Search

Deep convolutional neural networks (DCNNs) have exhibited excellent feature extraction and detail reconstruction capabilities for single image super-resolution (SISR). Nevertheless, most previous DCNN-based methods do not fully utilize the complementary strengths between feature maps, channels, and pixels. Therefore, it hinders the ability of DCNNs to represent abundant features. To tackle the aforementioned issues, we present a Cascaded Visual Attention Network for SISR called CVANet, which simulates the visual attention mechanism of the human eyes to focus on the reconstruction process of details. Specifically, we first designed a trainable feature attention module (FAM) for feature-level attention learning. Afterward, we introduce a channel attention module (CAM) to reinforce feature maps under channel-level attention learning. Meanwhile, we propose a pixel attention module (PAM) that adaptively selects representative features from the previous layers, which are utilized to generate a high-resolution image. Satisfactory, our CVANet can effectively improve the resolution of images by exploring the feature representation capabilities of different modules and the visual perception properties of the human eyes. Extensive experiments with different methods on four benchmarks demonstrate that our CVANet outperforms the state-of-the-art (SOTA) methods in subjective visual perception, PSNR, and SSIM.The code will be made available https://github.com/WilyZhao8/CVANet. [ABSTRACT FROM AUTHOR]

Cancer originates from somatic cells that have accumulated mutations. These mutations alter the phenotype of the cells, allowing them to escape homeostatic regulation that maintains normal cell numbers. The emergence of malignancies is an evolutionary process in which the random accumulation of somatic mutations and sequential selection of dominant clones cause cancer cells to proliferate. The development of technologies such as high-throughput sequencing has provided a powerful means to measure subclonal evolutionary dynamics across space and time. Here, we review the patterns that may be observed in cancer evolution and the methods available for quantifying the evolutionary dynamics of cancer. An improved understanding of the evolutionary trajectories of cancer will enable us to explore the molecular mechanism of tumorigenesis and to design tailored treatment strategies. [ABSTRACT FROM AUTHOR]

A phenotype may be associated with multiple genes that interact with each other in the form of a gene module or network. How to identify these relationships is one important aspect of comparative transcriptomics. However, it is still a challenge to align gene modules associated with different phenotypes. Although several studies attempted to address this issue in different aspects, a general framework is still needed. In this study, we introduce Module Alignment of TranscripTomE (MATTE), a novel approach to analyze transcriptomics data and identify differences in a modular manner. MATTE assumes that gene interactions modulate a phenotype and models phenotype differences as gene location changes. Specifically, we first represented genes by a relative differential expression to reduce the influence of noise in omics data. Meanwhile, clustering and aligning are combined to depict gene differences in a modular way robustly. The results show that MATTE outperformed state-of-the-art methods in identifying differentially expressed genes under noise in gene expression. In particular, MATTE could also deal with single-cell ribonucleic acid-seq data to extract the best cell-type marker genes compared to other methods. Additionally, we demonstrate how MATTE supports the discovery of biologically significant genes and modules, and facilitates downstream analyses to gain insight into breast cancer. The source code of MATTE and case analysis are available at https://github.com/zjupgx/MATTE. [ABSTRACT FROM AUTHOR]

Herein, we report the function of copper sites in Cu-SSZ-13, Cu-ZSM-5 and Cu-Beta catalysts with the same Si/Al ratio (14) and Cu/Al ratio (0.4) on selective catalytic reduction of NO with NH3 (NH3-SCR) and reveal the relationship between active sites (Cu sites, acid sites) and catalytic activity. The results show that the amount of isolated Cu2+ ions in the catalysts directly determines the formation of strong Lewis acid sites and reaction intermediate NO3− ions, thus affecting the low-temperature SCR performance, while the amount of highly stable Cu+ ions and Brønsted acid sites is related to the high-temperature SCR performance of the catalysts. Consequently, it contains enough isolated Cu2+ ions, highly stable Cu+ ions and Brønsted acid sites, which endows Cu-SSZ-13 with excellent NH3-SCR activity. [ABSTRACT FROM AUTHOR]

Motivation Identifying genes that play a causal role in cancer evolution remains one of the biggest challenges in cancer biology. With the accumulation of high-throughput multi-omics data over decades, it becomes a great challenge to effectively integrate these data into the identification of cancer driver genes. Results Here, we propose MODIG, a graph attention network (GAT)-based framework to identify cancer driver genes by combining multi-omics pan-cancer data (mutations, copy number variants, gene expression and methylation levels) with multi-dimensional gene networks. First, we established diverse types of gene relationship maps based on protein–protein interactions, gene sequence similarity, KEGG pathway co-occurrence, gene co-expression patterns and gene ontology. Then, we constructed a multi-dimensional gene network consisting of approximately 20 000 genes as nodes and five types of gene associations as multiplex edges. We applied a GAT to model within-dimension interactions to generate a gene representation for each dimension based on this graph. Moreover, we introduced a joint learning module to fuse multiple dimension-specific representations to generate general gene representations. Finally, we used the obtained gene representation to perform a semi-supervised driver gene identification task. The experiment results show that MODIG outperforms the baseline models in terms of area under precision-recall curves and area under the receiver operating characteristic curves. Availability and implementation The MODIG program is available at https://github.com/zjupgx/modig. The code and data underlying this article are also available on Zenodo, at https://doi.org/10.5281/zenodo.7057241. Supplementary information Supplementary data are available at Bioinformatics online. [ABSTRACT FROM AUTHOR]

Automation is increasingly at the forefront of transportation research, with the potential to bring fully autonomous vehicles to our roads in the coming years. This comprehensive survey provides a holistic look at the essential components and cutting-edge technologies that are driving the development and implementation of autonomous driving. It starts by evaluating two critical system architectures that are fundamental to the operation of autonomous vehicles: the layered and end-to-end structures. It then examines the critical areas of scene perception and localization, emphasizing the importance of sensor technologies. These technologies are vital for tasks such as object detection and semantic segmentation, which allow vehicles to understand and navigate their environment. A special focus is given to the complex topic of object detection, along with suggestions for how it can be enhanced. The survey then proceeds to provide detailed discussions on path planning, trajectory prediction, and decision-making processes. These elements are crucial for the smooth navigation of autonomous vehicles, and the survey highlights the role of artificial intelligence (AI) and machine learning in these processes. Overall, the survey presents the rapid progress in the field of autonomous driving, offering a comprehensive assessment of the technologies and innovations that are essential for moving toward a safe and efficient autonomous future. [ABSTRACT FROM AUTHOR]

Siamese-architecture-based self-supervised learning usually suffers from relatively high computational consumption and unsatisfactory performance because of its slow convergence and poor feature extraction capability. To alleviate these issues, we propose a self-supervised method, called SSL 2 , that is both efficient and effective. Specifically, a global and local sampling method is proposed to increase the number of samples while maintaining connections between semantic features. More significantly, SSL 2 maintains low computational complexity and facilitates the establishment of mapping relationships between global comprehensive and local detailed features compared with other methods. In addition, an i nformation r etainer p rojection h ead (IRPH) is proposed to further balance the information between detailed inconsistency and semantic consistency. Finally, hybrid tasks are embedded into SSL 2 to optimize the model so that it can effectively leverage the data provided by global and local sampling. Extensive qualitative and quantitative evaluations of various types of benchmarks illustrate that SSL 2 outperforms existing self-supervised frameworks in commonly used computer vision tasks. Specifically, SSL 2 achieved satisfactory performance with linear classification on ImageNet, outperforming MoCo-v2 by 2.2% with fewer calculations, and it also achieved competitive results compared with other state-of-the-art methods. • A new single-branch self-supervised learning framework is designed. • A novel sampling strategy is designed to make full use of the information. • A head is designed to balance detailed inconsistency and semantic consistency. • Well-balanced hybrid loss functions are introduced. [ABSTRACT FROM AUTHOR]

Objectives: Extracorporeal membrane oxygenation (ECMO) plays an important role in providing temporary life support for patients with severe cardiac or pulmonary failure, but requires strict anticoagulation and monitoring. This network meta‐analysis systematically explored the most effective anticoagulation and monitoring strategies for patients receiving ECMO. Methods: MEDLINE, Embase, Web of Science, and the Cochrane Central Register of Controlled Trials were searched up to January 31, 2023, for studies comparing unfractionated heparin (UFH), argatroban (Arg), bivalirudin (Biv), and/or nafamostat mesylate (NM) in patients receiving ECMO. The primary outcomes included device‐related thrombosis, patient‐related thrombosis, and major bleeding events. The secondary outcomes included ECMO survival, ECMO duration, and in‐hospital mortality. Results: A total of 2522 patients from 23 trials were included in the study. Biv was associated with a decreased risk of device‐related thrombosis (odd ratio [OR] 0.51, 95% confidence interval [CI]: 0.33–0.84) compared with UFH, whereas NM (OR 2.2, 95% CI: 0.24–65.0) and Arg (OR 0.92, 95% CI: 0.43–2.0) did not reduce the risk of device‐related thrombosis compared with UFH. Biv was superior to Arg in decreasing the risk of device‐related thrombosis (OR 0.14, 95% CI: 0.03–0.51). Biv reduced the risk of patient‐related thrombosis compared with UFH (OR 0.44, 95% CI: 0.18–0.85); NM (OR 0.65, 95% CI: 0.14–3.3) and Arg (OR 3.1, 95% CI: 0.94–12.0) did not decrease risk of patient‐related thrombosis compared with UFH. No significant difference was observed in the risk of major bleeding between three alternatives and UFH: Biv (OR 0.54, 95% CI: 0.23–1.3), Arg (OR 1.3, 95% CI: 0.34–5.8), and NM (OR 0.60, 95% CI: 0.13–2.6). NM showed a reduced risk of in‐hospital mortality compared with UFH (OR 0.27, 95% CI: 0.091–0.77), whereas Arg (OR 0.43, 95% CI: 0.15–1.2) and Biv (OR 0.75, 95% CI: 0.52–1.1) did not decrease risk of in‐hospital mortality. Conclusions: Compared with UFH and Arg, Biv reduces the risk of thrombosis and appears to be a better choice for patients requiring ECMO. NM was associated with a reduced risk of in‐hospital mortality. [ABSTRACT FROM AUTHOR]

• Use the new model (the 3D-3T model) to estimate plant transpiration. • The Normalized Difference Ratio Index (NDRI) is proposed for the first time. • The 2D-3T model was converted to the 5P-3T model. • The temperature of the canopy was accurately obtained by a portable thermal imager. Crop transpiration is the main dissipation way of agricultural water. Accurate and effective estimation of crop transpiration is of great significance for the study of the carbon-water cycle relationship and efficient water-saving irrigation. This study uses infrared thermal imaging technology to accurately extract the average temperature of the plant canopy as a key input for the new model (three-dimensional temperature (3D-3T) model), which is an extension of the three-temperature (3T) model and applied to estimate the canopy transpiration rate of typical citrus trees (variety named Guoqing No. 1) from 8:00 to 18:00 on May 7 to 11, 2021. And a processing method for extracting green plants in visible-light images based on the Normalized Difference Ratio Index (NDRI) is proposed for the first time, which can distinguish green plants from other ground objects, backgrounds, and shadows in the visible-light image, and then help to accurately extract the average canopy temperature from infrared thermal images. For this experiment, the canopy structure characteristics of the selected experimental subjects and the measurement methods used were considered, so the 3D-3T model was converted to a 5-point three-temperature model (5P-3T model) to estimate the canopy transpiration rate of citrus trees and evaluate the plant water status, and then the sap flow measurement was used for validation. The results showed that: (1) The 3T model overestimated the transpiration rate of the citrus tree, while the 5P-3T model underestimated it. The estimated values of the 5P-3T model have a good correlation with the measured values of the packaged stem sap flow gauge method, with a determination coefficient (R2) of 0.72 and root mean square error (RMSE) of 0.35 mm/d. (2) In this model, the plant transpiration transfer coefficient (h at ) can evaluate the water shortage of plants, where the value of h at increased with the aggravation of plant water stress. (3) The citrus trees have lower transpiration rates under water stress, ranging from 0.06 to 0.12 mm/h in the daytime (from 8-10 May 2021). The model used in this study only needs the input of temperature information and net radiation, and would not depend on aerodynamic resistance. The method in this paper can realize the non-contact and non-invasive measurement of crop transpiration, and it provides a feasible way to use a portable thermal imager to accurately obtain the temperature for estimating plant canopy transpiration rate. [ABSTRACT FROM AUTHOR]

Ideal wheat grain quality detection systems aim for rapid and precise recognition of flaws. Computer vision and machine learning have been widely studied as potential alternatives to human inspection. Numerous deep learning-based detection and classification methods have been proposed in the last few years. However, these methods suffer from poor performance and extremely long time consumption due to the weak feature extraction ability and high computational overhead. To solve these challenges, we present the first comprehensive study and analysis of wheat grain detection using improved YOLOv5. Specifically, we design a machine vision system and construct a Wheat Grain Detection Benchmark (WGDB) including 1746 images with 7844 bounding boxes, all of which have been independently classified. Utilizing this dataset, we conduct a comprehensive study of the most advanced objection detection methods. In addition, we propose a Wheat Grain Detection Network (called WGNet) trained on this benchmark as a baseline, aiming to solve the degradation issues by employing sparse network pruning and a hybrid attention module. Extensive experiments demonstrate the limitations of existing methods and the improved performance of our method, which achieves state-of-the-art precision with the fastest inference speed. The constructed benchmark and the improved experiments shed light on future research in wheat grain detection. The dataset and code will be available at WGNet. • We construct a large-scale wheat grain detection benchmark. • We conduct a comprehensive study with the constructed dataset. • We propose a light-weight Wheat Grain Detection network by employing sparse network pruning and a hybrid attention module. [ABSTRACT FROM AUTHOR]

Convolutional neural network (CNN)-based self-supervised visual representation learning (SSL) is a long-standing problem achieving significant successes with traditional handcrafted pretext tasks and contrastive learning. However, existing SSL methods typically suffer from high computational overhead and poor performance due to sluggish convergence speeds and poor detail extraction capabilities. In this work, to address these issues and improve the robustness, we provide a new self-supervised architecture for incorporating a single-branch backbone with hybrid tasks into the representation learning process. Specifically, our method takes advantage of features from both intra- and inter-images by using discrete montage images. Then a single backbone with a novel A daptive D ecouple C onfusion (ADC) module is proposed to improve the feature extraction capabilities and alleviate the confusion regions in montage images. Besides, both concatenated discrete vectors and patch-based global average pooled vectors in latent space are utilized to learn local detailed features and maintain semantic consistency simultaneously. Moreover, our method is optimized by hybrid tasks and enjoys faster convergence speed due to these improvements. Extensive experiments on several datasets demonstrate the effectiveness and robustness of our method. The proposed method has 2.0% improved in linear classification to the conventional single-branch methods. [Display omitted] • A new framework SSH that leverages concatenated discrete and PGAP vectors is proposed. • An A daptive D ecouple C onfusion (ADC) module is proposed to avoid degenerate performance in existing jigsaw-based methods. • We demonstrate the advantages of the above contributions through extensive experiments. [ABSTRACT FROM AUTHOR]

• We developed a cost-effective method for precision 2D stage error calibration. • Images of an arbitrarily textured plane are captured and analysed in the method. • A modified bundle adjustment algorithm was developed for global optimisation. • 3 μm accuracy within a 100 mm range was demonstrated using experiments. Positioning errors are a primary performance indicator of precision stages. Calibration of stage positioning errors usually requires expensive, specialised instruments or artefacts. In this paper, we introduce a simple and cost-effective, machine vision-based method for stage error calibration. The novelty is that we used an arbitrarily textured plane, for example an array of coins, as the imaging target, as opposed to costly precision machined artefacts. Positioning errors were extracted from overlapped texture images using a fast and globally optimised feature-matching algorithm. A computer simulation demonstrates that the method is accurate and numerically stable in terms of sensing noise. Trials with a coordinate measuring machine reveal that the method can achieve sub-pixel accuracy, with a maximum measurement error of approximately 3 μm within a 100 mm range. [ABSTRACT FROM AUTHOR]

Capturing the comprehensive information of various sizes and shapes of images in the same convolution layer is typically a challenging task in computer vision. There are two main kinds of methods for capturing those features. The first uses the inception structure and its variants. The second utilizes larger convolution kernels on specific layers or stacks with more convolution blocks. However, these methods can result in computationally intensive or vanishing gradients. In this paper, to accommodate feature distributions with different sizes, shapes and reduce computational cost, we propose a width- and depth-aware module named the WD-module to match feature distributions. Moreover, the proposed WD-module consumes less computational cost and parameters compared with traditional residual convolution layers. To verify the effectiveness of our proposed method, a size- and shape-aware backbone network named S2A-Net was built, which was obtained by stacking the WD-modules. By visualizing heat maps and features, the proposed S2A-Net can adapt to objects with different sizes and shapes in visual recognition tasks and learn more comprehensive characteristics. Experimental results show that the proposed method has higher accuracy in image recognition and outperforms other state-of-the-art networks with the same numbers of layers. • A width-aware and deep-aware module is build. • A size-aware and shape-aware network for visual recognition is build. • Compared with other state-of-the-art methods, the proposed method can achieve better results while consume less parameters and FLOPs. [ABSTRACT FROM AUTHOR]

Objectives: Regulatory T (Treg) cells regulate immunity in autoimmune diseases and cancers. However, immunotherapies that target tumor‐infiltrating Treg cells often induce unwanted immune responses and tissue inflammation. Our research focussed on exploring the expression pattern of CD177 in tumor‐infiltrating Treg cells with the aim of identifying a potential target that can enhance immunotherapy effectiveness. Methods: Single‐cell RNA sequencing (scRNA‐seq) data and survival data were obtained from public databases. Twenty‐one colorectal cancer patient samples, including fresh tumor tissues, peritumoral tissues and peripheral blood mononuclear cells (PBMCs), were analysed using flow cytometry. The transendothelial activity of CD177+ Treg cells was substantiated using in vitro experiments. Results: ScRNA‐seq and flow cytometry results indicated that CD177 was exclusively expressed in intratumoral Treg cells. CD177+ Treg cells exhibited greater activation status and expressed elevated Treg cell canonical markers and immune checkpoint molecules than CD177− Treg cells. We further discovered that both intratumoral CD177+ Treg cells and CD177‐overexpressing induced Treg (iTreg) cells had lower levels of PD‐1 than their CD177− counterparts. Moreover, CD177 overexpression significantly enhanced the transendothelial migration of Treg cells in vitro. Conclusions: These results demonstrated that Treg cells with higher CD177 levels exhibited an enhanced activation status and transendothelial migration capacity. Our findings suggest that CD177 may serve as an immunotherapeutic target and that overexpression of CD177 may improve the efficacy of chimeric antigen receptor T (CAR‐T) cell therapy. [ABSTRACT FROM AUTHOR]

Existing immune checkpoint inhibitors focus on activating T cells and show limited effectiveness in gastric cancer (GC). SIGLEC10 is identified as a novel tumor-associated macrophage-related immune checkpoint in other cancer types. However, its immunosuppressive role and clinical significance in GC remain unclear. In this study, we find a dominant expression of SIGLEC10 on CD68+ macrophages in GC. SIGLEC10 can suppress the proliferation and function of tumor-infiltrating CD8+ T cells in vitro via the Akt/P38/Erk signaling pathway. Furthermore, in ex vivo and in vivo models, SIGLEC10 blockade promotes CD8+ T cell effector function. Finally, SIGLEC10+ macrophages are positively correlated with the adverse prognosis of GC. Our study highlights that SIGLEC10 directly suppresses T cell function and serves as a promising target for immunotherapy and suggests SIGLEC10+ macrophages as a novel potential predictor of the clinical prognosis of GC. [ABSTRACT FROM AUTHOR]

Background: Peritoneal metastasis (PM) frequently occurs in patients with gastric cancer (GC) and is a major cause of mortality. Risk stratification for PM can optimize decision making in GC treatment. Methods: A total of 25 GC patients (13 with synchronous, 6 with metachronous PM and 6 PM-free) were included in this study. Quantitative proteomics by high-depth tandem mass tags labeling and whole-exome sequencing were conducted in primary GC and PM samples. Proteomic signature and prognostic model were established by machine learning algorithms in PM and PM-free GC, then validated in two external cohorts. Tumor-infiltrating immune cells in GC were analyzed by CIBERSORT. Results: Heterogeneity between paired primary and PM samples was observed at both genomic and proteomic levels. Compared to primary GC, proteome of PM samples was enriched in RNA binding and extracellular exosomes. 641 differently expressed proteins (DEPs) between primary GC of PM group and PM-free group were screened, which were enriched in extracellular exosome and cell adhesion pathways. Subsequently, a ten-protein signature was derived based on DEPs by machine learning. This signature was significantly associated with patient prognosis in internal cohort and two external proteomic datasets of diffuse and mixed type GC. Tumor-infiltrating immune cell analysis showed that the signature was associated with immune microenvironment of GC. Conclusions: We characterized proteomic features that were informative for PM progression of GC. A protein signature associated with immune microenvironment and patient outcome was derived, and it could guide risk stratification and individualized treatment. [ABSTRACT FROM AUTHOR]

With rapid economic development, extensive human activity has changed landscape patterns (LPs) dramatically, which has further influenced hydrological processes. However, the effects of LPs changes on hydrological processes, especially for the streamflow–sediment relationship in the subtropical monsoon climate zone, have not been reported. In our study, 10 watersheds with different sizes in the subtropical monsoon climate zone of southeastern China were chosen as the study area, and the effect of the 14 most commonly used landscape metrics (LMs) on 4 typical hydrological indices (water yields (WY), the runoff coefficient (RC), the soil erosion modulus (SEM), and the suspended sediment concentration (SSC)) were analyzed based on land use maps and hydrological data from 1990 to 2019. The results reveal that the LP characteristics within the study area have changed significantly. The number of patches and landscape shape indices were significantly positively correlated with watershed size (p < 0.01). For most watersheds, the largest patch index was negatively correlated with WY, RC, and SEM, and the perimeter area fractal dimension was positively correlated with WY, RC, SEM, and SSC. The effects of several LMs on the hydrological indices had scale effects. WY/RC and the interspersion and juxtaposition index were negatively correlated in most larger watersheds but were positively correlated in most smaller watersheds. Similar results were found for Shannon's diversity/evenness index and SEM. In general, an increase in a small patch of landscape and in landscape diversity would increase WY, the fragmentation of LPs would result in more soil erosion, and LPs would affect the relationship between streamflow and sediment yield. As a result, a proper decrease in landscape fragmentation and physical connectivity in the subtropical monsoon climate zone of southeastern China would benefit soil erosion prevention. These results enhance the knowledge about the relationship between LPs and hydrological processes in the subtropical monsoon climate zone of southeastern China and benefit local water and soil conservation efforts. [ABSTRACT FROM AUTHOR]

Background: Gastric cancer (GC) is one of the leading causes of cancer deaths with high heterogeneity. There is currently a paucity of clinically applicable molecular classification system to guide precise medicine. Methods: A total of 70 Chinese patients with GC were included in this study and whole-exome sequencing was performed. Unsupervised clustering was undertaken to identify genomic subgroups, based on mutational signature, copy number variation, neoantigen, clonality, and essential genomic alterations. Subgroups were characterized by clinicopathological factors, molecular features, and prognosis. Results: We identified 32 significantly mutated genes (SMGs), including TP53, ARID1A, PIK3CA, CDH1, and RHOA. Of these, PREX2, PIEZO1, and FSIP2 have not been previously reported in GC. Using a novel genome-based classification method that integrated multidimensional genomic features, we categorized GC into four subtypes with distinct clinical phenotypes and prognosis. Subtype 1, which was predominantly Lauren intestinal type, harbored recurrent TP53 mutation and ERBB2 amplification, high tumor mutation burden (TMB)/tumor neoantigen burden (TNB), and intratumoral heterogeneity, with a liver metastasis tendency. Subtype 2 tended to occur at an elder age, accompanying with frequent TP53 and SYNE1 mutations, high TMB/TNB, and was associated with poor prognosis. Subtype 3 and subtype 4 included patients with mainly diffuse/mixed type tumors, high frequency of peritoneal metastasis, and genomical stability, whereas subtype 4 was associated with a favorable prognosis. Conclusions: By integrating multidimensional genomic characteristics, we proposed a novel genomic classification system of GC associated with clinical phenotypes and provided a new insight to facilitate genome-guided risk stratification and disease management. [ABSTRACT FROM AUTHOR]

Hyperspectral imaging techniques are widely used for rapid, efficient, and non-destructive identification of wheat varieties. However, the interference of noise in hyperspectral images and the underutilization of spatial information by most methods are two challenging issues in identifying wheat varieties. In this paper, we present a new approach called Cost-sensitive K-Nearest Neighbor using Hyperspectral imaging (CSKNN) to address these issues. First, we fuse 128 bands acquired by hyperspectral imaging equipment to obtain hyperspectral images of wheat grains, and employ a central regionalization strategy to extract the region of interest. We then use a smoothing denoising strategy to remove noise from the hyperspectral images and improve the saliency of the object grains. Furthermore, we consider the characteristics of different bands and use linear discriminant analysis to compress features, reducing intra-class differences and increasing inter-class differences. Finally, we propose a Cost-sensitive KNN for training and testing of wheat varieties. Our experiments on different strains and varieties of wheat datasets in the same region show that our CSKNN achieves high classification accuracies of 98.09% and 97.45%, outperforming state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Hyperspectral image (HSI) based methods are widely used in identifying seeds varieties with high accuracy. However, the excessive spectral bands in HSI may contain redundant information and degrade the model performance. To address this challenge, we propose a novel feature extraction method called low rank tensor approximation (LRTA). Unlike traditional methods, LRTA extracts joint discriminative information of all wheat seeds from hyperspectral images in a 3-order tensor form, preserving intrinsic information. Our model has three key steps: extracting the region of interest from hyperspectral images and presenting average spectral information in tensor form, using LRTA to extract jointly discriminative information, and feeding this information into a classifier to identify seeds varieties. Experiments on our proposed dataset show that the proposed method has 4% improved to the conventional methods on average. Our method shows promise for improving the accuracy of seeds identification while reducing the dimensionality of HSI. [ABSTRACT FROM AUTHOR]

• A error analyzing toolbox for VIS is proposed, bining false prediction into 7 types. • TIVE weights error type's damage to mAP and evaluates over different temporal length. • We investigate the spatial and temporal localization abilities of popular VIS models. In this paper, we introduce TIVE, a T oolbox for I dentifying V ideo instance segmentation E rrors. By directly operating output prediction files, TIVE defines isolated error types and weights each type's damage to mAP, for the purpose of distinguishing model characters. By decomposing localization quality in spatial–temporal dimensions, model's potential drawbacks on spatial segmentation and temporal association can be revealed. TIVE can also report mAP over instance temporal length for real applications. We conduct extensive experiments by the toolbox to further illustrate how spatial segmentation and temporal association affect each other. We expect the analysis of TIVE can give the researchers more insights, guiding the community to promote more meaningful explorations for video instance segmentation. The proposed toolbox is available at https://github.com/wenhe-jia/TIVE. [ABSTRACT FROM AUTHOR]

Self-supervised visual representation learning focuses on capturing comprehensive features via exploiting the unlabeled datasets. However, existing contrastive learning based SSL frameworks are subjected to higher computational consumption and unsatisfactory performance. To handle these issues, we present a novel single-branch SSL method that incorporates an adaptive feature selection and activation module and a progressive cascade dimensionality reduction module, called APNet. Specifically, our method first fully exploits the unlabeled datasets and extracts intra- and inter-image information via introducing montage image. In addition, a novel adaptive feature selection and activation module is designed to generate the most comprehensive features. Besides, a progressive cascade dimensionality reduction module is proposed to capture the most representative features from latent vectors through cascade dimensionality increasing–decreasing operations. Extensive experiments have demonstrated the robustness and effectiveness of APNet. Specifically, APNet exceeds MoCo-v3 by 3.1% on the ImageNet-100 dataset, and consumes only half of the calculation. Code is available at https://github.com/AI-TYQ/APNet. • We embed feature selection and dimensionality reduction modules into SSL framework. • We design an attention module that selects and activates representative features. • We introduce a dimensionality reduction module to retain discriminative features. • We demonstrate the advantages of the above contributions through experiments. [Display omitted] [ABSTRACT FROM AUTHOR]

• The characteristics of rainfall erosivity in the Huaihe River Basin were analyzed. • 87% rainfall erosivity concentrated during May – September in the Huaihe River Basin. • Rainfall erosivity in the Huaihe River Basin was highest in the upper basin. • Climatic indexes were correlated with rainfall erosivity with different lag times. • Some advice was provide for local human activities based on soil erosion prevention. Rainfall erosivity (RE) is an important factor in the soil erosion process, which cannot be altered by human intervention alone. The Huaihe River Basin (HRB) is a large agricultural watershed, suffering severe water erosion, investigating the spatio-temporal variation of RE is essential for local soil erosion prevention. The linear regression, Yue-Pilon method, and the Hurst exponent were used in analyzing the spatio-temporal variation within the HRB during 1960–2018. The slip correlation analysis and F-test were applied to obtain the correlation between climatic indices [the Arctic Oscillation (AO), the Southern Oscillation (SOI), the North Pacific Index (NPI), and the Niño 4SST index (SST)] and RE in the HRB. Results revealed that the annual average RE were 4280, 5061, 4068, 4886, and 4089 MJ mm hm−2 h−1 within the HRB, upper-HRB, middle-HRB, lower-HRB, and the Yishusi River Watershed, respectively. The annual average RE will increase within the upper-HRB and lower-HRB and decrease within the middle-HRB and the Yishusi River Watershed in the future. The seasonal average RE was ranked as summer > autumn = spring > winter in the HRB. The spatiotemporal difference was significant in the HRB, and different sub-regions exhibited a different trend in seasonal RE, except for the winter RE that increased significantly in all sub-regions (p < 0.05). The highest monthly RE occurred in July, with RE during May–September accounting for approximately 87% of the annual RE in the HRB. The AO and NPI had significant correlations with RE (p < 0.05) both on the annual and monthly scales with different lag times. The monthly AO, SOI, SST, and NPI were significantly correlated with RE in the long term with different lag times in different months (p < 0.05). These findings could provide potential predictive factors for RE prediction and help prevent soil erosion within the HRB. [ABSTRACT FROM AUTHOR]

Complete surgical resection, accessible therapeutic targets and effective tyrosine kinase inhibitors (TKIs) have not completely cured gastrointestinal stromal tumours (GISTs), with most patients suffering from residual tumours and recurrence. The existence of nerve infiltration in GIST provides a way for tumour cells to escape local resection and systemic targeted therapy, which may challenge the previous understanding of its behaviour patterns and inspire the development of more radical excision and more precise targeted therapy. Moreover, tumour dormancy has emerged as a major cause of drug resistance and tumour relapse. Among these pathways, the nerve-tumour regulatory axis GDNF-GFRA1 is activated in GISTs, assists tumour cells in achieving dormancy and protects them from apoptosis under environmental stress by enhancing autophagic flux. The concrete mechanism is that the GDNF-regulating interaction between GFRA1 and the lysosomal calcium channel MCOLN1 activates Ca2+-dependent TFEB signalling. Activated TFEB transcriptionally regulates intracellular lysosome levels, which could achieve feedback upregulation of cellular autophagy flux during TKI treatment. This dormancy-transition axis fills parts of the mechanistic vacancy before the onset of secondary mutations, and strategies for TKIs combined with targeting GFRA1-dependent autophagy have distinct promise as prospective clinical therapies. [ABSTRACT FROM AUTHOR]