Search NASA⌕ Search

SEARCH · Search NASA

Results for “convolution”

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 415 records · Page 23

Dose convolution filter: Incorporating spatial dose information into tissue response modeling

Purpose: A model is introduced to integrate biological factors such as cell migration and bystander effects into physical dose distributions, and to incorporate spatial dose information in plan analysis and optimization. Methods: The model consists of a dose convolution filter (DCF) with single parameter {sigma}. Tissue response is calculated by an existing NTCP model with DCF-applied dose distribution as input. The authors determined {sigma} of rat spinal cord from published data. The authors also simulated the GRID technique, in which an open field is collimated into many pencil beams. Results: After applying the DCF, the NTCP model successfully fits the rat spinal cord data with a predicted value of {sigma}=2.6{+-}0.5 mm, consistent with 2 mm migration distances of remyelinating cells. Moreover, it enables the appropriate prediction of a high relative seriality for spinal cord. The model also predicts the sparing of normal tissues by the GRID technique when the size of each pencil beam becomes comparable to {sigma}. Conclusions: The DCF model incorporates spatial dose information and offers an improved way to estimate tissue response from complex radiotherapy dose distributions. It does not alter the prediction of tissue response in large homogenous fields, but successfully predicts increased tissue tolerance in small or highly nonuniform fields.

61 RADIATION PROTECTION AND DOSIMETRY↗

A fast double template convolution isocenter evaluation algorithm with subpixel accuracy

Purpose: To design a fast Winston Lutz (fWL) algorithm for accurate analysis of radiation isocenter from images without edge detection or center of mass calculations. Methods: An algorithm has been developed to implement the Winston Lutz test for mechanical/radiation isocenter agreement using an electronic portal imaging device (EPID). The algorithm detects the position of the radiation shadow of a tungsten ball within a stereotactic cone. The fWL algorithm employs a double convolution to independently find the position of the sphere and cone centers. Subpixel estimation is used to achieve high accuracy. Results of the algorithm were compared to (1) a human observer with template guidance and (2) an edge detection/center of mass (edCOM) algorithm. Testing was performed with high resolution (0.05mm/px, film) and low resolution (0.78mm/px, EPID) image sets. Results: Sphere and cone center relative positions were calculated with the fWL algorithm for high resolution test images with an accuracy of 0.002{+-}0.061 mm compared to 0.042{+-}0.294 mm for the human observer, and 0.003{+-}0.038 mm for the edCOM algorithm. The fWL algorithm required 0.01 s per image compared to 5 s for the edCOM algorithm and 20 s for the human observer. For lower resolution images the fWL algorithm localized the centers with an accuracy of 0.083{+-}0.12 mm compared to 0.03{+-}0.5514 mm for the edCOM algorithm. Conclusions: A fast (subsecond) subpixel algorithm has been developed that can accurately determine the center locations of the ball and cone in Winston Lutz test images without edge detection or COM calculations.

60 APPLIED LIFE SCIENCES↗

A convolution model for obtaining the response of an ionization chamber in static non standard fields

Purpose: This work contains an alternative methodology for obtaining correction factors for ionization chamber (IC) dosimetry of small fields and composite fields such as IMRT. The method is based on the convolution/superposition (C/S) of an IC response function (RF) with the dose distribution in a certain plane which includes chamber position. This method is an alternative to the full Monte Carlo (MC) approach that has been used previously by many authors for the same objective. Methods: The readout of an IC at a point inside a phantom irradiated by a certain beam can be obtained as the convolution of the dose spatial distribution caused by the beam and the IC two-dimensional RF. The proposed methodology has been applied successfully to predict the response of a PTW 30013 IC when measuring different nonreference fields, namely: output factors of 6 MV small fields, beam profiles of cobalt 60 narrow fields and 6 MV radiosurgery segments. The two-dimensional RF of a PTW 30013 IC was obtained by MC simulation of the absorbed dose to cavity air when the IC was scanned by a 0.6 x 0.6 mm{sup 2} cross section parallel pencil beam at low depth in a water phantom. For each of the cases studied, the results of the IC direct measurement were compared with the corresponding obtained by the C/S method. Results: For all of the cases studied, the agreement between the IC direct measurement and the IC calculated response was excellent (better than 1.5%). Conclusions: This method could be implemented in TPS in order to calculate dosimetry correction factors when an experimental IMRT treatment verification with in-phantom ionization chamber is performed. The miss-response of the IC due to the nonreference conditions could be quickly corrected by this method rather than employing MC derived correction factors. This method can be considered as an alternative to the plan-class associated correction factors proposed recently as part of an IAEA work group on nonstandard field dosimetry.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

WE-A-17A-08: Evaluation of the OncentraBrachy Collapsed Cone Convolution Algorithm for Ir-192 Source Using Phantom and Real-Patient Heterogeneous Geometries

Purpose: To evaluate the commercially released Collapsed Cone convolution-based(CCC) dose calculation module of the Elekta OncentraBrachy(OcB) treatment planning system(TPS). Methods: An allwater phantom was used to perform TG43 benchmarks with single source and seventeen sources, separately. Furthermore, four real-patient heterogeneous geometries (chestwall, lung, breast and prostate) were used. They were selected based on their clinical representativity of a class of clinical anatomies that pose clear challenges. The plans were used as is(no modification). For each case, TG43 and CCC calculations were performed in the OcB TPS, with TG186-recommended materials properly assigned to ROIs. For comparison, Monte Carlo simulation was run for each case with the same material scheme and grid mesh as TPS calculations. Both modes of CCC (standard and high quality) were tested. Results: For the benchmark case, the CCC dose, when divided by that of TG43, yields hot-n-cold spots in a radial pattern. The pattern of the high mode is denser than that of the standard mode and is representative of angular dicretization. The total deviation ((hot-cold)/TG43) is 18% for standard mode and 11% for high mode. Seventeen dwell positions help to reduce “ray-effect”, with the total deviation to 6% (standard) and 5% (high), respectively. For the four patient cases, CCC produces, as expected, more realistic dose distributions than TG43. A close agreement was observed between CCC and MC for all isodose lines, from 20% and up; the 10% isodose line of CCC appears shifted compared to that of MC. The DVH plots show dose deviations of CCC from MC in small volume, high dose regions (>100% isodose). For patient cases, the difference between standard and high modes is almost undiscernable. Conclusion: OncentraBrachy CCC algorithm marks a significant dosimetry improvement relative to TG43 in real-patient cases. Further researches are recommended regarding the clinical implications of the above observations. Support provided by a CIHR grant and CCC system provided by Elekta-Nucletron.

60 APPLIED LIFE SCIENCES↗

Automated detection of pulmonary nodules in PET/CT images: Ensemble false-positive reduction using a convolutional neural network technique

Purpose: Automated detection of solitary pulmonary nodules using positron emission tomography (PET) and computed tomography (CT) images shows good sensitivity; however, it is difficult to detect nodules in contact with normal organs, and additional efforts are needed so that the number of false positives (FPs) can be further reduced. In this paper, the authors propose an improved FP-reduction method for the detection of pulmonary nodules in PET/CT images by means of convolutional neural networks (CNNs). Methods: The overall scheme detects pulmonary nodules using both CT and PET images. In the CT images, a massive region is first detected using an active contour filter, which is a type of contrast enhancement filter that has a deformable kernel shape. Subsequently, high-uptake regions detected by the PET images are merged with the regions detected by the CT images. FP candidates are eliminated using an ensemble method; it consists of two feature extractions, one by shape/metabolic feature analysis and the other by a CNN, followed by a two-step classifier, one step being rule based and the other being based on support vector machines. Results: The authors evaluated the detection performance using 104 PET/CT images collected by a cancer-screening program. The sensitivity in detecting candidates at an initial stage was 97.2%, with 72.8 FPs/case. After performing the proposed FP-reduction method, the sensitivity of detection was 90.1%, with 4.9 FPs/case; the proposed method eliminated approximately half the FPs existing in the previous study. Conclusions: An improved FP-reduction scheme using CNN technique has been developed for the detection of pulmonary nodules in PET/CT images. The authors’ ensemble FP-reduction method eliminated 93% of the FPs; their proposed method using CNN technique eliminates approximately half the FPs existing in the previous study. These results indicate that their method may be useful in the computer-aided detection of pulmonary nodules using PET/CT images.

60 APPLIED LIFE SCIENCES↗

SU-F-T-377: Monte Carlo Re-Evaluation of Volumetric-Modulated Arc Plans of Advanced Stage Nasopharygeal Cancers Optimized with Convolution-Superposition Algorithm

Background: Commercial treatment planning system Pinnacle3 (Philips, Fitchburg, WI, USA) employs a convolution-superposition algorithm for volumetric-modulated arc radiotherapy (VMAT) optimization and dose calculation. Study of Monte Carlo (MC) dose recalculation of VMAT plans for advanced-stage nasopharyngeal cancers (NPC) is currently limited. Methods: Twenty-nine VMAT prescribed 70Gy, 60Gy, and 54Gy to the planning target volumes (PTVs) were included. These clinical plans achieved with a CS dose engine on Pinnacle3 v9.0 were recalculated by the Monaco TPS v5.0 (Elekta, Maryland Heights, MO, USA) with a XVMC-based MC dose engine. The MC virtual source model was built using the same measurement beam dataset as for the Pinnacle beam model. All MC recalculation were based on absorbed dose to medium in medium (Dm,m). Differences in dose constraint parameters per our institution protocol (Supplementary Table 1) were analyzed. Results: Only differences in maximum dose to left brachial plexus, left temporal lobe and PTV54Gy were found to be statistically insignificant (p> 0.05). Dosimetric differences of other tumor targets and normal organs are found in supplementary Table 1. Generally, doses outside the PTV in the normal organs are lower with MC than with CS. This is also true in the PTV54-70Gy doses but higher dose in the nasal cavity near the bone interfaces is consistently predicted by MC, possibly due to the increased backscattering of short-range scattered photons and the secondary electrons that is not properly modeled by the CS. The straight shoulders of the PTV dose volume histograms (DVH) initially resulted from the CS optimization are merely preserved after MC recalculation. Conclusion: Significant dosimetric differences in VMAT NPC plans were observed between CS and MC calculations. Adjustments of the planning dose constraints to incorporate the physics differences from conventional CS algorithm should be made when VMAT optimization is carried out directly with MC dose engine.

60 APPLIED LIFE SCIENCES↗

A Novel Pruning Method for Convolutional Neural Networks Based off Identifying Critical Filters

Convolutional Neural Networks (CNNs) are one of the most extensively used tools in machine learning, but they are still not well understood and in many cases they are over-parameterized, leading to slow inference and impeding their deployment on low-power devices. In the last few years, many methods for decreasing the number of parameters in a network by pruning its output channels have been suggested, but a very recent work has argued that random pruning of channels performs on-par with state-of-the-art pruning methods. While random and other pruning methods might be effectively used for lowering the number of parameters in a CNN, none of these methods can be used to gain any further understanding of the model that the CNN has built. In this work, we propose a novel method for pruning a network, that at the same time can lead to a better understanding of what the individual filters of the network learn about the data. The method proposed aims to keep only the filters that are "important" for a class. We define a filter as important for a class if its removal has the highest negative impact on the accuracy for that class. We demonstrate that our method is better than random pruning on two networks used on the EMNIST and CIFAR10 datasets. By analyzing the important filters, we find that the important filters in the pruned networks learn features which are more general across classes. We demonstrate the importance and applicability of that observation in two transfer-learning tasks.

Dimovska, Mihaela↗

Evaluating Automated Face Identity-Masking Methods with Human Perception and a Deep Convolutional Neural Network

Face de-identification (or “masking”) algorithms have been developed in response to the prevalent use of video recordings in public places. Here, we evaluated the success of face identity masking for human perceivers and a deep convolutional neural network (DCNN). Eight de-identification algorithms were applied to videos of drivers’ faces, while they actively operated a motor vehicle. These masks were pre-selected to be applicable to low-quality video and to maintain coarse information about facial actions. Humans studied high-resolution images to learn driver identities and were tested on their recognition of active drivers in low-resolution videos. Faces in the videos were either unmasked or were masked by one of the eight algorithms. When participants were tested immediately after learning (Experiment 1), all masks reduced identification, with six of eight masks reducing identification to extremely poor performance. In a second experiment, two of the most effective masks were tested after a delay of 7 or 28 days. The delay did not further reduce identification of the masked faces. In all masked conditions, participants maintained stringent decision criteria, with low confidence in recognition, further indicating the effectiveness of the masks. Next, the DCNN performed an identity-matching task between high-resolution images and masked videos—a task analogous to that done by humans. The pattern of accuracy for the DCNN mirrored some, but not all, aspects of human performance, highlighting the need to test the effectiveness of identity masking for both humans and machines. The DCNN was also tested on its ability to match identity between masked and unmasked versions of the same video, based only on the face. DCNN performance for the eight masks offers insight into the nature of the information in faces that is coded in these networks.

97 MATHEMATICS AND COMPUTING↗

Remote Sensor Design for Visual Recognition with Convolutional Neural Networks

While deep learning technologies for computer vision have developed rapidly since 2012, modeling of remote sensing systems has remained focused around human vision. In particular, remote sensing systems are usually constructed to optimize sensing cost-quality trade-offs with respect to human image interpretability. While some recent studies have explored remote sensing system design as a function of simple computer vision algorithm performance, there has been little work relating this design to the state-of-the-art in computer vision: deep learning with convolutional neural networks. We develop experimental systems to conduct this analysis, showing results with modern deep learning algorithms and recent overhead image data. Our results are compared to standard image quality measurements based on human visual perception, and we conclude not only that machine and human interpretability differ significantly, but that computer vision performance is largely self-consistent across a range of disparate conditions. This repository contains all the code required to replicate the results of our research paper, and the corresponding docker environment in which that code can be executed. All code is written in Python3 and utilizes the PyTorch library (v1.0) for neural network model training and evaluation. Jupyter notebooks are used to visualize images transformed with our code, in addition to plotting experimental results. We also include the specific parameter files used to execute our experiments, with the intent that these experiments can be replicated.

Zelinski, michaelE.↗

Multi-Task Convolutional Neural Networks for Natural Text Classification

Information extraction and coding of free-text pathology reports is an important activity for cancer registries to support national cancer surveillance. Cancer registrars must process high volumes of pathology reports on an annual basis. We investigated an automated approach using a convolutional neural networks (CNNs) for extracting the primary site, laterality, histology, behavior, and histological grade from unstructured cancer pathology text reports. Our multi-task learning (MTL) with hard parameter sharing approach is used to train a multi-task MT-CNN model for all the tasks. The performance of our proposed approach was compared against a state of-the-art CNN and the commonly used SVM classifier. We observed that the proposed model consistently outperformed the baseline models, especially for the less prevalent classes. Also, the results demonstrate the potential of the proposed method for handling class imbalance within each task.

Yoon, HongJun↗

GRAPH CONVOLUTIONAL NEURAL NETWORK (GCNN)FOR POLYMER PROPERTY PREDICTION

SF-22-121 Graph convolutional neural networks (GCNN) are powerful machine learning models that apply a message-passing algorithm on adjacent nodes in a graph. Due to the natural representation of molecules via discrete graphs (where nodes encode atoms and edges encode bonds), GCNNs are an intuitive machine learning model chemical space informatics, including the prediction of polymer properties. Here, molecules are represented in feature and adjacency matrices where they are directly used as inputs to a GCNN.

FEINSTEIN, JEREMY↗

Convolution Neural Network for Fault Identification in Distribution Feeder with High Penetration Solar PV

Identification and zonal classification of the faults is a decisive factor in the relay’s decision to trip or not. Different types of fault like three-phase, line-to-line-to-ground and single-line-to-ground can occur at various locations in the feeder. These faults are seen as the variation in the instantaneous values of three-phase voltages and currents, i.e., waveforms, that are measured at the relay location. The objective of this work is to develop a machine learning model that can identify a fault and classify it to various protection zones based on measured waveforms. In this work, a data-driven relay based on Convolutional Neural Network (CNN) is proposed for fault identification in distribution feeders with high penetration solar PV. The proposed CNN model takes local current and voltage waveforms as input and classify it into fault, no-fault or a capacitor switching. Further, the CNN also attempts to identify fault zones based on the images of waveforms. The overall testing accuracy of the trained model exceeds 95%.

Ramesh, Meghana↗

Detecting Living-off-the-land Attacks Using K-means And Graph Convolutional Networks

The code ingests Zeek logs derived from network packet captures and goes through data preprocessing before it gets passed into a K-Means model that labels each device as either a client or server. Graph Convolutional Network (GCN) model is used to obtain the embeddings to represent the features in lower dimension. Last, K-means cluster analysis is used to cluster the embeddings for each class.

Quach, Anna [Idaho National Laboratory (INL), Idah↗

PERSIANN-CNN: Precipitation Estimation from Remotely Sensed Information Using Artificial Neural Networks–Convolutional Neural Networks

Abstract Accurate and timely precipitation estimates are critical for monitoring and forecasting natural disasters such as floods. Despite having high-resolution satellite information, precipitation estimation from remotely sensed data still suffers from methodological limitations. State-of-the-art deep learning algorithms, renowned for their skill in learning accurate patterns within large and complex datasets, appear well suited to the task of precipitation estimation, given the ample amount of high-resolution satellite data. In this study, the effectiveness of applying convolutional neural networks (CNNs) together with the infrared (IR) and water vapor (WV) channels from geostationary satellites for estimating precipitation rate is explored. The proposed model performances are evaluated during summer 2012 and 2013 over central CONUS at the spatial resolution of 0.08° and at an hourly time scale. Precipitation Estimation from Remotely Sensed Information Using Artificial Neural Networks (PERSIANN)–Cloud Classification System (CCS), which is an operational satellite-based product, and PERSIANN–Stacked Denoising Autoencoder (PERSIANN-SDAE) are employed as baseline models. Results demonstrate that the proposed model (PERSIANN-CNN) provides more accurate rainfall estimates compared to the baseline models at various temporal and spatial scales. Specifically, PERSIANN-CNN outperforms PERSIANN-CCS (and PERSIANN-SDAE) by 54% (and 23%) in the critical success index (CSI), demonstrating the detection skills of the model. Furthermore, the root-mean-square error (RMSE) of the rainfall estimates with respect to the National Centers for Environmental Prediction (NCEP) Stage IV gauge–radar data, for PERSIANN-CNN was lower than that of PERSIANN-CCS (PERSIANN-SDAE) by 37% (14%), showing the estimation accuracy of the proposed model.

54 ENVIRONMENTAL SCIENCES↗

Classification of Cloud Particle Imagery from Aircraft Platforms Using Convolutional Neural Networks

Abstract A vast amount of ice crystal imagery exists from a variety of field campaign initiatives that can be utilized for cloud microphysical research. Here, nine convolutional neural networks are used to classify particles into nine regimes on over 10 million images from the Cloud Particle Imager probe, including liquid and frozen states and particles with evidence of riming. A transfer learning approach proves that the Visual Geometry Group (VGG-16) network best classifies imagery with respect to multiple performance metrics. Classification accuracies on a validation dataset reach 97% and surpass traditional automated classification. Furthermore, after initial model training and preprocessing, 10 000 images can be classified in approximately 35 s using 20 central processing unit cores and two graphics processing units, which reaches real-time classification capabilities. Statistical analysis of the classified images indicates that a large portion (57%) of the dataset is unusable, meaning the images are too blurry or represent indistinguishable small fragments. In addition, 19% of the dataset is classified as liquid drops. After removal of fragments, blurry images, and cloud drops, 38% of the remaining ice particles are largely intersecting the image border (≥10% cutoff) and therefore are considered unusable because of the inability to properly classify and dimensionalize. After this filtering, an unprecedented database of 1 560 364 images across all campaigns is available for parameter extraction and bulk statistics on specific particle types in a wide variety of storm systems, which can act to improve the current state of microphysical parameterizations.

54 ENVIRONMENTAL SCIENCES↗

Two-Stream Multi-Channel Convolutional Neural Network for Multi-Lane Traffic Speed Prediction Considering Traffic Volume Impact

Traffic speed prediction is a critically important component of intelligent transportation systems. Recently, with the rapid development of deep learning and transportation data science, a growing body of new traffic speed prediction models have been designed that achieved high accuracy and large-scale prediction. However, existing studies have two major limitations. First, they predict aggregated traffic speed rather than lane-level traffic speed; second, most studies ignore the impact of other traffic flow parameters in speed prediction. To address these issues, the authors propose a two-stream multi-channel convolutional neural network (TM-CNN) model for multi-lane traffic speed prediction considering traffic volume impact. In this model, the authors first introduce a new data conversion method that converts raw traffic speed data and volume data into spatial–temporal multi-channel matrices. Then the authors carefully design a two-stream deep neural network to effectively learn the features and correlations between individual lanes, in the spatial–temporal dimensions, and between speed and volume. Accordingly, a new loss function that considers the volume impact in speed prediction is developed. A case study using 1-year data validates the TM-CNN model and demonstrates its superiority. This paper contributes to two research areas: (1) traffic speed prediction, and (2) multi-lane traffic flow study.

97 MATHEMATICS AND COMPUTING↗

TULIP: An RNA-seq-based Primary Tumor Type Prediction Tool Using Convolutional Neural Networks

Background: With cancer as one of the leading causes of death worldwide, accurate primary tumor type prediction is critical in identifying genetic factors that can inhibit or slow tumor progression. There have been efforts to categorize primary tumor types with gene expression data using machine learning, and more recently with deep learning, in the last several years. Methods In this paper, we developed four 1-dimensional (1D) Convolutional Neural Network (CNN) models to classify RNA-seq count data as one of 17 highly represented primary tumor types or 32 primary tumor types regardless of imbalanced representation. Additionally, we adapted the models to take as input either all Ensembl genes (60,483) or protein coding genes only (19,758). Unlike previous work, we avoided selection bias by not filtering genes based on expression values. RNA-seq count data expressed as FPKM-UQ of 9,025 and 10,940 samples from The Cancer Genome Atlas (TCGA) were downloaded from the Genomic Data Commons (GDC) corresponding to 17 and 32 primary tumor types respectively for training and validating the models. Results: All 4 1D-CNN models had an overall accuracy of 94.7% to 97.6% on the test dataset. Further evaluation indicates that the models with protein coding genes only as features performed with better accuracy compared to the models with all Ensembl genes for both 17 and 32 primary tumor types. For all models, the accuracy by primary tumor type was above 80% for most primary tumor types. Conclusions: We packaged all 4 models as a Python-based deep learning classification tool called TULIP (TUmor CLassIfication Predictor) for performing quality control on primary tumor samples and characterizing cancer samples of unknown tumor type. Further optimization of the models is needed to improve the accuracy of certain primary tumor types.

Jones, Sara↗

DeepDist: real-value inter-residue distance prediction with deep residual convolutional network

Abstract Background Driven by deep learning, inter-residue contact/distance prediction has been significantly improved and substantially enhanced ab initio protein structure prediction. Currently, most of the distance prediction methods classify inter-residue distances into multiple distance intervals instead of directly predicting real-value distances. The output of the former has to be converted into real-value distances to be used in tertiary structure prediction. Results To explore the potentials of predicting real-value inter-residue distances, we develop a multi-task deep learning distance predictor (DeepDist) based on new residual convolutional network architectures to simultaneously predict real-value inter-residue distances and classify them into multiple distance intervals. Tested on 43 CASP13 hard domains, DeepDist achieves comparable performance in real-value distance prediction and multi-class distance prediction. The average mean square error (MSE) of DeepDist’s real-value distance prediction is 0.896 Å 2 when filtering out the predicted distance ≥ 16 Å, which is lower than 1.003 Å 2 of DeepDist’s multi-class distance prediction. When distance predictions are converted into contact predictions at 8 Å threshold (the standard threshold in the field), the precision of top L/5 and L/2 contact predictions of DeepDist’s multi-class distance prediction is 79.3% and 66.1%, respectively, higher than 78.6% and 64.5% of its real-value distance prediction and the best results in the CASP13 experiment. Conclusions DeepDist can predict inter-residue distances well and improve binary contact prediction over the existing state-of-the-art methods. Moreover, the predicted real-value distances can be directly used to reconstruct protein tertiary structures better than multi-class distance predictions due to the lower MSE. Finally, we demonstrate that predicting the real-value distance map and multi-class distance map at the same time performs better than predicting real-value distances alone.

59 BASIC BIOLOGICAL SCIENCES↗