models for predictive control of microstructures and thereby properties of additively manufactured components.

36 MATERIALS SCIENCE↗
DOE OSTI2014

TU-A-9A-10: Verification of Photoacoustic Computed Tomography Perfusion Imaging Using DCE-CT

Purpose: We propose to verify quantifiable perfusion information generated by a Photoacoustic Computed Tomography (PCT) scanner using Dynamic Contrast-Enhanced CT (DCE-CT), and to investigate physicsbased models of acoustic properties of tissue and photon transport to improve quantification. These corrections first necessitate a skin identifying algorithms to reduce speed-of-sound blurring and empirical photon correction methods. Methods: Xenograft mice (n=5) of breast cancer was imaged using DCE-CT which was followed by DCE-PCT. To obtain CT perfusion data, each mouse was i.v. injected (0.2mL Isovue @0.5mL/min) and subsequent radio-opaque time curves fit to a 2-compartmental model on a voxel-wise basis. For DCE-PCT, different concentrations of ICG (250, 125, and 62.5 micro-Molar) were injected at the same rate, but also acquired at different sampling rates (3, 6, and 12 seconds). The time intensity curves from PCT were fit to a 1-compartmental model on a voxel by voxel basis. The images were coregistered (Oncentra) based on the structural similarities of the tumor vasculature after which we compared both the contrastenhanced dynamics and the vascular physiology. Results: Moderate to high doses of ICG impact the washin phase of the PCT contrast due to photon losses as a function of depth. A semi-automatic algorithm has been developed to identify the skin margin, and subsequent MC and empirical models of photon transport and variations in speed-of-sound are being evaluated. Conclusion: From our results we find that there is a need to apply photon and speed-of-sound corrections to our PCT data to improve the quantifiable image data at depth in the tumor for PCT. The dose and injection rate may help in reducing large systematic effects. Our project is partially funded by a NIH SBIR grant.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2016

A novel approach to assess the treatment response using Gaussian random field in PET

Purpose: The assessment of early therapeutic response to anticancer therapy is vital for treatment planning and patient management in clinic. With the development of personal treatment plan, the early treatment response, especially before any anatomically apparent changes after treatment, becomes urgent need in clinic. Positron emission tomography (PET) imaging serves an important role in clinical oncology for tumor detection, staging, and therapy response assessment. Many studies on therapy response involve interpretation of differences between two PET images, usually in terms of standardized uptake values (SUVs). However, the quantitative accuracy of this measurement is limited. This work proposes a statistically robust approach for therapy response assessment based on Gaussian random field (GRF) to provide a statistically more meaningful scale to evaluate therapy effects. Methods: The authors propose a new criterion for therapeutic assessment by incorporating image noise into traditional SUV method. An analytical method based on the approximate expressions of the Fisher information matrix was applied to model the variance of individual pixels in reconstructed images. A zero mean unit variance GRF under the null hypothesis (no response to therapy) was obtained by normalizing each pixel of the post-therapy image with the mean and standard deviation of the pretherapy image. The performance of the proposed method was evaluated by Monte Carlo simulation, where XCAT phantoms (128{sup 2} pixels) with lesions of various diameters (2–6 mm), multiple tumor-to-background contrasts (3–10), and different changes in intensity (6.25%–30%) were used. The receiver operating characteristic curves and the corresponding areas under the curve were computed for both the proposed method and the traditional methods whose figure of merit is the percentage change of SUVs. The formula for the false positive rate (FPR) estimation was developed for the proposed therapy response assessment utilizing local average method based on random field. The accuracy of the estimation was validated in terms of Euler distance and correlation coefficient. Results: It is shown that the performance of therapy response assessment is significantly improved by the introduction of variance with a higher area under the curve (97.3%) than SUVmean (91.4%) and SUVmax (82.0%). In addition, the FPR estimation serves as a good prediction for the specificity of the proposed method, consistent with simulation outcome with ∼1 correlation coefficient. Conclusions: In this work, the authors developed a method to evaluate therapy response from PET images, which were modeled as Gaussian random field. The digital phantom simulations demonstrated that the proposed method achieved a large reduction in statistical variability through incorporating knowledge of the variance of the original Gaussian random field. The proposed method has the potential to enable prediction of early treatment response and shows promise for application to clinical practice. In future work, the authors will report on the robustness of the estimation theory for application to clinical practice of therapy response evaluation, which pertains to binary discrimination tasks at a fixed location in the image such as detection of small and weak lesion.

61 RADIATION PROTECTION AND DOSIMETRY↗
NASA NTRS2001

Legato: Personal Computer Software for Analyzing Pressure-Sensitive Paint Data

'Legato' is personal computer software for analyzing radiometric pressure-sensitive paint (PSP) data. The software is written in the C programming language and executes under Windows 95/98/NT operating systems. It includes all operations normally required to convert pressure-paint image intensities to normalized pressure distributions mapped to physical coordinates of the test article. The program can analyze data from both single- and bi-luminophore paints and provides for both in situ and a priori paint calibration. In addition, there are functions for determining paint calibration coefficients from calibration-chamber data. The software is designed as a self-contained, interactive research tool that requires as input only the bare minimum of information needed to accomplish each function, e.g., images, model geometry, and paint calibration coefficients (for a priori calibration) or pressure-tap data (for in situ calibration). The program includes functions that can be used to generate needed model geometry files for simple model geometries (e.g., airfoils, trapezoidal wings, rotor blades) based on the model planform and airfoil section. All data files except images are in ASCII format and thus are easily created, read, and edited. The program does not use database files. This simplifies setup but makes the program inappropriate for analyzing massive amounts of data from production wind tunnels. Program output consists of Cartesian plots, false-colored real and virtual images, pressure distributions mapped to the surface of the model, assorted ASCII data files, and a text file of tabulated results. Graphical output is displayed on the computer screen and can be saved as publication-quality (PostScript) files.

Schairer, Edward T.↗
NASA NTRS2018

Remote Sensing Based Yield Monitoring: Application to Winter Wheat in United States and Ukraine

Accurate and timely crop yield forecasts are critical for making informed agricultural policies and investments, as well as increasing market efficiency and stability. Earth observation data from space can contribute to agricultural monitoring, including crop yield assessment and forecasting. In this study, we present a new crop yield model based on the Difference Vegetation Index (DVI) extracted from Moderate Resolution Imaging Spectroradiometer (MODIS) data at 1 km resolution and the un-mixing of DVI at coarse resolution to a pure wheat signal (100 percent of wheat within the pixel). The model was applied to estimate the national and subnational winter wheat yield in the United States and Ukraine from 2001 to 2017. The model at the subnational level shows very good performance for both countries with a coefficient of determination higher than 0.7 and a root mean square error (RMSE) of lower than 0.6 t/ha (tonnes per hectare) (15-18 percent). At the national level for the United States (US) and Ukraine the model provides a strong coefficient of determination of 0.81 and 0.86, respectively, which demonstrates good performance at this scale. The model was also able to capture low winter wheat yields during years with extreme weather events, for example 2002 in US and 2003 in Ukraine. The RMSE of the model for the US at the national scale is 0.11 t/ha (3.7 percent) while for Ukraine it is 0.27 t/ha (8.4 percent).

DVI↗
DOE OSTI2013

Compensator models for fluence field modulated computed tomography

Purpose: Fluence field modulated computed tomography (FFMCT) presents a novel approach for acquiring CT images, whereby a patient model guides dynamically changing fluence patterns in an attempt to achieve task-based, user-prescribed, regional variations in image quality, while also controlling dose to the patient. This work aims to compare the relative effectiveness of FFMCT applied to different thoracic imaging tasks (routine diagnostic CT, lung cancer screening, and cardiac CT) when the modulator is subject to limiting constraints, such as might be present in realistic implementations.Methods: An image quality plan was defined for a simulated anthropomorphic chest slice, including regions of high and low image quality, for each of the thoracic imaging tasks. Modulated fluence patterns were generated using a simulated annealing optimization script, which attempts to achieve the image quality plan under a global dosimetric constraint. Optimization was repeated under different types of modulation constraints (e.g., fixed or gantry angle dependent patterns, continuous or comprised of discrete apertures) with the most limiting case being a fixed conventional bowtie filter. For each thoracic imaging task, an image quality map (IQM{sub sd}) representing the regionally varying standard deviation is predicted for each modulation method and compared to the prescribed image quality plan as well as against results from uniform fluence fields. Relative integral dose measures were also compared.Results: Each IQM{sub sd} resulting from FFMCT showed improved agreement with planned objectives compared to those from uniform fluence fields for all cases. Dynamically changing modulation patterns yielded better uniformity, improved image quality, and lower dose compared to fixed filter patterns with optimized tube current. For the latter fixed filter cases, the optimal choice of tube current modulation was found to depend heavily on the task. Average integral dose reduction compared to a uniform fluence field ranged from 10% using a bowtie filter to 40% or greater using an idealized modulator.Conclusions: The results support that FFMCT may achieve regionally varying image quality distributions in good agreement with user-prescribed values, while limiting dose. The imposition of constraints inhibits dose reduction capacity and agreement with image quality plans but still yields significant improvement over what is afforded by conventional dose minimization techniques. These results suggest that FFMCT can be implemented effectively even when the modulator has limited modulation capabilities.

61 RADIATION PROTECTION AND DOSIMETRY↗
DOE OSTI2011

A HIGH SPATIAL RESOLUTION STUDY OF THE {lambda} = 3 mm CONTINUUM OF ORION-KL

Recent interferometric observations have called into question the traditional view of the Orion-KL region, which displays one of the most well-defined cases of chemical differentiation in a star-forming region. Previous lower-resolution images of Orion-KL show emission signatures for oxygen-bearing organic molecules toward the Orion Compact Ridge and emission for nitrogen-bearing organic molecules toward the Orion Hot Core. However, more recent observations at higher spatial resolution indicate that the bulk of the molecular emission arises from many smaller, compact clumps that are spatially distinct from the traditional Hot Core and Compact Ridge sources. It is this type of observational information that is critical for guiding astrochemical models, as the spatial distribution of molecules and their relation to energetic sources will govern the chemical mechanisms at play in star-forming regions. We have conducted millimeter-imaging studies of Orion-KL with various beam sizes using the Combined Array for Research in Millimeter-Wave Astronomy in order to investigate the continuum structure. These {lambda} = 3 mm observations have synthesized beam sizes of {approx}0.''5-5.''0. These observations reveal the complex continuum structure of this region, which stands in sharp contrast to the previous structural models assumed for Orion-KL based on lower spatial resolution images. The new results indicate that the spatial scaling previously used in determination of molecular abundances for this region are in need of complete revision. Here we present the results of the continuum observations, discuss the sizes and structures of the detected sources, and suggest an observational strategy for determining the proper spatial scaling to accurately determine molecular abundances in the Orion-KL region.

79 ASTRONOMY AND ASTROPHYSICS↗
NASA NTRS1980

A theoretical study of microwave beam absorption by a rectenna

The rectenna's microwave power beam absorption limit was theoretically confirmed by two mathematical models descriptive of the microwave absorption process; first one model was based on the current sheet equivalency of a large planar array above a reflector and the second model, which was based on the properties of a waveguide with special imaging characteristics, quantified the electromagnetic modes (field configurations) in the immediate vicinity of a Rectenna element spacing which permit total power beam absorption by preventing unwanted modes from propagating (scattering) were derived using these models. Several factors causing unwanted scattering are discussed.

Ott, J. H.↗
DOE OSTI2021

SGD-Net: Efficient Model-Based Deep Learning with Theoretical Guarantees

Deep unfolding networks have recently gained popularity for solving imaging inverse problems. However, the computational and memory complexity of data-consistency layers within traditional deep unfolding networks scales with the number of measurements, limiting their applicability to large-scale imaging inverse problems. We propose SGD-Net as a new methodology for improving the efficiency of deep unfolding through stochastic approximations of the data-consistency layers. Our theoretical analysis shows that SGD-Net can be trained to approximate batch deep unfolding networks to an arbitrary precision. Our simulations on intensity diffraction tomography and sparse-view computed tomography show that SGD-Net can match the performance of the traditional batch network at a fraction of training and testing complexity.Deep unfolding networks have recently gained popularity for solving imaging inverse problems. However, the computational and memory complexity of data-consistency layers within traditional deep unfolding networks scales with the number of measurements, limiting their applicability to large-scale imaging inverse problems. We propose SGD-Net as a new methodology for improving the efficiency of deep unfolding through stochastic approximations of the data-consistency layers. Our theoretical analysis shows that SGD-Net can be trained to approximate batch deep unfolding networks to an arbitrary precision. Our simulations on intensity diffraction tomography and sparse-view computed tomography show that SGD-Net can match the performance of the traditional batch network at a fraction of training and testing complexity.

97 MATHEMATICS AND COMPUTING↗