Search

Despite that convolutional neural networks (CNNs) have shown high-quality reconstruction for single image dehazing, recovering natural and realistic dehazed results remains a challenging problem due to semantic confusion in the hazy scene. In this article, we show that it is possible to recover textures faithfully by incorporating semantic prior into dehazing network since objects in haze-free images tend to show certain shapes, textures, and colors. We propose a semantic-aware dehazing network (SDNet) in which the semantic prior is taken as a color constraint for dehazing, benefiting the acquisition of a reasonable scene configuration. In addition, we design a densely connected block to capture global and local information for dehazing and semantic prior estimation. To eliminate the unnatural appearance of some objects, we propose to fuse the features from shallow and deep layers adaptively. Experimental results demonstrate that our proposed model performs favorably against the state-of-the-art single image dehazing approaches.

Feature selection (FS), which aims to identify the most informative subset of input features, is an important approach to dimensionality reduction. In this article, a novel FS framework is proposed for both unsupervised and semisupervised scenarios. To make efficient use of data distribution to evaluate features, the framework combines data structure learning (as referred to as data distribution modeling) and FS in a unified formulation such that the data structure learning improves the results of FS and vice versa. Moreover, two types of data structures, namely the soft and hard data structures, are learned and used in the proposed FS framework. The soft data structure refers to the pairwise weights among data samples, and the hard data structure refers to the estimated labels obtained from clustering or semisupervised classification. Both of these data structures are naturally formulated as regularization terms in the proposed framework. In the optimization process, the soft and hard data structures are learned from data represented by the selected features, and then, the most informative features are reselected by referring to the data structures. In this way, the framework uses the interactions between data structure learning and FS to select the most discriminative and informative features. Following the proposed framework, a new semisupervised FS (SSFS) method is derived and studied in depth. Experiments on real-world data sets demonstrate the effectiveness of the proposed method.

Categorizing aerial photographs with varied weather/lighting conditions and sophisticated geomorphic factors is a key module in autonomous navigation, environmental evaluation, and so on. Previous image recognizers cannot fulfill this task due to three challenges: 1) localizing visually/semantically salient regions within each aerial photograph in a weakly annotated context due to the unaffordable human resources required for pixel-level annotation; 2) aerial photographs are generally with multiple informative attributes (e.g., clarity and reflectivity), and we have to encode them for better aerial photograph modeling; and 3) designing a cross-domain knowledge transferal module to enhance aerial photograph perception since multiresolution aerial photographs are taken asynchronistically and are mutually complementary. To handle the above problems, we propose to optimize aerial photograph's feature learning by leveraging the low-resolution spatial composition to enhance the deep learning of perceptual features with a high resolution. More specifically, we first extract many BING-based object patches (Cheng et al., 2014) from each aerial photograph. A weakly supervised ranking algorithm selects a few semantically salient ones by seamlessly incorporating multiple aerial photograph attributes. Toward an interpretable aerial photograph recognizer indicative to human visual perception, we construct a gaze shifting path (GSP) by linking the top-ranking object patches and, subsequently, derive the deep GSP feature. Finally, a cross-domain multilabel SVM is formulated to categorize each aerial photograph. It leverages the global feature from low-resolution counterparts to optimize the deep GSP feature from a high-resolution aerial photograph. Comparative results on our compiled million-scale aerial photograph set have demonstrated the competitiveness of our approach. Besides, the eye-tracking experiment has shown that our ranking-based GSPs are over 92% consistent with the real human gaze shifting sequences.

Single image haze removal, which is to recover the clear version of a hazy image, is a challenging trask in computer vision. In this paper, an additive haze model is proposed to approximate the hazy image formation process. In contrast with the traditional optical model, it regards the haze as an additive layer to a clean image. The model thus avoids estimating the medium transmission rate and the global atmospherical light. In addition, based on a critical observation that haze changes gradually and smoothly accross the image, a haze smoothness prior is proposed to constrain this model. This prior assumes that the haze layer is much smoother than the clear image. Benefiting from this prior, we can directly separate the clean image from a single hazy image. Experimental results and comparisons with synthetic images and real-world images demonstrate that the proposed method outperforms state-of-the-art single image haze removal algorithms.

Video deblurring, which aims at restoring the sharp video from blurry video, is drawing increasing attention in the field of computer vision. In this paper, a method called Multi-Attention Convolutional Neural Network (MACNN) consisting of the temporal-spatial attention module, the frame channel attention module, and the feature extraction-reconstruction module is proposed. First, we use the temporal-spatial attention module and the frame channel attention module to capture features with temporal and spatial information existing across neighboring frames. Then, these captured features are fused and reconstructed to restore the sharp frame. Last but not least, we train MACNN together with a content loss and a perceptual loss in an end-to-end manner to recover realistic video details. Both quantitative and qualitative evaluation results on standard benchmarks demonstrate the proposed MACNN is superior to the state-of-the-art methods in terms of accuracy, efficiency, and visual effect.

Single image dehazing, which aims at restoring a haze-free image from its correspondingly unconstrained hazy scene, is highly challenging and has gained immense popularity in recent years. However, the images generated using existing haze-removal methods often contain haze, artifacts, and color distortions, which severely degrade the visual quality of the final images. To this end, we propose a network combining multi-scale hierarchical feature fusion and mixed convolution attention to progressively and adaptively enhance the dehazing performance. The haze levels and image structure information are accurately estimated by fusing multi-scale hierarchical features, thus the model restores images with less remaining haze. The proposed attention mechanism is capable of reducing feature redundancy, learning compact internal representations, highlighting task-relevant features and further helping the model to estimate images with sharper textural details and more vivid colors. Therefore, with the application of multi-scale features extracted from both diverse layers and filters, the dehazing performance is significantly improved. Furthermore, a deep semantic loss function is proposed to highlight more semantic information in deep features. The experimental results show that the proposed method outperforms state-of-the-art haze removal algorithms.

Establishing reliable correspondences by a deep neural network is an important task in computer vision, and it generally requires permutation-equivariant architecture and rich contextual information. In this paper, we design a Permutation-Equivariant Split Attention Network (called PESA-Net), to gather rich contextual information for the feature matching task. Specifically, we propose a novel “Split–Squeeze–Excitation–Union” (SSEU) module. The SSEU module not only generates multiple paths to exploit the geometrical context of putative correspondences from different aspects, but also adaptively captures channel-wise global information by explicitly modeling the interdependencies between the channels of features. In addition, we further construct a block by fusing the SSEU module, Multi-Layer Perceptron and some normalizations. The proposed PESA-Net is able to effectively infer the probabilities of correspondences being inliers or outliers and simultaneously recover the relative pose by essential matrix. Experimental results demonstrate that the proposed PESA-Net relative surpasses state-of-the-art approaches for pose estimation and outlier rejection on both outdoor scenes and indoor scenes (i.e., YFCC100M and SUN3D). Source codes: https://github.com/x-gb/PESA-Net .

In this paper, we tackle the problem of single image dehazing with a convolutional neural network. Within this network, we develop a multi-level fusion module to utilize both low-level and high-level features. The low-level features help to recover finer details, and the high-level features discover abstract semantics. They are complementary in the restoring of clear images. Moreover, a Residual Mixed-convolution Attention Module (RMAM) with an attention block is proposed to guide the network to focus on important features in the learning process. In this RMAM, group convolution, depth-wise convolution, and point-wise convolution are mixed, and thus it is much faster than its counterparts. With these two modules, we thus have an end-to-end network without explicitly estimating the atmospheric light intensity and the transmission map in the classical atmosphere scattering model. Both qualitative and quantitative experimental studies are carried out on public datasets including RESIDE, DCPDN-TestA, and the real-world dataset. The extensive results demonstrate both the effectiveness and efficiency of the proposed solution to single image dehazing.

Having stepped into the era of information explosion, storing, processing and analyzing the vast data sometimes are quite intractable problems. However, it is impossible for personal computer or devices to tackle with such heavy workloads. Then, companies that provides cloud computational services come into business. Aiming at minimizing the expenditures, the most important part is how many cloud services the customers should reserve in advance because different amounts they consume will yield different expense. The emerging machine learning method provides a powerful tool to address such a prediction problem. In this paper, we propose a hierarchical Pythagorean fuzzy deep neural network to forecast the quantity of requisite cloud services. Besides the employment of fuzzy logic, the neural representation is also utilized as a complementary method. The knowledge acquired from fuzzy and neural perspectives are coalesced as the final transformed data to be put into the learning systems. Based on the anticipation of the deep neural network, the consumers are able to decide the amount of cloud services to purchase. Numerical results based on the real data set from Carnegie Mellon University demonstrate that the proposed model yields the economical predictions and outperforms the prediction by the traditional neural network.

Background This study sought to examine the factors influencing acute kidney injury (AKI) in the perioperative period after lung transplantation (LT). Methods We collected and analyzed the clinical data of 25 patients, who were diagnosed with AKI in the perioperative period after LT at Sichuan Provincial People's Hospital from July 1, 2018 to June 30, 2020. Based on the clinical outcomes, the patients were divided into an AKI group and a non-AKI group. Differences between the two groups were compared, including differences in mechanical ventilation (MV) time and intensive care unit (ICU) stay time, the mode of transplantation, the total amount of dehydration in the first week after surgery, use of potential kidney damaging drugs, and whether extracorporeal membrane oxygenation (ECMO) was used. Results Nineteen patients (76%) were diagnosed with AKI in the perioperative period after LT. There were no statistically significant differences between the two groups in terms of basic information, the mode of transplantation, the total amount of dehydration in the first week after surgery, the daily dose of tacrolimus, whether ganciclovir was used, whether voriconazole was used, whether ECMO was used, and mortality (P>0.05). However, the MV time and ICU stay time of the AKI group was longer than that of the non-AKI group (P=0.006, 0.011, respectively). Analysis within the group shows there were no significant differences in terms of mortality, the MV time, and the ICU stay time between the AKI stage two group and the AKI stage three group (all P>0.05). A multi-factor analysis was conducted in which AKI was the dependent variable, whether an amount of dehydration greater than 2,000 mL, a body mass index (BMI) greater than 18, and the use of ganciclovir and voriconazole had been examined as an independent variable; however, none of these were found to be risk factors associated with AKI. Conclusions The incidence of AKI in the perioperative period after LT is high. AKI in the perioperative period after LT prolonged patients' MV time and ICU stay time. In the perioperative management of LT, it is necessary to consider kidney protection to reduce the risk of AKI.

Video deblurring is a challenging low-level vision task due to variant blur artifacts caused by factors such as depth variations, high-speed movements and camera shakes. Although significant efforts have been devoted to addressing this task, two challenges of capturing temporal patterns and spatial topologies still remain. In this paper, an attention-based interframe compensation scheme is proposed to address the first challenge. The proposed scheme replaces frames in blurry sequences with newly restored frames, and estimates temporal patterns among the replaced sequence to restore the whole sequence. After each replacement, an attention block is employed to exploit dependencies among restored and blurry frames to capture stable temporal patterns. To tackle the second challenge, we propose an adaptive residual block that dynamically fuses multi-level features via learning location-specific weights. Comprehensive experimental results demonstrate that the proposed method achieves state-of-the-art performance in terms of accuracy, visual effect and model size.

Video deblurring is still a challenging low-level vision task since spatio-temporal characteristics across both the spatial and temporal domains are difficult to model. In this article, to model the temporal information, we develop a non-local block which estimates inter-frame similarity and inter-frame difference. Specially, for modeling the spatial characteristics and restoring sharp frame details, we propose a recursive block that iteratively refines feature maps generated at the last iteration. In addition, a novel temporal loss function is introduced to ensure the temporal consistency of generated frames. Experimental results on public datasets demonstrate that our method achieves state-of-the-art performance both quantitatively and qualitatively.

Context Pien-Tze-Huang (PTH) is traditionally applied to treat various inflammation-related diseases including stroke. However, literature regarding the anti-inflammatory effects and possible mechanisms of PTH in ischaemic stroke is unavailable. Objective This study investigates the anti-inflammatory effects and its underlying mechanism of PTH on ischaemic stroke. Materials and methods Cerebral ischaemia-reperfusion injury was induced through 2 h middle cerebral artery occlusion (MCAO) followed by 24 h reperfusion in male Sprague-Dawley (SD) rats receiving oral pre-treatment with PTH (180 mg/kg) for 4 days. TLR4 antagonist TAK-242 (3 mg/kg) was injected intraperitoneally at 1.5 h after MCAO. MRI, HE staining, qRT-PCR, western blot, and immunofluorescence methods were employed. Results PTH treatment markedly reduced cerebral infarct volume (by 51%), improved neurological function (by 33%), and ameliorated brain histopathological damage in MCAO rats. It also reduced the levels of four inflammatory mediators including IL-1β (by 70%), IL-6 (by 78%), TNF-α (by 60%) and MCP-1 (by 58%); inhibited microglia and astrocyte activation; and decreased protein expression of iNOS and COX-2 in injured brains. Moreover, PTH down-regulated the protein expressions of TLR4, MyD88, and TRAF6; reduced the expression and nuclear translocation of NF-κB; and lowered the protein expressions of p-ERK1/2, p-JNK, and p-p38. Similar effects were observed in MCAO rats with TAK-242 treatment. However, combined administration of PTH and TAK-242 did not significantly reinforce the anti-inflammatory effects of PTH. Discussion and conclusion PTH improved cerebral ischaemia-reperfusion injury by inhibiting neuroinflammation partly via the TLR4/NF-κB/MAPK signalling pathway, which will help guide its clinical application.

Three-dimensional printing has received tremendous interest because of the various potential applications in robotics, electronics, aerospace, and biomedicine. Current initiatives in the development of 3D-printed robotics involve soft bionic structures with functional materials. In this paper, a 3D-printed bionic robotic fish has been designed and fabricated by direct ink writing (DIW) 3D printing. The robotic fish consists of four main parts: two-laminated caudal fins, rigid fish body with permanent magnet counted in tail, tiny battery, and controller. The robot is modeled based body/caudal actuation of coaxial circuit actuator. In a driven process, electric energy is transformed into mechanical energy of caudal fins, generating robotic fish swimming movement. With the assistance of 3D printing technique, an underwater robot of well-optimized shape and light weight is manufactured with total size of 200 mm (length) × 50 mm (width) × 30 mm (height). The embedded battery offers power for whole movements, and Arduino controls the waveform of signal. Overall, this robotic fish combines 3D-printed functional structures and embedded electronics in a total compact size with good mobility, showing intensive applications of underwater activities as well as other special tasks.

Deep neural networks have demonstrated excellent performance in most computer vision tasks in recent years. However, they are vulnerable to adversarial perturbations generated by adversarial attacks. These human-imperceptible perturbations often lead to severe distortion in the high-dimensional intermediate feature space, which is one of the major reasons for the vulnerabilities in deep neural networks. Therefore, input images with perturbations can completely change the predictions of the networks in the decision space. To overcome this drawback, we propose to progressively align the intermediate feature representations extracted from the adversarial domain with feature representations extracted from a clean domain through domain adaptation. The difference between two feature distributions can be accurately measured via an optimal transport-based Wasserstein distance. Thus, the deep networks are forced to learn robust and domain-invariant feature representations, so that the gap between the different domains is minimized and that the networks are no longer easily fooled by diverse adversaries. Extensive evaluations are conducted on four classification benchmark datasets in white-box attack scenarios. The evaluation results demonstrate a significant performance improvement over several state-of-the-art defense methods.

Learning-based methods have attracted considerable interest in image dehazing. However, most existing methods are not well adapted to different hazy conditions, especially when dealing with the heavily hazy scene. There is often a significant amount of haze that remains in the images recovered by most methods. To address this issue, we propose an end-to-end Haze Concentration Adaptive Network (HCAN), including a pyramid feature extractor (PFE), a feature enhancement module (FEM), and a multi-scale feature attention module (MSFAM) for image dehazing. Specifically, PFE based on the feature pyramid structure leverages complementary features from different CNN layers to help the clear image prediction. Then, FEM fuses four kinds of images with different haze density (i.e., three recovered images in the FEM with light haze density, and the input hazy image with strong haze condition) to guide the network to adaptively perceive images under different haze conditions. Finally, MSFAM is designed under two principles, multi-scale structure and attention mechanism. It is used to help the network produce a clear image with more details, and ease the network training. Comprehensive experiments demonstrate that the proposed HCAN performs favorably against the state-of-the-art methods in terms of PSNR, SSIM, and visual effect. The results, per-trained models and code are available at https://github.com/TaoWangzj/HCAN .

Background The objective of the present study was to determine the effect of allisartan, a new angiotensin II type 1 receptor antagonist on vascular remodeling through voltage gated potassium channels (Kv7) in hypertensive rats. Methods The study included a total of 47 Sprague Dawley (SD) rats. The animals were randomized to sham operation (n = 14), untreated hypertensive control group (n = 18) and allisartan treatment group (n = 15). Using renal artery stenosis, hypertension was induced in animals. Single dose of allisartan was administered intra-gastrically to animals in the allisartan treatment group and match placebo in the other 2 groups. Wire myography was used to measure the muscle tension in isolated mesenteric arteries from the animals. Real-time polymerase chain reaction was used to quantify the expression of Kv7 channel mRNA subunits. Results After 4 weeks of treatment, a significant decrease in mean arterial, systolic and diastolic blood pressure (SBP and DBP) was observed in allisartan treatment group compared to hypertension control group. The median arterial wall thickness and area/diameter ratio reduced significantly in treatment group compared to untreated hypertension group (P Conclusions From the results, allisartan was found to lower BP and preserve vascular remodeling through Kv7 channels.

Driving behavior plays a key role in the interaction between vehicle and driver in transportation systems. Some applications about driving behavior in Advanced Driver Assistance Systems (ADAS) improve driving safety significantly. This paper introduces the driving context and models driving behavior in a combination of cognitive perspective and data-driven perspective. First, we use a cognitive fusion method by adding a delay time module to fuse the environmental information and inside information. To better capture the driving context relationship between outside and inside features, we transfer the behavior prediction task to the sequence labeling task by introducing the visual inertia hypothesis. We propose the Predictive-Bi-LSTM-CRF algorithm which used the Bidirectional Long-Short Term Memory Networks (Bi-LSTM) and Conditional Random Field (CRF) as the loss layer to model the driving behavior. Besides, we define a new comprehensive evaluation metric for the prediction task considering F1-score and the prediction time before maneuver together. Our experiment results achieve the state of art performance on the Brain4Cars dataset and demonstrate the applicability of our theory.

As a pre-processing step of computer vision applications, single image dehazing remains challenging due to existing inefficiencies in the restoration of content and details. In this paper, the self-supporting dehazing network (SSDN) is proposed to overcome these two problems. For the restoration of image content, the self-filtering block is introduced to remove redundant features, hence improving the representation abilities of learned features. For the recovery of image details, a novel self-supporting module is proposed as a crucial component of the proposed SSDN. With this module, the complementary information among support images that are transformed from multi-level features is explored. By incorporating such information, the self-supporting module can learn more intrinsic image characteristics and generate fine-detail images. Experimental results demonstrate that the proposed SSDN outperforms state-of-the-art dehazing methods in terms of both quantitative accuracy and qualitative visual effect.

Recently, deep learning has achieved great success in visual tracking tasks, particularly in single-object tracking. This paper provides a comprehensive review of state-of-the-art single-object tracking algorithms based on deep learning. First, we introduce basic knowledge of deep visual tracking, including fundamental concepts, existing algorithms, and previous reviews. Second, we briefly review existing deep learning methods by categorizing them into data-invariant and data-adaptive methods based on whether they can dynamically change their model parameters or architectures. Then, we conclude with the general components of deep trackers. In this way, we systematically analyze the novelties of several recently proposed deep trackers. Thereafter, popular datasets such as Object Tracking Benchmark (OTB) and Visual Object Tracking (VOT) are discussed, along with the performances of several deep trackers. Finally, based on observations and experimental results, we discuss three different characteristics of deep trackers, i.e., the relationships between their general components, exploration of more effective tracking frameworks, and interpretability of their motion estimation components.

Background Increased incidence of lung cancer in older adults is attributed to increased life expectancy, increased risk for many types of cancer, age-associated comorbidities and physical performance status. Contraindicatory tumor resection is suggested to benefit survival outcomes in stage IV non-small-cell lung cancer (NSCLC). We analyzed the clinical characteristics of older adults ≥70 years old with stage IV NSCLC and investigated whether radical local treatment may benefit this population. Methods This retrospective, population-based cohort study analyzed patient data from the USA Surveillance, Epidemiology and End Results (SEER) Program during 2004–2016. Eligible patients were aged ≥70 years and diagnosed with stage IV NSCLC. Primary endpoints were overall survival (OS) and lung-cancer-specific survival (LCSS). Propensity-score matching (PSM) and Cox regression analysis were performed to assess the prognostic role of surgical resection of primary tumor or metastasis. Results Among 54,310 stage IV NSCLC older patients, 7.50% received radical local treatment of the primary tumor or metastasis. PSM resulted in a balanced study population consisting of a treatment group (n=4,037) and a matched no-local-treatment group (n=15,658). After adjusting for confounders, radical local treatment was significantly associated with increased LCSS [hazard ratio (HR): 0.85, 95% confidence interval (95% CI): 0.72–1.00] but decreased OS (HR: 1.97, 95% CI: 1.31–2.97, P=0.0012). Conclusions In older adults ≥70 years old with stage IV NSCLC, radical local treatment is associated with increased LCSS but decreased OS. Additional prospective studies are warranted to confirm the benefit of radical local treatment for primary or metastatic NSCLC.

Abnormal driving may cause serious danger to both the driver and the public. Existing dectors of abnormal driving behaviour are mainly based on shallow models, which require large quantities of labelled data. The aquisition and labelling of abnormal driving data are, however, difficult, labour-intensive and time-consuming. This situation inspires us to rethink the abnormal driving detection problem and to apply deep architecture models. In this study, we establish a novel deep-learning-based model for abnormal driving detection. A stacked sparse autoencoders model is used to learn generic driving behavior features. The model is trained in a greedy layer-wise fashion. As far as the authors know, this is the first time that a deep learning approach is applied using autoencoders as building blocks to represent driving features for abnormal driving detection. In addition, a model for denoising is added to the algorithm to increase the robustness of feature expression. The dropout technology is introduced into the entire training process to avoid overfitting. Experiments carried out on our self-created driving behaviour dataset demonstrate that the proposed scheme achieves a superior performance for abnormal driving detection compared to the state-of-the-art.

Chronic obstructive pulmonary disease (COPD) is a common airway disease, and epithelial mesenchymal transition (EMT) is participated in the pathogenesis of COPD. However, the role of CD147 in COPD remains largely unknown. In order to clarify the role of CD147 in EMT induced by cigarette smoke, we established animal and cell model of EMT by mean of cigarette smoke exposure and detected the expressions of CD147 and EMT markers

The effective training of supervised Person Re-identification (Re-ID) models requires sufficient pairwise labeled data. However, when there is limited annotation resource, it is difficult to collect pairwise labeled data. We consider a challenging and practical problem called Early Active Learning, which is applied to the early stage of experiments when there is no pre-labeled sample available as references for human annotating. Previous early active learning methods suffer from two limitations for Re-ID. First, these instance-based algorithms select instances rather than pairs, which can result in missing optimal pairs for Re-ID. Second, most of these methods only consider the representativeness of instances, which can result in selecting less diverse and less informative pairs. To overcome these limitations, we propose a novel pair-based active learning for Re-ID. Our algorithm selects pairs instead of instances from the entire dataset for annotation. Besides representativeness, we further take into account the uncertainty and the diversity in terms of pairwise relations. Therefore, our algorithm can produce the most representative, informative, and diverse pairs for Re-ID data annotation. Extensive experimental results on five benchmark Re-ID datasets have demonstrated the superiority of the proposed pair-based early active learning algorithm.

The camera shake and high-speed motion of objects often produce a blurry video. However, it is hard to recover sharp videos using existing single or multiple image deblurring methods, as the blur artifacts in blurry videos are both temporally and spatially varying. In this paper, we propose a temporally and spatially variant recurrent neural network for video deblurring, in which both temporally and spatially variants employ ConvGRU blocks and a weight generator to capture spatio-temporal features. Meanwhile, the proposed model is trained in an end-to-end manner, where the model input and output are set to the same number. Thus, our model does not reduce the number of frames both in training and testing stages, which is important in practical applications. Quantitative and qualitative evaluations on standard benchmark datasets demonstrate that the proposed method outperforms the current state-of-the-art methods.

When a human perceives videos composed of the same images in various orders, such as normal order, reverse order, and random order, the human visual attention system perceives them as different visual inputs. This means that the temporal change in the image sequence exerts a considerable influence on the human visual system. However, most state-of-the-art computational visual attention models have not considered the temporal cues adequately. Motivated by this deficiency, we propose a novel temporally irreversible visual attention model considering the following three aspects. First, the central bias of human dynamic vision is incorporated into the model to manifest this tendency. Second, the depth and directional motion-sensitive neurons are fused to discern different motion patterns. Third, the rarity factor is integrated into the model to mimic the attention shift when human observer perceives new emerging motion cues. The proposed model demonstrates its competitiveness to select attentive events in our experiments, in both laboratory setting and real driving video clips. When compared with recent visual attention models, the proposed model achieves the highest score in similarity with human dynamic vision. The proposed model could be one of the fundamental building blocks for any visual attention systems coping with dynamic scenes.

The position of abdominal organs, and mechanisms by which these are centrally connected, are currently described in peritoneal terms. As part of the peritoneal model of abdominal anatomy, there are multiple mesenteries. Recent findings point to an alternative model in which digestive organs are connected to a single mesentery. Given that direct evidence of this is currently lacking, we investigated the development and shape of the entire mesentery. Here we confirm that, within the abdomen, there is one mesentery in which all abdominal digestive organs develop and remain connected to. We show that all abdominopelvic organs are organised into two, discrete anatomical domains, the mesenteric and non-mesenteric domain. A similar organisation occurs across a range of animal species. The findings clarify the anatomical foundation of the abdomen; at the foundation level, the abdomen comprises a visceral (i.e. mesenteric) and somatic (i.e. musculoskeletal) frame. The organisation at that level is a fundamental order that explains the positional anatomy of all abdominopelvic organs, vasculature and peritoneum. Collectively, the findings provide a novel start point from which to systemically characterise the abdomen and its contents., Byrnes et al. reconstruct the developing mesentery from digitized embryonic datasets and human and animal cadavers using 3D digital and printed models. They confirm the mesentery remains a continuous organ in and on which all abdominal digestive organs develop and that at the foundation level, the abdomen comprises a mesenteric and non-mesenteric domain.

In actor-critic reinforcement learning (RL) algorithms, function estimation errors are known to cause ineffective random exploration at the beginning of training, and lead to overestimated value estimates and suboptimal policies. In this paper, we address the problem by executing advantage rectification with imperfect demonstrations, thus reducing the function estimation errors. Pretraining with expert demonstrations has been widely adopted to accelerate the learning process of deep reinforcement learning when simulations are expensive to obtain. However, existing methods, such as behavior cloning, often assume the demonstrations contain other information or labels with regard to performances, such as optimal assumption, which is usually incorrect and useless in the real world. In this paper, we explicitly handle imperfect demonstrations within the actor-critic RL frameworks, and propose a new method called learning from imperfect demonstrations with advantage rectification (LIDAR). LIDAR utilizes a rectified loss function to merely learn from selective demonstrations, which is derived from a minimal assumption that the demonstrating policies have better performances than our current policy. LIDAR learns from contradictions caused by estimation errors, and in turn reduces estimation errors. We apply LIDAR to three popular actor-critic algorithms, DDPG, TD3 and SAC, and experiments show that our method can observably reduce the function estimation errors, effectively leverage demonstrations far from the optimal, and outperform state-of-the-art baselines consistently in all the scenarios.

Multi-task long-range autonomous driving in urban areas is a challenging task. Traditional methods are not applicable to situations with high-dimensional observations. Current reinforcement learning (RL) based algorithms are able to train agents to utilize high-dimensional signals for vehicle control and accomplish short-range tasks. However, current RL methods do not work well on multi-task long-range urban driving tasks. In this paper, we propose a novel Hierarchical-planning and Spatial-encoding-based Multi-task Reinforcement Learning (HSMRL) model for the autonomous driving problem in urban areas. For long-range tasks, global routes are planned in real-time based on the A* algorithm, and short-range tasks are constructed based on the global routes, through which the short-range tasks are regarded as sub-tasks of long-range tasks. Then the short-range task information is efficiently encoded using spatial encoding, and the short-range tasks are accomplished by RL algorithms. We also prove that under our settings, multiple autonomous driving tasks are capable of training a shared policy. Experiments show that our method can stably accomplish the multi-task long-range driving tasks and performs significantly better than the state-of-the-art methods.

ObjectiveTo describe how mothers of late preterm infants experienced the provision of intermittent kangaroo mother care (KMC) in four postnatal wards in different hospitals in China, under a pilot KMC project.DesignA concurrent mixed-methods approach incorporating quantitative maternal questionnaires and qualitative semistructured interviews.SettingFour postnatal wards in level-III hospitals based in different provinces of Southeast and Northwest China.ParticipantsAll 752 mothers who provided intermittent KMC to their late preterm newborns in the four participating postnatal wards consented to participate in the study (quantitative component), as well as six nurses, two obstetricians and two mothers from two of the participating postnatal wards (qualitative component).Outcome measuresMaternal KMC experiences during a hospital stay, patients’ perceptions of KMC initiation, processes, benefits and challenges.ResultsMost mothers had not heard of KMC before being introduced to it in the postnatal ward. On average, mothers and newborns stayed in postnatal wards for 3.6 days; during their stay, mothers provided an average of 3.5 KMC sessions, which is an average of 1.1 sessions a day. Each KMC session lasted an average of 68 min, though there was much variation in the length of a session. Common reasons given for discontinuing a KMC session included restroom use, infant crying and perceived time limitations. Some mothers would have preferred to provide KMC for longer periods of time and nurses encouraged this. Most mothers experienced no difficulty providing KMC, received support from family and medical staff and intended to continue with KMC postdischarge.ConclusionIn order to improve the maternal experience of KMC, it is recommended that raising awareness of KMC should be included in antenatal care and after birth. Longer periods of KMC provision should be encouraged, greater privacy should be provided for mothers providing KMC in postnatal wards and family members should be encouraged to support KMC.

Backgrounds Pneumocystis pneumonia (PCP) remains an important cause of morbidity and mortality in non-HIV immunocompromised patients especially solid organ transplant（SOT）recipients. There is a lack of data on the effects of PCP on the outcomes of them.MethodWe retrospectively analyzed the clinical data of 34 non-HIV immunocompromised patients who were diagnosed with PJP by metagenomic next-generation sequencing(mNGS) combined with clinical symptoms and CT imaging changes admitted to our hospital from October 2018 to December 2020. We use univariable analysis and multi-logistic regression analysis to screen the main risk factors that affect the prognosis.ResultsA total of 65 patients found Pneumocystis jiroveci through mNGS from the BALF. Among them 34 non-HIV immunocompromised patients who were diagnosed with PCP were included. The mortality rate was 41.2%. Among them, 19 were kidney recipients, 1 liver recipient, 5 cases of connective tissue disease，4 cases of blood system diseases, 5 cases of others (including tumors and skin diseases). The ICU occupancy rate was 85.3%. The univariable analysis between the two groups found that the main risk factors affecting the outcome included age, body mass index (BMI), albumin (ALB), SOFA score on the first day of entering the ICU, the interval time between symptom onset and the start of TMP/SMX treatment, and invasive mechanical ventilation. However, patients who accepted early combination seems to get better outcome.ConclusionOld age, low BMI, hypoalbuminemia, mechanical ventilation, and delayed treatment may be associated with a worse outcome. Early diagnosis and treatment are key factors to improve the outcome of PCP. We should actively correct the malnutrition condition, and give early combination of TMP/SMX and echinocandin therapy may improve the prognosis.

An image can be decomposed into two parts: the basic content and details, which usually correspond to the low-frequency and high-frequency information of the image. For a hazy image, these two parts are often affected by haze in different levels, e.g. , high-frequency parts are often affected more serious than low-frequency parts. In this paper, we approach the single image dehazing problem as two restoration problems of recovering basic content and image details, and propose a Dual-Path Recurrent Network (DPRN) to simultaneously tackle these two problems. Specifically, the core structure of DPRN is a dual-path block, which uses two parallel branches to learn the characteristics of the basic content and details of hazy images. Each branch consists of several Convolutional LSTM blocks and convolution layers. Moreover, a parallel interaction function is incorporated into the dual-path block, thus enables each branch to dynamically fuse the intermediate features of both the basic content and image details. In this way, both branches can benefit from each other, and recover the basic content and image details alternately, therefore alleviating the color distortion problem in the dehazing process. Experimental results show that the proposed DPRN outperforms state-of-the-art image dehazing methods in terms of both quantitative accuracy and qualitative visual effect.

High-definition and arbitrary grayscale hydrogel paintings that can appear reversibly with hydration/dehydration are realized through spatially grayscale exposure. Spatio-temporally grayscale images are used to guide the exposure of a hydrogel processor to dictate the gradient cross-linking density spatially, which thereafter results in the heterogeneity of hydrogels in swelling ratio, mechanical properties, and especially visible light transmittance, leading to swelling-induced patterns by gradient and local visible light scattering difference based on tunable mesh size and microphase separation. The resultant grayscale hydrogel patterns, with visible light transmittance adjustable, are reversible during hydration (in 1-2 s) and dehydration and possess the feature size of 70 μm and more pattern information compared with previous hydrogel patterning. Uniquely, the patterns can be realized not only on the outmost surface of hydrogels as usual but inside. Combining the unique grayscale exposure with three-dimensional printing technology, arbitrary hydrogel patterns that have plenty of details and even vary at different layers are fabricated readily, indicating its broad potential in smart anti-counterfeiting of security field, mechanics, engineering, and many others.

BACKGROUND The aim of this study was to investigate whether additional volume reduction by ultrafiltration can improve blood pressure in patients with intradialytic hypertension (IDH) defined as at least 10 mmHg systolic blood pressure (SBP) rise during hemodialysis. PATIENTS AND METHODS This prospective, open-label, single-center study included 11 IDH patients with normal predialytic blood pressure (BP) (group A), 11 IDH patients with high predialytic BP (group B), and 18 patients without IDH as control. Serum angiotensin-II, aldosterone (ALD), angiotensin-converting enzyme, endothelin-1, nitric oxide, and asymmetric dimethylarginine were measured before and after the treatments. RESULTS Basic angiotensin-converting enzyme, ALD, endothelin-1, and asymmetric dimethylarginine serum levels were significantly increased in group B compared with control (P < 0.05). On comparing the results from the first and 13th dialysis sessions in group A, the dry weight was reduced by - 0.15 ± 0.16 kg after 12 sessions and the predialytic SBP increased by 3.18 ± 6.25 mmHg before and by 7.37 ± 14.90 mmHg at 4 h during the 13th session. In group B, the dry weight was reduced by 0.67 ± 0.53 kg (P = 0.006 vs. group A) at the 13th session and they had - 12.09 ± 16.20 mmHg less SBP before (P = 0.009 vs. group A) and - 11.82 ± 14.66 mmHg at 4 h of the 13th session. The decrease in dry weight was associated with significantly higher decreases in angiotensin-II and ALD serum levels in group B compared with group A. CONCLUSION Reducing fluid overload in IDH patients with high predialytic BP can effectively improve their BP, but had no effect on BP in normal predialytic BP IDH cases.

Salidroside, an active constituent of Rhodiola rosea, is neuroprotective after transient middle cerebral artery occlusion (tMCAO). However, its effects in other experimental stroke models are less understood. Here, we investigated the effect of daily intraperitoneal injections of salidroside in rats after permanent MCAO (pMCAO). Cerebral infarct volumes at 1 day after pMCAO were significantly reduced by treatment with 100 mg/kg/day salidroside, but not by 25 or 50 mg/kg/day, and this benefit of salidroside increased significantly over at least 7 days of treatment, when it was also accompanied by decreased neurological deficit scores. These observations led us to investigate the underlying mechanism of action of salidroside. 100 mg/kg salidroside for 1 day increased NeuN, Nrf2, and its downstream mediator HO-1, while it reduced nuclear NFκB p50, IL-6, and TNFα. Brusatol, a Nrf2 inhibitor, blocked the actions of salidroside on Nrf2, NFκB p50, IL-6, and TNFα. Salidroside also increased the ratio of p-PKB/PKB at 1 day after pMCAO even in the presence of brusatol. LY294002, a PI3K inhibitor, prevented all these effects of salidroside, including those on NeuN, p-PKB/PKB, Nrf2, HO-1, and pro-inflammatory mediators. In contrast, salidroside had no significant effect on the level of cerebral complement C3 after pMCAO, or on the activity of C3 as measured by the expression of cerebral Egr1. Our findings therefore suggest that salidroside reduces neuroinflammation and neural damage by regulating the PI3K/PKB/Nrf2/NFκB signaling pathway after pMCAO, and that this neuroprotective effect does not involve modulation of complement C3 activity.

Recent 3D visual quality assessment methods still have difficulties in providing the best viewing experience from the viewer’s perspective due to the ambiguous understanding of human stereo vision. One of the key reasons is that the disparity gradient, which affects human depth perception, is hard to control for the input stereo image pair. In this paper, we mathematically formulated the human disparity gradient and optimized the disparity gradients for each stereo image pair. Considering that the disparity gradient needs to be limited to a specific range to satisfy the human visual preference and comfortableness, we proposed a new quantitative definition of disparity gradient and trained the optimal disparity gradients were learned from the pilot study to enhance the viewing experience. Extensive subjective evaluations have demonstrated the competitiveness of this proposed method for the improvement of the viewing experience.

Detecting smoke from visual sensors is crucial for fire warnings. However, visually detecting smoke is still challenging because the chrominance, shape, transparency, texture, motion, and so on, of smoke vary over a wide range. A deep neural network (DNN) could detect smoke with higher accuracy, but the models are too large to run on limited resource platforms. To make the models smaller and improve the accuracy, three novel strategies are proposed in this paper: 1) feed the networks with blocks rather than the original image; 2) feed the network with condensed data rather than videos to extract dynamic characteristics; and 3) the framework has two stages, the first stage focuses on every block to find thin smoke, and the second stage focuses on the ascending and expanding motions in videos. It is not an end-to-end network, and every subnetwork works on a small matrix, even the one at the second stage, which operates on the whole image. Therefore, the network based on these strategies, the slight smoke perceptual network (SSPN), is small enough to be integrated on a raspberry pi. The experiments demonstrate that SSPN outperforms existing traditional methods and the methods based on DNN and achieves the highest accuracy and the highest sensitivity.

Background: Thoracic wall tumors can leave large defects in the thoracic wall after tumor resection. Currently, the shape of the materials commonly used for thoracic wall repair, including dacron mesh and titanium alloy mesh, cannot readily conform to the shapes of defect sites. In this study, we aimed to retrospectively review and evaluate the outcomes of applying three-dimensional (3D) printing technology in assisting in thoracic wall tumor resection and thoracic wall construction. Methods: Six patients with thoracic wall tumors underwent thin-slice CT scanning. We 3D reconstructed pleural tumors and adjacent structures with Amira software and 3D printed them. Preoperative simulation, surgical rehearsal, and surgical planning were performed, and 3D conformal titanium plates were created based on 3D reconstruction models and sutured to the defect sites of the thoracic wall. We also retrospectively reviewed 10 patients who underwent this surgery with conventional methods. All of the demographic data, clinical data, and laboratory findings (non-normally distributed variables) were compared between these two groups. Results: 3D reconstructions of the tumors and their adjacent structures were successfully performed, and 3D printing physical models and conformal titanium plates were also successfully obtained. The plate afforded accurate matching, less bleeding, fewer postoperative complications, and less pain. Conclusions: This 3D printing technology can aid in preoperative rehearsal, surgical planning, and the manufacturing of 3D implants. The 3D titanium plate has such advantages over traditional implants as having good fit and hardness, improving the surgical accuracy and curative effect, and reducing complications, such as bleeding and pain.

In a typical multi-kW industrial fiber laser, the influence of the main amplifier to master oscillator power ratio on laser performance has been fully demonstrated. An optimized output power ratio is obtained to be ~5, which can provide the maximum main amplifier slope efficiency of ~78.2% at multi-kW laser output. In this case, the laser setup was kept at ∼2.6kW for 1h and the output power presented a relatively small power degradation of 0.36%. It was found the injected seed power has no remarkable effect on the optimal slope efficiency for industry fiber laser. The present results indicated that a suitable power ratio between the main amplifier and the oscillator is of great importance for the power scaling up and high stability of industry fiber laser.

Accurately classifying sceneries with different spatial configurations is an indispensable technique in computer vision and intelligent systems, for example, scene parsing, robot motion planning, and autonomous driving. Remarkable performance has been achieved by the deep recognition models in the past decade. As far as we know, however, these deep architectures are incapable of explicitly encoding the human visual perception, that is, the sequence of gaze movements and the subsequent cognitive processes. In this article, a biologically inspired deep model is proposed for scene classification, where the human gaze behaviors are robustly discovered and represented by a unified deep active learning (UDAL) framework. More specifically, to characterize objects' components with varied sizes, an objectness measure is employed to decompose each scenery into a set of semantically aware object patches. To represent each region at a low level, a local-global feature fusion scheme is developed which optimally integrates multimodal features by automatically calculating each feature's weight. To mimic the human visual perception of various sceneries, we develop the UDAL that hierarchically represents the human gaze behavior by recognizing semantically important regions within the scenery. Importantly, UDAL combines the semantically salient region detection and the deep gaze shifting path (GSP) representation learning into a principled framework, where only the partial semantic tags are required. Meanwhile, by incorporating the sparsity penalty, the contaminated/redundant low-level regional features can be intelligently avoided. Finally, the learned deep GSP features from the entire scene images are integrated to form an image kernel machine, which is subsequently fed into a kernel SVM to classify different sceneries. Experimental evaluations on six well-known scenery sets (including remote sensing images) have shown the competitiveness of our approach.

IntroductionShared decision making is endorsed by guidelines for both acute kidney injury and critical care medicine. However, there is still a huge need for effective interventions, especially those focusing on decisions about renal replacement therapy for intensive care unit (ICU) patients with acute kidney injury. The decision aids provide evidence-based support for shared decision making, to achieve better decisions through enhanced knowledge of treatment options and treatment aligns with patients’ preferences and values. Therefore, our objectives are to develop and evaluate a decision aid systematically and rigorously for family surrogate decision makers of ICU patients with acute kidney injury who need renal replacement therapy.Methods and analysisWe will use a systematic development process that focuses on user-centred design to develop and evaluate the decision aid in three phases: (1) development of a draft prototype for the decision aid based on extensive literature reviews, interviews with key stakeholders and evidence synthesis; (2) alpha testing (‘near live’ usability) the decision aid during simulated clinical encounters to test its comprehensibility, acceptability and usability and (3) beta testing (‘live’ usability) to examine the aid’s clinical feasibility. User testing will be conducted using mixed-methods approach to support iterative revision of the decision aid. The IPDASi (V.4.0) will be used for following qualitative assessment. All interviews will be analysed by Colaizzi’s seven-step approach to qualitative analysis. The coding scheme will use to analyse user interactions. Questionnaire surveys will be analysed using paired sample t-tests when related to the before-and-after survey, otherwise using one-sample t-test.Ethics and disseminationEthical approval for this research was obtained from the Ethics Committee of the First Affiliated Hospital of Army Medical University, PLA (Ref: KY2020104). All participants will sign a formal informed consent form. The findings will be published in peer-reviewed journals and reported in appropriate meetings.Trial registration numberChiCTR2000031613.

Objective The primary aim of our study was to explore the mechanisms through which long non-coding RNA (lncRNA)-mediated sirtuin-1 (SIRT1) signaling regulates type II alveolar epithelial cell (AECII) senescence induced by a cigarette smoke-media suspension (CSM). Methods Pharmacological SIRT1 activation was induced using SRT2104 and senescence-associated lncRNA 1 (SAL-RNA1) was overexpressed. The expression of SIRT1, FOXO3a, p53, p21, MMP-9, and TIMP-1 in different groups was detected by qRT-PCR and Western blotting; the activity of SA-β gal was detected by staining; the binding of SIRT1 to FOXO3a and p53 gene transcription promoters was detected by Chip. Results We found that CSM increased AECII senescence, while SAL-RNA1 overexpression and SIRT1 activation significantly decreased levels of AECII senescence induced by CSM. Using chromatin immunoprecipitation, we found that SIRT1 bound differentially to transcriptional complexes on the FOXO3a and p53 promoters. Conclusion Our results suggested that lncRNA-SAL1-mediated SIRT1 signaling reduces senescence of AECIIs induced by CSM. These findings suggest a new therapeutic target to limit the irreversible apoptosis of lung epithelial cells in COPD patients.

We consider the problem of referring segmentation in images and videos with natural language. Given an input image (or video) and a referring expression, the goal is to segment the entity referred by the expression in the image or video. In this paper, we propose a cross-modal self-attention (CMSA) module to utilize fine details of individual words and the input image or video, which effectively captures the long-range dependencies between linguistic and visual features. Our model can adaptively focus on informative words in the referring expression and important regions in the visual input. We further propose a gated multi-level fusion (GMLF) module to selectively integrate self-attentive cross-modal features corresponding to different levels of visual features. This module controls the feature fusion of information flow of features at different levels with high-level and low-level semantic information related to different attentive words. Besides, we introduce cross-frame self-attention (CFSA) module to effectively integrate temporal information in consecutive frames which extends our method in the case of referring segmentation in videos. Experiments on benchmark datasets of four referring image datasets and two actor and action video segmentation datasets consistently demonstrate that our proposed approach outperforms existing state-of-the-art methods., 14 pages, 8 figures. arXiv admin note: substantial text overlap with arXiv:1904.04745

Recently, deep convolution neural networks have achieved remarkable performance in the task of single image super-resolution (SISR). However, effectiveness of existing networks highly relies on their receptive field, which always increases with the depth of the network. In this work, we propose a novel module, named as residual group, to effectively learn feature maps by using dynamic receptive field. This residual group firstly uses a selective kernel convolution layer to dynamically learn multi-scale information from its input features. Then, several residual blocks are employed to further refine the learned feature. In addition, we also propose a selective feature fusion module to fuse appearance information in multi-level features. Within this module, the low-level features and high-level features are selectively fused to complement the high-level ones. Finally, by combining these two methods, we introduce a multi-level feature fusion network (MLFFN) for single image super-resolution (SISR). Through comprehensive experiments, we demonstrate that the proposed MLFFN achieves state-of-the-art performance both quantitatively and qualitatively.

Linear discriminant analysis (LDA) is a popular technique to learn the most discriminative features for multi-class classification. A vast majority of existing LDA algorithms are prone to be dominated by the class with very large deviation from the others, i.e., edge class, which occurs frequently in multi-class classification. First, the existence of edge classes often makes the total mean biased in the calculation of between-class scatter matrix. Second, the exploitation of l2-norm based between-class distance criterion magnifies the extremely large distance corresponding to edge class. In this regard, a novel self-weighted robust LDA with l21-norm based pairwise between-class distance criterion, called SWRLDA, is proposed for multi-class classification especially with edge classes. SWRLDA can automatically avoid the optimal mean calculation and simultaneously learn adaptive weights for each class pair without setting any additional parameter. An efficient re-weighted algorithm is exploited to derive the global optimum of the challenging l21-norm maximization problem. The proposed SWRLDA is easy to implement, and converges fast in practice. Extensive experiments demonstrate that SWRLDA performs favorably against other compared methods on both synthetic and real-world datasets, while presenting superior computational efficiency in comparison with other techniques., 17 pages, has been accepted by ACM TIST

Background China has an extremely low exclusive breastfeeding rate. Kangaroo mother care (KMC) has been shown to increase the exclusive breastfeeding rate among infants born extremely or very preterm. However, there is limited evidence surrounding intermittent KMC and exclusive breastfeeding in late preterm infants. In our study we investigated the association between the provision of intermittent KMC and breastfeeding practice for late preterm infants in four hospitals in different provinces of China. Methods Intermittent KMC was recommended to the mothers of all preterm infants admitted to the postnatal wards of participating hospitals between March 2018 and March 2019. Those who agreed to practice KMC were enrolled in the “KMC group”, those who did not were enrolled in the “No KMC group”. Basic maternal socio-demographic information was collected, feeding practice; outcome and method, were recorded daily whilst in hospital. A follow-up survey of feeding practice was conducted 42 days after discharge. Calculations for feeding practice were performed separately for both groups. Logistics regression was used to analyze the association between KMC and feeding outcome and method, adjusting for socio-demographic covariates. Results Among the 844 mothers participating in the study, 627 (74.3%) chose to perform KMC. More of the mothers who provided KMC were exclusively breast milk feeding their infants in the 24 h before hospital discharge (54.6%) and at follow-up (57.3%), compared to mothers who did not provide KMC (34.6% at discharge and 33.2% at follow-up,). Mothers in the KMC group were more likely to be breastfeeding (method) than mothers in the No KMC group (65.3% vs. 52.1% at discharge, and 83.1% vs. 67.3% at follow up). Logistic regression indicated that compared with the No KMC group, mothers who provided KMC were twice as likely to be exclusively breast milk feeding their infants at discharge (OR = 2.15 (95% CI 1.53, 3.02)), use breastfeeding method at discharge as opposed to other means such as bottle or cup feeding (OR = 1.61 (95% CI 1.15, 2.25)), be exclusive breast milk feeding at follow-up (OR = 2.55 (95% CI 1.81, 3.61)), and use breastfeeding method at follow-up (OR = 2.09 (95% CI 1.44, 3.02)). Conclusions Intermittent KMC was associated with a nearly doubled increase in exclusive breast milk feeding (outcome) and breastfeeding (method) at both discharge and 42 days after discharge for late preterm infants. This is especially important in China where exclusive breastfeeding rates are low, intermittent KMC provides a feasible means to increase the likelihood of these vulnerable infants receiving the benefits of exclusive breastmilk.

Background and Aims We studied the downstream and mechanism of β-arrestins signaling in renal fibrosis process and the role of lysyl oxidase (LOX) in the AT1R-β-arrestins pathway. Method The mechanism of β-arrestins signaling was studied in normal rat kidney tubule epithelial cells (NRK-52E) treated with SII in vitro. BAPN or placebo was administered during ischemia reperfusion (IR)-induced fibrosis progression. Collagen crosslinking and fibrosis progression were assessed histologically and biochemically. Results The mRNA and protein levels of β-arrestin-1 and β-arrestin-2 were significantly upregulated in renal fibrosis model both in vitro and in vivo. SII activated the ERK-STAT3 PY705 but not STAT3-Try727 in nucleus of NRK-52E cells, which effects were abolished when transfection of siRNA targeting β-arrestin-1 and β-arrestin-2 or pretreated with PD98059 (MEK inhibitor). LOX was strongly induced in fibrotic kidney and NRK-52E cells treated with SII. Active LOX significantly increased collagen crosslinking. In established IR-28d renal fibrosis, LOX inhibition promoted fibrosis reversal and with a 25% decrease insoluble collagen. Gene silencing of β-arrestin-1 + 2 or STAT3 apparently inhibited SII-induced LOX expression in vitro. Besides, chromatin immunoprecipitation (ChIP) assay clearly demonstrating the interaction between STAT3 and the LOX promoter, which indicated LOX is a direct target gene of SII-β-arrestins-STAT3 signaling. Conclusion The ERK/STAT3 was downstream of AT1R-β-arrestins, ERK entered the nucleus and activated STAT3-PY705. LOX mediates collagen crosslinking and fibrotic matrix stabilization during renal fibrosis via the AT1R-β-arrestins-ERK-STAT3-PY705 signaling. By blocking this profibrotic pathway, therapeutic LOX inhibition attenuates the fibrosis and suggesting target the LOX has significant potency for the treatment of patients with fibrotic kidney disorders.