Underwater image super-resolution using multi-stage information distillation networks

Recently, single image super-resolution (SISR) has been widely applied in the fields of underwater robot vision and obtained remarkable performance. However, most current methods generally suffered from the problem of a heavy burden on computational resources with large model sizes, which limited their real-world underwater robotic applications. In this paper, we introduce and tackle the super resolution (SR) problem for underwater robot vision and provide an efficient solution for near real-time applications. We present a novel lightweight multi-stage information distillation network, named MSIDN, for better balancing performance against applicability, which aggregates the local distilled features from different stages for more powerful feature representation. Moreover, a novel recursive residual feature distillation (RRFD) module is constructed to progressively extract useful features with a modest number of parameters in each stage. We also propose a channel interaction & distillation (CI&D) module that employs channel split operation on the preceding features to produce two-part features and utilizes the inter channel-wise interaction information between them to generate the distilled features, which can effectively extract the useful information of current stage without extra parameters. Besides, we present USR-2K dataset, a collection of over 1.6K samples for large-scale underwater image SR training, and a testset with an additional 400 samples for benchmark evaluation. Extensive experiments on several standard benchmark datasets show that the proposed MSIDN can provide state-of-the-art or even better performance in both quantitative and qualitative measurements.     

WavNet — Visual saliency detection using Discrete Wavelet Convolutional Neural Network

In the recent advancements in image and video analysis, the detection of salient regions in the image becomes the initial step. This plays a crucial role in deciding the performance of such algorithms. In this work, a Multi-Resolution Feature Extraction (MRFE) technique that makes use of Discrete Wavelet Convolutional Neural Network (DWCNN) for generating features is employed. An Enhanced Feature Extraction (EFE) module extracts additional features from the high level features of the DWCNN, which are used to frame both channel as well as spatial attention models for yielding contextual attention maps. A new hybrid loss function is also proposed, which is a combination of Balanced Cross Entropy (BCE) loss and Edge based Structural Similarity (ESSIM) loss that effectively identifies and segments the salient regions with clear boundaries. The method is tested exhaustively with five different benchmark datasets and is proved superior to the existing state-of-the-art methods with a minimum Mean Absolute error (MAE) of 0.03 and F-measure of 0.956.     

Modeling Sialidosis with Neural Precursor Cells Derived from Patient-Derived Induced Pluripotent Stem Cells

Sialidosis, caused by a genetic deficiency of the lysosomal sialidase gene (NEU1), is a systemic disease involving various tissues and organs, including the nervous system. Understanding the neurological dysfunction and pathology associated with sialidosis remains a challenge, partially due to the lack of a human model system. In this study, we have generated two types of induced pluripotent stem cells (iPSCs) with sialidosis-specific NEU1^G227R and NEU1^V275A/R347Q mutations (sialidosis-iPSCs), and further differentiated them into neural precursor cells (iNPCs). Characterization of NEU1^G227R- and NEU1^V275A/R347Q- mutated iNPCs derived from sialidosis-iPSCs (sialidosis-iNPCs) validated that sialidosis-iNPCs faithfully recapitulate key disease-specific phenotypes, including reduced NEU1 activity and impaired lysosomal and autophagic function. In particular, these cells showed defective differentiation into oligodendrocytes and astrocytes, while their neuronal differentiation was not notably affected. Importantly, we found that the phenotypic defects of sialidosis-iNPCs, such as impaired differentiation capacity, could be effectively rescued by the induction of autophagy with rapamycin. Our results demonstrate the first use of a sialidosis-iNPC model with NEU1^G227R- and NEU1^V275A/R347Q- mutation(s) to study the neurological defects of sialidosis, particularly those related to a defective autophagy–lysosome pathway, and may help accelerate the development of new drugs and therapeutics to combat sialidosis and other LSDs.     

Targeted editing of intronic-splicing silencer enhancement of SMN2 Exon 7 inclusion by CRISPR/Case 9

Spinal muscular atrophy (SMA) is an autosomal recessive hereditary neuromuscular disease. Exon 7 and 8 of survival of motor neuron 1 (SMN1) gene or only exon 7 homology deletion leads to the failure to produce a full-length SMN gene. The copy number of SMN2 gene with high homology of SMN1 affects the degree of disease and was the target gene for targeting therapy, in which splicing silencer in intron 7 was the key to suppress the inclusion of exon 7. In this study, we projected to use CRISPR/Case 9 for the targeted editing of intronic-splicing silencer (ISS) sequence to promote the inclusion of SMN2 exon 7 and increase the production of SMN2 full-length (FL) gene expression. It happens that there was a protospacer adjacent motif (PAM) at one end of the ISS sequence according to the design of sgRNA. The recombinant vector of sgRNA HSMN2 CRISPR/Case 9 was constructed and transfected into HEK293 cells. Sequencing results showed that the ISS sequence could be edited accurately and targeting in the predicted direction, in which deleting small fragments, inserting small amounts and mutation. Quantitative analysis of RT-PCR products by restriction enzyme of DdeI digestion showed that the FL of SMN2 increased by 8% (P < 0.05). In the primary cultured chondrocytes of SMA mice, in which sgRNA HSMN2 CRISPR/Case9 recombinant vector transfection could increase the SMN2 FL gene by 23% (P < 0.05) and significantly improve SMN protein levels (P < 0.05). CRISPR/Case 9 is an effective tool for gene editing and therapy of hereditary diseases, but it is rarely reported in the treatment of SMA diseases. This study shows that CRISPR/Case 9 was first used for the precision target of ISS sequence editing, which can effectively promote the production of SMN2 FL gene expressions, in which there was an important clinical reference value.     

Portrait of Cancer Stem Cells on Colorectal Cancer: Molecular Biomarkers,Signaling Pathways and miRNAome

Colorectal cancer (CRC) is a leading cause of cancer death worldwide, and about 20% is metastatic at diagnosis and untreatable. Increasing evidence suggests that the heterogeneous nature of CRC is related to colorectal cancer stem cells (CCSCs), a small cells population with stemness behaviors and responsible for tumor progression, recurrence, and therapy resistance. Growing knowledge of stem cells (SCs) biology has rapidly improved uncovering the molecular mechanisms and possible crosstalk/feedback loops between signaling pathways that directly influence intestinal homeostasis and tumorigenesis. The generation of CCSCs is probably connected to genetic changes in members of signaling pathways, which control self-renewal and pluripotency in SCs and then establish function and phenotype of CCSCs. Particularly, various deregulated CCSC-related miRNAs have been reported to modulate stemness features, controlling CCSCs functions such as regulation of cell cycle genes expression, epithelial-mesenchymal transition, metastasization, and drug-resistance mechanisms. Primarily, CCSC-related miRNAs work by regulating mainly signal pathways known to be involved in CCSCs biology. This review intends to summarize the epigenetic findings linked to miRNAome in the maintenance and regulation of CCSCs, including their relationships with different signaling pathways, which should help to identify specific diagnostic, prognostic, and predictive biomarkers for CRC, but also develop innovative CCSCs-targeted therapies.     

Transmission line detection in aerial images: An instance segmentation approach based on multitask neural networks

Camera-based transmission line detection (TLD) is a fundamental and crucial task for automatically patrolling powerlines by aircraft. Motivated by instance segmentation, a TLD algorithm is proposed in this paper with a novel deep neural network, i.e., CableNet. The network structure is designed based on fully convolutional networks (FCNs) with two major improvements, considering the specific appearance characteristics of transmission lines. First, overlaying dilated convolutional layers and spatial convolutional layers are configured to better represent continuous long and thin cable shapes. Second, two branches of outputs are arranged to generate multidimensional feature maps for instance segmentation. Thus, cable pixels can be detected and assigned cable IDs simultaneously. Multiple experiments are conducted on aerial images, and the results show that the proposed algorithm obtains reliable detection performance and is superior to traditional TLD methods. Meanwhile, segmented pixels can be accurately identified as cable instances, contributing to line fitting for further applications.     

Electrochemical performance of in situ LiFePO4 modified by N-doped graphene for Li-ion batteries

LiFePO₄ modified by N-doped graphene (NG) with a three-dimensional conductive network structure was synthesized via a one-step in situ hydrothermal method. The effects of N amount of NG on the phase structure, morphology, and electrochemical properties of LiFePO₄ are investigated in this study. X-ray diffraction (XRD) results show that doping suitable N amounts in NG do not alter the crystal structure of LiFePO_4, and scanning electron microscopy (SEM) images show that NG can slightly reduce the particle size of LiFePO₄. The high-resolution transmission electron microscopy (HRTEM) results show that the LiFePO₄ particles are well covered and connected by NG. The electrochemical performance confirms that LiFePO₄ modified by 20% N-doped graphene (named LFP/NG-4) displays a perfect specific capacity of 166.6 mAh·g^?1 at a rate of 0.2C and can reach 125 mAh·g^?1 at a rate of 5 C. Electrochemical impedance spectroscopy (EIS) results illustrate that the charge transfer resistance value of the LFP/NG-4 composite is only 58.6 Ω, which is very low compared with LiFePO₄. Cyclic voltammetry (CV) tests indicate that the addition of 20% N-doped graphene can effectively reduce electrode polarization and improve reversibility. The LFP/NG-4 composite with a three-dimensional conductive network structure can be regarded as a promising cathode material for Li-ion batteries.     

The P-graph approach for systematic synthesis of wastewater treatment networks

Wastewater treatment consists of three or four sequential stages: preliminary, primary, secondary, and tertiary. Each stage can comprise multiple alternative technologies that can perform the same tasks with different efficiencies, operating times, and costs. Thus, we propose a systematic approach for designing wastewater treatment networks by utilizing principles of mathematical modeling and generating an exhaustive enumeration of all the possible technologies and their connections during the early stages of designing a treatment facility. Some of these structures are nonintuitive and include recycling, reprocessing, bypasses, and multiple technologies in parallel or series to remove the same contaminant. The nonintuitive structures with multiple technologies may provide a measure of resilience compared to typical heuristic designs. Thus, the combination of P-graph methodology and the sequence of treatment technologies predicted via the optimization algorithm from the maximal structure is based on holistic considerations and does not lead to suboptimal solutions.     

5G技术在医院智能服务中的应用

纵观当今的发展格局,是科学技术高速发展的时代。随着5G时代的来临,网络技术在各个领域中扮演着越来越重要的角色。特别是在医疗方面,为医院的日常工作奠定了坚实的基础。今天就5G网络技术在医院只能服务中的应用展开谈论。     

基于GA-BP的汽车风振噪声声品质预测模型

目前对于汽车风振噪声的优化研究主要以声压级（Sound pressure level，SPL）作为单一评价指标，既不能全面反映噪声的物理属性，也无法考虑人耳对噪声的主观认知过程。为准确评价风振噪声，引入声品质，运用大涡模拟（Large eddy simulation，LES）对风振噪声进行数值仿真，根据实车道路试验判断仿真的准确性；对仿真结果进行声品质客观评价与主观评价，综合声品质客观评价参数与声品质主观评价试验结果建立BP神经网络预测模型；利用遗传算法（Genetic algorithm，GA），进一步对BP神经网络的结构参数进行优化，建立GA-BP声品质预测模型。研究结果表明，GA-BP声品质预测模型在训练速度和预测精度上都优于BP神经网络预测模型。预测模型基于声品质主客观评价结果，其预测值可以代替传统的声压级评价指标，为风振噪声提供更为准确合理的评价。