Search

Objective: To investigate the incidence rate, clinical and gene mutation characteristics of multiple acyl-CoA dehydrogenase deficiency (MADD) in newborns in Zhejiang province. Methods: A total of 3 896 789 newborns were screened for MADD using tandem mass spectrometry in Zhejiang Neonatal Screening Center during January 2009 and December 2020. Patients of MADD were confirmed by urine organic acid and electron transferring flavoprotein ( ETF) or electron transferring flavoprotein dehydrogenase ( ETFDH ) gene detection. MADD patients were given diet and life management, supplemented with L-carnitine, riboflavin and coenzyme Q 10 treatment, and their growth and intellectual development were evaluated during the followed up. Results: Thirteen patients with MADD were diagnosed, with an incidence of 1/299 753. One patient was type Ⅱ, and the rest were type Ⅲ. Patients were followed up for 3–45 months, 1 case died, 4 cases had acute metabolic disorders with hypoglycemia as the main manifestation due to infection, 1 case had hypotonia, and the rest 7 cases developed well. Patients had raised levels of C4–C18:1 acylcarnitines in the initial screening. Thirteen children were genetically tested, 1 case with compound heterozygous mutation in the ETFA gene, 1 case with homozygous mutation in the ETFA gene, 1 case with compound heterozygous mutation in the ETFB gene, 8 cases with compound heterozygous mutation and 1 case with homozygous mutation in the ETFDH gene, 1 case that only 1 locus of ETFDH gene was detected. The c.250G>A was the hotspot mutation in this study. Conclusion: The clinical manifestations of MADD are highly heterogeneous. The neonatal-onset form is serious, and late onset form usually has no obvious clinical symptoms. C4–C18:1 acylcarnitines usually increased in the initial screening, and the hotspot gene mutation is c.250G>A.

Since the traditional mobile edge computing (MEC) server is fixed at the edge of the network, it is likely to cause severe propagation loss between the edge server and users. However, ground users can be served when the MEC server is integrated into the unmanned aerial vehicle (UAV) base station (BS). Nevertheless, there are three main factors, including the limited system radio resources, the finite UAV energy and severe interference between the uplink and downlink, that make it challenging for the system to guarantee quality-of-experience of users in the downlink while ensuring the minimum amount of offloading data for users in the uplink. Therefore, this paper investigates computation-efficiency maximization. We jointly optimize computing scheduling, UAV 3D trajectory, bandwidth allocation and transmission power control to maximize the amount of offloaded data, minimize UAV energy consumption, and simultaneously guarantee the QoE of users in the downlink. Due to the nonconvex and nonconcave objective function and the coupling between variables, a multi-stage alternative optimization algorithm is proposed to solve the problem. The simulation results have demonstrated that the computation-efficiency obtained by the proposed scheme is higher than that of other benchmark schemes and meets the QoE demands of users under insufficient resources.

Objectives Determination of methylmalonic acid, 2-methylcitric acid, and total homocysteine in dried blood spots by liquid chromatography–tandem mass spectrometry has usually been used as a second-tier test to improve performance of newborn screening for propionylcarnitine-related disorders. However, factors that potentially affect its detection results have not been investigated, and we aimed to evaluate these influencing factors and explore their potential utility in newborn screening and initial follow-up for propionylcarnitine-related disorders. Methods This study comprised a prospective group (1998 healthy infants, to establish cutoff values and investigate the influencing factors) and a retrospective group (804 suspected positive cases screened from 381, 399 newborns for propionylcarnitine-related disorders by tandem mass spectrometry, to evaluate the performance of newborn screening and initial follow-up). Results Cutoff values for methylmalonic acid, 2-methylcitric acid, and total homocysteine were 2.12, 0.70, and 10.05 µmol/l, respectively. Concentration of methylmalonic acid, 2-methylcitric acid, and total homocysteine in dried blood spots is not impacted by sex, age, birth weight, gestational age, or dried blood spot storage time. A total of 75 of 804 cases were screened positive by combined tandem mass spectrometry and liquid chromatography–tandem mass spectrometry, thus eliminating 90% of the false positives without compromising sensitivity. Eighteen propionylcarnitine-related disorders were successfully identified, including one CblX case missed in the initial follow-up by tandem mass spectrometry. Conclusions Methylmalonic acid, 2-methylcitric acid, and total homocysteine detected in dried blood spots by liquid chromatography–tandem mass spectrometry is a reliable, specific, and sensitive approach for identifying propionylcarnitine-related disorders. We recommend this assay should be performed rather than tandem mass spectrometry in follow-up for propionylcarnitine-related disorders besides second-tier tests in newborn screening.

In the past several years, unmanned aerial vehicle (UAV) have been employed to provide enhanced coverage or relay service to mobile users in a scenario with limited or even no infrastructure since they can be deployed to almost everywhere and can be manipulated at anytime. This paper studies UAV as aerial base station (BS) enabled wireless communication system, where a UAV is dispatched to provide wireless communication service to a set of ground users with difference quality-of-experience (QoE) requirements. In real world, user requirements are randomly and unevenly distributed. In addition, UAV communication coverage and on-board energy are limited and system resources are also limited (e.g., transmission power, spectrum). In order to meet the QoE of all users with limited system resources and limited UAV energy, we jointly optimize user communication scheduling, UAV trajectory, transmit power and bandwidth allocation to maximize energy-efficiency and satisfy user QoE requirement. The formulated problem is mixes integer non-convex and non-concave so it is difficult to solve. In this paper, we solvevv the problem with two steps as follows. Firstly, we transform the objective function into a tractable form. Secondly, we divide the optimal problem into four sub-optimal problems, and then use a powerful iterative algorithm with the Dinkelbach and block coordinate descent to solve the optimal problem. That is to say, the user scheduling, UAV trajectory, transmission power and bandwidth allocation are alternately optimized in each iteration. Extensive simulation results present that our proposed method can obtain higher energy efficiency than that of baselines. Specifically, the energy efficiency obtained by our proposed method is 12.5% higher than the approach that only maximizes throughput.

We present a case of an asymptomatic chronic middle cerebral artery (MCA) occlusion that was incidentally detected on transcranial color-coded sonography (TCCS) and confirmed by magnetic resonance angiography (MRA). Besides the presence of accelerated blood flow velocity in the ipsilateral anterior cerebral artery (ACA) and posterior cerebral artery (PCA), which is known as signs of pial collaterals, TCCS specifically visualized some deep anastomoses and new vessels running along the occluded MCA stem course. Therefore, TCCS showed transcortical collaterals indirectly and deep anastomoses and neovascularization directly, providing at least as much information as MRA in explaining patient’s lack of symptoms and predicting his outcome. However, during TCCS procedures, we should be aware of the possibility of mistaking neovascularization as a patent MCA stem, tortuous MCA, or indistinct MCA influenced by poor temporal bone window.

Curriculum Learning (CL) selects the training samples from the easy to difficult to boost the classification results. Most existing variates of CL measure the difficulty level of an example in an intuitive way, i.e., the loss value between the prediction and the ground truth. This way ignores the different distances to every class boundary, which is implied in the prediction vectors. In this paper, we propose a novel CL framework, named Balance Loss Curriculum Learning(BLCL), to reveal the comprehensive difficulty level of an example and improve the curriculum process based on the deep architectures. We follow the teacher-student learning module and train the teacher & student networks simultaneously. The predictions learned by the teacher network indicates the boundary information and is set as the difficulty measurement to guide the student network. The student network focuses on the easy samples according to the boundary information by decreasing the loss from difficult examples as well as increasing the loss from easy examples. Through extensive and rigor experiments, we demonstrate the effectiveness and generality of our framework.

To evaluate the feasibility of incorporating genetic screening for neonatal intrahepatic cholestasis, caused by citrin deficiency (NICCD), into the current newborn screening (NBS) program. We designed a high-throughput iPLEX genotyping assay to detect 28 SLC25A13 mutations in the Chinese population. From March 2018 to June 2018, 237 630 newborns were screened by tandem mass spectrometry at six hospitals. Newborns with citrulline levels between 1/2 cutoff and cutoff values of the upper limit were recruited for genetic screening using the newly developed assay. The sensitivity and specificity of the iPLEX genotyping assay both reached 100% in clinical practice. Overall, 29 364 (12.4%) newborns received further genetic screening. Five patients with conclusive genotypes were successfully identified. The most common SLC25A13 mutation was c.851_854del, with an allele frequency of 60%. In total, 658 individuals with one mutant allele were identified as carriers. Eighteen different mutations were observed, yielding a carrier rate of 1/45. Notably, Quanzhou in southern China had a carrier rate of up to 1/28, whereas Jining in northern China had a carrier rate higher than that of other southern and border cities. The high throughput iPLEX genotyping assay is an effective and reliable approach for NICCD genotyping. The combined genetic screening could identify an additional subgroup of patients with NICCD, undetectable by conventional NBS. Therefore, this study demonstrates the viability of incorporating genetic screening for NICCD into the current NBS program.

The combination of visual recognition and language understanding is aim to build a commonly shared space between heterogeneous data of vision and text, such as the tasks of image captioning and visual question answering (VQA). Most existing approaches convert an image into a semantic visual feature vector via deep convolutional neural networks (CNN), while keep the sequential property of text data and represent it with Recurrent Neural Networks(RNN). The key to analyse multi-source heterogeneous data is to construct the inherent correlations between data. In order to reduce the heterogeneous gap among the vision and language, in this work, we represent the image in a sequential way as well as the text. We utilize the objects in the visual scenes and convert the image to a sequence of detected objects and their locations. Then we analogize a sequence of objects(visual language) to a sequence of words(natural language). We take the order of objects into account and evaluate different permutations and combinations of objects. Experimental results on the image captioning and VQA benchmarks demonstrate our hypothesis it’s beneficial to appropriately arrange objects sequence on the Vision-to-Language(V2L) problems.

Generative Adversarial Networks (GANs) have been widely used for the image-to-image translation task. While these models rely heavily on the labeled image pairs, recently some GAN variants have been proposed to tackle the unpaired image translation task. These models exploited supervision at the domain level with a reconstruction process for unpaired image translation. On the other hand, parallel works have shown that leveraging perceptual loss functions based on high level deep features could enhance the generated image quality. Nevertheless, as these GAN-based models either depended on the pretrained deep network structure or relied on the labeled image pairs, they could not be directly applied to the unpaired image translation task. Moreover, despite the improvement of the introduced perceptual losses from deep neural networks, few researchers have explored the possibility of improving the generated image quality from classical image quality measures. To tackle the above two challenges, in this paper, we propose a unified quality-aware GAN-based framework for unpaired image-to-image translation, where a quality-aware loss is explicitly incorporated by comparing each source image and the reconstructed image at the domain level. Specifically, we design two detailed implementations of the quality loss. The first method is based on a classical image quality assessment measure by defining a classical quality-aware loss. The second method proposes an adaptive deep network based loss. Finally, extensive experimental results on many real-world datasets clearly show the quality improvement of our proposed framework, and the superiority of leveraging classical image quality measures for unpaired image translation compared to the deep network based model., IEEE Transactions on Multimedia

Background The inverted classroom model differs from the traditional teaching model as it reverses the pattern of knowledge transfer and internalization. In recent years, this new teaching model has received much attention in undergraduate medical education. Pathophysiology is a course in the undergraduate Chinese medical curriculum that is critical in bridging basic medical science and clinical medicine. Objective The purpose of this study was to investigate the application of inverted classroom in delivering the course on pathophysiology to Chinese undergraduate medical students. Methods In the spring semester of 2018, inverted classroom teaching was implemented for second-year clinical medicine students at the College of Medicine at Nanchang University. The topics of hypoxia and respiratory failure were selected for the inverted classroom study. The effect of the inverted classroom on teaching pathophysiology was evaluated using classroom performance metrics, a final examination, and questionnaires. Results This study found that students in the inverted classroom group achieved higher scores in their in-course assessments (82.35 [SD 11.45] vs 81.33 [SD 9.51], respectively) and in their final exams (73.41 [SD 10.37] vs 71.13 [SD 11.22], respectively) than those in the traditional lecture-based group, but the scores were not significantly different (P=.13, unpaired two-tailed t test). There was also no significant difference in the distribution of the score segments in the class quiz (P=.09, chi-square test) and in the final exams (P=.25, chi-square test) between the 2 groups. Further, most of the students reported that the inverted classroom increased their learning motivation, made them more confident, and helped them understand the content on pathophysiology better. The students in the inverted classroom also improved in their problem-solving skills and teamwork abilities. However, some students from the inverted classroom group also reported that the self-learning and preparatory work before class increased their learning burden. Conclusions This study shows the feasibility and promise of inverted classroom for teaching pathophysiology to undergraduate Chinese medical students. The inverted classroom improves students’ learning interests and attitudes toward learning. However, further studies are required to assess the benefits of broader acceptance and implementation of the inverted classroom among Chinese undergraduate medical students.

Background Chronic alcohol intake is associated with an increased risk of alcoholic cardiomyopathy, which may present with pathological changes such as myocardial insulin resistance, leading to ventricular dilation and cardiac dysfunction. Although a correlation between microRNA-155 (miR-155) and insulin signaling has been identified, the underlying mechanism has not been elucidated to date. The purpose of the study was to determine whether overexpression of miR-155-5p in vivo could ameliorate chronic alcohol-induced myocardial insulin resistance and cardiac dysfunction. Material and Methods Wistar rats were fed with either alcohol or water for 20 weeks to establish chronic alcohol intakes model. Then the alcohol group were divided into three groups: model group, miRNA-155 group and AAV-NC group. Rats undergoing alcohol treatment were injected with AAV-miRNA-155 (adeno-associated virus 9) or its negative control AAV-NC, respectively. Gene expression was determined by real-time PCR, and protein expression was determined by western blot. Echocardiography was performed to assess terminal cardiac function. Insulin responsiveness was determined through the quantification of phosphorylated insulin receptor substrate 1 (ser 307) and phosphorylated insulin receptor (Tyr 1185) levels. Results We found that cardiac function was attenuated in chronic alcohol intake rats, with an activated mammalian target of rapamycin (mTOR) signaling pathway, accompanied by an increase in p-IRS1(ser 307) and a decrease in p-IR (Tyr 1185) level in myocardial tissue. Also, alcohol drinking significantly up-regulated miR-155-5p level and its overexpression decreased p-IRS1 (ser 307) and increased p-IR (Tyr 1185) levels, and meanwhile inhibited the mTOR signaling pathway. Conclusion miR-155-5p upregulation ameliorates myocardial insulin resistance via the mTOR signaling in chronic alcohol drinking rats. We propose that miR-155 may serve as a novel potential therapeutic target for alcoholic heart disease.

Image dehazing is a fundamental task for the computer vision and multimedia and usually in the face of the challenge from two aspects, i) the uneven distribution of arbitrary haze and ii) the distortion of image pixels caused by the hazed image. In this paper, we propose an end-to-end trainable framework, named Weighted-Fusion Network with Poly-Scale Convolution (WFN-PSC), to address these dehazing issues. The proposed method is designed based on the Poly-Scale Convolution (PSConv). It can extract the image feature from different scales without upsampling and downsampled, which avoids the image distortion. Beyond this, we design the spatial and channel weighted-fusion modules to make the WFN-PSC model focus on the hard dehazing parts of image from two dimensions. Specifically, we design three Part Architectures followed by the channel weighted-fusion module. Each Part Architecture consists of three PSConv residual blocks and a spatial weighted-fusion module. The experiments on the benchmark demonstrate the dehazing effectiveness of the proposed method. Furthermore, considering that image dehazing is a low-level task in the computer vision, we evaluate the dehazed image on the object detection task and the results show that the proposed method can be a good pre-processing to assist the high-level computer vision task.

Objectives To study the effect of endoscopic endonasal surgery on nasal function for the treatment of clival chordoma. Methods Pre and post-operative computed tomography (CT) scans of a case of chordoma treated with an endoscopic endonasal approach (EEA) were collected retrospectively, and models of the nasal cavity were reconstructed so that a subsequent numerical simulation of nasal airflow characteristics, warming, and humidification could be conducted. Results Middle turbinectomy resulted in redistribution of airflow within the nasal cavity, and the most significant changes occurred in the middle section. Consistent with the results of airflow evaluation, it was found that the change in nasal anatomical structure significantly reduced warming and humidification. Nasal humidification decreased substantially when postoperative loss of mucosa was taken into consideration. The H2O mass fraction of pharynx in inspiration phase were significantly correlated with airway surface-to-volume ratio (SVR). Conclusions The EEA for chordoma significantly affected nasal function. Attention should be paid to the protection of nasal structure and the associated mucosa.

Current state-of-the-art image captioning models adopt autoregressive decoders, \ie they generate each word by conditioning on previously generated words, which leads to heavy latency during inference. To tackle this issue, non-autoregressive image captioning models have recently been proposed to significantly accelerate the speed of inference by generating all words in parallel. However, these non-autoregressive models inevitably suffer from large generation quality degradation since they remove words dependence excessively. To make a better trade-off between speed and quality, we introduce a semi-autoregressive model for image captioning~(dubbed as SATIC), which keeps the autoregressive property in global but generates words parallelly in local . Based on Transformer, there are only a few modifications needed to implement SATIC. Experimental results on the MSCOCO image captioning benchmark show that SATIC can achieve a good trade-off without bells and whistles. Code is available at {\color{magenta}\url{https://github.com/YuanEZhou/satic}}., Comment: Revised

Image captioning task requires a comprehensive understanding of visual content and has received a significant amount of attention. Recent studies have revealed that modelling relationships between visual objects imply a high-level semantic feature. However, most existing relationship modelling methods for image captioning heavily rely on the object detection results and handcrafted structured label to build the graph model. In this paper, we explore the relationships in a text-guided way via the descriptions from similar images to provide the context clues. We propose a novel framework named Text-Guided Graph (TGG) to employ image-related text to help build the relationship between objects in the image and incorporate the high-level graph information and captions associated with a certain image. Experiments conducted on the MS COCO dataset demonstrate the effectiveness of our text-guided graph model under various standard evaluation metrics.

Guojin Xia,1,2 Zhenzhen Hu,3 Fuqing Zhou,1,2 Wenfeng Duan,1,2 Min Wang,1,2 Honghan Gong,1,2 Yulin He,1,2 Yanxing Guan4 1Department of Radiology, The First Affiliated Hospital of Nanchang University, Jiangxi, People’s Republic of China; 2Jiangxi Province Medical Imaging Research Institute, Jiangxi, People’s Republic of China; 3Department of Thoracic Oncology, Jiangxi Cancer Hospital, Jiangxi, People’s Republic of China; 4Department of Nuclear Medicine, The First Affiliated Hospital of Nanchang University, Jiangxi, People’s Republic of ChinaCorrespondence: Yulin He; Yanxing Guan 17 Yongwaizheng Street, Nanchang City, Jiangxi Province, People’s Republic of ChinaTel +86-791-88692585Email 33heyulin@163.com; yanxingguan2000@aliyun.comPurpose: We explored changes in spontaneous brain connectivity in patients with diffuse axonal injury (DAI), assessed via functional connectivity density (FCD) tests using different frequency bands.Patients and Methods: In all, 23 patients with DAI (17 males and 6 females) and 23 healthy controls (HCs; 17 males and 6 females) were included. Functional magnetic resonance imaging scans were performed when the participants were in a resting state and the FCD levels in three frequency bands (slow-4: 0.027– 0.073 Hz, slow-5: 0.01– 0.027 Hz, and typical: 0.01– 0.08 Hz) were measured. In addition, Pearson’s correlation coefficient was used to explore the relationship between clinical indices and brain regions with abnormal FCD values.Results: Compared to HCs, DAI patients had significantly greater FCD values in the right extranuclear/limbic lobe/cingulate gyrus and left limbic lobe/hippocampus/parahippocampal gyrus, and significantly lower FCD values in the left precuneus/posterior cingulate gyrus, in the slow-4 band. In the slow-5 band, the DAI patients had higher FCD values in the left inferior temporal gyrus/superior temporal gyrus, left parahippocampal gyrus/limbic lobe, left extranuclear/cingulate gyrus, and right medial frontal gyrus, and lower values in the right inferior frontal gyrus, right inferior parietal lobule, and left cingulate gyrus/limbic lobe. Moreover, compared to HCs, the values in the typical band were higher in the right extranuclear/limbic lobe/hippocampus/parahippocampal gyrus, but were significantly lower in the right precuneus/posterior cingulate gyrus and right inferior parietal lobule/supramarginal gyrus. The abnormal FCD values of these brain regions were linearly correlated with different clinical scale scores.Conclusion: DAI patients had abnormal FCD values in various brain regions, indicating disruption to the brain functional network. Moreover, the values were frequency dependent. Our results provide new evidence for the pathogenesis of functional impairment and may explain the neuropathological or compensatory mechanism of the disease.Keywords: diffuse axonal injury, functional connectivity density, functional magnetic resonance imaging, frequency bands

Cognitive radio networks (CRNs), which allow secondary users (SUs) to dynamically access a network without affecting the primary users (PUs), have been widely regarded as an effective approach to mitigate the shortage of spectrum resources and the inefficiency of spectrum utilization. However, the SUs suffer from frequent spectrum handoffs and transmission limitations. In this paper, considering the quality of service (QoS) requirements of PUs and SUs, we propose a novel dynamic flow-adaptive spectrum leasing with channel aggregation. Specifically, we design an adaptive leasing algorithm, which adaptively adjusts the portion of leased channels based on the number of ongoing and buffered PU flows. Furthermore, in the leased spectrum band, the SU flows with access priority employ dynamic spectrum access of channel aggregation, which enables one flow to occupy multiple channels for transmission in a dynamically changing environment. For performance evaluation, the continuous time Markov chain (CTMC) is developed to model our proposed strategy and conduct theoretical analyses. Numerical results demonstrate that the proposed strategy effectively improves the spectrum utilization and network capacity, while significantly reducing the forced termination probability and blocking probability of SU flows.

Testing for primary carnitine deficiency (PCD) has been implemented in many newborn screening (NBS) programs, but few large-scale studies on NBS for PCD have been reported in China. This study aimed to assess the incidence and biochemical, clinical, and genetic characteristics of PCD discovered by NBS. Dried blood spots from newborns were analyzed by tandem mass spectrometry (MS/MS) and suspected positive patients were further tested using molecular genetic analysis. Infants who carried two variants in SLC22A5 or those with extremely low free carnitine levels during recall were referred for follow-up and treatment. Over 3.4 million newborns were screened and 113 newborns were diagnosed with PCD, yielding a positive predictive value of 1.93%. In addition, 63 mothers with PCD were identified. The incidence of PCD in newborns and mothers in Zhejiang was 1:30,182 and 1:54,137, respectively. Thirty-seven distinct variants were identified in SLC22A5 of which 10 were novel. c.1400C > G (p.S467C) was the most prevalent variant in both newborns and mothers with PCD, while c.760C > T (p.R254*), which is reportedly common in other Chinese regions, was rarely detected in maternal PCD patients. This study reports the largest series of patients with PCD detected by NBS and identifies 10 novel variants, expanding the variant spectrum of SLC22A5.

Although drugs such as barbiturates and benzodiazepines are often used for the treatment of insomnia, they are associated with various side effects such as habituations, tolerance and addiction. Alternatively, natural products with minimal unwanted effects have been preferred for the treatment of acute and/or mild insomnia, with additional benefits of overall health-promotion. Basic and clinical researches on the mechanisms of action of natural products have been carried out so far in insomnia treatments. Recent studies have been focusing on diverse chemical components available in natural products, with an interest of developing drugs that can improve sleep duration and quality. In the last 15 years, our co-workers have been actively looking for candidate substances from natural products that can relieve insomnia. This review is, therefore, intended to bring pharmacological data regarding to the effects of natural products on sleep duration and quality, mainly through the activation of GABAA receptors. It is imperative that phytochemicals will provide useful information during electroencephalography (EEG) analysis and serve as an alternative medications for insomnia patients who are reluctant to use conventional drugs.

Age estimation based on the human face remains a significant problem in computer vision and pattern recognition. In order to estimate an accurate age or age group of a facial image, most of the existing algorithms require a huge face data set attached with age labels. This imposes a constraint on the utilization of the immensely unlabeled or weakly labeled training data, e.g., the huge amount of human photos in the social networks. These images may provide no age label, but it is easy to derive the age difference for an image pair of the same person. To improve the age estimation accuracy, we propose a novel learning scheme to take advantage of these weakly labeled data through the deep convolutional neural networks. For each image pair, Kullback-Leibler divergence is employed to embed the age difference information. The entropy loss and the cross entropy loss are adaptively applied on each image to make the distribution exhibit a single peak value. The combination of these losses is designed to drive the neural network to understand the age gradually from only the age difference information. We also contribute a data set, including more than 100 000 face images attached with their taken dates. Each image is both labeled with the timestamp and people identity. Experimental results on two aging face databases show the advantages of the proposed age difference learning system, and the state-of-the-art performance is gained.

Visual attention not only improves the performance of image captioners, but also serves as a visual interpretation to qualitatively measure the caption rationality and model transparency. Specifically, we expect that a captioner can fix its attentive gaze on the correct objects while generating the corresponding words. This ability is also known as grounded image captioning. However, the grounding accuracy of existing captioners is far from satisfactory. To improve the grounding accuracy while retaining the captioning quality, it is expensive to collect the word-region alignment as strong supervision. To this end, we propose a Part-of-Speech (POS) enhanced image-text matching model (SCAN \cite{lee2018stacked}): POS-SCAN, as the effective knowledge distillation for more grounded image captioning. The benefits are two-fold: 1) given a sentence and an image, POS-SCAN can ground the objects more accurately than SCAN; 2) POS-SCAN serves as a word-region alignment regularization for the captioner's visual attention module. By showing benchmark experimental results, we demonstrate that conventional image captioners equipped with POS-SCAN can significantly improve the grounding accuracy without strong supervision. Last but not the least, we explore the indispensable Self-Critical Sequence Training (SCST) \cite{Rennie_2017_CVPR} in the context of grounded image captioning and show that the image-text matching score can serve as a reward for more grounded captioning \footnote{https://github.com/YuanEZhou/Grounded-Image-Captioning}., Comment: Accepted by CVPR 2020

Knowledge distillation is an effective way to transfer the knowledge from a pre-trained teacher model to a student model. Co-distillation, as an online variant of distillation, further accelerates the training process and paves a new way to explore the “dark knowledge” by training n models in parallel. In this paper, we explore the “divergent examples”, which can make the classifiers have different predictions and thus induce the “dark knowledge”, and we propose a novel approach named Adversarial Co-distillation Networks (ACNs) to enhance the “dark knowledge” by generating extra divergent examples. Note that we do not involve any extra dataset, and we only utilize the standard training set to train the entire framework. ACNs are end-to-end frameworks composed of two parts: an adversarial phase consisting of Generative Adversarial Networks (GANs) to generate the divergent examples and a co-distillation phase consisting of multiple classifiers to learn the divergent examples. These two phases are learned in an iterative and adversarial way. To guarantee the quality of the divergent examples and the stability of ACNs, we further design “Weakly Residual Connection” module and “Restricted Adversarial Search” module to assist in the training process. Extensive experiments with various deep architectures on different datasets well demonstrate the effectiveness of our approach.

Wideband sensing-based cognitive radio with simultaneous wireless information and power transfer can be designed for efficient spectrum and energy usage. We first aim to maximize the sum energy harvested by all the energy harvesters subject to constraints on rate, transmit power, interference, and subchannel assignment. Due to the non-convexity of the formulated problem, we relax the integer variable and introduce an auxiliary variable to transform the original problem into a convex problem. Then, the Lagrangian and subgradient methods are adopted to obtain the optimal solutions. However, this scheme may lead to a severe fairness issue among different links. In view of this fact, we further propose an energy-harvesting scheme for max–min fairness. In particular, we aim to maximize the energy harvested by the worst case individual link. We show that this problem can be solved in a similar manner to the problem of sum harvested energy maximization. Simulation results are presented to verify the convergence and fairness performance of the proposed algorithm, and to reveal a novel tradeoff between the network harvested energy and sensing time.

Extracranial internal carotid artery aneurysms (EICAA) are rare and can elicit various neurologic symptoms. Here, we present a case of a saccular EICAA compressing its proximal parent internal carotid artery (ICA). Ultrasonography demonstrated the proximal ICA stenosis and the "tardus-parvus" Doppler waveform downstream. The patient underwent aneurysmectomy and graft interposition. The histologic analysis highly supported an atypical fibromuscular dysplasia. Although this patient only showed a neck mass, the reduced ipsilateral cerebral blood supply was a potential cause for neurologic symptoms. © 2016 Wiley Periodicals, Inc. J Clin Ultrasound 45:116-120, 2017.

Recently, image inpainting task has revived with the help of deep learning techniques. Deep neural networks, especially the generative adversarial networks~(GANs) make it possible to recover the missing details in images. Due to the lack of sufficient context information, most existing methods fail to get satisfactory inpainting results. This work investigates a more challenging problem, e.g., the newly-emerging semantic image inpainting - a task to fill in large holes in natural images. In this paper, we propose an end-to-end framework named progressive generative networks~(PGN), which regards the semantic image inpainting task as a curriculum learning problem. Specifically, we divide the hole filling process into several different phases and each phase aims to finish a course of the entire curriculum. After that, an LSTM framework is used to string all the phases together. By introducing this learning strategy, our approach is able to progressively shrink the large corrupted regions in natural images and yields promising inpainting results. Moreover, the proposed approach is quite fast to evaluate as the entire hole filling is performed in a single forward pass. Extensive experiments on Paris Street View and ImageNet dataset clearly demonstrate the superiority of our approach. Code for our models is available at https://github.com/crashmoon/Progressive-Generative-Networks.

Attention mechanism plays an important role in understanding images and demonstrates its effectiveness in generating natural language descriptions of images. In recent years, with the advance of deep neural networks, visual attention has been well exploited in the encoder-decoder neural network-based framework. On the one hand, existing study shows that guidance captions can help attend to relevant image regions and suppress unimportant ones during the image encoding stage, especially for cluttered images. On the other hand, visual attention has been well exploited during the decoding stage. Observing the naturally complementary property between them, we propose a two-side attention model which combines the attention mechanism seamlessly and associated with a coarse to fine attention mechanism. The original text-guided attention model operates on region-level image feature, which is lack of definite semantic information and causes unsatisfied attention visualization. We alleviate the problem by enabling attention to be calculated at object-level image feature, which helps to obtain performance improvement and more interpretable attention visualization. Experiments conducted on MSCOCO datasets demonstrate the consistent improvement on text-guided attention model for image captioning.

Age estimation is a difficult task which requires the automatic detection and interpretation of facial features. Recently, Convolutional Neural Networks (CNNs) have made remarkable improvement on learning age patterns from benchmark datasets. However, for a face ``in the wild'' (from a video frame or Internet), the existing algorithms are not as accurate as for a frontal and neutral face. In addition, with the increasing number of in-the- wild aging data, the computation speed of existing deep learning platforms becomes another crucial issue. In this paper, we propose a high-efficient age estimation system with joint optimization of age estimation algorithm and deep learning system. Cooperated with the city surveillance network, this system can provide age group analysis for intelligent demographics. First, we build a three- tier fog computing architecture including an edge, a fog and a cloud layer, which directly processes age estimation from raw videos. Second, we optimize the age estimation algorithm based on CNNs with label distribution and K-L divergence distance embedded in the fog layer and evaluate the model on the latest wild aging dataset. Experimental results demonstrate that: 1. our system collects the demographics data dynamically at far-distance without contact, and makes the city population analysis automatically; and 2. the age model training has been speed-up without losing training progress or model quality. To our best knowledge, this is the first intelligent demographics system which has potential applications in improving the efficiency of smart cities and urban living.

Human group, which indicates the people who share similar characteristics, is used to categorize humans into distinct populations or groups. In recent years, with the explosive growth of image, new concepts of human group are blooming in social networks . People in the same human group can be categorized by their facial and clothes appearance characteristics. In this work, we propose an approach to understanding the new concepts of human group with few positive samples. To this end, we construct visual models crossing two modalities related to human images and surrounding texts. Two convolutional neural networks based on face and upper body are constructed separately. Two different convolutional neural networks (CNNs) architectures are explored for visual pre-traing. To assist the human group recognition, we also merge global convolutional feature of the image. The surrounding texts are represented by semantical vectors and utilized as image labels. We transform words in the text into fixed length vectors by the skip-gram model. Then the texts corresponding to each image are converted into one feature vector by sparse coding and max pooling. Given a few positive samples of new concepts of human group, the visual model can be improved to understand the semantical meaning of the new label. The experimental results demonstrate the effectiveness of the proposed visual model and show the excellent learning capacity with few samples.

Video captioning is a crucial task for video understanding and has attracted much attention recently. Regions-of-Interest (ROI) of video always contains the most interesting information for the audience. Different from the ROI of images, the ROI of videos has the property of temporally-continuity (e.g. a moving object, or an action in video clips), which is the focus of people’s attention. Inspired by this insight we propose an approach to automatically trace the Spatial-Temporal Saliency content for video captioning by catching the temporal structure of ROI candidates. To this aim, we employ a set of modules named tracing LSTMs, each of which traces a single ROI candidate of feature maps across the entire video. The temporal structure of global features and ROI features are combined to obtain a rough understanding of video content and information of ROI, which is set as the initial states of the decoder to generate captions. We verify the effectiveness of our method on the public benchmark: the Microsoft Video Description Corpus (MSVD). The experimental results demonstrate that catching temporal ROI information by tracing LSTMs enhances the representation of input videos and achieves the state-of-the-art results.

Prior results on performance analysis for cognitive relay networks mainly involve perfect channel state information (CSI), which is not readily available. Thus, the effects of outdated CSI on the performance of a multi-antenna multi-relay cognitive radio system are investigated in this work. To exploit the benefits of multiple antennas, collaborative zero-forcing beamforming at the secondary source is proposed to enhance the system performance under the outdated CSI. At the destination, the maximum ratio combining (MRC) diversity scheme is adopted. The Nth best relay selection strategy is applied before the secondary data transmission process. Closed-form expressions for the exact and asymptotic outage probabilities are derived for the secondary user over the Rayleigh fading in the first hop and Nakagami-m fading in the second hop with and without feedback delay. These analytical results can reveal the system diversity order and coding gain. Monte Carlo simulations are carried out to verify the correctness of our analysis. These new analytical results can reveal insights into the characteristics of the proposed system over the outdated fading channels, and can provide useful design criteria for relay-assisted spectrum sharing networks.

In this paper, we propose a novel text-art system: input a source picture and some keywords introducing the information about the picture, and the output is the so-called PicWords in the form of the source picture composed of the introduction keywords. Different from traditional text-graphics which are created by highly skilled artists and involve a huge amount of tedious manual work, PicWords is an automatic non-photorealistic rendering (NPR) packing system. Given a source picture, we first generate its silhouette, which is a binary image containing a Yang part and a Yin part. Yang part is for keywords placing while the Yin part can be ignored. Next, the Yang part is further over-segmented into small patches, each of which serves as a container for one keyword. To make sure that more important keywords are put into more salient and larger image patches, we rank both the patches and keywords and construct a correspondence between the patch list and keyword list. Then, mean value coordinates method is used for the keyword-patch warping. Finally, certain post-processing techniques are adopted to improve the aesthetics of PicWords. Extensive experimental results well demonstrate the effectiveness of the proposed PicWords system.

In this paper we address the problem of automatically parsing the fashion images with weak supervision from the user-generated color-category tags such as “red jeans” and “white T-shirt”. This problem is very challenging due to the large diversity of fashion items and the absence of pixel-level tags, which make the traditional fully supervised algorithms inapplicable. To solve the problem, we propose to combine the human pose estimation module, the MRF-based color and category inference module and the (super)pixel-level category classifier learning module to generate multiple well-performing category classifiers, which can be directly applied to parse the fashion items in the images. Besides, all the training images are parsed with color-category labels and the human poses of the images are estimated during the model learning phase in this work. We also construct a new fashion image dataset called Colorful-Fashion, in which all 2,682 images are labeled with pixel-level color-category labels. Extensive experiments on this dataset clearly show the effectiveness of the proposed method for the weakly supervised fashion parsing task.

Methods Six age groups were defined: 3-4 d (Group A), 5-7 d (Group B), 8-10 d (Group C), 11-15 d (Group D), 16-20 d (Group E), and 21-28 d (Group F). Amino acid, free carnitine, and acylcarnitine concentrations of dried whole blood spots were obtained from the MS/MS analysis. Linear regression analyses of all results of highly diseaserelated analytes were carried out to identify covariates associated with sampling time in Matlab. For the respective regression coefficients (RC), plus represents the analyte levels elevate as the sampling time prolongs, and vice versa for minus.

Cerebral hyperperfusion syndrome (CHS) is a well-documented complication after carotid endarterectomy or stenting. In contrast, CHS following vertebral revascularization is extremely rare. Here we present a case of a 77-year-old man with high-grade vertebral stenosis who subsequently underwent balloon angioplasty, complicated by hemorrhagic CHS manifesting as cortical blindness, although strict postoperative blood pressure control was administered. To our knowledge, cortical blindness as a presentation of hemorrhagic CHS has not previously been reported. This study highlights the fact that identifying high-risk patients, as well as making an individual therapeutic plan, is important prior to revascularization. Further studies are needed to elucidate the exact mechanism of this condition and thereby prevent it.

Coded-cooperation cognitive relay networks can improve both spectrum utilization and bandwidth efficiency. However, the performance analysis of such a system has not been well investigated. Therefore, in this paper, the outage performance of the coded-cooperation cognitive multirelay network with a direct link is studied. Both the Nth best relay selection scheme, where the Nth best relay is selected, and the sum relays scheme, where the relays that successfully decode the first time-slot information all participate in data transmission, are considered under the interference temperature and the maximum transmit power constraint to protect the primary user in a cognitive radio network. Analytical exact outage probability and closed-form asymptotic outage probability expressions are derived for the secondary user. The diversity and coding gains are deduced from them. Monte Carlo simulations are carried out to verify the correctness of the theoretic analysis. The results show that the outage performance of both the best relay selection scheme and the sum relays scheme is close to each other, and the primary network parameters only affect the coding gain and are not related to the diversity order.

Perfluorooctanoic acid (PFOA) is widely present in the environment and has been reported to induce hepatic toxicity in animals and humans. In this study, mice were orally administered different concentrations of PFOA (2.5, 5, or 10 mg/kg/day). Histological examination showed that the exposure to PFOA for 14 consecutive days led to serious hepatocellular injury and obvious inflammatory cell infiltration. In addition, malondialdehyde formation and hydrogen peroxide generation, indicators of oxidative stress, were significantly induced by PFOA treatment in the liver of mice. Furthermore, hepatic levels of interleukin-6, cyclooxygenase-2, and C-reactive protein, markers of inflammatory response, were markedly increased by exposure to PFOA in mice. These results demonstrated that PFOA-induced hepatic toxicity may be involved in oxidative stress and inflammatory response in mice.

To preserve the precious traditional heritage Chinese shadow puppetry, we propose the puppetry eHeritage, including a creator module and a manipulator module. The creator module accepts a frontal view face image and a profile face image of the user as input, and automatically generates the corresponding puppet, which looks like the original person and meanwhile has some typical characteristics of traditional Chinese shadow puppetry. In order to create the puppet, we first extract the central profile curve and warp the reference puppet eye and eyebrow to the shape of the frontal view eye and eyebrow. Then we transfer the puppet texture to the real face area. The manipulator module can accept the script provided by the user as input and automatically generate the motion sequences. Technically, we first learn atomic motions from a set of shadow puppetry videos. A scripting system converts the user's input to atomic motions, and finally synthesizes the animation based on the atomic motion instances. For better visual effects, we propose the sparsity optimization over simplexes formulation to automatically assemble weighted instances of different atomic actions into a smooth shadow puppetry animation sequence. We evaluate the performance of the creator module and the manipulator module sequentially. Extensive experimental results on the creation of puppetry characters and puppetry plays well demonstrate the effectiveness of the proposed system.

Cordycepin (3′-deoxyadenosine) is a naturally occurring adenosine analogue and one of the bioactive constituents isolated fromCordyceps militaris/Cordyceps sinensis, species of the fungal genusCordyceps. It has traditionally been a prized Chinese folk medicine for the human well-being. Because of similarity of chemical structure of adenosine, cordycepin has been focused on the diverse effects of the central nervous systems (CNSs), like sleep regulation. Therefore, this study was undertaken to know whether cordycepin increases the natural sleep in rats, and its effect is mediated by adenosine receptors (ARs). Sleep was recorded using electroencephalogram (EEG) for 4 hours after oral administration of cordycepin in rats. Sleep architecture and EEG power spectra were analyzed. Cordycepin reduced sleep-wake cycles and increased nonrapid eye movement (NREM) sleep. Interestingly, cordycepin increasedθ(theta) waves power density during NREM sleep. In addition, the protein levels of AR subtypes (A1,A2A,andA2B) were increased after the administration of cordycepin, especially in the rat hypothalamus which plays an important role in sleep regulation. Therefore, we suggest that cordycepin increases theta waves power density during NREM sleep via nonspecific AR in rats. In addition, this experiment can provide basic evidence that cordycepin may be helpful for sleep-disturbed subjects.

Objective To evaluate whether saw palmetto extract (SPE) relaxes corpus cavernosum and explore the underlying mechanisms. Methods Forty Sprague-Dawley rats and 30 New Zealand rabbits were randomly allocated into 3 SPE-treated groups (low-, middle-, and high-dose) and 1 saline-treated control group. SPE was administered intragastrically for 7 consecutive days. Another 23 rats treated with sildenafil were used to appraise the erectile response to electrical stimulation of nerves in the corpus cavernosum. The erectile functions of rats and rabbits were evaluated 24 hours after the last SPE administration or 15 minutes after intragastric sildenafil. Outcome measures included corpus cavernosum electrical activity recording, phosphodiesterase 5 (PDE5) activity detected by the colorimetric quantitative method, and messenger ribonucleic acid (mRNA) expression level for PDE5 and inducible nitric oxide synthase (iNOS) determined using real-time polymerase chain reaction. Results In the SPE-treated animals, the relaxant response to electrical stimulation of nerves in the corpus cavernosum, reflected by the amplitude of the electrical activity within the cavernosum, was significantly and dose-dependently augmented. Similar effects were observed in the sildenafil-treated rats. PDE5 activity in rat and rabbit corpus cavernosum tissues was significantly and dose-dependently inhibited in SPE-treated animals, whereas the iNOS mRNA level increased compared with the saline group. PDE5 mRNA, however, was only significantly enhanced in the rats treated with the middle dose of SPE. Conclusion The results suggest that SPE may have potential application value for the prevention or treatment of erectile dysfunction through an increase in iNOS mRNA expression and inhibition of PDE5 activity in corpus cavernosum smooth muscles.