Independent Test Data Research Articles

Variability of serum novel serum peptide biomarkers correlates with the disease states of multiple myeloma

BackgroundThe bone marrow microenvironment provides an optimal substrate for multiple myeloma (MM) initiation and progression. The soluble component of MM niche is dynamic with the disease states of MM. We formerly employed proteomic profiling to construct a MM model. Four peptides constituting the model were selected by supervised neural network algorithm (SNN).Methods62 Newly diagnosed (ND) MM and 64 healthy controls (HCs) were picked up for validating the distinguishing capability of the SNN model. Nano-liquid chromatography-electrospray ionization-tandem mass spectrometry was used for peptide identification. MM in different disease states and HCs were choosed for peptides relative intensities comparison. Western blot and ELISA were employed to validate the variability.ResultsThe sensitivity and specificity of the independent testing data set for blind validation were 93.55% and 92.19%. The relative intensities of three out of the four peptides were increased in ND and refractory and relapse patients but decreased to that level of HCs in complete remission and very good partial remission patients. Relative intensity of the remaining peptide was negatively associated with MM remission. The peptides sequencing results showed that they were derived from dihydropyrimidinase-like 2, fibrinogen alpha chain, platelet factor 4 and alpha-fetoprotein.ConclusionsThe potential value of the four peptides in monitoring MM treatment response was arised from their close correlation with MM disease states.

Quantifying how landscape composition and configuration affect urban land surface temperatures using machine learning and neutral landscapes

The urban heat island effect is an important 21st century issue because it intersects with the complex challenges of urban population growth, global climate change, public health and increasing energy demand for cooling. While the effects of urban landscape composition on land surface temperature (LST) are well-studied, less attention has been paid to the spatial arrangement of land cover types especially in smaller, often more diverse cities. Landscape configuration is important because it offers the potential to provide refuge from excessive heat for both people and buildings.We present a novel approach to quantifying how both composition and configuration affect LST derived from Landsat imagery in Southampton, UK. First, we trained a machine-learning (generalized boosted regression) model to predict LST from landscape covariates that included the characteristics of the immediate pixel and its surroundings. The model achieved a correlation between predicted and measured LST of 0.956 on independent test data (n = 102,935) and included predictors for both the immediate and adjacent land use. In contrast to other studies, we found adjacency effects to be stronger than immediate effects at 30 m resolution. Next, we used a landscape generation tool (Landscape Generator) to alter landscape configuration by varying natural and built patch sizes and arrangements while holding composition constant. The generated neutral landscapes were then fed into the machine learning model to predict patterns of LST.When we manipulated landscape configuration, the average city temperature remained the same but the local minima varied by 0.9 °C and the maxima by 4.2 °C. The effects on LST and heat island metrics correlated with landscape fragmentation indices. Moreover, the surface temperature of buildings could be reduced by up to 2.1 °C through landscape manipulation.We found that the optimum mix of land use types is neither at the land-sharing nor land-sparing extremes, but a balance between the two. In our city, maximum cooling was achieved when ~60% of land was left natural and distributed in 7–8 patches km−2 although this could be location dependent and further work is needed. Opportunities for urban cooling should be required in the planning process and must consider both composition and configuration at the landscape scale if cities are to build capacity for a growing population and climate change.

Robust metamodels for accurate quantitative estimation of turbulent flow in pipe bends

Pipe bends are inevitable in industrial piping systems, turbomachinery, heat exchangers, etc. Computational fluid dynamics (CFD), which is commonly employed to understand the flow behavior in such systems has very accurate estimation but is computationally cost intensive. Thus, in this paper, an efficient computational approach for such computationally expensive problems is presented. Using genetic programming (GP), metamodels are built using a small number of samples points from the CFD data. These GP metamodels are then shown to be able to replace the actual CFD models with considerable accuracy. The applicability and suitability of the GP metamodels are validated using a variety of statistical metrics on the training as well as independent test data. It is shown that the use of metamodels leads to significant savings in computational cost.

A Comparative Assessment of Geostatistical, Machine Learning, and Hybrid Approaches for Mapping Topsoil Organic Carbon Content

Accurate digital soil mapping (DSM) of soil organic carbon (SOC) is still a challenging subject because of its spatial variability and dependency. This study is aimed at comparing six typical methods in three types of DSM techniques for SOC mapping in an area surrounding Changchun in Northeast China. The methods include ordinary kriging (OK) and geographically weighted regression (GWR) from geostatistics, support vector machines for regression (SVR) and artificial neural networks (ANN) from machine learning, and geographically weighted regression kriging (GWRK) and artificial neural networks kriging (ANNK) from hybrid approaches. The hybrid approaches, in particular, integrated the GWR from geostatistics and ANN from machine learning with the estimation of residuals by ordinary kriging, respectively. Environmental variables, including soil properties, climatic, topographic, and remote sensing data, were used for modeling. The mapping results of SOC content from different models were validated by independent testing data based on values of the mean error, root mean squared error and coefficient of determination. The prediction maps depicted spatial variation and patterns of SOC content of the study area. The results showed the accuracy ranking of the compared methods in decreasing order was ANNK, SVR, ANN, GWRK, OK, and GWR. Two-step hybrid approaches performed better than the corresponding individual models, and non-linear models performed better than the linear models. When considering the uncertainty and efficiency, ML and two-step approach are more suitable than geostatistics in regional landscapes with the high heterogeneity. The study concludes that ANNK is a promising approach for mapping SOC content at a local scale.

Landslide susceptibility: a statistically-based assessment on a depositional pyroclastic ramp

This study aimed to produce a high-quality landslide susceptibility map for Teziutlan municipality, a landslide-prone region in Mexico, which is characterised by a depositional pyroclastic ramp. The heterogeneous quality of available topographic information (i.e. higher resolution digital elevation model only for a sub-region) encouraged to confront modelling results based on two different study area delineations and two raster resolutions. Input data was based on the larger modelling region L15 (163 km2) and smaller S (70 km2; located inside L15) with an associated raster cell size of 15 m (region L15 and S15) and 5 m (region S5). The resulting three data sets (L15, S15 and S5) were included into three differently flexible modelling techniques (Generalized Linear Model - GLM, General Additive Model - GAM, Support Vector Machine-SVM) to produce nine landslide susceptibility models. Preceding variable selection was performed heuristically and supported by an exploratory data analysis. The final models were based on the explanatory variables slope angle, slope aspect, lithology, relative slope position, elevation, convergence index, distance to streams, distance to springs and topographic wetness index. The ability of the models to classify independent test data was elaborated using a k-fold cross validation procedure and the AUROC (Area Under the Receiver Operating Characteristic) metric. In general, all produced landslide susceptibility maps depicted the hillslopes of the ravines, which cut the pyroclastic ramp, as prone to landsliding. The modelling results showed that predictive performances (i.e. AUROC values) slightly increased with an increasing flexibility of the applied modelling technique. Thus, SVM performed best, while the GAM outperformed the GLM. This tendency was most distinctive when modelling with the largest landslide sample size (i.e. data set L15; n = 662 landslides). Non-linear classifiers (GAMs, SVMs) performed slightly better when trained on the basis of lower raster resolution (data set S15) compared to the 5 m counterparts (data set S5). Highest predictive performance was obtained for the model based on data set L15 and the SVM classifier (median AUROC: 0.82). However, SVMs also indicated the highest degree of model overfitting. This study indicates that the decision to delineate a study area, the selection of a raster resolution as well as the chosen classification technique can affect varying aspects of subsequent modelling results. The results do not support the assumption that a higher raster resolution (i.e. a more detailed digital representation of the terrain) inevitably leads to better performing or geomorphically more plausible landslide susceptibility maps.

NCBoost classifies pathogenic non-coding variants in Mendelian diseases through supervised learning on purifying selection signals in humans

State-of-the-art methods assessing pathogenic non-coding variants have mostly been characterized on common disease-associated polymorphisms, yet with modest accuracy and strong positional biases. In this study, we curated 737 high-confidence pathogenic non-coding variants associated with monogenic Mendelian diseases. In addition to interspecies conservation, a comprehensive set of recent and ongoing purifying selection signals in humans is explored, accounting for lineage-specific regulatory elements. Supervised learning using gradient tree boosting on such features achieves a high predictive performance and overcomes positional bias. NCBoost performs consistently across diverse learning and independent testing data sets and outperforms other existing reference methods.

Machine-learning error models for approximate solutions to parameterized systems of nonlinear equations

This work proposes a machine-learning framework for constructing statistical models of errors incurred by approximate solutions to parameterized systems of nonlinear equations. These approximate solutions may arise from early termination of an iterative method, a lower-fidelity model, or a projection-based reduced-order model, for example. The proposed statistical model comprises the sum of a deterministic regression-function model and a stochastic noise model. The method constructs the regression-function model by applying regression techniques from machine learning (e.g., support vector regression, artificial neural networks) to map features (i.e., error indicators such as sampled elements of the residual) to a prediction of the approximate-solution error. The method constructs the noise model as a mean-zero Gaussian random variable whose variance is computed as the sample variance of the approximate-solution error on a test set; this variance can be interpreted as the epistemic uncertainty introduced by the approximate solution. This work considers a wide range of feature-engineering methods, data-set-construction techniques, and regression techniques that aim to ensure that (1) the features are cheaply computable, (2) the noise model exhibits low variance (i.e., low epistemic uncertainty introduced), and (3) the regression model generalizes to independent test data. Numerical experiments performed on several computational-mechanics problems and types of approximate solutions demonstrate the ability of the method to generate statistical models of the error that satisfy these criteria and significantly outperform more commonly adopted approaches for error modeling.

Serum Metabolome Is Associated With the Nasopharyngeal Microbiota and Disease Severity Among Infants With Bronchiolitis.

Emerging evidence suggests relationships between the nasopharyngeal metabolome and both the microbiota and severity of bronchiolitis. However, the influence of host systemic metabolism on disease pathobiology remains unclear. We aimed to examine metabolome profiles and their association with more-severe disease, defined by use of positive pressure ventilation (PPV), in infants hospitalized for bronchiolitis. In 140 infants with bronchiolitis, metabolomic profiling was performed on serum; samples from 70 were in a training data set, and samples from 70 were in an independent test data set. We also profiled the nasopharyngeal airway microbiota and examined its association with the serum metabolites. Serum metabolome profiles differed by bronchiolitis severity (P < .001). In total, 20 metabolites in the training data set were significantly associated with the risk of PPV, of which 18 remained significant following adjustment for confounders (false-discovery rate [FDR], < 0.10). Phosphatidylcholine metabolites were associated with higher risks of PPV use, while metabolites from the plasmalogen subpathway were associated with lower risks. The test data set validated these findings (FDR < 0.05). Streptococcus abundance was positively associated with metabolites that are associated with higher risks of PPV. Serum metabolomic signatures were associated with both the nasopharyngeal microbiota and the severity of bronchiolitis. Our findings advance research into the complex interrelations between the airway microbiome, host systemic response, and pathobiology of bronchiolitis.

Usability of one-class classification in mapping and detecting changes in bare peat surfaces in the tundra

ABSTRACTArctic areas have experienced greening and changes in permafrost caused by climate change during recent decades. However, there has been a lack of automated methods in mapping changes in fine-scale patterns of permafrost landscapes. We mapped areal coverage of bare peat areas and changes in them in a peat plateau located in north-western Russia between 2007 and 2015. We utilized QuickBird and WorldView-3 satellite image data in an object-based setting. We compared four different one-class classifiers (one-class support vector machine, binary support vector machine, random forest, rotation forest) both in a fully supervised binary setting and with positive and unlabelled training data. There was notable variation in classification performance. The bare peat area F-score varied between 0.77 and 0.96 when evaluated by cross-validated training data and between 0.22 and 0.57 when evaluated by independent test data. Overall, random forest performed the most robustly but all classifiers performed well in some classifications. During the 8 year period, there was a 21%–26% decrease in the bare peat areal coverage. We conclude that (1) tested classifiers can be used in one-class settings and (2) there is a need to develop methods for tracking changes in single land cover types.

Automatic Prostate Zonal Segmentation Using Fully Convolutional Network With Feature Pyramid Attention

Our main objective is to develop a novel deep learning-based algorithm for automatic segmentation of prostate zone and to evaluate the proposed algorithm on an additional independent testing data in comparison with inter-reader consistency between two experts. With IRB approval and HIPAA compliance, we designed a novel convolutional neural network (CNN) for automatic segmentation of the prostatic transition zone (TZ) and peripheral zone (PZ) on T2-weighted (T2w) MRI. The total study cohort included 359 patients from two sources; 313 from a deidentified publicly available dataset (SPIE-AAPM-NCI PROSTATEX challenge) and 46 from a large U.S. tertiary referral center with 3T MRI (external testing dataset (ETD)). The TZ and PZ contours were manually annotated by research fellows, supervised by genitourinary (GU) radiologists. The model was developed using 250 patients and tested internally using the remaining 63 patients from the PROSTATEX (internal testing dataset (ITD)) and tested again (n=46) externally using the ETD. The Dice Similarity Coefficient (DSC) was used to evaluate the segmentation performance. DSCs for PZ and TZ were 0.74 and 0.86 in the ITD respectively. In the ETD, DSCs for PZ and TZ were 0.74 and 0.792, respectively. The inter-reader consistency (Expert 2 vs. Expert 1) were 0.71 (PZ) and 0.75 (TZ). This novel DL algorithm enabled automatic segmentation of PZ and TZ with high accuracy on both ITD and ETD without a performance difference for PZ and less than 10% TZ difference. In the ETD, the proposed method can be comparable to experts in the segmentation of prostate zones.

23 - MACHINE LEARNING OPTIMISED FOR PERSONALIZED MEDICINE: PREDICTING LIFETIME AND RECURRENT DEPRESSION IN THE GENERATION SCOTLAND: SCOTTISH FAMILY HEALTH STUDY

Background Major Depressive Disorder (MDD) is one of the most common psychiatric disorders, with a prevalence of ~15%. Diagnosis is often inaccurate, based on self-report and is largely under-diagnosed. Data-driven approaches to predict lifetime risk for MDD and single versus recurrent MDD, especially using a small number of highly-accurate predictors that could be easily collected in clinic, would be a step forward in realising the promises of personalised medicine. We applied machine learning algorithms (MLAs) to the Generation Scotland cohort study (N > 21,000), a deeply-phenotyped cohort. The cohort was first divided into a training (63%) and held-out test (37%) set for both lifetime and single versus recurrent MDD analyses. Methods In the training set, 10-fold cross-validation was used to estimate optimal hyperparameters for the following algorithms: Random Forest, Conditional Inference Forest, Gradient Descent Boosting (GBM), Support Vector Machines (with linear, polynomial and radial basis function kernels), Neural Networks, C5.0, Elastic Net and Forward Stepwise Regression. Using the optimal hyperparameters, we ran each MLA on the full training set, then ran the independent test data through the best model derived from the training data to predict case-control or single-recurrent status. Receiving Operator Characteristic Area Under the Curve (AUC) values were used to assess performance of the algorithms on the test data. To obtain a reduced set of predictors with optimised predictive value, the Markov Chain 4 (MC4) algorithm was used. MC4 is a rank aggregation algorithm, originally designed for meta-Internet search rankings. The aggregated rankings of predictors across all MLAs were then used to determine the smallest set with equal or better predictive value on the test data versus using all of the predictors. Results AUC values for the prediction of lifetime MDD were in the range of 0.81–0.84 for all MLAs; similarly, the AUC range for single versus recurrent MDD were between 0.69 and 0.76. All AUCs were significantly better than expected by chance (AUC = 0.5). Using the MC4 ranked variables and the best-performing model (GBM), we found that equal performance was obtained using only 20 variables for lifetime MDD (AUC = 0.84) compared to 155 variables in the full set, and 10 for single versus recurrent MDD (AUC = 0.76) compared to 180 variables in the full set. Discussion The MC4 ranked subsets performed equally well to the full subset, although it is likely other subsets exist that could perform similarly. The subsets identified for lifetime MDD included neuroticism, general psychological distress, age, income, family history of depression, living alone, sex, home ownership, smoking, education, pain status, mania and schizotypy. Age at MDD onset, neuroticism, psychological distress, measures of cognition, age, smoking and home ownership were the most relevant predictors for whether someone would have a single episode of MDD or recurrent episodes. These highly-predictive questionnaires and demographic information can be easily collected in clinic to assist in accurate diagnosis and preventative treatment for recurrent MDD.

Detection of spoilage associated bacteria using Raman-microspectroscopy combined with multivariate statistical analysis

Raman-Microspectroscopy with subsequent chemometric evaluation was used for the rapid and non-destructive differentiation of seven important spoilage related microorganisms, namely Brochothrix thermosphacta DSM 20171, Pseudomonas fluorescens DSM 4358, Pseudomonas fluorescens DSM 50090, Micrococcus luteus, Escherichia coli HB101, Escherichia coli TOP10 and Bacillus thuringiensis israelensis DSM 5724. Therefore fast collected spectra directly from rapid surface blots without any pretreatments like purification or singulation steps were used. To estimate and classify the Raman-spectroscopic data at genera and strain level an adequate preprocessing together with a subsequent chemometric evaluation consisting of principal component analysis and discriminant analysis was used. Thereby, importance was attached to a balanced data set, as this makes the multivariate analysis of the data significantly more resilient and meaningful. The analysis showed that the differentiation of spoilage related microorganisms on genera and strain level was successful and the classification of independent test data showed only an error rate of 3.5%.

New Criteria for Safety Distances During Pneumatic Pressure Testing of Vessels and Pipes

The industrial gas and liquefied natural gas (LNG) industries routinely perform equipment pressure testing pneumatically, rather than hydraulically, due to the necessity for keep piping and equipment dry. Hazards associated with potential failure of pneumatically pressurized equipment under test are well understood, but today, there is no commonly recognized industry guidance on recommended safety distances to protect personnel during such tests. Air Products, Air Liquide and BakerRisk have worked together to develop such guidance for the testing of vessels, process pipes and pipelines, and present in this article new, simple to use correlations, with results also presented in the form of lookup tables. These new correlations are based on the application of established methods, validated against previously published independent test data and new test data presented here for the first time. © 2018 American Institute of Chemical Engineers Process Saf Prog 38: e12025, 2019

A novel bagging ensemble approach for predicting summertime ground-level ozone concentration

ABSTRACTOzone pollution appears as a major air quality issue, e.g. for the protection of human health and vegetation. Formation of ground level ozone is a complex photochemical phenomenon and involves numerous intricate factors most of which are interrelated with each other. Machine learning techniques can be adopted to predict the ground level ozone. The main objective of the present study is to develop the state-of-the-art ensemble bagging approach to model the summer time ground level ozone in an industrial area comprising a hazardous waste management facility. In this study, the feasibility of using ensemble model with seven meteorological parameters as input variables to predict the surface level O3 concentration. Multilayer perceptron, RTree, REPTree, and Random forest were employed as the base learners. The error measures used for checking the performance of each model includes IoAd, R2, and PEP. The model results were validated against an independent test data set. Bagged random forest predicted the ground level ozone better with higher Nash-Sutcliffe coefficient 0.93. This study scaffolded the current research gap in big data analysis identified with air pollutant prediction.Implications: The main focus of this paper is to model the summer time ground level O3 concentration in an Industrial area comprising of hazardous waste management facility. Comparison study was made between the base classifiers and the ensemble classifiers. Most of the conventional models can well predict the average concentrations. In this case the peak concentrations are of importance as it has serious effect on human health and environment. The models developed should also be homoscedastic.

Added Value of Radiomics on Mammography for Breast Cancer Diagnosis: A Feasibility Study

BackgroundThis study aimed to evaluate whether radiomics can improve the diagnostic performance of mammography compared with that obtained by experienced radiologists. MethodsThis retrospective study included 173 patients (with 74 benign and 99 malignant lesions) who underwent mammography examination before neoadjuvant chemotherapy. Radiomic features were extracted from the mammography image of each patient. Several preprocessing methods, including centering and normalization, were used along with statistical analysis to reduce and select radiomic features. Four machine learning algorithms, namely, support vector machine, logistic regression, K-nearest neighbor, and Bayes classification, were applied to construct a predictive model. An independent testing data set was used to validate the prediction ability of the model. The classification performance was compared with the diagnostic predictions of two breast radiologists who had access to the same mammography cases. ResultsA total of 51 radiomic features remained after the preprocessing. Logistic regression classification presented the best differentiation ability among the four regression models. The diagnostic accuracy, specificity, and sensitivity of the logistic regression model for the training data set were 0.978, 0.975, and 0.983, respectively. The diagnostic accuracy, specificity, and sensitivity for the testing data set were 0.886, 0.900, and 0.867, respectively. The accuracy, specificity, and sensitivity of the combined reading of the two radiologists were 0.772, 0.710, 0.862 in the training data set and 0.769, 0.695, 0.858 in the testing data set, respectively. ConclusionsMammography images could be captured and quantified by radiomics, which offers a good diagnostic ability for benign and malignant breast tumors and provides complementary information to radiologists.

Gene2vec: gene subsequence embedding for prediction of mammalian N6-methyladenosine sites from mRNA.

N6-Methyladenosine (m6A) refers to methylation modification of the adenosine nucleotide acid at the nitrogen-6 position. Many conventional computational methods for identifying N6-methyladenosine sites are limited by the small amount of data available. Taking advantage of the thousands of m6A sites detected by high-throughput sequencing, it is now possible to discover the characteristics of m6A sequences using deep learning techniques. To the best of our knowledge, our work is the first attempt to use word embedding and deep neural networks for m6A prediction from mRNA sequences. Using four deep neural networks, we developed a model inferred from a larger sequence shifting window that can predict m6A accurately and robustly. Four prediction schemes were built with various RNA sequence representations and optimized convolutional neural networks. The soft voting results from the four deep networks were shown to outperform all of the state-of-the-art methods. We evaluated these predictors mentioned above on a rigorous independent test data set and proved that our proposed method outperforms the state-of-the-art predictors. The training, independent, and cross-species testing data sets are much larger than in previous studies, which could help to avoid the problem of overfitting. Furthermore, an online prediction web server implementing the four proposed predictors has been built and is available at http://server.malab.cn/Gene2vec/.

Ultra-high strength circular short CFST columns: Axisymmetric analysis, behaviour and design

A nonlinear axisymmetric simulation of axially loaded ultra-high strength circular short Concrete-Filled Steel Tubular (CFST) columns is reported in this paper. An efficient material law of Ultra-High Strength Concrete (UHSC) is employed in this axisymmetric simulation to consider the confining effects which increase the ultimate strain, stress and axial ductility. The peak loads and load-strain responses of short CFST columns predicted by the axisymmetric model are verified with the independent test data and those predicted by the three-dimensional (3D) model. The structural characteristics of circular short CFST columns having different material and geometric parameters are subsequently examined by the axisymmetric analysis. The numerical results indicate that the column size can approximately be reduced by 50% when UHSC is used in the composite construction instead of Normal Strength Concrete (NSC). Comparisons of the current test data are made with the current structural codes including AS/NZS 5100.6, Eurocode 4, AISC and GB 50936. The comparisons demonstrate that the Australian/New Zealand and European design codes provide consistent predictions for the ultimate axial strengths of ultra-high strength circular short CFST columns. On the other hand, the North American design code is conservative, whereas the Chinese code is unconservative.

More Than Meets the Eye: Using Sentinel-2 to Map Small Plantations in Complex Forest Landscapes

Many tropical forest landscapes are now complex mosaics of intact forests, recovering forests, tree crops, agroforestry, pasture, and crops. The small patch size of each land cover type contributes to making them difficult to separate using satellite remote sensing data. We used Sentinel-2 data to conduct supervised classifications covering seven classes, including oil palm, rubber, and betel nut plantations in Southern Myanmar, based on an extensive training dataset derived from expert interpretation of WorldView-3 and UAV data. We used a Random Forest classifier with all 13 Sentinel-2 bands, as well as vegetation and texture indices, over an area of 13,330 ha. The median overall accuracy of 1000 iterations was &gt;95% (95.5%–96.0%) against independent test data, even though the tree crop classes appear visually very similar at a 20 m resolution. We conclude that the Sentinel-2 data, which are freely available with very frequent (five day) revisits, are able to differentiate these similar tree crop types. We suspect that this is due to the large number of spectral bands in Sentinel-2 data, indicating great potential for the wider application of Sentinel-2 data for the classification of small land parcels without needing to resort to object-based classification of higher resolution data.

Machine learning-based prediction of clinical pain using multimodal neuroimaging and autonomic metrics.

Although self-report pain ratings are the gold standard in clinical pain assessment, they are inherently subjective in nature and significantly influenced by multidimensional contextual variables. Although objective biomarkers for pain could substantially aid pain diagnosis and development of novel therapies, reliable markers for clinical pain have been elusive. In this study, individualized physical maneuvers were used to exacerbate clinical pain in patients with chronic low back pain (N = 53), thereby experimentally producing lower and higher pain states. Multivariate machine-learning models were then built from brain imaging (resting-state blood-oxygenation-level-dependent and arterial spin labeling functional imaging) and autonomic activity (heart rate variability) features to predict within-patient clinical pain intensity states (ie, lower vs higher pain) and were then applied to predict between-patient clinical pain ratings with independent training and testing data sets. Within-patient classification between lower and higher clinical pain intensity states showed best performance (accuracy = 92.45%, area under the curve = 0.97) when all 3 multimodal parameters were combined. Between-patient prediction of clinical pain intensity using independent training and testing data sets also demonstrated significant prediction across pain ratings using the combined model (Pearson's r = 0.63). Classification of increased pain was weighted by elevated cerebral blood flow in the thalamus, and prefrontal and posterior cingulate cortices, and increased primary somatosensory connectivity to frontoinsular cortex. Our machine-learning approach introduces a model with putative biomarkers for clinical pain and multiple clinical applications alongside self-report, from pain assessment in noncommunicative patients to identification of objective pain endophenotypes that can be used in future longitudinal research aimed at discovery of new approaches to combat chronic pain.

Deep Learning-based Method for Fully Automatic Quantification of Left Ventricle Function from Cine MR Images: A Multivendor, Multicenter Study.

Purpose To develop a deep learning-based method for fully automated quantification of left ventricular (LV) function from short-axis cine MR images and to evaluate its performance in a multivendor and multicenter setting. Materials and Methods This retrospective study included cine MRI data sets obtained from three major MRI vendors in four medical centers from 2008 to 2016. Three convolutional neural networks (CNNs) with the U-NET architecture were trained on data sets of increasing variability: (a) a single-vendor, single-center, homogeneous cohort of 100 patients (CNN1); (b) a single-vendor, multicenter, heterogeneous cohort of 200 patients (CNN2); and (c) a multivendor, multicenter, heterogeneous cohort of 400 patients (CNN3). All CNNs were tested on an independent multivendor, multicenter data set of 196 patients. CNN performance was evaluated with respect to the manual annotations from three experienced observers in terms of (a) LV detection accuracy, (b) LV segmentation accuracy, and (c) LV functional parameter accuracy. Automatic and manual results were compared with the paired Wilcoxon test, Pearson correlation, and Bland-Altman analysis. Results CNN3 achieved the highest performance on the independent testing data set. The average perpendicular distance compared with manual analysis was 1.1 mm ± 0.3 for CNN3, compared with 1.5 mm ± 1.0 for CNN1 (P < .05) and 1.3 mm ± 0.6 for CNN2 (P < .05). The LV function parameters derived from CNN3 showed a high correlation (r2 ≥ 0.98) and agreement with those obtained by experts for data sets from different vendors and centers. Conclusion A deep learning-based method trained on a data set with high variability can achieve fully automated and accurate cine MRI analysis on multivendor, multicenter cine MRI data. © RSNA, 2018 See also the editorial by Colletti in this issue.