Search

With the goal of developing an accurate and fast image reconstruction algorithm for ultrasound computed tomography, we combine elements of model- and data-driven approaches and propose a learned method which addresses the disadvantages of both approaches. We design a deep neural network which accounts for a nonlinear forward operator and primal-dual algorithm by its inherent network architecture. The network is trained end-to-end, with ultrasound pressure field data as input to get directly an optimized reconstruction of speed of sound and attenuation images. The training and test data are based on a set of Optical and Acoustic Breast Phantom Database, where we use the image as ground truth and simulate pressure field data according to our forward model. Extensive experiments show that our method achieves significant improvements over state-of-the-art reconstruction methods in this field. Experiments show that the proposed algorithm improves the measures structural similarity measure (SSIM) from 0.74 to 0.95 and root mean squared error (RMSE) from 0.13 to 0.09 on average concerning the speed of sound reconstruction, while it improves the SSIM from 0.60 to 0.94 and RMSE from 0.24 to 0.10 on average in attenuation reconstruction.

This Letter presents a novel ultra-wide-band polarization-selective reflectarray with true-time delay. The reflectarray can be perceived as a combination of polarization grids and Vivaldi antennas, so is characterized by large bandwidth, low cross polarization, and polarization selectivity. A reflectarray of elements arranged in a 41 41 square grid operating from 6 to 32 GHz is designed, fabricated, and tested. The feed is provided by four pyramidal horn antennas operating at the working frequency (C, X, Ku/K, and Ka band). In measurements, the reflectarray shows stable radiation patterns from 6 to 32 GHz (more than 5:1 frequency range) and has a maximum aperture efficiency of 40%. It also has the low cross polarization below 28 dB across the entire band and has the characteristic of polarization selectivity.

OBJECTIVES We aimed to explore the impact of 7-Fr sheaths on the incidence of early radial artery occlusion (RAO) after transradial coronary intervention (TRI) in Chinese patients. BACKGROUND RAO precludes future use of the vessel for vascular access. Transradial catheterization is usually performed via 5-Fr or 6-Fr catheters; 7-Fr sheath insertion enables complex coronary interventions but may increase the RAO risk. METHODS We prospectively enrolled 130 consecutive patients undergoing complex TRI using 7-Fr sheaths. Radial artery ultrasound assessment was performed before and after TRI. Early RAO was defined as the absence of flow on ultrasound within 6-24 hr after TRI. Multivariate logistic regression was used to determine the factors related to early RAO after TRI. RESULTS 7-Fr sheaths were mainly used for chronic total occlusion (44.6%), bifurcation (30.0%), and tortuous calcification (25.4%) lesions. All patients were successfully sheathed. Percutaneous coronary intervention (PCI) procedural success was 96.2%; 119 patients (91.5%) had preserved radial artery patency after TRI. All 11 RAO cases (8.5%) were asymptomatic. The radial artery diameter was significantly larger postoperatively (3.1 ± 0.4 mm) than preoperatively (2.6 ± 0.5 mm) (p

With the vigorous development of deep learning and the widespread use of mobile robots, automatic driving has gradually become a research hotspot. Environment perception is the most important part of automatic driving technology, and the purpose of environment perception is to distinguish the environmental content. Therefore, accurate and efficient image semantic segmentation method is becoming more and more important. In this paper, we introduce a deep semantic segmentation solution: IIE-SegNet: Deep semantic segmentation network with enhanced boundary based on image information entropy. At present, deep learning based on semantic segmentation solutions has some problems, such as low segmentation accuracy for small-scale objects and unclear boundary of segmented objects. Our method preserves the boundary of the segmentation object, and has higher segmentation accuracy for small-scale objects. In our method, the features of the underlying pooling layer are added to the ASPP structure of the encoding module, and the image information entropy of the previous pooling layers is introduced into the decoding module. We also introduce focal loss to solve the problem of imbalance between positive and negative samples. Finally, the test results on the extended Pascal VOC 2012 test set, abbreviated to Exp-Pascal VOC 2012 test set show that the proposed method has better performance on Exp-Pascal VOC 2012 test set compared with the advanced methods at the present stage, the segmentation accuracy of small-scale targets is higher, and the boundary is clearer.

A novel serrated ground plane (NSGP) for na low-loss reflectarray antenna is presented in this paper. Compared with a conventional smooth ground plane (CGP), the NSGP consists of a series of serrated elements, which can reflect the incident waves in the main beam direction, so the losses of the reflectarray can be effectively reduced. The principle and losses of reflectarray antennas are studied and analyzed. Then, a low-loss NSGP is proposed, and two design methods for the NSGP are given in this paper. Finally, a reflectarray antenna with elements arranged in a 15×15 grid is designed, simulated and measured with the NSGP and CGP respectively within the frequency from 12.88 to 13.88GHz. The results show that the reflectarray antenna with the NSGP can effectively utilize the reflected waves and has a maximum higher gain of 0.681 dB compared with the gain of the reflectarray with the CGP. This NSGP has a potential to be used in the high accurate design of the reflectarray which requires to realize beam forming, low-loss, high-efficiency, etc.

Visual odometry (VO) is one of the essential techniques in mobile robots field; an accurate VO system is of great significance for mobile robot simultaneous localization and mapping. As for traditional monocular VO systems, they work by presuming the monocular scale is 1 (scale = 1), or relying on ground truth (GT) to estimate scale. As a result, the traditional monocular VO systems estimate the pose state with big drift or cannot work on the image sequence without GT. Although some classical monocular VO systems have been proposed, they still have imperfect performance or even unable to work in some extreme scene conditions, such as scene is monotony without obvious texture information or camera large-scale displacement motion. As for learning-based VO system, it is realized by training deep neural networks in supervised or self-supervised manner to end-to-end estimate the pose state; however, the accuracy of pose estimation entirely depends on the ability of networks. Although the ability of networks can be improved by increasing the number of training data sets and optimizing the network structure, it is inevitable to encounter problems such as insufficient generalization ability and insufficient accuracy on rotational pose estimation. In this article, a monocular VO system named DL_Hybrid is proposed, which takes full advantage of DL networks used in image processing and geometric localization theory based on hybrid pose estimation methods. The DL_Hybrid VO system can estimate a six-DoF pose one-frame-by-one-frame and recover camera trajectory, and it can extract accurate key points from per-frame even in extreme scene condition, and it has good performance even in the extreme moving condition, such as camera rotation-only action or static action, also it can work well in the condition of camera large-scale displacement motion. The real scale is also accurately estimated without depending on GT, and the pose estimation method is designed based on hybrid 2d–2d and 3d–2d localization theory to make the DL_Hyrid VO system to estimate translational and rotational information with accuracy and robustness. Experimental results show that the proposed DL_Hybrid VO system has a better performance than traditional and learning-based VO systems.

Highly efficient and stable blue organic lighting-emitting diodes (OLEDs) continue to be a challenge. The substitution strategy has been proven to be one of the most effective chemical approaches to improve the electroluminescence (EL) performance and the operational stability of blue OLEDs based on thermally activated delay fluorescent (TADF) emitters. Herein, the effects of tert-butyl and 9-phenyl-9-fluorenyl substituents on the physicochemical properties and EL performance are systematically studied for aromatic-imide-based TADF molecules. The robust 9-phenyl-9-fluorenyl units outperform the tert-butyl groups in improving the thermal and electrochemical stability. Consequently, sky blue OLEDs based on 4-DPFCzAIAd containing 9-phenyl-9-fluorenyl substituents exhibit excellent comprehensive EL performance with a maximum external quantum efficiency of 28.2% and a long LT50 (time to 50% of initial luminance) of 178 h at an initial luminance of 500 cd m−2, which is much better than that of the device with a tert-butyl substituted emitter. These results demonstrate that the 9-phenyl-9-fluorenyl substituent is superior to the tert-butyl group in constructing TADF emitters for efficient and stable blue OLEDs.

The extent and quality, i.e. consistency, of the fusion bonding between filaments that occurs during the fused filament fabrication (FFF) process play an important role in determining the mechanical performance of 3D printed polymeric parts. This present research is designed to quantify these parameters of the fusion bonding between individual filaments deposited during the 3D printing process. Polyamide 6 (PA6) specimens were prepared from both 3D printed single-layer and single-filament films. Also, to provide a benchmark material, 3D printed PA6 films prepared by the FFF process were further processed by using compression moulding (CM) to investigate whether the fusion bonding could be enhanced. An essential work of fracture method was employed to evaluate the fusion bonding between the filaments by determining the value of the specific essential work of fracture, we, and the associated experimental reproducibility. The higher value of we, and the relatively low degree of scatter determined for the benchmark material revealed that the extent and quality of fusion bonding attained when using only the FFF process clearly needed to be improved when parts are fabricated for practical applications. To study the effect of temperature on the fusion bonding, the structural detail and fracture surfaces of thin polymer films were studied, with the thermal history of the deposited filaments also being measured using a thermal camera.

Mode-locked fibre lasers (MLFLs) are fundamental building blocks of many photonic systems used in industrial, scientific and biomedical applications. To date, 1–2 μm MLFLs have been well developed; however, passively mode-locked fibre lasers in the visible region (380–760 nm) have never been reported. Here, we address this challenge by demonstrating an all-fibre visible-wavelength passively mode-locked picosecond laser at 635 nm. The 635 nm mode-locked laser with an all-fibre figure-eight cavity uses a Pr/Yb codoped ZBLAN fibre as the visible gain medium and a nonlinear amplifying loop mirror as the mode-locking element. First, we theoretically predict and analyse the formation and evolution of 635 nm mode-locked pulses in the dissipative soliton resonance (DSR) regime by solving the Ginzburg-Landau equation. Then, we experimentally demonstrate the stable generation of 635 nm DSR mode-locked pulses with a pulse duration as short as ~96 ps, a radio-frequency signal-to-noise ratio of 67 dB and a narrow spectral bandwidth of 1 nm) and modulated optical spectrum. This work represents an important step towards miniaturized ultrafast fibre lasers in the visible spectral region., Photonics: Lighting the way ultrafast compact visible light lasers Chinese scientists have demonstrated a low-cost and compact ultrafast passively mode-locked laser that operates in the visible light spectrum and could see use in a range of industrial, scientific, and biomedical applications. Although Mode-locked fibre lasers (MLFLs) are the fundamental building blocks of many photonic systems, ultrafast lasers in the visible light spectral region are costly and challenging to make. For the first time, Zhengqian Luo and colleagues from Xiamen University in China have demonstrated a visible-wavelength passively mode-locked all-fibre laser that operates in the dissipative soliton resonance regime. The laser generates picosecond pulses of light at 635 nanometres and represents an essential step towards miniaturized ultrafast fibre lasers in the visible light range. The work lays the foundations for photonic devices for use in applications such as material processing, medicine, spectroscopy, and optical communications.

Monitoring the particulate composition changes during the flocculation process is still challenging for the research community. We use an experimental setup based on polarized light scattering to measure the polarization states of the scattered light of the individual particles. We build a classifier based on the support vector machine and feed it with the measured parameters. Results show that the classifier can effectively classify the particulate compositions, such as the sediment particles, flocculants, and flocs, which can be used to monitor the particulate composition changes during the flocculation process. Discussions on the intensity and polarization parameters find that the polarization parameters play a vital role in the classification of the particulate compositions in the flocculation suspensions. Additionally, the further analysis of the experimental data and the related simulations show that the degree of polarization can be an indicator of the flocculation process. We prove that the method based on polarized light scattering may be a potential in situ monitoring tool in the future for the study of the flocculation process.

We report the first demonstration of a compact all-fiber visible-wavelength laser directly emitting cylindrical vector or vortex beams. The visible all-fiber cavity is constructed by a few-mode Pr/Yb-codoped ZBLAN fiber pumped by a 850 nm laser diode, a 634.6 nm few-mode fiber Bragg grating, and a visible-wavelength fiber mirror. Direct oscillation of high-order modes (LP 11 modes) at 634.6 nm wavelength is achieved using core-shifted fiber coupling method. The visible cylindrical vector beams with the radially or azimuthally polarized state can be observed. Furthermore, by adjusting the rotation of the intracavity polarization controller appropriately, two orthogonal modes (i.e., LP 11 even and LP 11 odd ) can introduce a π/2 phase difference, and the 634.6 nm vortex beams with ±1-order orbital angular momentum modes are also obtained.

Crime is one of the most important social problems in the country, affecting public safety, children development, and adult socioeconomic status. Understanding what factors cause higher crime rate is critical for policy makers in their efforts to reduce crime and increase citizens’ life quality. We tackle a fundamental problem in our paper: crime rate inference at the neighborhood level. Traditional approaches have used demographics and geographical influences to estimate crime rates in a region. With the fast development of positioning technology and prevalence of mobile devices, a large amount of modern urban data have been collected and such big data can provide new perspectives for understanding crime. In this paper, we use large-scale Point-Of-Interest data and taxi flow data in the city of Chicago, IL in the USA. We observe significantly improved performance in crime rate inference compared to using traditional features. Such an improvement is consistent over multiple years. We also show that these new features are significant in the feature importance analysis. The correlations between crime and various observed features are not constant over the whole city. In order to address this geospatial non-stationary property, we further employ the geographically weighted regression on top of negative binomial model (GWNBR). Experiments have shown that GWNBR outperforms the negative binomial model.

There are some questions in sonar images features extract, such as strong speckle noise, low image resolution, poor image quality, and difficult target segmentation. In order to overcome the shortcomings of traditional algorithms in sonar image feature extraction, we apply Mask RCNN instance segmentation network to sonar image feature extraction. The effectiveness of the deep learning algorithm is verified by experiments. First of all, we use online and off-line data enhancement method to expand the data set. Then, we improve the residual network, and adding convolutional block attention module (CBAM), group normalization (GN) and atrous spatial pyramid pooling (ASPP) module to network. Finally, we choose the best network structure and add focal loss function to improve semantic segmentation. The final experimental results show that the Average Precision (AP) and Mean Intersection over Union (mIoU) of the proposed network are improved compared with the original network on the side scan sonar dataset.

In the field of underwater maneuvering target tracking, the detection of sensors is not effective in real time. Under man-made or non-man-made interference, the observer only receives the noise value or abnormal value, that is, the sensor will fail for a short time. Once the sensor fails to obtain correct data continuously for a period of time, this is fatal to the filter. Aiming at the short-term failure of sensors, this paper proposes a tracking method for surface targets with compensation mechanism. This method uses the idea of deep learning to extract enough training samples from the observation data, and uses the samples to train the Long Short-Term Memory (LSTM) neural network with time series prediction ability offline. Secondly, the trained neural network model is used to compensate the sensor's observations online, combined with the unscented Kalman filter theory, to further correct the filter gain and covariance, thereby reducing the sensor's uncertainty measurement error and improving the target tracking filter Accuracy. The simulation results show that the unscented Kalman filter based on LSTM neural network compensation can predict the sensor measurement value, greatly reduce the target tracking error caused by the sensor uncertainty measurement problem, and meet the actual requirements.

Deep neural networks have proven to excel classical medical image reconstruction techniques. Some networks are based on fully connected (FC) layers to achieve domain transformation such as from the data acquisition domain to the image domain. However, FC layers result in huge numbers of parameters which take a lot of GPU memory. Hence, they do not scale well, and the overall performance is limited. For ultrasound computed tomography (USCT) application, we propose a memory-efficient convolutional network that reconstructs images from the frequency domain to image domain with much less parameters compared with multilayer perceptron, by using data-driven learning. Extensive experiments demonstrate that our method achieves high reconstruction quality. It improves the structural similarity measure (SSIM) from 0.73 to 0.99 when compared with state-of-the-art reconstruction methods in this field while reduces 2/3 parameters when compared with deep neural network with FC layers to reconstruct images from frequency domain to image domain.

Plasmonic chirality has drawn a lot of attention owing to its strong and tunable circular dichroism (CD) effects and its desirable applications in bio-sensing, chiral molecules analysis, etc. In this work, the CD response in the mid-infrared region of a metamaterial consisting of Z-shaped metal array and graphene ribbons has been studied theoretically. Benefiting from symmetry breaking and electromagnetic coupling effect, the strong CD signal is generated by the metamaterial and the maximum CD value could reach 22%, which is much larger than that of reported graphene-based chiral systems. In particular, the intensity and peak position of the CD response could be dynamically controlled by the Fermi level in a large range. Meanwhile, the CD response also could be actively controlled by changing the number and the width of graphene ribbons, the distance between the graphene ribbons and Z-shaped metal array, and so on. Furthermore, the CD effect is sensitive to the ambient medium. Therefore, the graphene ribbons and metal metamaterial could be potentially utilized in bio-sensing and related fields.

In this paper, without circular scanning of laser focus, annular ablation region was reported to be induced by femtosecond laser static irradiation of ZnO surface immersed in oil, and these annular ablation regions were all covered by laser-induced periodic surface structures (LIPSS) upon irradiation of either linearly-polarized light or circularly-polarized light. This novel phenomenon was found to be related to the bubbles in-situ generated and attached at laser focus inside oil during irradiation process. With increasing irradiation pulse number, the shallow annular ablation region transited to a deep annular groove, and the unablated region surrounded by the annular ablation region was finally ablated to a LIPSS-covered spherical-crown-shaped bump. Finite-difference-time-domain (FDTD) simulation shows that the bubble at the laser focus may act as a beam shaping device to partly transform a Gaussian-intensity-profile beam to an annular-intensity-profile beam and the increased bubble size induces a decrease of the radius of the annular beam, which can roughly explain the experimentally observed morphology transformation phenomenon. Further analysis of light field transformation based on matrix optics further demonstrates that the annular intensity profile is originated from hollow beams generated after incident light passing through the bubble-type radial polarizer.

In this paper, the Mach Zehnder interferometer is utilized to generate a collinear dual optical path with 800-nm center wavelength, 1 kHz repetition rate, 120-fs pulse duration and an energy ratio of 1:1 to scan 6H-SiC crystal at a scanning speed of 500 μm/s, resulting in laser-induced periodic surface structure (LIPSS). From the perspective of instantaneous electron density level, variation of the period of low spatial frequency LIPSS (LSFL) with interpulse delay time and polarization angle is studied. The scanning electron microscope (SEM) characterization of the line-scanned ablation zone combined with fast Fourier transformation (FFT) spectrum analysis shows that the period of LSFL decreases with the increase of delay time, and remains stable when the delay exceeds 240 fs. Meanwhile, the period of LSFL increases with the increase of polarization angle under the same delay time, showing an anisotropy that dependent on the polarization angle. The variation trend of the maximum electron density level with delay time calculated by electron rate equation is roughly consistent with the periodic variation trend. Interaction between incident light and transient metal-like SiC surface is simulated based on finite-difference time-domain method (FDTD). The simulation results show that the surface of the sample has a periodic light field enhancement structure controlled by the transient electron density, which explains the phenomenon that the period decreases with increasing delay time. It is further confirmed that the formation of LSFL is due to the interference of incident light and surface plasmon, and the period of LSFL can be controlled by adjusting the electron density level by delay time. The periodic polarization dependence of LSFL may be related to the relative orientation change of the laser pulse front tilt (PFT) and laser polarization.

A wideband magneto-electric (ME) dipole ridge gap waveguide (RGW) waveguide slotted array(WSA) antenna working at Ka band is presented in this paper. That antenna is designed with four layers: ME dipole layer, backed cavity layer, coupling layer, and waveguide network layer. Since this RGW technology does not require electrical contact between the different metal layers, the mechanical fabrication complexity are decreased. A transition from standard Ka-band waveguide port to RGW transmission line is designed, while a waveguide network is utilized as the power divider. The asymmetric RGW is also employed to have non uniform patterns. To validate the performance of the designed antennas, a 16* 8 array is designed in Ka band. Simulations of the whole array antenna demonstrate the wide bandwidth of the reflection coefficient, meanwhile, antenna efficiency is about 88% in the whole frequency band with maximum gain 30dB.

Recent studies have determined that chronic inflammation contributes to left ventricular (LV) remodeling following myocardial infarction (MI) leading to heart failure (HF). Evidence points to a critical role of T lymphocytes in driving the maladaptive inflammatory process associated with adverse tissue remodeling and cardiac dysfunction. FTY720, a non-selective sphingosine 1-phosphate (S1P) receptor modulator, elicits anti-inflammatory effects via inhibition of lymphocyte egress from secondary lymphoid organs and has been reported to improve cardiac remodeling and function post-MI. We hypothesized that APD588, a next-generation S1P receptor modulator with optimized S1P 1,5 receptor selectivity, would prevent cardiac remodeling and dysfunction following MI through modulation of T lymphocyte-mediated inflammatory responses. Efficacy of APD588 on cardiac function was assessed using C57BL/6N mice (n=7-10 mice/group) with experimental MI induced by coronary artery occlusion. Vehicle (VEH) or APD588 (0.3 mg/kg) was administered orally once daily starting at Day 3 post-MI and continued until Day 28. Blood samples were collected at Days 7 and 28 for lymphocyte count. Echocardiographic measurements, including LV ejection fraction (LVEF), were performed on Days 7, 14, and 28 post-MI. Administration of APD588 following MI resulted in a reduction of lymphocyte count at Days 7 and 28 post-MI, but only reached significance in the latter (VEH: 3.3 ± 0.5 x10 3 /μL; APD588: 1.3 ± 0.2 x10 3 /μL). Decreased cardiac function was observed at Day 7 post-MI (LVEF: 27.7 ± 3.6%, MI vs 45.7 ± 0.6%, Sham). No difference in LVEF was observed 7 days post-MI, however APD588 significantly improved LVEF at Day 28 post-MI, compared to vehicle (APD588: 35.7 ± 3.9%; VEH: 25.3 ± 3.9%). Therapies targeting inflammation in the pathophysiology of HF have shown limited success in the clinic. Novel strategies based on advancements in understanding the underlying immunobiology are warranted. Our results demonstrate that APD588 was effective at regulating lymphocyte activity and improved cardiac functional recovery following MI in mice. This supports the potential of S1P receptor modulation and other T lymphocyte-directed strategies as a targeted anti-inflammatory approach in HF.

Current methods for edge contour feature extraction for Unmanned Underwater Vehicle (UUV) side-scan sonar images have yet to solve the problems of low accuracy, discontinuous edges, and loss of detail. This paper proposes a new feature extraction method for UUV side-scan sonar images. By adding a batch normalization layer, the skip structure of the fully convolutional network (FCN) is improved, the redistribution of parameters in the skip structure is realized, and the training of the network is more sufficient. And we design a positive sample weighted loss function (WPSL) to improve the problem that the performance of the classification algorithm is degraded due to the imbalance of sample distribution in the data set. In this paper, an initial dataset is expanded by turning, rotating, and adding noise. An improved feature extraction network is then constructed, and the training of the improved FCN is completed by using a mini-batch gradient descent method, thus realizing accurate extraction of edge contour features of seabed topography. The experimental results show that the proposed method is more suitable to reject speckle noise than the traditional Canny and Fuzzy C-Means algorithms. Compared with current deep learning methods, the proposed method improves the ability to fuse detailed information and make discontinuous edges continuous. The mean intersection over union (IU) reaches 83.05%, which is 5.48% higher than the 77.57% before improvement.

This paper studies a single target tracking method based on deep learning. If the target's motion model is known, then we can use appropriate filtering methods to achieve target tracking based on the target's motion state. However, when the target motion is complex and the motion model is unknown, the filtering method is difficult to apply. To this end, we have designed two single target tracking methods based on recurrent neural networks, that is, a single target tracking algorithm based on Long and Short-Term Memory Network (LSTM) and Bidirectional Long and Short-Term Memory Network (Bi-LSTM). LSTM is a variant network based on RNN, and Bi-LSTM is based on LSTM plus One hidden layer, both LSTM and Bi-LSTM networks can process data on time series. Finally, tracking training and testing are performed for UUV under different maneuvering states to verify the effectiveness of the method, and the tracking effects and accuracy of the two models are analyzed.

Ultrasound Computed Tomography (USCT) is a new high resolution imaging method with high potential for breast cancer detection. USCT image reconstruction requires the iterative solution of a wave equation and its adjoint, being very time-consuming. To address this time consumption problem, a convolutional neural network with frequency pressure field as input is presented, outperforming classical iterative optimization strategies to solve the inverse problem. The Optical and Acoustic Breast Phantom Database (OA-Breast) is used to generate simulated USCT data using paraxial approximation forward model both for training and testing. Extensive experiments demonstrate that our neural network outperforms the classical strategies both in reconstruction time and quality.

This paper mainly studies the threat factors that affect the red and the blue Unmanned Underwater Vehicles (UUVs) in the confrontation environment. Based on these threat factors, completing threat assessment(TA) of two side UUVs , we design Bayesian network (BN), obtain the posterior information through network reasoning, and get the threaten degree of the current against situation. And through the design of the network structure, this paper defines the blue UUV as a "brave" attacker. Aiming at the problem of target selection of the blue UUV, the threat level of incoming targets is sorted by analytic hierarchy process (AHP). The simulation results show that the threat level derived from Bayesian network reasoning and the target obtained from AHP are consistent with the actual analysis results, which proves the correctness and effectiveness of the two methods.

Background: The association between sleep quality and risk of coronary heart disease (CHD) remains unclear in the elderly. Results: At eight-year follow up, a total of 411 (4.29%) participants developed CHD. Compared with good quality group, the multivariable hazard ratio [HR] (95% confidence interval [CI]) for CHD was 1.393 (1.005, 1.931) for intermediate quality group and 1.913 (1.206, 3.035) for poor quality group. Consistent results were observed in participants with normal sleep duration. Conclusions: Poor sleep quality may be a novel modifiable risk factor for CHD in the elderly independent of conventional cardiovascular risk factors, even when sleep duration was normal. Methods: The current study included 9570 CHD-free participants in the English Longitudinal Study of Ageing (ELSA) from wave 4 (2008 to 2009). Incident CHD included new onset angina or myocardial infarction. Sleep quality was measured by a four-item questionnaire. Score ranged from 1 (best) to 4 (poorest). Participants were divided into three groups: good quality (1 ≤ score

The increased availability of large-scale trajectory data provides rich information for the study of urban dynamics. For example, New York City Taxi 8 Limousine Commission regularly releases source/destination information of taxi trips, where 173 million taxi trips released for Year 2013 [29]. Such a big dataset provides us potential new perspectives to address the traditional traffic problems. In this article, we study the travel time estimation problem. Instead of following the traditional route-based travel time estimation, we propose to simply use a large amount of taxi trips without using the intermediate trajectory points to estimate the travel time between source and destination. Our experiments show very promising results. The proposed big-data-driven approach significantly outperforms both state-of-the-art route-based method and online map services. Our study indicates that novel simple approaches could be empowered by big data and these approaches could serve as new baselines for some traditional computational problems.

In this paper, we present an online obstacle avoidance planning method for unmanned underwater vehicle (UUV) based on clockwork recurrent neural network (CW-RNN) and long short-term memory (LSTM), respectively. In essence, UUV online obstacle avoidance planning is a spatiotemporal sequence planning problem with the spatiotemporal data sequence of sensors as input and control instruction to motion controller of UUV as output. And recurrent neural networks (RNNs) have proven to give state-of-the-art performance on many sequence labeling and sequence prediction tasks. In order to train the networks, a UUV obstacle avoidance dataset is generated and an offline training and testing is adopted in this paper. Finally, the proposed two types of RNN based online obstacle avoidance planners are compared in path cost, obstacle avoidance planning success rate, training time, time-consumption, learning, and generalization, respectively. And the good performance of the proposed methods is demonstrated with a series of simulation experiments in different environments.

A novel triple-layer wideband transmitarray antenna using finger-type slot elements is presented in this paper. The unit-cell contains five longitudinal slots and a transverse slot on every thin metal layer, the simulated transmission coefficients of the unit-cell indicate that low loss and wideband properties are achieved at Ku-band. A wideband TA with diameter of 7.1 wavelengths using the proposed unit-cell is designed and simulated by commercial software Ansys 18.0. Simulated gain of 21.2 dBi at the center working frequency of 13.58 GHz and -1dB gain bandwidth of 24.3% are achieved at Ku band. The simulated radiation patterns of TA show that low side lobe level and low cross polarization level are reached in broadband range.

The existence of drug resistance is the main reason for chemotherapeutic failure in malignancies. Long non-coding RNA (lncRNA) maternally expressed gene 3 (MEG3) is implicated with tumorigenesis and chemoresistance. In the present study, we aimed to identify the role of MEG3 in oxaliplatin-resistant colorectal cancer (CRC) and its potential mechanisms. MEG3 was down-regulated in oxaliplatin-resistant CRC tissues and cell lines. Low MEG3 expression was correlated with poor prognosis of CRC patients. Overexpression of MEG3 improved oxaliplatin sensitivity of HT29/OXA and HCT116/OXA cells. MEG3 suppressed miR-141 expression in HCT116/OXA cells. Moreover, MEG3 elevated PDCD4 expression through targeting miR-141, acting as a competing endogenous RNA (ceRNA). miR-141 inhibition or PDCD4 up-regulation could mimic the functional role in oxaliplatin resistance, which was counteracted by overexpression of MEG3. Collectively, MEG3 facilitated the sensitivity of CRC cells to oxaliplatin by regulating miR-141/PDCD4 axis, providing a novel therapeutic strategy for CRC.

Inferring social relationships from user location data has become increasingly important for real-world applications, such as recommendation, advertisement targeting, and transportation scheduling. Most existing mobility relationship measures are based on pairwise meeting frequency, that it, the more frequently two users meet (i.e., co-locate at the same time), the more likely that they are friends. However, such frequency-based methods suffer greatly from data sparsity challenge. Due to data collection limitation and bias in the real world (e.g., check-in data), the observed meeting events between two users might be very few. On the other hand, existing methods focus too much on the interactions between two users, but fail to incorporate the whole social network structure. For example, the relationship propagation is not well utilized in existing methods. In this paper, we propose to construct a user graph based on their spatial-temporal interactions and employ graph embedding technique to learn user representations from such a graph. The similarity measure of such representations can well describe mobility relationship and it is particularly useful to describe the similarity for user pairs with low or even zero meeting frequency. Furthermore, we introduce semantic information on meeting events by using point-of-interest (POI) categorical information. Additionally, when part of the social graph is available as friendship ground truth, we can easily encode such online social network information through a joint graph embedding. Experiments on two real-world datasets demonstrate the effectiveness of our proposed method.

Exposure to ambient air pollution has been associated with greater risk of elevated blood pressure (BP) in adults and children. Recent evidence suggests that air pollution exposure in pregnancy may also portend increased risk for the next generation; however, few studies have examined this relationship. We conducted a prospective study of 1293 mothers in the Boston Birth Cohort (enrolled 1998–2012) and their children who had follow-up visits between 3 and 9 years of age and complete exposure and outcome data. Our primary exposure, ambient particulate matter ≤2.5 µm (PM 2.5 ) concentration during pregnancy, was estimated by matching mother’s residential address to the US Environmental Protection Agency’s air quality monitors. We defined our primary outcome child systolic BP (SBP) percentile according to US reference (Fourth Report) and classified elevated BP as SBP ≥90th percentile. Our multivariable-adjusted cubic spline showed a sharp increase in offspring SBP percentile and risk for elevated BP when third-trimester PM 2.5 concentration was ≥13 μg/m 3 . The highest versus lowest tertile of third-trimester PM 2.5 exposure was associated with a 4.85 (95% confidence interval: 1.38–8.37) percentile increase in child SBP or a 1.61 (95% confidence interval: 1.13–2.30) times higher risk of child elevated BP. A 5-μg/m 3 increment in PM 2.5 during the third trimester was associated with a 3.49 (95% confidence interval: 0.71–6.26) percentile increase in child SBP or a 1.47 (95% confidence interval: 1.17–1.85) times higher risk of elevated BP. Our findings suggest that exposure to ambient PM 2.5 during the third trimester of pregnancy is associated with elevated BP in children, ages 3 to 9 years.

Children born small for gestational age (SGA) are more likely to develop high blood pressure. In prior studies, longer sleep duration is associated with lower BP, and SGA is associated with shorter sleep duration in childhood. We investigated whether sleep duration in early childhood modifies the association between SGA and higher childhood SBP in 1178 children recruited at birth and followed up to age 9 years. We ascertained birthweight and gestational age from medical records. We derived child sleep duration from maternal questionnaire interview. We calculated child SBP percentile according to U.S. reference data. We defined elevated SBP as SBP ≥75th percentile. In this sample, 154 (13.1%) children were born SGA. Children born SGA had higher SBP percentiles and higher risk of elevated SBP. Among children born SGA, those in the highest compared to the lowest tertile for sleep had a 12.28 lower (−22.00, −2.57) SBP percentile and 0.44 (0.25 to 0.79) times lower risk of developing elevated SBP. Our data are consistent with an interaction between SGA and sleep duration on childhood elevated SBP (Pinteraction = 0.0056). In conclusion, in this prospective birth cohort, longer sleep duration in early childhood may mitigate the blood pressure-raising effect of being born small.

This article aims to explain how and why a Hong Kong comedy film, A Chinese Ghost Story (1987), which did not have access to the movie theatres in mainland China, has generated a phenomenal cult fo...

In the present work, continuous carbon fibre (CCF) reinforced polyamide 6 (PA6) composites were fabricated using the Fused Filament Fabrication (FFF) 3D printing technique. The sliding friction and wear behaviour of the printed CCF/PA6 composites was investigated as a function of varying fibre orientations (i.e. normal, parallel, and anti-parallel) and layer deposition directions (i.e. parallel and perpendicular) with respect to the sliding direction. It emerged that fibre orientation played a key role in determining the wear resistance of 3D printed composites. Severe fibre breakage was observed when the orientation of fibre was in normal direction against the sliding direction, associated with a higher wear rate than that obtained in parallel or anti-parallel direction. To better understand the effects of voids on wear mechanism, tribo-tests were further carried out with the specimens that underwent compression moulding (CM) post-treatment process. The results confirmed that the internal defects could cause adverse impacts on wear performance, depending on the fibre orientations. In particular, it was found that fibre distribution, as well as the fibre/matrix interface within the filament, was the crucial parameter affecting the overall tribological performance of the printed structures. The findings add to our understanding of designing and fabricating wear-resistant polymer composites using 3D printing technologies.

Cyanobacterial blooms have become an urgent threat to the aquatic ecosystem, but early warning of the blooms is still challenging for the research community. In this paper, a method based on polarized light scattering and powered by machine learning is proposed to in-situ early warn the cyanobacterial blooms. In this work, the wild types of Microcystis are treated and the cells are individually measured to obtain their polarization parameters. The experimental results show that machine learning algorithms can be used to well identify the states of the Microcystis cells, and the compositions of the mixed samples can be effectively retrieved by this method. Subsequently, one application strategy is suggested to early warn the blooms, which is potential and powerful to achieve the in-situ early warning of cyanobacterial blooms in the future.

Electrospray is a simple and versatile approach to deposit thin-films. Traditionally, electrospray is achieved through capillary nozzle electrode to create fluid jet. Here, we report a novel needle...

Two high-gain terahertz (THz) offset reflector antennas are proposed. The silicon carbide material and thermally stable carbon fiber-reinforced plastic are selected as reflectors for high specific stiffness and low thermal expansion coefficient which are helpful for space use. Dual polarization is achieved by introducing a wire-grid polarizer which can separate the two THz polarization waves provided by multimode horns fabricated in electroforming technology. Furthermore, three-coordinate test system and theodolites are employed to detect the surface error of dishes and alignment of antenna subsystem. The physical optics analysis based on the measured reflector surface data is introduced referring to the large aperture antenna patterns by virtue of its accuracy and high efficiency. The results of this novel method are compared to that of the near-field test. Measurements show that the sidelobe levels (SLLs) of the single polarization silicon-based reflector are below −32 dB while the gain is 55.3 dB at 330 GHz, and the dual polarization carbon fiber-reinforced plastic reflector antenna suppresses the SLLs to −28 and −29 dB in two principal planes while the gain is 57.2 dB at the same frequency, respectively. The good performances of the proposed antennas indicate the potential for the application of space THz radiometer.

Background The prevalence of childhood elevated blood pressure (BP) has increased in the United States, particularly among African Americans. The influence of maternal plasma folate levels, alone or in combination with maternal cardiometabolic risk factors (hypertensive disorders, diabetes, and prepregnancy obesity), on child systolic BP (SBP) has not been examined in a prospective birth cohort. We hypothesize that adequate maternal folate levels can reduce the risk of elevated SBP in children born to mothers with cardiometabolic risk factors. Methods This study included 1,290 mother-child dyads (875 African Americans (67.8%)) recruited at birth and followed prospectively up to age 9 years from 2003 to 2014 at the Boston Medical Center. Child SBP percentile was calculated according to US reference data and elevated SBP was defined as SBP ≥75th percentile. Results Maternal folate levels, overall, were not associated with child SBP. However, we found a significant multiplicative interaction between maternal cardiometabolic risk factors and maternal folate levels (Pinteraction = 0.015) on childhood elevated SBP. Among children born to mothers with any cardiometabolic risk factors, those whose mothers had folate levels above (vs. below) the median had 40% lower odds of elevated childhood SBP (odds ratio = 0.60, 95% confidence interval: 0.40-0.90). The associations did not differ appreciably in analyses restricted to African Americans, and they were not explained by gestational age, size at birth, prenatal folate intake, or breastfeeding. Conclusions Findings from our urban minority birth cohort suggest that higher levels of maternal folate may help counteract the adverse associations of maternal cardiometabolic risk factors on child SBP.

Deep learning-based visual odometry systems have shown promising performance compared with geometric-based visual odometry systems. In this paper, we propose a new framework of deep neural network, named Deep Siamese convolutional neural network (DSCNN), and design a DL-based monocular VO relying on DSCNN. The proposed DSCNN-VO not only considers positive order information of image sequence but also focuses on the reverse order information. It employs supervised data-driven training without relying on any modules in traditional visual odometry algorithm to make the DSCNN to learn the geometry information between consecutive images and estimate a six-DoF pose and recover trajectory using a monocular camera. After the DSCNN is trained, the output of DSCNN-VO is a relative pose. Then, trajectory is recovered by translating the relative pose to the absolute pose. Finally, compared with other DL-based VO systems, we demonstrate the proposed DSCNN-VO achieve a more accurate performance in terms of pose estimation and trajectory recovering through experiments. Meanwhile, we discuss the loss function of DSCNN and find a best scale factor to balance the translation error and rotation error.

Image reconstruction of ultrasound computed tomography based on the wave equation is able to show much more structural details than simpler ray-based image reconstruction methods. However, to invert the wave-based forward model is computationally demanding. To address this problem, we develop an efficient fully learned image reconstruction method based on a convolutional neural network. The image is reconstructed via one forward propagation of the network given input sensor data, which is much faster than the reconstruction using conventional iterative optimization methods. To transform the ultrasound measured data in the sensor domain into the reconstructed image in the image domain, we apply multiple down-scaling and up-scaling convolutional units to efficiently increase the number of hidden layers with a large receptive and projective field that can cover all elements in inputs and outputs, respectively. For dataset generation, a paraxial approximation forward model is used to simulate ultrasound measurement data. The neural network is trained with a dataset derived from natural images in ImageNet and tested with a dataset derived from medical images in OA-Breast Phantom dataset. Test results show the superior efficiency of the proposed neural network to other reconstruction algorithms including popular neural networks. When compared with conventional iterative optimization algorithms, our neural network can reconstruct a 110 × 86 image more than 20 times faster on a CPU and 1000 times faster on a GPU with comparable image quality and is also more robust to noise.

We report a compact 2166 nm germania-fiber short-pulsed Raman laser based on the cavity matching scheme. The all-fiber Raman cavity is formed by a pair of 2166 nm fiber Bragg gratings. High-power noise-like pulses from a 1981 nm fiber laser are used to pump a 22 m germania-core fiber for providing Raman gain at ∼2166 nm, and readily realizes the Raman-cavity synchronization with high mismatching tolerance. Stable Raman pulses at 2166 nm are therefore generated with the tunable pulse width of 0.9-4.4 ns and the large pulse energy up to 12.15 nJ. This is, to the best of our knowledge, the first demonstration of all-fiber short-pulsed Raman laser in the mid-infrared region.

A low loss multi-beam transmitarray antenna (MBTA) working at 340 GHz is presented in this paper. In order to eliminate the substrate loss to the terahertz wave, a substrateless all metal unit cell using double hexagon ring slots is designed and calculated. The simulation results show that low transmission loss (mostly better than −1 dB) and full transmission phase range (360°) are achieved. A 14 × 20 focal plane multi-beam transmitarray fed by seven horn feeds is designed by using bifocal theory, which could effectively lower the scanning loss when the feed offset the center of the MBTA focal plane. The MBTA is simulated by commercial electromagnetic software Ansys 18.1, the simulation results show that scanning beams could cover from −15° to +15° with gains more than 25.3 dB, side lobe levels (SLL) lower than −19.3 dB and scanning loss superior to −0.6 dB The terahertz MBTA prototype is currently under fabrication, lithography and deep reactive ion etching (DRIE) processes will be applied to realize microfabrication of the prototype, the fabrication error could reach ±1 μm The microfabrication processes and measured results of the MBTA will be shown on the conference.

Machine learning is nowadays one of the most efficient and popular tool and theory which has influenced many of the engineering fields. The traditional failure analysis is also based on statistical learning and reliability data, these methods can be used to assess characteristics over the design life, predict reliability, assess the exchange effect, product life prognosis and help to failure analysis. These two subjects have the natural connection, so this paper presents a very general overview on reliability and machine learning, which will demonstrate how the machine learning tools used for classical reliability system and failure analysis. We especially state some algorithms such as Bayesian networks and its’ method to reliability area. Then we can see how a typical engineering area can benefit from the machine learning.

Nowadays, with the products and systems becomes more complex, it will be more difficulty to process the failure analysis in the dependability. Besides we can see the great progress in the field of machine learning, and the joint applications of machine learning and other fields are becoming more widespread. In this paper we propose a new function which combine the expert system and Bayesian networks to failure analysis. This method starts with data mining which will collect the clear, complete data to a data warehouse. Then it takes advantage of the Fault Tree and mapping it to Bayesian networks. At last we use the Bayesian networks and the statistical data to compute the probability of the failure in order to take analysis. We give an example to explain how this model works, and it will get a well performance because of it uses the data from failure cases and other related data.

An effective method to calculate the statistical Mueller matrix (SMM) of suspended particles based on polarized light scattering is presented that takes advantage of the Stokes vectors measurement of individual particles. The calculation method of the SMM is derived based on statistics. Experimental results of Microcystis samples confirm that the SMM can characterize cells of different states. Then, pairwise contrast experiments indicate the great prospect of the SMM applied on the discrimination of suspended particles. It helps to find the optimal incident polarization state to discriminate suspended particles, and it has optimal discrimination ability. The parameter derived from the SMM can simultaneously discriminate particles including microalgae, microplastics, and sand-like particles.

A firm trades off information transfer costs and agency costs to determine the distribution of decision rights to lower-level management. Using the replacement of municipal officials as an exogenous shock of political uncertainty, we find a business group is more likely to delegate decision rights of cash resources to subsidiaries when political uncertainty is high. The association is only evident among non-state-owned-enterprises (non-SOEs) and not SOEs. Additional tests show that the delegation of decision rights is more pronounced among firms with weak internal control and those located in cities where local governments have replaced their party secretary. We also find that firm decentralization can enhance investment efficiency and the marginal value of cash in the presence of high political uncertainty. Overall, our findings support the decentralization theory, and provide causal evidence on how political uncertainty influences allocation of decision rights among listed firms in an emerging economy with large state ownership.

This paper proposes a method based on the improved genetic algorithm for cooperative collision avoidance by multiple unmanned surface vehicles (hereinafter referred to as multiple USVs). In the multiple USVs collaborative task mode in a complex environment featuring obstacles, we establish models of the multiple USVs system and sensor detection; divide scenarios involving multiple USVs encounters and design corresponding collision avoidance strategies; and calculate the motion parameters and risk of collision to determine whether to take measures to avoid collisions. Following this, we choose the genetic algorithm (GA) as core algorithm to plan for collision avoidance, improve it through retention, deletion, and replacement, use the analytic hierarchy process to build a fitness, iteratively optimize the adjustment of velocity and heading, and calculate the current best path for collision avoidance for multiple USVs. Finally, we built a simulation platform for multiple USVs collision avoidance planning based on the QT software, and designed typical cases to verify the proposed method with and without communication conditions. The results show that the proposed method can be used for the safe operation of multiple USVs.