Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Support vector machine”

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
All sourcesNASA NTRSDOE OSTI
DOE OSTI2020

Detection and Diagnosis of Data Integrity Attacks in Solar Farms Based on Multilayer Long Short-Term Memory Network

Photovoltaic (PV) systems are becoming more vulnerable to cyber threats. In response to this emerging concern, developing cyber-secure power electronics converters has received increased attention from the IEEE Power Electronics Society that recently launched a cyber-physical-security initiative. Here this letter proposes a deep sequence learning based diagnosis solution for data integrity attacks on PV systems in smart grids, including dc–dc and dc–ac converters. The multilayer long short-term memory networks are used to leverage time-series electric waveform data from current and voltage sensors in PV systems. The proposed method has been evaluated in a PV smart grid benchmark model with extensive quantitative analysis. As a comparison, we have evaluated classic data-driven methods, including K-nearest neighbor, decision tree, support vector machine, artificial neural network, and convolutional neural network. Comparison results verify performances of the proposed method for detection and diagnosis of various data integrity attacks on PV systems.

42 ENGINEERING↗
DOE OSTI2021

Fault Detection Utilizing Convolution Neural Network on Timeseries Synchrophasor Data From Phasor Measurement Units

An end-to-end supervised learning method is proposed for fault detection in the electric grid using Big Data from multiple Phasor Measurement Units (PMUs). The approach consists of preprocessing steps aimed at reducing data noise and dimensionality, followed by utilization of six classification models considered for detecting faults. Three of the models were variants of Convolutional Neural Network (CNN) architectures that consider a single type of measurement (voltage, current or frequency) at all PMUs or all types together also at all PMUs. CNN based models were compared to traditional methods of Logistic Regression (LR), Multi-layer Perceptron (MLP) and Support Vector Machine (SVM). Evaluation was conducted on two-year data measured by PMUs at 37 locations in a large electric grid. Here, the response variable for classification were extracted from the grid-wide outage event log. Experiments show that CNN-based models outperformed traditional methods on one year out-of-sample outage detection over the entire grid.

42 ENGINEERING↗
DOE OSTI2024

Scalable Risk Assessment of Rare Events in Power Systems With Uncertain Wind Generation and Loads

Risk assessment of rare events has become increasingly important in power system planning and operation with the increasing integration of renewable energy and the presence of system uncertainties. However, quantifying the risk posed by rare events via the traditional method, i.e., Monte Carlo sampling (MCS), incurs substantial computational expense stemming from the vast ensemble of power flow simulations. To accelerate the assessment, this paper proposes a Deep Neural Network (DNN)-kernelized vector-valued Gaussian Process (VVGP) approach with excellent computational efficiency while maintaining high accuracy. Consequently, serving as a surrogate model for the power flow solver, the DNN-kernelized VVGP enables significantly faster but accurate risk assessment compared to the power flow solver. The developed surrogate model evaluates low-order N - k events that contain more than 90% instances by adeptly capturing the topological features while the high-order N - k events are assessed via a power flow solver, thereby striking a balance between computational efficiency and uncertainty quantification accuracy. Moreover, the model incorporates a Support Vector Machine (SVM) classifier to resample concerning low-probability tail events to counteract the biases potentially introduced during the DNN-kernelized VVGP evaluations. Simulations conducted on the modified IEEE 24-bus, 118-bus, and European 1354-bus systems demonstrate that the proposed method maintains the accuracy benchmark set by MCS while significantly reducing computational demands in large-scale power systems as compared to other state-of-the-art methods.

17 WIND ENERGY↗
DOE OSTI2018

Unsupervised Clustering-Based Short-Term Solar Forecasting

Solar forecasting accuracy is highly affected by weather conditions, therefore, weather awareness forecasting models are expected to improve the forecasting performance. However, it may not be available or reliable to classify different forecasting tasks by only using predefined meteorological weather categorization. In this paper, an unsupervised clustering-based (UC-based) solar forecasting method is developed for short-term (1-h-ahead) global horizontal irradiance (GHI) forecasting. This UC-based method consists of three parts: GHI time series unsupervised clustering, pattern recognition, and UC-based forecasting. The daily GHI time series is first clustered by an Optimized Cross-validated ClUsteRing (OCCUR) method, which determines the optimal number of clusters and best clustering results. Then, support vector machine pattern recognition is adopted to recognize the category of a certain day using the first four hours' data in the forecasting stage. GHI forecasts are generated by the most suitable models in different clusters, which are built by a two-layer machine learning based multi-model (M3) forecasting framework. The developed UC-M3 method is validated by using 1-year of data with 13 solar features from three information sources. Numerical results show that 1) UC-based models outperform non-UC (all-in-one) models with the same M3 architecture by approximately 20%; and 2) M3-based models also outperform the single-algorithm machine learning models by approximately 20%.

14 SOLAR ENERGY↗
DOE OSTI2024

Cy-Phy ADS: Cyber Physical Anomaly Detection Framework for EV Charging Systems

Today’s large-scale Electric Vehicle (EV) infrastructures are heavily dependent on information communication technologies to maintain their operation and to support communication within sub-system components as well as the outside world. These technologies are vulnerable to various cyber and physical threats. Timely identification and mitigation of these threats are critical for improving human safety, avoiding economic losses, and preventing catastrophic system failures. By addressing this, our work presents a ResNet Autoencoder (AE) based Cyber-Physical Anomaly Detection System (Cy-Phy ADS) for detecting anomalies in EV Controller Area Network (CAN) protocol communication. It consists of four main components: Cyber-Physical Feature Extractor, ResNet AE-based Anomaly Detection Framework, Cyber-Physical Health Metric (CPHM), and Visualization Dashboard. The presented framework was trained and tested using CAN data collected from the EV charging system testbed at the Idaho National Laboratory. The presented Cy-Phy ADS compared against six widely used unsupervised anomaly detection algorithms: One Class Support Vector Machine (OCSVM), Variational Autoencoder (VAE), LSTM Autoencoder (LSTM AE), Isolation Forest (IForest), Principle Component Analysis (PCA) and Local Outlier Factor (LOF). Here the presented approach showed the highest accuracy among the compared methods. Further, the proposed approach showed comparable performance in terms of precision, F1, and False positive rate. It also showed the lowest training and inference time compared to the neural network-based baseline algorithms compared against with. Additionally, the Cy-Phy ADS has advantages such as unsupervised training, the ability to provide a holistic metric for system health characterization, and non-linear feature extraction.

99 GENERAL AND MISCELLANEOUS↗
DOE OSTI2018

Rapid determination of soil classes in soil profiles using vis–NIR spectroscopy and multiple objectives mixed support vector classification

Summary Visible‐near infrared (vis–NIR) spectroscopy can reveal various soil properties and facilitate soil classification. However, few studies have attempted to classify vertical soil profiles that contain several genetic horizons. Here, we propose the ‘multiple objectives mixed support vector classification’ (MOM–SVC) method to classify soil profiles. A total of 130 soil profiles were collected from genetic horizons in Zhejiang Province, China. After laboratory analysis, soil profiles were classified according to the Chinese Soil Taxonomy system. Vis–NIR spectra were recorded from each genetic horizon of each soil profile and were then pre‐processed. We performed the MOM–SVC method as follows: (i) created a support vector machine (SVM) model (one‐versus‐one approach) using spectral data from all soil horizons in calibration profiles, (ii) applied the SVM model on each horizon of the profile to be predicted, (iii) extracted ‘votes’ from each horizon and mixed (or summarized) them into the votes of each profile to be predicted and (iv) classified each profile by the majority‐voting method. We also investigated whether the additional input of auxiliary soil information (e.g. moist soil colour, soil organic matter and soil texture), which could be measured easily or be well predicted by vis–NIR spectroscopy, could improve the accuracy of soil classification when combined with it. Independent validation results showed that the MOM–SVC method performed better at the soil order level than at the suborder level. Adding auxiliary soil information to the classification model improved the overall accuracy of classification at the soil order level. The proposed MOM–SVC method provides a fast objective diagnostic of soil classes for use in soil surveys and can help to update soil databases when a more objective soil classification system is developed. Highlights The MOM–SVC method can be used to classify soil profiles objectively with a variety of soil horizons. Stratified random sampling was used to quantify prediction uncertainty in classification MOM–SVC can predict soil orders with greater accuracy than suborders. Adding auxiliary soil information into the classification model improved prediction accuracy.

Chen, S.↗
DOE OSTI2017

Learning an Eddy Viscosity Model Using Shrinkage and Bayesian Calibration: A Jet-in-Crossflow Case Study

In this paper, we demonstrate a statistical procedure for learning a high-order eddy viscosity model (EVM) from experimental data and using it to improve the predictive skill of a Reynolds-averaged Navier–Stokes (RANS) simulator. The method is tested in a three-dimensional (3D), transonic jet-in-crossflow (JIC) configuration. The process starts with a cubic eddy viscosity model (CEVM) developed for incompressible flows. It is fitted to limited experimental JIC data using shrinkage regression. The shrinkage process removes all the terms from the model, except an intercept, a linear term, and a quadratic one involving the square of the vorticity. The shrunk eddy viscosity model is implemented in an RANS simulator and calibrated, using vorticity measurements, to infer three parameters. The calibration is Bayesian and is solved using a Markov chain Monte Carlo (MCMC) method. A 3D probability density distribution for the inferred parameters is constructed, thus quantifying the uncertainty in the estimate. The phenomenal cost of using a 3D flow simulator inside an MCMC loop is mitigated by using surrogate models (“curve-fits”). A support vector machine classifier (SVMC) is used to impose our prior belief regarding parameter values, specifically to exclude nonphysical parameter combinations. The calibrated model is compared, in terms of its predictive skill, to simulations using uncalibrated linear and CEVMs. Finally, we find that the calibrated model, with one quadratic term, is more accurate than the uncalibrated simulator. The model is also checked at a flow condition at which the model was not calibrated.

42 ENGINEERING↗
DOE OSTI2022

Segmenting water and shadow regions within WorldView imagery using local binary patterns

Many shadow detection algorithms pertaining to remotely sensed imagery exist. Several of these algorithms exploit the spectral characteristics of shadows within imagery to identify shaded regions. However, these algorithms can have problems when water is also present within the imagery because water shares similar spectral characteristics with shadows. Many of these algorithms are applied to small image subsets instead of the image as a whole and frequently are applied to urban environments that require additional use of the normalized difference water index or other features to affect the removal of water from the shadow mask. This diversity of the image scene content coupled with the complexity and wide variety of environmental conditions that satellite imagery can acquire makes reliably separating shadow and water within larger images a complex problem. Thresholding various spectral indices to produce segmentation maps can be challenging when large, complex, and often imbalanced scenes are captured, which may require manual adjustment of algorithmic parameters for different image areas. Here, we present a methodology that makes use of the near-infrared channel using a watershed segmentation algorithm, local binary pattern measurements, and a support vector machine to classify shadow and water within a full WorldView scene. Results are promising with initial accuracies above 97% for both shadow and water.

47 OTHER INSTRUMENTATION↗
DOE OSTI2014

SU-E-J-107: Supervised Learning Model of Aligned Collagen for Human Breast Carcinoma Prognosis

Purpose: Our goal is to develop and apply a set of optical and computational tools to enable large-scale investigations of the interaction between collagen and tumor cells. Methods: We have built a novel imaging system for automating the capture of whole-slide second harmonic generation (SHG) images of collagen in registry with bright field (BF) images of hematoxylin and eosin stained tissue. To analyze our images, we have integrated a suite of supervised learning tools that semi-automatically model and score collagen interactions with tumor cells via a variety of metrics, a method we call Electronic Tumor Associated Collagen Signatures (eTACS). This group of tools first segments regions of epithelial cells and collagen fibers from BF and SHG images respectively. We then associate fibers with groups of epithelial cells and finally compute features based on the angle of interaction and density of the collagen surrounding the epithelial cell clusters. These features are then processed with a support vector machine to separate cancer patients into high and low risk groups. Results: We validated our model by showing that eTACS produces classifications that have statistically significant correlation with manual classifications. In addition, our system generated classification scores that accurately predicted breast cancer patient survival in a cohort of 196 patients. Feature rank analysis revealed that TACS positive fibers are more well aligned with each other, generally lower density, and terminate within or near groups of epithelial cells. Conclusion: We are working to apply our model to predict survival in larger cohorts of breast cancer patients with a diversity of breast cancer types, predict response to treatments such as COX2 inhibitors, and to study collagen architecture changes in other cancer types. In the future, our system may be used to provide metastatic potential information to cancer patients to augment existing clinical assays.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2015

Computer-aided detection of renal calculi from noncontrast CT images using TV-flow and MSER features

Purpose: Renal calculi are common extracolonic incidental findings on computed tomographic colonography (CTC). This work aims to develop a fully automated computer-aided diagnosis system to accurately detect renal calculi on CTC images. Methods: The authors developed a total variation (TV) flow method to reduce image noise within the kidneys while maintaining the characteristic appearance of renal calculi. Maximally stable extremal region (MSER) features were then calculated to robustly identify calculi candidates. Finally, the authors computed texture and shape features that were imported to support vector machines for calculus classification. The method was validated on a dataset of 192 patients and compared to a baseline approach that detects calculi by thresholding. The authors also compared their method with the detection approaches using anisotropic diffusion and nonsmoothing. Results: At a false positive rate of 8 per patient, the sensitivities of the new method and the baseline thresholding approach were 69% and 35% (p < 1e − 3) on all calculi from 1 to 433 mm{sup 3} in the testing dataset. The sensitivities of the detection methods using anisotropic diffusion and nonsmoothing were 36% and 0%, respectively. The sensitivity of the new method increased to 90% if only larger and more clinically relevant calculi were considered. Conclusions: Experimental results demonstrated that TV-flow and MSER features are efficient means to robustly and accurately detect renal calculi on low-dose, high noise CTC images. Thus, the proposed method can potentially improve diagnosis.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2015

Predicting adenocarcinoma recurrence using computational texture models of nodule components in lung CT

Purpose: To investigate the importance of presurgical computed tomography (CT) intensity and texture information from ground-glass opacities (GGO) and solid nodule components for the prediction of adenocarcinoma recurrence. Methods: For this study, 101 patients with surgically resected stage I adenocarcinoma were selected. During the follow-up period, 17 patients had disease recurrence with six associated cancer-related deaths. GGO and solid tumor components were delineated on presurgical CT scans by a radiologist. Computational texture models of GGO and solid regions were built using linear combinations of steerable Riesz wavelets learned with linear support vector machines (SVMs). Unlike other traditional texture attributes, the proposed texture models are designed to encode local image scales and directions that are specific to GGO and solid tissue. The responses of the locally steered models were used as texture attributes and compared to the responses of unaligned Riesz wavelets. The texture attributes were combined with CT intensities to predict tumor recurrence and patient hazard according to disease-free survival (DFS) time. Two families of predictive models were compared: LASSO and SVMs, and their survival counterparts: Cox-LASSO and survival SVMs. Results: The best-performing predictive model of patient hazard was associated with a concordance index (C-index) of 0.81 ± 0.02 and was based on the combination of the steered models and CT intensities with survival SVMs. The same feature group and the LASSO model yielded the highest area under the receiver operating characteristic curve (AUC) of 0.8 ± 0.01 for predicting tumor recurrence, although no statistically significant difference was found when compared to using intensity features solely. For all models, the performance was found to be significantly higher when image attributes were based on the solid components solely versus using the entire tumors (p < 3.08 × 10{sup −5}). Conclusions: This study constitutes a novel perspective on how to interpret imaging information from CT examinations by suggesting that most of the information related to adenocarcinoma aggressiveness is related to the intensity and morphological properties of solid components of the tumor. The prediction of adenocarcinoma relapse was found to have low specificity but very high sensitivity. Our results could be useful in clinical practice to identify patients for which no recurrence is expected with a very high confidence using a presurgical CT scan only. It also provided an accurate estimation of the risk of recurrence after a given duration t from surgical resection (i.e., C-index = 0.81 ± 0.02)

60 APPLIED LIFE SCIENCES↗
DOE OSTI2015

Quantitative analysis of breast echotexture patterns in automated breast ultrasound images

Purpose: Breast tissue composition is considered to be associated with breast cancer risk. This study aimed to develop a computer-aided classification (CAC) system to automatically classify echotexture patterns as heterogeneous or homogeneous using automated breast ultrasound (ABUS) images. Methods: A CAC system was proposed that can recognize breast echotexture patterns in ABUS images. For each case, the echotexture pattern was assessed by two expert radiologists and classified as heterogeneous or homogeneous. After neutrosophic image transformation and fuzzy c-mean clusterings, the lower and upper boundaries of the fibroglandular tissues were defined. Then, the number of hypoechoic regions and histogram features were extracted from the fibroglandular tissues, and the support vector machine model with the leave-one-out cross-validation method was utilized as the classifier. The authors’ database included a total of 208 ABUS images of the breasts of 104 females. Results: The accuracies of the proposed system for the classification of heterogeneous and homogeneous echotexture patterns were 93.48% (43/46) and 92.59% (150/162), respectively, with an overall Az (area under the receiver operating characteristic curve) of 0.9786. The agreement between the radiologists and the proposed system was almost perfect, with a kappa value of 0.814. Conclusions: The use of ABUS and the proposed method can provide quantitative information on the echotexture patterns of the breast and can be used to evaluate whether breast echotexture patterns are associated with breast cancer risk in the future.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2015

TU-G-204-02: Automatic Sclerotic Bone Metastases Detection in the Pelvic Region From Dual Energy CT

Purpose: To automatically detect sclerotic bone metastases in the pelvic region using dual energy computed tomography (DECT). Methods: We developed a two stage algorithm to automatically detect sclerotic bone metastases in the pelvis from DECT for patients with multiple bone metastatic lesions and with hip implants. The first stage consists of extracting the bone and marrow regions by using a support vector machine (SVM) classifier. We employed a novel representation of the DECT images using multi-material decomposition, which represents each voxel as a mixture of different physical materials (e.g. bone+water+fat). Following the extraction of bone and marrow, in the second stage, a bi -histogram equalization method was employed to enhance the contrast to reveal the bone metastases. Next, meanshift segmentation was performed to separate the voxels by their intensity levels. Finally, shape-based filtering was performed to extract the possible locations of the metastatic lesions using multiple shape criteria. We used the following shape parameters: area, eccentricity, major and minor axis, perimeter and skeleton. Results: A radiologist with several years of experience with DECT manually labeled 64 regions consisting of metastatic lesions from 10 different patients. However, the patients had many more metastasic lesions throughout the pelvis. Our method correctly identified 46 of the marked 64 regions (72%). In addition, our method also identified several other lesions, which can then be validated by the radiologist. The missed lesions were typically very large elongated regions consisting of several islands of very small (<4mm) lesions. Conclusion: We developed an algorithm to automatically detect sclerotic lesions in the pelvic region from DECT. Preliminary assessment shows that our algorithm generated lesions agreeing with the radiologist generated candidate regions. Furthermore, our method reveals additional lesions that can be inspected by the radiologist, thereby, reducing radiologist effort in identifying all the lesions with poor contrast from the DECT images.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2015

A two-stage method for microcalcification cluster segmentation in mammography by deformable models

Purpose: Segmentation of microcalcification (MC) clusters in x-ray mammography is a difficult task for radiologists. Accurate segmentation is prerequisite for quantitative image analysis of MC clusters and subsequent feature extraction and classification in computer-aided diagnosis schemes. Methods: In this study, a two-stage semiautomated segmentation method of MC clusters is investigated. The first stage is targeted to accurate and time efficient segmentation of the majority of the particles of a MC cluster, by means of a level set method. The second stage is targeted to shape refinement of selected individual MCs, by means of an active contour model. Both methods are applied in the framework of a rich scale-space representation, provided by the wavelet transform at integer scales. Segmentation reliability of the proposed method in terms of inter and intraobserver agreements was evaluated in a case sample of 80 MC clusters originating from the digital database for screening mammography, corresponding to 4 morphology types (punctate: 22, fine linear branching: 16, pleomorphic: 18, and amorphous: 24) of MC clusters, assessing radiologists’ segmentations quantitatively by two distance metrics (Hausdorff distance—HDIST{sub cluster}, average of minimum distance—AMINDIST{sub cluster}) and the area overlap measure (AOM{sub cluster}). The effect of the proposed segmentation method on MC cluster characterization accuracy was evaluated in a case sample of 162 pleomorphic MC clusters (72 malignant and 90 benign). Ten MC cluster features, targeted to capture morphologic properties of individual MCs in a cluster (area, major length, perimeter, compactness, and spread), were extracted and a correlation-based feature selection method yielded a feature subset to feed in a support vector machine classifier. Classification performance of the MC cluster features was estimated by means of the area under receiver operating characteristic curve (Az ± Standard Error) utilizing tenfold cross-validation methodology. A previously developed B-spline active rays segmentation method was also considered for comparison purposes. Results: Interobserver and intraobserver segmentation agreements (median and [25%, 75%] quartile range) were substantial with respect to the distance metrics HDIST{sub cluster} (2.3 [1.8, 2.9] and 2.5 [2.1, 3.2] pixels) and AMINDIST{sub cluster} (0.8 [0.6, 1.0] and 1.0 [0.8, 1.2] pixels), while moderate with respect to AOM{sub cluster} (0.64 [0.55, 0.71] and 0.59 [0.52, 0.66]). The proposed segmentation method outperformed (0.80 ± 0.04) statistically significantly (Mann-Whitney U-test, p < 0.05) the B-spline active rays segmentation method (0.69 ± 0.04), suggesting the significance of the proposed semiautomated method. Conclusions: Results indicate a reliable semiautomated segmentation method for MC clusters offered by deformable models, which could be utilized in MC cluster quantitative image analysis.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2015

A minimum spanning forest based classification method for dedicated breast CT images

Purpose: To develop and test an automated algorithm to classify different types of tissue in dedicated breast CT images. Methods: Images of a single breast of five different patients were acquired with a dedicated breast CT clinical prototype. The breast CT images were processed by a multiscale bilateral filter to reduce noise while keeping edge information and were corrected to overcome cupping artifacts. As skin and glandular tissue have similar CT values on breast CT images, morphologic processing is used to identify the skin based on its position information. A support vector machine (SVM) is trained and the resulting model used to create a pixelwise classification map of fat and glandular tissue. By combining the results of the skin mask with the SVM results, the breast tissue is classified as skin, fat, and glandular tissue. This map is then used to identify markers for a minimum spanning forest that is grown to segment the image using spatial and intensity information. To evaluate the authors’ classification method, they use DICE overlap ratios to compare the results of the automated classification to those obtained by manual segmentation on five patient images. Results: Comparison between the automatic and the manual segmentation shows that the minimum spanning forest based classification method was able to successfully classify dedicated breast CT image with average DICE ratios of 96.9%, 89.8%, and 89.5% for fat, glandular, and skin tissue, respectively. Conclusions: A 2D minimum spanning forest based classification method was proposed and evaluated for classifying the fat, skin, and glandular tissue in dedicated breast CT images. The classification method can be used for dense breast tissue quantification, radiation dose assessment, and other applications in breast imaging.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2016

Computer-aided diagnosis for classifying benign versus malignant thyroid nodules based on ultrasound images: A comparison with radiologist-based assessments

Purpose: To develop a semiautomated computer-aided diagnosis (CAD) system for thyroid cancer using two-dimensional ultrasound images that can be used to yield a second opinion in the clinic to differentiate malignant and benign lesions. Methods: A total of 118 ultrasound images that included axial and longitudinal images from patients with biopsy-confirmed malignant (n = 30) and benign (n = 29) nodules were collected. Thyroid CAD software was developed to extract quantitative features from these images based on thyroid nodule segmentation in which adaptive diffusion flow for active contours was used. Various features, including histogram, intensity differences, elliptical fit, gray-level co-occurrence matrixes, and gray-level run-length matrixes, were evaluated for each region imaged. Based on these imaging features, a support vector machine (SVM) classifier was used to differentiate benign and malignant nodules. Leave-one-out cross-validation with sequential forward feature selection was performed to evaluate the overall accuracy of this method. Additionally, analyses with contingency tables and receiver operating characteristic (ROC) curves were performed to compare the performance of CAD with visual inspection by expert radiologists based on established gold standards. Results: Most univariate features for this proposed CAD system attained accuracies that ranged from 78.0% to 83.1%. When optimal SVM parameters that were established using a grid search method with features that radiologists use for visual inspection were employed, the authors could attain rates of accuracy that ranged from 72.9% to 84.7%. Using leave-one-out cross-validation results in a multivariate analysis of various features, the highest accuracy achieved using the proposed CAD system was 98.3%, whereas visual inspection by radiologists reached 94.9% accuracy. To obtain the highest accuracies, “axial ratio” and “max probability” in axial images were most frequently included in the optimal feature sets for the authors’ proposed CAD system, while “shape” and “calcification” in longitudinal images were most frequently included in the optimal feature sets for visual inspection by radiologists. The computed areas under curves in the ROC analysis were 0.986 and 0.979 for the proposed CAD system and visual inspection by radiologists, respectively; no significant difference was detected between these groups. Conclusions: The use of thyroid CAD to differentiate malignant from benign lesions shows accuracy similar to that obtained via visual inspection by radiologists. Thyroid CAD might be considered a viable way to generate a second opinion for radiologists in clinical practice.

61 RADIATION PROTECTION AND DOSIMETRY↗
DOE OSTI2016

Automatic coronary calcium scoring using noncontrast and contrast CT images

Purpose: Calcium scoring is widely used to assess the risk of coronary heart disease (CHD). Accurate coronary artery calcification detection in noncontrast CT image is a prerequisite step for coronary calcium scoring. Currently, calcified lesions in the coronary arteries are manually identified by radiologists in clinical practice. Thus, in this paper, a fully automatic calcium scoring method was developed to alleviate the work load of the radiologists or cardiologists. Methods: The challenge of automatic coronary calcification detection is to discriminate the calcification in the coronary arteries from the calcification in the other tissues. Since the anatomy of coronary arteries is difficult to be observed in the noncontrast CT images, the contrast CT image of the same patient is used to extract the regions of the aorta, heart, and coronary arteries. Then, a patient-specific region-of-interest (ROI) is generated in the noncontrast CT image according to the segmentation results in the contrast CT image. This patient-specific ROI focuses on the regions in the neighborhood of coronary arteries for calcification detection, which can eliminate the calcifications in the surrounding tissues. A support vector machine classifier is applied finally to refine the results by removing possible image noise. Furthermore, the calcified lesions in the noncontrast images belonging to the different main coronary arteries are identified automatically using the labeling results of the extracted coronary arteries. Results: Forty datasets from four different CT machine vendors were used to evaluate their algorithm, which were provided by the MICCAI 2014 Coronary Calcium Scoring (orCaScore) Challenge. The sensitivity and positive predictive value for the volume of detected calcifications are 0.989 and 0.948. Only one patient out of 40 patients had been assigned to the wrong risk category defined according to Agatston scores (0, 1–100, 101–300, >300) by comparing with the ground truth. Conclusions: The calcified lesions in the noncontrast CT images can be detected automatically by using the segmentation results of the aorta, heart, and coronary arteries obtained in the contrast CT images with a very high accuracy.

60 APPLIED LIFE SCIENCES↗