Browse Articles

YOLO-HPSD: A high-precision ship target detection model based on YOLOv10

Ship target detection is crucial in maritime traffic management, smart ports, autonomous ship systems, environmental monitoring, and ship scheduling. Accurate detection of various ships on the water can significantly enhance maritime traffic safety, reduce accidents, and improve the efficiency of port and waterway management. This study proposes a high-precision ship target detection algorithm based on YOLOv10, named YOLO-HPSD (High-precision Ship Target Detection). To meet the high-precision requirements in practical applications, several precision-enhancement strategies are introduced based on YOLOv10. To optimize the feature fusion process, the Iterative Attentional Feature Fusion (iAFF) is integrated with the C2F module in the backbone, resulting in the development of a novel C2F_iAFF module that utilizes a multi-scale channel attention mechanism. Meanwhile, the Mixed Local Channel Attention (MLCA) is introduced after the C2F module at the network neck, which improves the model’s ability to integrate both local and global information. Additionally, the BiFPN module is incorporated after the connection operation at the network neck, utilizing learnable weights to optimize the importance of different input features, thereby further enhancing multi-scale feature fusion. The experimental results demonstrate that YOLO-HPSD achieves excellent detection performance on the ship dataset, with an F1-score of 97.88% and mAP@0.5 of 98.86%. Compared to YOLOv10n, the F1-score, and mAP@0.5 have improved by 1.22% and 0.31%, respectively. Furthermore, the detection time for a single image is only 20.6 ms. These results indicate that the model not only ensures high detection speed but also delivers high-accuracy ship target detection. This study provides technical support for real-time ship target detection and the development of edge computing devices.

Application of improved graph convolutional network for cortical surface parcellation

Abstract Accurate cortical surface parcellation is essential for elucidating brain organizational principles, functional mechanisms, and the neural substrates underlying higher cognitive and emotional processes. However, the cortical surface is a highly folded complex geometry, and large regional variations make the analysis of surface data challenging. Current methods rely on geometric simplification, such as spherical expansion, which takes hours for spherical mapping and registration, a popular but costly process that does not take full advantage of inherent structural information. In this study, we propose an Attention-guided Deep Graph Convolutional network (ADGCN) for end-to-end parcellation on primitive cortical surface manifolds. ADGCN consists of a deep graph convolutional layer with a symmetrical U-shaped structure, which enables it to effectively transmit detailed information of the original brain map and learn the complex graph structure, help the network enhance feature extraction capability. What’s more, we introduce the Squeeze and Excitation (SE) module, which enables the network to better capture key features, suppress unimportant features, and significantly improve parcellation performance with a small amount of computation. We evaluated the model on a public dataset of 100 artificially labeled brain surfaces. Compared with other methods, the proposed network achieves Dice coefficient of 88.53% and an accuracy of 90.27%. The network can segment the cortex directly in the original domain, and has the advantages of high efficiency, simple operation and strong interpretability. This approach facilitates the investigation of cortical changes during development, aging, and disease progression, with the potential to enhance the accuracy of neurological disease diagnosis and the objectivity of treatment efficacy evaluation.

Impact of contaminant size and density on their incorporation into sea ice

Abstract Sea ice accumulates contaminants and redistributes them laterally as the ice drifts and vertically as it melts. Contaminant incorporation into sea ice must be better understood to resolve contaminant cycling and exposure to polar organisms. Here we develop an experimental method that mimics the formation of young sea ice and enables the quantification of model contaminants separately in the ice matrix and brine. Several limitations inherent in field studies are overcome using this approach. Results show that dissolved contaminants (<1 nm) and dispersed colloidal contaminants (1 nm–1 μm) follow the same behavior as sea salts. When colloids aggregate they follow a similar transport pathway to high-density particulate contaminants (>1 μm). While high-density particles are depleted in sea ice and low-density particles are enriched relative to their initial concentration in seawater, both are engulfed and can travel in wide brine channels. These results can also help to predict the incorporation of natural species in sea ice.

Prevention and inhibition of post-harvest browning in longkong pericarp using Prunus Persica resin coating during ambient storage

Background Longkong fruit is highly perishable due to rapid pericarp browning, limiting its post-harvest shelf life to less than 7 days under ambient conditions. Objectives This study evaluated the efficacy of Prunus persica resin (PPR) coatings at 1%, 2%, 3%, and 4% concentrations in preserving fruit quality over 14 days of ambient storage, with assessments conducted at two-day intervals. Methods Longkong fruits were treated with PPR coatings and stored under ambient conditions. Parameters including pericarp browning, decay rate, moisture loss, color retention, biochemical properties, and enzymatic activities were analyzed. Results PPR-coated samples exhibited superior performance compared to control samples, with higher PPR concentrations effectively mitigating pericarp browning, decay rate, and moisture loss while maintaining fruit color and biochemical properties. Control fruits became unacceptable by day 8, exhibiting a browning index exceeding 4.51 and a decay rate of 82.56%, whereas fruits treated with higher PPR concentrations (>3%) maintained lower browning indices (<3.47) and decay rates (<58.99%) by day 14. Color retention was significantly enhanced in 4% PPR-treated fruits, which maintained lightness, redness, and yellowness values, while untreated fruits exhibited severe discoloration. Moisture loss in untreated fruits reached 35.31% by day 8, whereas 4% PPR-treated fruits limited moisture loss to 20% throughout the storage period. Additionally, high concentrations of PPR coatings preserved higher total phenolic content, flavonoid levels, and antioxidant activities. PPR-coated longkong fruit effectively suppressed the activity of browning-related enzymes such as polyphenol oxidase, peroxidase, and cinnamate-4-hydroxylase, alongside reductions in membrane-degrading enzymes lipoxygenase and phospholipase D. Conclusion These findings indicate that PPR coatings, particularly at 4% concentration, effectively extend longkong fruit shelf life for up to 14 days, providing a natural, biodegradable post-harvest solution. This approach holds significant potential for reducing food waste, supporting sustainable agricultural practices, and enhancing marketability across longkong fruit supply chains, particularly benefiting small-scale farmers in tropical regions.

A novel sodium Iron silicate composite with chitosan for efficient removal of Cd(II) ions from water

Multi-omic analysis reveals a CAF-stemness-governed classification in HCC liver transplant recipients beyond the Milan criteria

A comprehensive protocol for PDMS fabrication for use in cell culture

Cells exhibit remarkable sensitivity to the mechanical properties of their surrounding matrix, particularly stiffness changes, a phenomenon known as cellular mechanotransduction. In vivo, tissues exhibit a wide range of stiffness, from kilopascals (kPa) to megapascals (MPa), which can alter with aging and disease. Traditional cell culture methods employ plastic substrates with stiffness in the gigapascal range, which does not accurately mimic the physiological conditions of most biological tissues. Therefore, employing substrates that can be engineered to span a wide range of stiffnesses, closely resembling the native tissue environment, is crucial for obtaining results that more accurately reflect cellular responses in vivo. Polydimethylsiloxane (PDMS) substrates are widely used in cell culture for their ability to simulate tissue stiffness, but their optimization presents several challenges. Fabrication requires precise control over mixing, weighing, and curing to ensure reproducible mechanical properties. Inconsistent preparation can lead to improperly cured PDMS substrates, compromising experimental outcomes. Additionally, PDMS’s inherent hydrophobicity poses challenges for cell attachment, necessitating surface modifications to enhance adhesion. Moreover, the risk of contamination during the sterilization process necessitates stringent protocols to maintain cell culture integrity. These challenges are further compounded by substrate autofluorescence which can cause difficulties when imaging cells. The aim of this study is to develop a standardized method for fabricating PDMS substrates with tuneable stiffness, ranging from kPa to MPa, suitable for diverse cell types using standard laboratory equipment. This method aims to minimize the complexity and equipment required for PDMS fabrication, ensuring reproducibility and ease of use. Achieving consistent and contaminant-free PDMS substrates will facilitate a broader adoption of these substrates in mechanobiology research and improve the relevance of in vitro models to in vivo conditions. Ultimately, contributing to a more comprehensive understanding of cellular responses to mechanical cues in health and disease.

Reward monitoring in the frontopolar cortex of macaques

Abstract Reward processing involves several prefrontal cortex areas, enabling individuals to learn from behavioral outcomes and shape decisions. However, the role of the frontopolar cortex (FPC) in these processes remains unclear due to limited single-neuron research. In this study, we recorded neural activity from the FPC of two macaques performing a fast-learning task, the object-in-place reward task, which examined how reward size affects learning. Results showed that FPC feedback monitoring activity extends to the value of specific choices. Moreover, once the association between scenes and reward had been learned, FPC neural activity before choice reflected the future animal’s behavior to stay or to switch on their previous behavioral strategy, i.e., to choose the same target or the other one. These results suggest that FPC neurons integrated information for action monitoring and later reprocessed it to decide the best behavioral strategy to adopt, determining whether to maintain or change the action plan.

Earth-abundant Ni-Zn nanocrystals for efficient alkyne semihydrogenation catalysis

Abstract The development of catalysts that are based on earth-abundant metals remains a grand challenge. Alloy nanocrystals (NCs) form an emerging class of heterogeneous catalysts, offering the promise of small, uniform catalysts with composition-control. Here, we report the synthesis of small Ni and bimetallic Ni-X (X= Zn, Ga, In) NCs for alkyne semihydrogenation catalysis. We show that Ni3Zn NCs are particularly reactive and selective under mild reaction conditions and at low loadings. While bimetallic NCs are all more selective than pure Ni NCs, Ni-Zn NCs also maintain excellent reactivity compared to Ni-Ga and Ni-In alloys. Ab-initio calculations can explain the differences in reactivity, indicating that, unlike Ga and In, Zn atoms interact with the substrates. We further show that Ni3Zn NCs are robust and tolerate a broad range of substrates, which may be linked to the favorable amine-terminated surface.

Psychometric assessment of the Persian translated version of the “medical artificial intlligence readiness scale for medical students”

Background Artificial intelligence (AI) has recently entered the medical field, but the level of readiness of medical students for it is not obvious. A tool with appropriate psychometric properties for use in different languages and for international comparison is required to measure this readiness. Medical Artificial Intelligence Readiness Scale for medical students (MAIRS-MS) is most complete scale for this purpose till now. Objectives The purpose was to evaluate the Psychometric properties of the Persian-translated version of the MAIRS-MS and verify the replication of the original factor structure in Persian. Materials and methods This study was conducted at Guilan University of Medical Sciences in 2023. Validation of the Persian translated scale (P-MAIRS-MS) was performed by determining the face, content, and construct validity and reliability, impact Score, CVI, CVR, Cronbach’s alpha, McDonald’s omega, and ICC, and performing confirmatory factor analysis (CFA). AMOS26 and SPSS26 software were used. Results The translated scale had good quantitative and qualitative face and content validity (all items had the Impact Score higher than 1.5, CVI >= 0.8 and CVR>= 0.8). CFA confirmed the appropriate fit of the four-factor model (χ2/df = 1.963, RMSEA-0.063, CFI = 0.939, GFI = 0.901). Convergent validity was suitable in the first- and second-order CFA (AVE > 0.5, CR > 0.7 CR > AVE for each factor except Ability). Cronbach’s alpha (α=0.938) and McDonald’s omega (ω= 0.938), and ICC (0.992) indicated acceptable reliability and reproducibility of the scale. Conclusion The P-MAIRS-MS demonstrated good psychometric properties and can be used for measuring and international comparing the medical students’ readiness for AI.

Study on the impact of engineering construction on environmental risk assessment in alpine canyon area: a case study of Chaya County, Xizang

Test of the physical significance of Bell non-locality

Tribo-synergism in titanium complex grease using micro and nano particles

Micro-nano additive-enhanced lubricating greases are pivotal for extreme-condition tribology, yet optimizing synergistic additive concentrations remains constrained by conventional experimental designs. This study employs a central composite design (CCD) coupled with MATLAB response surface methodology to precisely determine optimal concentrations of nano-graphite (N-G), graphene (GN), and potassium borate (PB) in titanium complex grease. Fifteen formulations were tested under progressive loads (98–598 N) via four-ball tribometry, with SEM/XPS characterizing wear mechanisms. The synergistic grease (G-MX: 0.83 wt% N-G, 0.05 wt% GN, 2.59 wt% PB) reduced the average friction coefficient by 45.3% and wear scar diameter by 23.3% versus base grease, surpassing single-additive variants. The CCD-MATLAB framework addressed sampling limitations of prior orthogonal methods, enabling optimization beyond discrete testing points. Mechanistic analysis revealed a dual lubrication regime: physically adsorbed films (soap molecules and refined PB particles) dominated at low loads, while chemically bonded tribofilms (Fe₃C, B₂O₃, TiO₂) ensured wear resistance under extreme pressures.

Association between the big five personalities and academic performance among grade 12 students at international high school in Thailand

Structural proteomics defines a sequential priming mechanism for the progesterone receptor

Construction and application of SARS-CoV-2 protein ontology (CoVPO)

The emergence of the SARS-CoV-2 virus and the resulting COVID-19 pandemic brought forth an urgent need for an in-depth molecular understanding, organization, and data integration to expedite therapeutic and preventive strategies. An essential approach to achieving this is through the development of a well-structured ontology of SARS-CoV-2 proteins. In response, this paper introduces CoVPO, a SARS-CoV-2 protein ontology that improves upon existing ontologies on protein function annotation and viral-to-viral protein interactions, highlighting their limited scope in covering all aspects of SARS-CoV-2 proteins. CoVPO extends classes from other relevant ontologies. Terms, annotations, and synonyms are added with proper definitions, clear origins, and an interaction map of viral-to-viral protein interactions is captured. We demonstrate CoVPO’s application in an information retrieval system, expanding user queries by adding related terms or concepts. This approach helps overcome issues like term mismatch and improves the retrieval of relevant documents. The feasibility and superiority of the domain ontology model are demonstrated through experiments, showing that it outperforms traditional keyword-based searches and provides grounds for further research and discussion in the SARS-CoV-2 protein domain.

Comprehensive metabolomics study identifies SN-38 organ specific toxicity in mice

SARS-CoV-2 ORF3a drives dynamic dense body formation for optimal viral infectivity

Treatment outcomes among young persons living with HIV who transitioned to adult care in southern Nigeria: A retrospective cohort study

Background In October 2019, a peer-based transition preparedness model was introduced as part of peer club activities to prepare young persons living with HIV (YLHIV) for adult care. This study compared the 12 and 24 months treatment outcomes of YLHIV who transitioned to adult care in primary, secondary and tertiary health facilities in Southern Nigeria, following the introduction of this model. Materials and methods This was a retrospective cohort study using data extracted from the medical records of YLHIV who transitioned to adult care at 25 years in 2018 and in 2021 across 155 healthcare facilities in southern Nigeria. Baseline data at transition, as well as 12 and 24 months post-transition data were extracted for comparison between those who were transitioned before (2018 cohort) and after (2021 cohort) the transition preparedness model was introduced. Logistics regression analysis was used to compare client continuity on treatment and undetectable viral load between the two groups at 12 and 24 months after transitioning to adult care. Results Most of the1,555 YLHIV who transitioned to adult care in 2018 (n = 343, 22.1%) and 2021 (n = 1,212, 77.9%) were females (91.0% in 2018 v.82.6% in 2021) and initiated ART at 20 years or older (92.7% v. 95.7%). A higher proportion of those in the 2021 cohort were continuously retained both at 12 months and 24 months post-transitioning compared to those in the 2018 cohort (12 months: 96.7% vs 80.2%, p < 0.001; 24 months: 92.7% vs 77.6% p < 0.001). Similarly, the proportion of YLHIV with undetectable viral load in the 2021 cohort was significantly higher than those in the 2018 cohort at both 12 months (96.1% vs 60.1%, p < 0.001) and 24 months (93.3% vs 80.6%, p < 0.001), respectively. Conclusion Peer-based transition preparedness model improved treatment outcomes of YLHIV who transition to adult care. Programs should implement tailored, peer-based interventions to address gaps in service delivery.

Polarity-tunable dye-sensitized optoelectronic artificial synapses for physical reservoir computing-based machine vision

Abstract Conventional machine vision systems process huge time-series data per second, presenting significant challenges for edge-device applications due to limitations in data transfer and storage. Inspired by the human visual system, artificial optoelectronic synapses replicating synaptic responses have emerged as promising solutions. However, achieving color recognition comparable to human vision remains challenging. Moreover, most optoelectronic artificial synapses rely on photocurrent-based operation, producing low current values and necessitating external circuits. This study reports a self-powered optoelectronic artificial synapse capable of distinguishing wavelengths with a resolution of 10 nm by integrating dye-sensitized solar cells. The device exhibits synaptic responses to light pulses and bipolar responses when exposed to different wavelengths. The wavelength-dependent bipolar behavior enables exceptional separation capabilities, achieving six-bit resolution with 64 distinct states and supporting multiple logic operations, including AND, OR, and XOR, within a single device. Additionally, the device leverages distinct responses to red, green, and blue light irradiation for physical reservoir computing, facilitating the classification of color-coded human motion with an accuracy of 82%. These findings advance the development of optoelectronic artificial synapses for precise, human-eye-like color discrimination.