Search

In contrast to fully connected networks, Convolutional Neural Networks (CNNs) achieve efficiency by learning weights associated with local filters with a finite spatial extent. Theoretically, an implication of this fact is that a filter may know what it is looking at, but not where it is positioned in the image. In this paper, we first test this hypothesis and reveal that a surprising degree of absolute position information is encoded in commonly used CNNs. We show that zero padding drives CNNs to encode position information in their internal representations, while a lack of padding precludes position encoding. This observation gives rise to deeper questions about the role of position information in CNNs: (i) What boundary heuristics enable optimal position encoding for downstream tasks? (ii) Does position encoding affect the learning of semantic representations? (iii) Does position encoding always improve performance? To provide answers, we perform the largest case study to date on the role that padding and border heuristics play in CNNs. We design novel tasks that allow us to quantify boundary effects as a function of the distance to the border. Numerous semantic objectives reveal the effect of the border on semantic representations. Finally, we demonstrate the implications of these findings on multiple real-world tasks to show that position information can both help or hurt performance. [ABSTRACT FROM AUTHOR]

Deep learning (DL) has emerged as a leading approach in accelerating MRI. It employs deep neural networks to extract knowledge from available datasets and then applies the trained networks to reconstruct accurate images from limited measurements. Unlike natural image restoration problems, MRI involves physics‐based imaging processes, unique data properties, and diverse imaging tasks. This domain knowledge needs to be integrated with data‐driven approaches. Our review will introduce the significant challenges faced by such knowledge‐driven DL approaches in the context of fast MRI along with several notable solutions, which include learning neural networks and addressing different imaging application scenarios. The traits and trends of these techniques have also been given which have shifted from supervised learning to semi‐supervised learning, and finally, to unsupervised learning methods. In addition, MR vendors' choices of DL reconstruction have been provided along with some discussions on open questions and future directions, which are critical for the reliable imaging systems. [ABSTRACT FROM AUTHOR]

In a monocropping pattern for the crop domestication process, the interspecific and intraspecific interaction modes can be changed. In domesticated crops, the intraspecific interactions may have detrimental effects. Moreover, soil‐borne diseases in consecutive monocropped medicinal crops can be a critical challenge that has been difficult to solve. This study has been conducted to evaluate the effect of monoculture and intercropped cultivation on soil‐borne diseases caused by root secretions in American ginseng. The results of the correlation analysis showed that the polyethylenes in the roots had a more negative correlation with the occurrence of soil‐borne diseases than ginsenosides. Therefore, polyethylenes can be considered as an important factor in replanting failure of American ginseng. Furthermore, polyethylenes were able to remain in the soil and have a greater impact on the incidence of soil‐borne diseases than ginsenosides. [ABSTRACT FROM AUTHOR]

Soil sealing treatment is an effective chemical weed control method in rapeseed field. To evaluate the effects of different soil-applied herbicides on soil microorganisms in rapeseed field, 16S rRNA amplicon sequencing was used to analyze the changes in soil microbial structure and diversity after treatment with four commonly used soil-applied herbicides alachlor, acetochlor, metochlor and s-metolachlor. The results showed that soil-applied herbicide treatments could significantly reduce soil bacterial diversity in rapeseed fields, and alachlor had the greatest effect on soil bacterial diversity. However, after a period of herbicide treatment, the diversity and richness of the soil bacterial community gradually recovered. Soil-applied herbicide treatment had great influence on Actinomycetes, Acidobacteria and others in rapeseed soil, and alachlor treatment showed the greatest change in soil bacteria at phylum level compared with other herbicides. After 60 days of treatment with soil-applied herbicides, the abundances of functional bacteria related to pollutant degradation, denitrification, photosynthesis, pathogen growth inhibition and rhizosphere growth promotion were increased, whereas functional bacteria associated with the absorption of nutrients and excretion of metabolites were reduced. Overall, the results indicated that the soil-applied herbicides had a certain effect on soil microbial diversity and community structure in rapeseed field, but the effect of herbicides on soil microbial diversity gradually decreased along with the time of treatment. [ABSTRACT FROM AUTHOR]

This study aimed to explore the effects of thinning years on the water holding capacity of litter layer and soil layer in Chinese fir forest, thus to provide a scientific basis for evaluating the response of water conservation function of Chinese fir forest to thinning. In Jingning County, southwest Zhejiang Province, the Chinese fir forest with thinning for 1 year, 3 years, 8 years and unthinning was selected as the research object, and the litter and soil samples were collected. Then, the moisture content, water holding capacity, soil bulk density, non-capillary porosity of litter were analyzed and measured, and the relevant indexes were calculated. The results showed that there was no significant difference in the litter stock on hand of Chinese fir forest among different thinning years (P>0.05). The relationship between litter water holding capacity, water holding rate and water immersion time was in accordance with logarithmic function and power function, respectively. The maximum water holding capacity, maximum retention rate and effective retention rate of litter layer in Chinese fir forest with thinning for 8 years were significantly higher than those in the unthinned forest (P<0.05). The maximum water holding capacity and effective retention rate of litter layer in Chinese fir forest with thinning for 3 years and 8 years were significantly higher than those in unthinned forest (P<0.05). After thinning for 8 years, the soil bulk density of 0-10 cm soil layer in Chinese fir forest was significantly decreased (P<0.05), the non-capillary porosity was significantly increased, and the non-capillary water holding capacity was significantly increased by 22.4% (P<0.05). However, there was no significant effect on the related indicators of soil water storage performance in 10-30 cm soil layer (P>0.05). In conclusion, the water conservation function of litter layer and soil layer in Chinese fir forest was enhanced with the extension of thinning years. [ABSTRACT FROM AUTHOR]

Annually, hundreds of individuals tragically lose their lives at sea due to shipwrecks or aircraft accidents. For search and rescue personnel, the task of locating the debris of a downed aircraft in the vastness of the ocean presents a formidable challenge. A primary task these teams face is determining the search area, which is a critical step in the rescue operation. The movement of aircraft wreckage on the ocean surface is extremely complex, influenced by the combined effects of surface winds, waves, and currents. Establishing an appropriate drift motion prediction model is instrumental in accurately determining the search area for the wreckage. This article initially conducts maritime drift observation experiments on wreckage, and based on the results of these experiments, analyzes the drift characteristics and patterns of the debris. Subsequently, employing a wealth of observational experimental data, three types of drift prediction models for the wreckage are established using the least squares method. These models include the AP98 model, the dynamics model, and an improved model. In conclusion, the effectiveness and accuracy of the three models is evaluated and analyzed using Monte Carlo techniques. The results indicate that the probability of positive crosswind leeway (CWL) is 47.4%, while the probability of negative crosswind leeway (CWL) is 52.6%. The jibing frequency is 7.7% per hour, and the maximum leeway divergence angle observed is 40.4 degrees. Among the three drift prediction models, the refined AP98 drift model demonstrates the highest forecasting precision. The findings of this study offer a more accurate drift prediction model for the search of an aircraft lost at sea. These results hold significant guiding importance for maritime search and rescue operations in the South China Sea. [ABSTRACT FROM AUTHOR]

Background: Wave gradient encoding can adequately utilize coil sensitivity profiles to facilitate higher accelerations in parallel magnetic resonance imaging (pMRI). However, there are limitations in mainstream pMRI and a few deep learning (DL) methods for recovering missing data under wave encoding framework: the former is prone to introduce errors from the auto‐calibration signals (ACS) signal acquisition and is time‐consuming, while the latter requires a large amount of training data. Purpose: To tackle the above issues, an untrained neural network (UNN) model incorporating wave‐encoded physical properties and deep generative model, named WDGM, was proposed with additional ACS‐ and training data‐free. Methods: Generally, the proposed method can provide powerful missing data interpolation capability using the wave physical encoding framework and designed UNN to characterize the MR image (k‐space data) priors. Specifically, the MRI reconstruction combining physical wave encoding and elaborate UNN is modeled as a generalized minimization problem. The designation of UNN is driven by the coil sensitivity maps (CSM) smoothness and k‐space linear predictability. And then, the iterative paradigm to recover the full k‐space signal is determined by the projected gradient descent, and the complex computation is unrolled to the network with optimized parameters by the optimizer. Simulated wave encoding and in vivo experiments are exploited to demonstrate the feasibility of the proposed method. The best quantitative metrics RMSE/SSIM/PSNR of 0.0413, 0.9514, and 37.4862 gave competitive results in all experiments with at least six‐fold acceleration, respectively. Results: In vivo experiments of human brains and knees showed that the proposed method can achieve comparable reconstruction quality and even has superiority relative to the comparison, especially at a high resolution of 0.67 mm and fewer ACS. In addition, the proposed method has a higher computational efficiency achieving a computation time of 9.6 s/per slice. Conclusions: The model proposed in this work addresses two limitations of MRI reconstruction in the wave encoding framework. The first is to eliminate the need for ACS signal acquisition to perform the time‐consuming calibration process and to avoid errors such as motion during the acquisition procedure. Furthermore, the proposed method has clinical application friendly without the need to prepare large training datasets, which is difficult in the clinical. All results of the proposed method demonstrate more confidence in both quantitative and qualitative metrics. In addition, the proposed method can achieve higher computational efficiency. [ABSTRACT FROM AUTHOR]

The irrational land use patterns in the Yellow River Delta (YRD) have resulted in an imbalance in ecological stoichiometry, leading to secondary salinization and soil degradation. However, there is limited knowledge about the long-term response of soil and enzyme stoichiometry to land use. This hampers our ability to optimize land use in the YRD to alleviate nutrient limitation and thus promote ecological stoichiometric balance. We investigated the stoichiometry of soil and enzyme carbon (C), nitrogen (N), and phosphorus (P) in three land use patterns (Alfalfa artificial grassland, AG; wheat–maize rotation field, WM; native grassland, PC) established for 19 years in the YRD. The results showed that the soil stoichiometry of the three land uses in the YRD was lower than the world and Chinese averages, indicating lower C and N levels. Nutrient limitations of soil microorganisms were C and P due to an enzyme C:N ratio greater than 1:1 and vector angle greater than 45°. The three land use patterns have different advantages in alleviating nutrient limitations in the YRD. AG promotes soil macroaggregate formation, reduces soil salt content, improves nutrient availability, and mitigates N limitation. This makes AG more conducive to improving the poor soil structure, high soil salinity, and stoichiometric imbalance in the YRD to mitigate local soil degradation and be suitable for long-term continuous cultivation. WM is beneficial for increasing soil total C content due to straw return. However, WM does not reduce soil salinity. WM is more suitable for intercropping or crop rotation to improve soil C content in the YRD. Although PC can alleviate soil microbial C limitation due to its significantly lower vector length than AG and WM, the low nutrient levels hindered its ability to alleviate local soil nutrient limitation. In conclusion, our study provides a theoretical basis for rational land use in the YRD to mitigate soil degradation. [ABSTRACT FROM AUTHOR]

Background: In the past two years, studies have found a significant increase in neutrophil extracellular traps (NETs) in patients with IgA vasculitis (IgAV), which is correlated with the severity of the disease. NETs have been reported as an intervention target in inflammatory and autoimmune diseases. This study aimed to investigate the effect of targeted degradation of NETs using DNase I in IgAV rat model. Methods: Twenty-four Sprague-Dawley rats were randomly divided into three groups: the IgAV model group, the DNase I intervention group and the normal control group, with an average of 8 rats in each group. The model group was established by using Indian ink, ovalbumin, and Freund's complete adjuvant. In the intervention group, DNase I was injected through tail vein 3 days before the end of established model. The circulating cell free-DNA (cf-DNA) and myeloperoxidase-DNA (MPO-DNA) were analyzed. The presence of NETs in the kidney, gastric antrum and descending duodenum were detected using multiple fluorescences immunohistochemistry and Western blots. Morphological changes of the tissues were observed. Results: After the intervention of DNase I, there was a significant reduction in cf-DNA and MPO-DNA levels in the intervention group compared to the IgAV model group (all P<0.001). The presence of NETs in renal, gastric, and duodenal tissues of the intervention group exhibited a significant decrease compared to the IgAV model group (P < 0.01). Moreover, the intervention group demonstrated significantly lower levels of renal MPO and citrullinated histone H3 (citH3) protein expression when compared to the IgAV model group (all P < 0.05). The HE staining results of intervention group demonstrated a significant reduction in congestion within glomerular and interstitial capillaries. Moreover, there was a notable improvement in gastric and intestinal mucosa necrosis, congestion and bleeding. Additionally, there was a substantial decrease in inflammatory cells infiltration. Conclusion: The degradation of NETs can be targeted by DNase I to mitigate tissue damage in IgAV rat models. Targeted regulation of NETs holds potential as a therapeutic approach for IgAV. [ABSTRACT FROM AUTHOR]

Magnetic resonance (MR) image reconstruction and super-resolution are two prominent techniques to restore high-quality images from undersampled or low-resolution k-space data to accelerate MR imaging. Combining undersampled and low-resolution acquisition can further improve the acceleration factor. Existing methods often treat the techniques of image reconstruction and super-resolution separately or combine them sequentially for image recovery, which can result in error propagation and suboptimal results. In this work, we propose a novel framework for joint image reconstruction and super-resolution, aiming to efficiently image recovery and enable fast imaging. Specifically, we designed a framework with a reconstruction module and a super-resolution module to formulate multi-task learning. The reconstruction module utilizes a model-based optimization approach, ensuring data fidelity with the acquired k-space data. Moreover, a deep spatial feature transform is employed to enhance the information transition between the two modules, facilitating better integration of image reconstruction and super-resolution. Experimental evaluations on two datasets demonstrate that our proposed method can provide superior performance both quantitatively and qualitatively. [ABSTRACT FROM AUTHOR]

Background: Neutrophil extracellular traps (NETs) have been found to play a role in the development of autoimmune diseases. In the past two years, studies have demonstrated a significantly increase of NETs in skin tissues during the early stages of IgAV, indicating their involvement in disease activity among children with IgAV. However, the presence of NETs in IgAV animal models has not yet been reported. The objective of this study is to investigate whether NETs are involved in the pathogenesis of IgA vasculitis (IgAV) rats. Methods: Twenty-four SD rats were randomly divided into three groups: the ovalbumin group, the gliadin group, and the control group. The IgAV rat models were established administering Indian ink with ovalbumin (ovalbumin group) or gliadin (gliadin group) with Freund's complete adjuvant. The cell-free DNA (cf-DNA) was quantified by using dsDNA quantification kit, while the levels of Immunoglobulins, complement C3 and myeloperoxidase-DNA (MPO-DNA) in serum were tested using enzyme linked immunosorbent assay (ELISA). The IgA, complement C3 and NETs in tissues were detected through multiple immunofluorescences. Results: Both the ovalbumin group and gliadin group showed IgA and C3 deposition in various tissues, including the glomerular mesangial region, skin, and digestive tract, while the control group showed no such deposition. The levels of circulatory cf-DNA and MPO-DNA, which are components of NETs, were significantly elevated in both ovalbumin and gliadin groups compared with the control group. Furthermore, the presence of NETs were found in gastrointestinal and renal tissues of the ovalbumin and gliadin groups, but not in the control group. Conclusions: IgAV model rat can be established through the combination of ovalbumin and gliadin with Indian ink and Freund's complete adjuvant. This study provides the first confirmation that NETs are involved in the pathogenesis of IgAV rat. [ABSTRACT FROM AUTHOR]

In order to explore the application effect of different light quality as the light source of tobacco growth, by taking Nicotiana benthamiana as the experimental material under the condition of artificial climate chamber, three light sources with different light quality, including LED-v1 (R1.2/B1, R3.4/FR1), LED-v2 (R7.4/B1, R2.9/FR1), and LED-v1/v2 (R2.3/B1, R3.1/FR1), were set up to analyze the response of tobacco growth indicators to different light quality. The results showed that the plant height, stem thick, leaf area, blade number and biomass of Nicotiana benthamiana treated with LED-v1/v2 were significantly better than those of other treatments. Low R/FR and low R/B light quality could increase the N content in the leaves of tobacco, help to synthesize the organic matters and improve the stress resistance, and also improve the economic quality of tobacco. Appropriately reducing the R/B and R/FR values of LED-v1/v2 light quality was more conducive to the growth of f tobacco and its quality improvement. There was no significant difference in the chlorophyll fluorescence parameters such as photosynthetic reaction centre II (ΦPSII)> apparent electron transport rate (ETR), PS II maximal photochemical efficiency (Fv/Fm) among different treatments. In summary, the most suitable artificial light source for the growth of tobacco was LED-v1/v2 (R2.3/B1, R3.1/FR1). [ABSTRACT FROM AUTHOR]

Vibrio alginolyticus is a Gram-negative bacterium commonly associated with mackerel poisoning. A bacteriophage that specifically targets and lyses this bacterium could be employed as a biocontrol agent for treating the bacterial infection or improving the shelf-life of mackerel products. However, only a few well-characterized V. alginolyticus phages have been reported in the literature. In this study, a novel lytic phage, named ΦImVa-1, specifically infecting V. alginolyticus strain ATCC 17749, was isolated from Indian mackerel. The phage has a short latent period of 15 min and a burst size of approximately 66 particles per infected bacterium. ΦImVa-1 remained stable for 2 h at a wide temperature (27–75 °C) and within a pH range of 5 to 10. Transmission electron microscopy revealed that ΦImVa-1 has an icosahedral head of approximately 60 nm in diameter with a short tail, resembling those in the Schitoviridae family. High throughput sequencing and bioinformatics analysis elucidated that ΦImVa-1 has a linear dsDNA genome of 77,479 base pairs (bp), with a G + C content of ~ 38.72% and 110 predicted gene coding regions (106 open reading frames and four tRNAs). The genome contains an extremely large virion-associated RNA polymerase gene and two smaller non-virion-associated RNA polymerase genes, which are hallmarks of schitoviruses. No antibiotic genes were found in the ΦImVa-1 genome. This is the first paper describing the biological properties, morphology, and the complete genome of a V. alginolyticus-infecting schitovirus. When raw mackerel fish flesh slices were treated with ΦImVa-1, the pathogen loads reduced significantly, demonstrating the potential of the phage as a biocontrol agent for V. alginolyticus strain ATCC 17749 in the food. Key points: • A novel schitovirus infecting Vibrio alginolyticus ATCC 17749 was isolated from Indian mackerel. • The complete genome of the phage was determined, analyzed, and compared with other phages. • The phage is heat stable making it a potential biocontrol agent in extreme environments. [ABSTRACT FROM AUTHOR]

By taking the public welfare forest land in Jiangshan City of Zhejiang Province as the research object, based on the measured data of fixed sample plots and sublot of public welfare forest, the water conservation capacity and value of litter layer and soil layer in different forest stands were quantitatively measured and analyzed. The results showed that the effective interception amount of litter in different forest types of public welfare forest in Jiangshan City was coniferous and broad-leaved mixed forest>broad-leaved forest>Chinese fir forest>pine forest>moso bamboo forest. In the same depth of soil layer, the difference of soil physical properties and water holding capacity was not obvious. The soil layer contributed more than 99 % of the water conservation quantity of forest land. The order of water conservation ability was coniferous and broad-leaved mixed forest>broad-leaved forest>moso bamboo forest> Chinese fir forest>coniferous mixed forest>pine forest>shrub forest. The total annual water conservation quantity of public welfare forests in the whole city was 18 7,7869 million tons, with the value quantity of 1.979 billion yuan. The water conservation quantity per unit area was about 2813 t ⋅ hm-2, and the value quantity per unit area was about 29,700 yuan ⋅ hm-2. In Jiangshan City, the water conservation capacity of public welfare forest was stronger in the east, while the water conservation capacity of public welfare forest in the north was relatively weak. [ABSTRACT FROM AUTHOR]

The determination of seepage field is the basis to design seepage drainage facilities and dam slopes in tailings pond. However, in the design of a dry-stack tailings pond with a long service life, previous research has been mostly limited to the influence of short-term rainfall, and a proper concept has not been formed to guide the calculation of the seepage field in the design of a dry-stack tailings pond under the cumulative effect of long-term periodic rainfall. The objective of this study is to propose a new concept to concisely determine the seepage field for the design of a dry-stack tailings pond under multi-year periodic rainfall. To this end, the calculation model of the seepage field of a dry-stack tailings pond under multiyear rainfall was established. The evolution process of the seepage field was studied by numerical simulation based on the calculation model and the final form of the seepage field evolution of a dry-stack tailings pond was found. Accordingly, a new concept, the relatively stable seepage field (RSSF) which can be used as the basis for seepage drainage facilities and the dam slope design of a dry-stack tailings pond, was proposed and named. Furthermore, the influencing factors of a relatively stable seepage field were studied by numerical simulation. The results show that: (1) the cumulative infiltration of long-term periodic rainfall is the main reason for the formation of the RSSF, and (2) under the condition of constant annual rainfall, the distribution of the RSSF has little to do with the selection of rainfall mode. Therefore, a new understanding has been formed from this article whereby the RSSF is an essential basis to be considered in the design of the dam slope and drainage system of a dry-stack tailings pond. [ABSTRACT FROM AUTHOR]

Lately, deep learning technology has been extensively investigated for accelerating dynamic magnetic resonance (MR) imaging, with encouraging progresses achieved. However, without fully sampled reference data for training, the current approaches may have limited abilities in recovering fine details or structures. To address this challenge, this paper proposes a self-supervised collaborative learning framework (SelfCoLearn) for accurate dynamic MR image reconstruction from undersampled k-space data directly. The proposed SelfCoLearn is equipped with three important components, namely, dual-network collaborative learning, reunderampling data augmentation and a special-designed co-training loss. The framework is flexible and can be integrated into various model-based iterative un-rolled networks. The proposed method has been evaluated on an in vivo dataset and was compared to four state-of-the-art methods. The results show that the proposed method possesses strong capabilities in capturing essential and inherent representations for direct reconstructions from the undersampled k-space data and thus enables high-quality and fast dynamic MR imaging. [ABSTRACT FROM AUTHOR]

Background: With rapid economic development, the world's average life expectancy is increasing, leading to the increasing prevalence of osteoporosis worldwide. However, due to the complexity and high cost of dual-energy x-ray absorptiometry (DXA) examination, DXA has not been widely used to diagnose osteoporosis. In addition, studies have shown that the psoas index measured at the third lumbar spine (L3) level is closely related to bone mineral density (BMD) and has an excellent predictive effect on osteoporosis. Therefore, this study developed a variety of machine learning (ML) models based on psoas muscle tissue at the L3 level of unenhanced abdominal computed tomography (CT) to predict osteoporosis.Methods: Medical professionals collected the CT images and the clinical characteristics data of patients over 40 years old who underwent DXA and abdominal CT examination in the Second Affiliated Hospital of Wenzhou Medical University database from January 2017 to January 2021. Using 3D Slicer software based on horizontal CT images of the L3, the specialist delineated three layers of the region of interest (ROI) along the bilateral psoas muscle edges. The PyRadiomics package in Python was used to extract the features of ROI. Then Mann-Whitney U test and the least absolute shrinkage and selection operator (LASSO) algorithm were used to reduce the dimension of the extracted features. Finally, six machine learning models, Gaussian naïve Bayes (GNB), random forest (RF), logistic regression (LR), support vector machines (SVM), Gradient boosting machine (GBM), and Extreme gradient boosting (XGBoost), were applied to train and validate these features to predict osteoporosis.Results: A total of 172 participants met the inclusion and exclusion criteria for the study. 82 participants were enrolled in the osteoporosis group, and 90 were in the non-osteoporosis group. Moreover, the two groups had no significant differences in age, BMI, sex, smoking, drinking, hypertension, and diabetes. Besides, 826 radiomic features were obtained from unenhanced abdominal CT images of osteoporotic and non-osteoporotic patients. Five hundred fifty radiomic features were screened out of 826 by the Mann-Whitney U test. Finally, 16 significant radiomic features were obtained by the LASSO algorithm. These 16 radiomic features were incorporated into six traditional machine learning models (GBM, GNB, LR, RF, SVM, and XGB). All six machine learning models could predict osteoporosis well in the validation set, with the area under the receiver operating characteristic (AUROC) values greater than or equal to 0.8. GBM is more effective in predicting osteoporosis, whose AUROC was 0.86, sensitivity 0.70, specificity 0.92, and accuracy 0.81 in validation sets.Conclusion: We developed six machine learning models to predict osteoporosis based on psoas muscle images of abdominal CT, and the GBM model had the best predictive performance. GBM model can better help clinicians to diagnose osteoporosis and provide timely anti-osteoporosis treatment for patients. In the future, the research team will strive to include participants from multiple institutions to conduct external validation of the ML model of this study. [ABSTRACT FROM AUTHOR]

A serious Juniperus rigida rust disease was found in Gansu Province, China. The disease incidence is approximately 80–90%. We also found rust disease on both Cotoneaster multiflorus and Sorbus in the same location. Two novel Gymnosporangium species were identified from the infected plants. Based on morphological observations and phylogenetic analyses, we describe the two new taxa as G. gansuense and G. granulatosporum. We also determined their life cycles. Moreover, this study documented a novel aeciospore surface structure with two different surface types on one aeciospore. We describe it here as "granular." [ABSTRACT FROM AUTHOR]

Background: Compared with open comminuted calcaneal fractures, less emphasis is placed on postoperative surgical site infection (SSI) of closed comminuted calcaneal fractures. This study aimed to identify the risk factors associated with SSI and build a nomogram model to visualize the risk factors for postoperative SSI.Methods: We retrospectively collected patients with closed comminuted calcaneal fractures from the Second Affiliated Hospital of Wenzhou Medical University database from 2017 to 2020. Risk factors were identified by logistics regression analysis, and the predictive value of risk factors was evaluated by ROC (receiver operating characteristic curve). Besides, the final risk factors were incorporated into R4.1.2 software to establish a visual nomogram prediction model.Results: The high-fall injury, operative time, prealbumin, aspartate aminotransferase (AST), and cystatin-C were independent predictors of SSI in calcaneal fracture patients, with OR values of 5.565 (95%CI 2.220-13.951), 1.044 (95%CI 1.023-1.064), 0.988 (95%CI 0.980-0.995), 1.035 (95%CI 1.004-1.067) and 0.010 (95%CI 0.001-0.185) (Ps < 0.05). Furthermore, ROC curve analysis showed that the AUC values of high-fall injury, operation time, prealbumin, AST, cystatin-C, and their composite indicator for predicting SSI were 0.680 (95%CI 0.593-0.766), 0.756 (95%CI 0.672-939), 0.331 (95%CI 0.243-0.419), 0.605 (95%CI 0.512-0.698), 0.319 (95%CI 0.226-0.413) and 0.860 (95%CI 0.794-0.926), respectively (Ps < 0.05). Moreover, the accuracy of the nomogram to predict SSI risk was 0.860.Conclusions: Our study findings suggest that clinicians should pay more attention to the preoperative prealbumin, AST, cystatin C, high-fall injury, and operative time for patients with closed comminuting calcaneal fractures to avoid the occurrence of postoperative SSI. Furthermore, our established nomogram to assess the risk of SSI in calcaneal fracture patients yielded good accuracy and can assist clinicians in taking appropriate measures to prevent SSI. [ABSTRACT FROM AUTHOR]

Response inhibition, a crucial component of executive function, is closely related to personal impulse control, social adaption, and mental health. Previous studies have found response inhibition deficit in patients with major depressive disorder, but whether it also exists in individuals with subclinical depression (SD) remains unclear. This study aimed to identify the ability of response inhibition to emotional face stimuli both under explicit and implicit conditions in individuals with SD. Thirty‐six subclinical depressed college students and 39 healthy individuals were recruited and administered the non‐emotional, explicit, and implicit emotional stop‐signal tasks (SSTs). Mixed‐model analyses of variance were used to analyze the differences between and within groups. In implicit emotional SST, the results showed a significant longer stop‐signal response time, a shorter stop‐signal delay time, a shorter go reaction time, and a similar proportion of stop success in the SD group compared to healthy controls. However, the above indices showed no significant difference between the two groups in the non‐emotional SST and explicit emotional SST. These findings suggest a possible defect of response inhibition in implicit emotional processing in individuals with SD, which may potentially serve as a marker of susceptibility to depression and thus be applied to early screening and intervention for major depressive disorder. [ABSTRACT FROM AUTHOR]

Image super-resolution (SR) methods can generate remote sensing images with high spatial resolution without increasing the cost of acquisition equipment, thereby providing a feasible way to improve the quality of remote sensing images. Clearly, image SR is a severe ill-posed problem. With the development of deep learning, the powerful fitting ability of deep neural networks has solved this problem to some extent. Since the texture information of various remote sensing images are totally different from each other, in this article, we proposed a network based on generative adversarial network (GAN) to achieve high-resolution remote sensing images, named multiattention GAN (MA-GAN). The main body of the generator in MA-GAN contains three blocks: pyramid convolutional residual dense (PCRD) block, attention-based upsampling (AUP) block, and attention-based fusion (AF) block. Specifically, the developed attention pyramid convolutional (AttPConv) operator in the PCRD block combines multiscale convolution and channel attention (CA) to automatically learn and adjust the scale of residuals for better representation. The established AUP block uses pixel attention (PA) to perform arbitrary scales of upsampling. The AF block uses branch attention (BA) to integrate upsampled low-resolution images with high-level features. Besides, the loss function takes both adversarial loss and feature loss into consideration to guide the learning procedure of the generator. We have compared our MA-GAN approach with several state-of-the-art methods on a number of remote sensing scenes, and the experimental results consistently demonstrate the effectiveness of the proposed MA-GAN. For study replication, the source code will be released at: https://github.com/ZhihaoWang1997/MA-GAN. [ABSTRACT FROM AUTHOR]

Purpose: To develop and evaluate an automatic segmentation method of arterial vessel walls and plaques, which is beneficial for facilitating the arterial morphological quantification in magnetic resonance vessel wall imaging (MRVWI). Methods: MRVWI images acquired from 124 patients with atherosclerotic plaques were included. A convolutional neural network-based deep learning model, namely VWISegNet, was used to extract the features from MRVWI images and calculate the category of each pixel to facilitate the segmentation of vessel wall. Two-dimensional (2D) cross-sectional slices reconstructed from all plaques and 7 main arterial segments of 115 patients were used to build and optimize the deep learning model. The model performance was evaluated on the remaining nine-patient test set using the Dice similarity coefficient (DSC) and average surface distance (ASD). Results: The proposed automatic segmentation method demonstrated satisfactory agreement with the manual method, with DSCs of 93.8% for lumen contours and 86.0% for outer wall contours, which were higher than those obtained from the traditional U-Net, Attention U-Net, and Inception U-Net on the same nine-subject test set. And all the ASD values were less than 0.198 mm. The Bland–Altman plots and scatter plots also showed that there was a good agreement between the methods. All intraclass correlation coefficient values between the automatic method and manual method were greater than 0.780, and greater than that between two manual reads. Conclusion: The proposed deep learning-based automatic segmentation method achieved good consistency with the manual methods in the segmentation of arterial vessel wall and plaque and is even more accurate than manual results, hence improved the convenience of arterial morphological quantification. [ABSTRACT FROM AUTHOR]

Purpose: To achieve simultaneous T1,w/proton density fat fraction (PDFF)/R2∗ mapping in abdomen within a single breadth‐hold, and validate the accuracy using state‐of‐art measurement. Theory and Methods: An optimized multiple echo gradient echo (GRE) sequence with dual flip‐angle acquisition was used to realize simultaneous water T1 (T1,w)/PDFF/R2∗ quantification. A new method, referred to as "solving the fat‐water ambiguity based on their T1 difference" (SORT), was proposed to address the fat‐water separation problem. This method was based on the finding that compared to the true solution, the wrong (or aliased) solution to fat‐water separation problem showed extra dependency on the applied flip angle due to the T1 difference between fat and water. The B1+ measurement sequence was applied to correct the B1+ inhomogeneity for T1,w relaxometry. The 2D parallel imaging was incorporated to enable the acquisition within a single breath‐hold in abdomen. Results: The multi‐parametric quantification results of the proposed method were consistent with the results of reference methods in phantom experiments (PDFF quantification: R2 = 0.993, mean error 0.73%; T1,w quantification: R2 = 0.999, mean error 4.3%; R2∗ quantification: R2 = 0.949, mean error 4.07 s−1). For volunteer studies, robust fat‐water separation was achieved without evident fat‐water swaps. Based on the accurate fat‐water separation, simultaneous T1,w/PDFF/R2∗ quantification was realized for whole liver within a single breath‐hold. Conclusion: The proposed method accurately quantified T1,w/PDFF/R2∗ for the whole liver within a single breath‐hold. This technique serves as a quantitative tool for disease management in patients with hepatic steatosis. [ABSTRACT FROM AUTHOR]

In order to understand the effect of exocarp removal on the seed germination and seedling growth of Ardisia violacea which was a rare and endangered plant, two treatments, peeled and unpeeled, were set up with four replicates to study the seed germination and growth traits of Ardisia violacea. The results showed that the emergence time of the unpeeled seeds of Ardisia violacea was delayed by 19 days, which was significantly later than that of the peeled seeds (P<0. 05) ; The seed emergence rate was 70. 3% - 75. 6 %, and there was no significant difference between the unpeeled and peeled seeds (P> 0. 05) ; The growth of Ardisia violacea seedlings showed an "S-shaped" curve of "slow-fast-slow". By the end of the growth, the plant height, ground diameter, leaf length and leaf width were respectively in the range of 5. 32 to 5. 34 cm, 0. 12 to 0. 13 cm, 4. 40 to 4. 96 cm, 2. 08 to 2. 12 cm, and the number of leaves ranged from 13. 4 to 14. 8, so there was no significant difference between the unpeeled and peeled treatments (P>-0. 05). In conclusion, the existence of pericarp had no significant effect on the germination and seedling growth of Ardisia violacea, and the seedlings could be directly sown. [ABSTRACT FROM AUTHOR]

The rapid development of remote sensing sensors makes the acquisition, analysis, and application of hyperspectral images (HSIs) more and more extensive. However, the limited sample sets, high-dimensional features, highly correlated bands, and mixing spectral information make the classification of HSIs a great challenge. In this article, an unsupervised multiscale and diverse feature learning (UMsDFL) method is proposed for HSI classification, which deeply considers the spatial–spectral features via convolutional neural networks (CNNs). Specifically, after employing the simple noniterative clustering (SNIC) algorithm with the heuristic calculation of superpixel size, the HSIs are segmented into superpixels for feature learning. The unsupervised network is designed with the convolutional encoder and decoder, the additional clustering branch, and the multilayer feature fusion to enhance the distinguishability of feature learning and the reusability of feature maps. Then, the spatial relationships and object attributes in large- and small-scale contexts are learned collaboratively through the unsupervised network to utilize the complementary multiscale characteristics. Moreover, the diverse features of hyperspectral information and nonsubsampled contourlet transform (NSCT) textures are learned simultaneously via the unsupervised network to alleviate the insufficiency of geometric representation. Finally, the random forest (RF) is adopted as the comprehensive classifier for land cover mapping based on the UMsDFL, and superpixel regularization is adopted to optimize the classification results. A series of experiments are performed on three real-world HSI datasets to demonstrate the effectiveness of our UMsDFL approach. The experimental results show that the proposed UMsDFL can achieve the overall accuracy of 79.23%, 96.49%, and 77.26% for Houston, Pavia, and Dioni datasets, respectively, when there are only five samples per class for training. [ABSTRACT FROM AUTHOR]

Dynamic MR image reconstruction from incomplete k‐space data has generated great research interest due to its capability in reducing scan time. Nevertheless, the reconstruction problem is still challenging due to its ill‐posed nature. Most existing methods either suffer from long iterative reconstruction time or explore limited prior knowledge. This paper proposes a dynamic MR imaging method with both k‐space and spatial prior knowledge integrated via multi‐supervised network training, dubbed as DIMENSION. Specifically, the DIMENSION architecture consists of a frequential prior network for updating the k‐space with its network prediction and a spatial prior network for capturing image structures and details. Furthermore, a multi‐supervised network training technique is developed to constrain the frequency domain information and the spatial domain information. The comparisons with classical k‐t FOCUSS, k‐t SLR, L+S and the state‐of‐the‐art CNN‐based method on in vivo datasets show our method can achieve improved reconstruction results in shorter time. [ABSTRACT FROM AUTHOR]

Purpose: To reconstruct aliasing‐free full field‐of‐view (FOV) images for reduced FOV (rFOV) parallel imaging (PI) with Cartesian and Wave sampling, which suffers from aliasing artifacts using existing PI methods. Theory and Methods: The sensitivity encoding method (SENSE) was extended to the Soft‐SENSE models supporting multiple‐set coil sensitivity maps (CSM) and point spread functions (PSF) for Cartesian and Wave sampled rFOV PI, respectively. The multiple‐set CSM and PSF were created from full FOV CSM and PSF according to the image folding process induced by rFOV sampling. The Soft‐SENSE reconstructions could be solved by the same algorithms for the conventional full FOV SENSE reconstruction. Results: Soft‐SENSE using multiple‐set full FOV CSM and PSF successfully reconstruct aliasing‐free full FOV image from rFOV PI data with Cartesian and Wave sampling. The proposed rFOV PI enables flexible control of the aliasing and achieves comparable geometry factors as the standard full FOV PI with the same net acceleration factor. Reduced FOV PI improves the computational efficiency of iterative compressed sensing (CS) and PI reconstruction, especially for high‐resolution volumetric imaging, thanks to the reduced fast Fourier transform (FFT) size. Moreover, rFOV PI reconstruction provides a potential alternative to the phase oversampling for the FOV aliasing problem. Conclusion: The proposed Soft‐SENSE using full FOV CSM and PSF could reconstruct aliasing‐free full FOV image for rFOV PI, and make it a viable solution enabling more flexible PI acceleration and effectively improving the computational efficiency of iterative CSPI reconstruction. [ABSTRACT FROM AUTHOR]

Due to the detailed spectral information through hundreds of narrow spectral bands provided by hyperspectral image (HSI) data, it can be employed to accurately classify diverse materials of interest, which is one of the core applications of hyperspectral remote sensing technology. In recent years, with the rapid development of deep learning, convolutional neural networks (CNNs) have been successfully applied in many fields, including HSI classification. However, the random gradient descent-based parameter updating scheme is too general and leading to the inefficiency of CNN models. Moreover, the high dimensionality and limited training samples of HSI data also exacerbate the overfitting problem. To tackle these issues, in this article, a novel deep network with multilayer and multibranch architecture, named 3-D Gabor CNN (3DG-CNN), is proposed for HSI classification. More precisely, since the predefined 3-D Gabor filters in multiple scales and orientations could well characterize the internal spatial–spectral structure of HSI data from various perspectives, the 3-D Gabor-modulated kernels (3-D GMKs) are employed to replace the random initialization kernels. Moreover, the specially designed multibranch architecture enables the network to better integrating the scalable property of 3-D Gabor filters; thus, the representative ability and robustness of the extracted features can be greatly improved. Alternatively, the number of network parameters is substantially reduced due to the incorporation of 3-D Gabor modulation, relieving the training complexity and also alleviating the training process from overfitting. Experimental results on four real HSI datasets (including two newly released ones in the literature) have demonstrated that the proposed 3DG-CNN model can achieve better performance than several widely used machine-learning-based and deep-learning-based approaches. For the sake of reproducibility, the codes of the proposed 3DG-CNN model are available at http://jiasen.tech/papers/. [ABSTRACT FROM AUTHOR]

As an important medium of human-computer interaction, flexible tactile sensors have a vital influence on the intelligence and safety of a system by accurately identifying the contact position and force. For now, most of the existing flexible capacitive pressure sensors adopt an array structure, which still cannot meet the requirements including simple manufacturing, simple wiring, and low cost. This paper innovatively proposes a flexible array-less capacitive tactile sensor, a 30 mm $\times30$ mm sensor prototype, consisting of flexible electrodes and a dielectric layer of a polydimethylsiloxane (PDMS) and multiwalled carbon nanotube (MWCNT) composite. Under the action of a contact force, the contact area is deformed, resulting in a capacitance change in the flexible sensor. First, Poisson’s equation is used to determine the relationship between the electric potential and position. Then, elastic contact mechanics is adopted to describe the relationship between the electric charges and force. Finally, the contact position and force are recognized by measuring the change in capacitance in real time, based on the classical principle of capacitance. Simulation and experimental results verified the effectiveness of the proposed recognition method, showing that the recognition resolution of contact position can reach $500\,\,\mu \text{m}\,\,\times 100\,\,\mu \text{m}$ , and the accuracy of contact force is 95.85%. Theoretically, the proposed flexible array-less sensor in this paper has infinite spatial resolution, and most importantly, it does not rely on complex manufacturing, redundant circuits, or sensor cell arrays. Therefore, it has broad application prospects in related fields such as electronic skin and wearable electronic devices. [ABSTRACT FROM AUTHOR]

Deep learning methods have achieved attractive performance in dynamic MR cine imaging. However, most of these methods are driven only by the sparse prior of MR images, while the important low-rank (LR) prior of dynamic MR cine images is not explored, which may limit further improvements in dynamic MR reconstruction. In this paper, a learned singular value thresholding (Learned-SVT) operator is proposed to explore low-rank priors in dynamic MR imaging to obtain improved reconstruction results. In particular, we put forward a model-based unrolling sparse and low-rank network for dynamic MR imaging, dubbed as SLR-Net. SLR-Net is defined over a deep network flow graph, which is unrolled from the iterative procedures in the iterative shrinkage-thresholding algorithm (ISTA) for optimizing a sparse and LR-based dynamic MRI model. Experimental results on a single-coil scenario show that the proposed SLR-Net can further improve the state-of-the-art compressed sensing (CS) methods and sparsity-driven deep learning-based methods with strong robustness to different undersampling patterns, both qualitatively and quantitatively. Besides, SLR-Net has been extended to a multi-coil scenario, and achieved excellent reconstruction results compared with a sparsity-driven multi-coil deep learning-based method under a high acceleration. Prospective reconstruction results on an open real-time dataset further demonstrate the capability and flexibility of the proposed method on real-time scenarios. [ABSTRACT FROM AUTHOR]

Hyperspectral images encompass abundant information and provide unique characteristics for material classification. However, the labeling of training samples can be challenging in hyperspectral image classification. To address this problem, this study proposes a framework named flexible Gabor-based superpixel-level unsupervised linear discriminant analysis (FG-Su ULDA) to extract the most informative and discriminating features for classification. First, a number of 3-D flexible Gabor filters are rigorously designed using an asymmetric sinusoidal wave to sufficiently characterize the spatial–spectral structure in hyperspectral images. Then, an unsupervised linear discriminant analysis strategy guided by the entropy rate superpixel (ERS) segmentation algorithm, called Su ULDA, is skillfully introduced to reduce the extracted large amount of FG features. The Su ULDA method not only boosts the classification capability but also increases the peculiarity of features, with the aid of superpixel information. Finally, the achieved features are imported to the popular support vector machine classifier. The proposed FG-Su ULDA framework is applied to four real hyperspectral image data sets, and the experiments constantly prove that our FG-Su ULDA is superior to several state-of-the-art methods in both classification performance and computational efficiency, especially with scarce training samples. The codes of this work are available at http://jiasen.tech/papers/ for the sake of reproducibility. [ABSTRACT FROM AUTHOR]

Due to the importance in many military and civilian applications, hyperspectral anomaly detection has attracted remarkable interest. Low-rank representation (LRR)-based anomaly detectors use the low-rank property to represent background pixels, and pixels that cannot be well represented are detected as anomalies. The ability of an LRR-based detector to separate background pixels and anomalous pixels depends on the dictionary representation ability, which usually can be enhanced by designing a proper prior for dictionary representation coefficients and constructing a better dictionary. However, it is not easy to handcraft effective and meaningful regularizers for dictionary coefficients. In this article, we propose a novel anomaly detection algorithm that uses a plug-and-play prior for representation coefficients and constructs a new dictionary based on clustering. Instead of cumbersomely handcrafting a regularizer for representation coefficients, we propose solving the anomaly detection problem using the plug-and-play framework, which enables us to plug state-of-the-art priors for representation coefficients. An effective convolutional neural network (CNN) denoiser is plugged into our framework to fully exploit the spatial correlation of representation coefficients. We also propose a modified background dictionary construction method, which carefully includes background pixels and excludes anomalous pixels from clustering results. We refer to the proposed anomaly detection method as plug-and-play denoising CNN regularized anomaly detection (DeCNN-AD) method. Extensive experiments were performed on five data sets in a comparison with eight state-of-the-art anomaly detection methods. The experimental results suggest that the proposed method is effective in anomaly detection and can produce better anomaly detection results than that of the comparison methods. The codes of this work will be available at https://github.com/FxyPd for the sake of reproducibility. [ABSTRACT FROM AUTHOR]

Objectives: This aim of this study was to determine whether neutrophil extracellular traps (NETs) are involved in the pathogenesis of IgA vasculitis (IgAV) and investigate whether the circulating NETs levels are associated with disease activity in children. Methods: We performed a case-control study and collected blood samples from 193 children with different stages of IgAV (61 were at the onset stage, 64 at the remission stage, 43 at the active stage, and 25 were undergoing drug withdrawal). A total of 192 healthy children were recruited as controls. Circulating cell free DNA (cf-DNA) was obtained from the plasma and quantified by using the Quant-iT PicoGreen DNA quantification kit. NETs-associated myeloperoxidase-DNA (MPO-DNA), citrullinated-histone H3 (cit-H3), neutrophil elastase (NE), and the deoxyribonuclease I (DNase I) concentrations were measured using enzyme-linked immunosorbent assays. The presence of NETs in the kidney and gastrointestinal tissues of onset and active IgAV patients was determined by multiple immunofluorescence staining in 15 IgAV nephritis patients and 9 IgAV patients without IgAV nephritis, respectively. NETs degradation potency of collected sera samples from IgAV patients were checked in vitro. Relationships between circulating levels of cf-DNA with MPO-DNA, NE, and DNase I and the patients were analyzed. Results: Circulating levels of cf-DNA in onset and active IgAV patients were significantly higher than those in remission and drug withdrawal patients as well as healthy controls. The results were similar for MPO-DNA and NE. The levels of circulating cf-DNA correlated significantly with MPO-DNA, NE and DNase I. A significantly decreased degradation of NETs from the onset and active IgAV patients was observed, but was normal in healthy controls. Furthermore, presence of NETs was also confirmed in all renal and gastrointestinal tissues obtained from the onset and active IgAV patients but not control samples. Conclusions: Our data showed that NETs were released into the circulation of IgAV patients and are involved in the disease activity. The circulating levels of NETs maybe used to assess disease severity in children with IgAV. [ABSTRACT FROM AUTHOR]

Purpose: To propose a novel model incorporating virtual conjugate coil (VCC) reconstruction and wave encoding (Wave) for improved parallel MRI. Theory and Methods: A novel model (VCC‐Wave) incorporating VCC and Wave is proposed. The correlation matrix of the encoding operator is introduced to analyze the encoding capability. In addition, simulation experiments are conducted to gain insights into VCC‐Wave. In vivo experiments are performed to compare VCC‐Wave with alternative methods. Results: The correlation matrix and the simulation experiments show that the proposed VCC‐Wave can utilize more priors of Wave under the VCC framework. In vivo experiments show that the proposed VCC‐Wave can achieve good image quality at a 6‐fold acceleration in high‐resolution and high‐bandwidth cases, indicating an improvement over the original Wave technique. Conclusion: The proposed VCC‐Wave can not only combine the advantages of both the VCC and Wave but also exploit more priors of Wave under the VCC framework. The improvement in VCC‐Wave alleviates the limitation of Wave in high‐resolution and high‐bandwidth cases. [ABSTRACT FROM AUTHOR]

Purpose: To develop a robust, accurate, and accelerated T1ρ quantification solution for submillimeter in vivo whole‐brain imaging. Methods: A multislice T1ρ mapping solution (MS‐T1ρ) was developed based on a two‐acquisition scheme using turbo spin echo with RF cycling to allow for whole‐brain coverage with 0.8‐mm in‐plane resolution. A compressed sensing–based fast imaging method, SCOPE, was used to accelerate the MS‐T1ρ acquisition time to a total scan time of 3 minutes 31 seconds. A phantom experiment was conducted to assess the accuracy of MS‐T1ρ by comparing the T1ρ value obtained using MS‐T1ρ with the reference value obtained using the standard single‐slice T1ρ mapping method. In vivo scans of 13 volunteers were acquired prospectively to validate the robustness of MS‐T1ρ. Results: In the phantom study, the T1ρ values obtained with MS‐T1ρ were in good agreement with the reference T1ρ values (R2 = 0.9991) and showed high consistency throughout all slices (coefficient of variation = 2.2 ± 2.43%). In the in vivo experiments, T1ρ maps were successfully acquired for all volunteers with no visually noticeable artifacts. There was no significant difference in T1ρ values between MS‐T1ρ acquisitions and fully sampled acquisitions for all brain tissues (p‐value >.05). In the intraclass correlation coefficient and Bland‐Altman analyses, the accelerated T1ρ measurements show moderate to good agreement to the fully sampled reference values. Conclusion: The proposed MS‐T1ρ solution allows for high‐resolution whole‐brain T1ρ mapping within 4 minutes and may provide a potential tool for investigating neural diseases. [ABSTRACT FROM AUTHOR]

In the hyperspectral image, each pixel corresponds to a small area on the Earth’s surface and represents the intrinsic characteristic of objects, which can be applied for recognition of land covers. Nevertheless, hyperspectral image processing should face some critical issues, and a small sample set problem may be the most challenging one in the research. Deep learning (DL), which has successfully been applied in many fields, has also been introduced for hyperspectral image classification. However, the large gap between the massive parameters to be tuned and limited labeled samples can lead to overfitting scenario, inevitably deteriorating the generalization ability of the DL model. In this article, a lightweight convolutional neural network (LWCNN) is proposed for hyperspectral image classification to mainly tackle the small sample set problem. Especially, spatial–spectral Schroedinger eigenmaps (SSSE) feature extraction is first adopted to obtain the joint spatial–spectral information, and the compressed dimensionality could significantly reduce the number of parameters in the following DL model. Second, a dual-scale convolution (DSC) module is carefully designed to address the SSSE features from a 1-D vector viewpoint (the number of parameters is further decreased), and the DSC procedure is successively employed to obtain the hierarchical structure description that could represent data distribution from different aspects. Subsequently, the feature vectors from all DSC layers are separately filtered by a new bichannel fusion (BCF) module, which could well encode both the intrinsic and contextual information inside DSC features. Finally, the filtered features are concatenated together and imported into a global average pooling classifier to achieve the predicted probability of each category. Experimental results on three famous hyperspectral image data sets illustrate that the developed LWCNN approach is advantageous in both the efficiency and robustness sides for hyperspectral image classification tasks and outperforms other state-of-the-art methods (both traditional-based and DL-based) with very limited labeled samples. [ABSTRACT FROM AUTHOR]

The use of carbon‐based nanomaterials as effective photocatalysts is an ideal alternative for environmental remediation. Here, (3‐aminopropyl)triethoxysilane‐functionalized carbon nanoparticles (SiCNPs) were prepared using a simple pyrolysis method with sodium citrate and urea as the precursors. The samples were characterized by X‐ray diffraction, transmission electron microscopy, ultraviolet–visible diffuse reflectance spectroscopy, X‐ray photoelectron spectroscopy, photoluminescence spectroscopy, and photo‐electrochemical measures. The obtained SiCNPs‐2.0 showed a better visible light response and more effective photocatalytic activity for degradation of methylene blue (MB) compared with pure CNPs. Under visible light irradiation, 98.8% of the MB was decomposed within 75 min when SiCNPs‐2.0 was used as the photocatalyst. The high photocatalytic activities of SiCNPs‐2.0 could be attributed to enhanced light absorption in the visible region, and improved photogenerated electron–hole separation efficiency. A possible photocatalytic mechanism for removal of MB over SiCNPs‐2.0 was proposed based on active species trapping experiments. Recycling experiments showed that SiCNPs‐2.0 had good stability during photocatalysis. This work provides a new easy method to synthesize carbon‐based nanomaterials and to catalytically degrade organic pollutants in water under visible light irradiation. [ABSTRACT FROM AUTHOR]

The rapid increase in the number of remote sensing sensors makes it possible to develop multisource feature extraction and fusion techniques to improve the classification accuracy of surface materials. It has been reported that light detection and ranging (LiDAR) data can contribute complementary information to hyperspectral images (HSIs). In this article, a multiple feature-based superpixel-level decision fusion (MFSuDF) method is proposed for HSIs and LiDAR data classification. Specifically, superpixel-guided kernel principal component analysis (KPCA) is first designed and applied to HSIs to both reduce the dimensions and compress the noise impact. Next, 2-D and 3-D Gabor filters are, respectively, employed on the KPCA-reduced HSIs and LiDAR data to obtain discriminative Gabor features, and the magnitude and phase information are both taken into account. Three different modules, including the raw data-based feature cube (concatenated KPCA-reduced HSIs and LiDAR data), the Gabor magnitude feature cube, and the Gabor phase feature cube (concatenation of the corresponding Gabor features extracted from the KPCA-reduced HSIs and LiDAR data), can be, thus, achieved. After that, random forest (RF) classifier and quadrant bit coding (QBC) are introduced to separately accomplish the classification task on the aforementioned three extracted feature cubes. Alternatively, two superpixel maps are generated by utilizing the multichannel simple noniterative clustering (SNIC) and entropy rate superpixel segmentation (ERS) algorithms on the combined HSIs and LiDAR data, which are then used to regularize the three classification maps. Finally, a weighted majority voting-based decision fusion strategy is incorporated to effectively enhance the joint use of the multisource data. The proposed approach is, thus, named MFSuDF. A series of experiments are conducted on three real-world data sets to demonstrate the effectiveness of the proposed MFSuDF approach. The experimental results show that our MFSuDF can achieve the overall accuracy of 73.64%, 93.88%, and 74.11% for Houston, Trento, and Missouri University and University of Florida (MUUFL) Gulport data sets, respectively, when there are only three samples per class for training. [ABSTRACT FROM AUTHOR]

Objective: The purpose of this article was to build a radio frequency (RF) coil system to achieve high vessel wall image quality with coverage extending from the aortic arch to the intracranial vessels. Methods: A 48-channel coil system was built and characterized at a 3 tesla (T) Magnetic Resonance Imaging (MRI) scanner (uMR 790, Shanghai United Imaging Healthcare, Shanghai, China). The coil's performance was compared with a commercially available 36-channel coil system. By human studies, signal-to-noise ratio (SNR) units were evaluated and g-factors were calculated in the transverse planes of the brain and neck regions. Results: The SNR was increased by at least 28% in the brain region and up to fourfold in the neck region. The average g-factor with the acceleration factor, R = 3, was lowered by 21% in the transverse plane of the neck region. Intracranial and carotid arterial wall images with an isotropic spatial resolution of 0.63 mm were acquired within 7.7 minutes and thoracic aorta wall images with an isotropic spatial resolution of 1.1 mm were acquired within 2.7 minutes with the 48-channel coil system. The vessel wall can be more clearly visualized with the 48-channel coil system compared with the 36-channel coil system. Conclusion: A 48-channel coil system was developed and demonstrated superior performance for vessel wall imaging at the intracranial and cervical carotid arteries compared with a commercial 36-channel coil. Significance: The 48-channel coil system is potentially useful for clinical diagnostics, especially when attempting to diagnose ischemic stroke. [ABSTRACT FROM AUTHOR]

As a typical graph-based semisupervised learning technique, the label propagation (LP) approach has gained much attention in recent years. The key to LP algorithms is the propagation capability and efficiency of the similarity matrix, which describes the similarity between two data points. Concerning hyperspectral image which often contains hundreds of thousands of pixels, the corresponding similarity matrix is particularly huge and thus the LP procedure is intractable. Fortunately, superpixel, which can effectively characterize the spatial semantic information of surface objects, provides a reasonable way to solve this problem. In this article, we propose an elaborate superpixel-based weighted LP approach, abbreviated as SuWLP, for hyperspectral image classification. First, the hyperspectral image is oversegmented by the entropy rate segmentation (ERS) method, and the internal consistency of each superpixel can be achieved. Second, a new similarity measure is designed to estimate the similarity between two superpixels, and a superpixel-based similarity matrix can be thus established. Third, after the training samples have been expanded based on the superpixel distribution, a weighted LP technique is designed to propagate the sample label at the superpixel level without any parameter tuning. Finally, the label of each superpixel maps back to the contained pixels. We compared our proposed SuWLP method with several state-of-the-art ones, and experimental results on three real hyperspectral data sets certify the effectiveness and efficiency of the superpixel-level LP strategy. [ABSTRACT FROM AUTHOR]

Ancient murals have been haunted by various problems such as color fading, surface layer turning crisp and even large-area peeling off. Virtually inpainting technologies are widely used to restore these damages. In general, when structure information are blurred or completely missing within a large region, the image inpainting would be more thorny. In this paper, we study mural image inpainting by incorporating structure information collected from the limners guidance or the line drawings, and propose a global and local feature weighted method based on structure guidance to repair the damaged murals of Yulin Grottoes and Mogao Grottoes, Gansu. Unlike traditional methods, a novel sparse representation model with elastic net regularization based on similarity-preserving overcomplete dictionary is formulated to enhance the global feature consistency, and then an estimated method of neighborhood similarity is presented to guarantee local feature consistency, finally, we apply a global feature patch and local feature patch weighted method to obtain the target patch. Experimental results on damaged murals demonstrate the proposed method outperforms state-of-the-art inpainting methods. [ABSTRACT FROM AUTHOR]

Purpose: To propose a parameter optimization framework on wave gradients of Wave‐CAIPI imaging for decreasing g‐factor penalty and reducing reconstruction artifacts. Theory and Methods: The influences of parameters on g‐factor are theoretically analyzed. The average g‐factor is chosen as a metric for parameter optimization, and then a fast calculation method is proposed to approximately and ultra‐fast calculate the average g‐factor. Based on this, a set of points in the function of the average g‐factor with respect to the wave gradient parameters is calculated, and the optimal wave gradient parameters are found according to these points. Results: In vivo human brain experiments were performed on 3T MR scanners for the comparison experiments. The results show that the proposed parameter optimization framework is able to efficiently obtain optimal wave gradient parameters, which can achieve decreased g‐factor penalty and less artifacts of reconstructions than the empirical parameters. Conclusion: The proposed parameter optimization framework is computationally efficient and can optimize the wave gradient parameters of Wave‐CAIPI imaging for better image quality than before. [ABSTRACT FROM AUTHOR]

Learning from a limited number of labeled samples (pixels) remains a key challenge in the hyperspectral image (HSI) classification. To address this issue, we propose a deep metric learning-based feature embedding model, which can meet the tasks both for same- and cross-scene HSI classifications. In the first task, when only a few labeled samples are available, we employ ideas from metric learning based on deep embedding features and make a similarity learning between pairs of samples. In this case, the proposed model can learn well to compare whether two samples belong to the same class. In another task, when an HSI image (target scene) that needs to be classified is not labeled at all, the embedding model can learn from another similar HSI image (source scene) with sufficient labeled samples and then transfer to the target model by using an unsupervised domain adaptation technique, which not only employs the adversarial approach to make the embedding features from the source and target samples indistinguishable but also encourages the target scene’s embeddings to form similar clusters with the source scene one. After the domain adaptation between the HSIs of the two scenes is finished, any traditional HSI classifier can be used. In a simple manner, the nearest neighbor (NN) algorithm is selected as the classifier for the classification tasks throughout this article. The experimental results from a series of popular HSIs demonstrate the advantages of the proposed model both in the same- and cross-scene classification tasks. [ABSTRACT FROM AUTHOR]