Search Results (7,411)

The betel palm is a critical economic crop on Hainan Island. Accurate and timely maps of betel palms are fundamental for the industry’s management and ecological environment evaluation. To date, mapping the spatial distribution of betel palms across a large regional scale remains a significant challenge. In this study, we propose an integrated framework that combines knowledge-based and machine learning approaches to produce a map of betel palms at 10 m spatial resolution based on Sentinel-1/2 data and Google Earth Engine (GEE) for 2023 on Hainan Island, which accounts for 95% of betel nut acreage in China. The forest map was initially delineated based on signature information and the Green Normalized Difference Vegetation Index (GNDVI) acquired from Sentinel-1 and Sentinel-2 data, respectively. Subsequently, patches of betel palms were extracted from the forest map using a random forest classifier and feature selection method via logistic regression (LR). The resultant 10 m betel palm map achieved user’s, producer’s, and overall accuracy of 86.89%, 88.81%, and 97.51%, respectively. According to the betel palm map in 2023, the total planted area was 189,805 hectares (ha), exhibiting high consistency with statistical data (R² = 0.74). The spatial distribution was primarily concentrated in eastern Hainan, reflecting favorable climatic and topographic conditions. The results demonstrate the significant potential of Sentinel-1/2 data for identifying betel palms in complex tropical regions characterized by diverse land cover types, fragmented cultivated land, and frequent cloud and rain interference. This study provides a reference framework for mapping tropical crops, and the findings are crucial for tropical agricultural management and optimization. Full article

Background: Ultrasonography is frequently used preoperatively to assess thyroid nodules. Hypoechogenicity, microcalcifications, high vascularity, or irregular tumor shape suggest malignancy. Methods: This is a retrospective analysis of 724 patients from 2008 to 2024 who underwent surgery for TC. Preoperative data, ultrasonographic findings, and histopathological results were collected. Ultrasonographic features indicative of possible malignancy included hypoechogenicity, microcalcifications, high vascularity, and irregular tumor shape. These were correlated with histopathologically seen extrathyroidal extension, capsular and vascular invasion, and lymph node metastasis. Results: A statistically significant association was seen for each of the evaluated ultrasonographic features (p < 0.05). More advanced TC had a greater number of suspicious ultrasonographic features averaging 3.05 to 3.12. Microcalcifications, high vascularity, and irregular tumor shape showed a strong correlation (r > 0.7) with all histopathological features. Hypoechogenicity had a strong correlation with lymph node metastasis and a moderate correlation (r = 0.5–0.7) with other features. Conclusions: Ultrasonographic features predict the likelihood of histopathological extrathyroidal extension, capsular and vascular invasion, and lymph node metastasis. Full article

Background/Objectives: Attention Deficit Hyperactivity Disorder (ADHD) is a common neurodevelopmental disorder frequently associated with other neuropsychiatric conditions, characterized by high clinical heterogeneity and a complex genetic background. Recent studies suggest that copy number variations (CNVs) may contribute to ADHD susceptibility, particularly when involving genes related to brain development, attention regulation, and impulse control. This study investigated the association between CNVs and ADHD phenotype by identifying patients with and without potential pathogenic CNVs. Methods: We evaluated 152 well-characterized ADHD pediatric patients through comprehensive clinical assessments, including dysmorphic features, brain MRI, EEG patterns, and cognitive testing. CNVs were identified using array Comparative Genomic Hybridization (array-CGH). Participants were classified as carrying potentially causative CNVs (PC-CNVs), non-causative CNVs (NC-CNVs), or without CNVs (W-CNVs) and statistically compared across clinical and neurodevelopmental measures. Results: CNVs were identified in 81 participants (53%), comprising 13 with PC-CNVs (8.5%) and 68 with NC-CNVs (44.7%). ADHD symptoms were pronounced across all groups, but PC-CNVs showed a higher burden of comorbidities, suggesting a stronger genetic contribution to ADHD complexity. Significant differences were observed in oppositional behavior, inattentive symptoms, brain MRI findings, and developmental language anomalies. Several CNVs involved genes previously implicated in neurodevelopmental disorders, supporting a potential genetic contribution to the clinical complexity of ADHD. Conclusions: This exploratory study supports the role of CNVs in ADHD susceptibility and highlights the value of genetic screening for understanding clinical variability. Larger studies are needed to clarify genotype–phenotype correlations in ADHD and to guide personalized clinical management. Full article

Background/Objectives: Extended-Spectrum Beta-Lactamase-producing Escherichia coli (ESBL-E. coli) are widely circulating in livestock of low- and middle-income countries. However, the drivers of their circulation remain largely unknown. Small-scale farms in Peru exhibit an unusually high prevalence of fecal carriage of ESBL-E. coli in their livestock. The objective of this study was to compare the prevalence of ESBL-E. coli fecal carriage in dairy cows, pigs, and poultry in the Lima and Ica regions of Peru and to identify the drivers associated with the observed prevalence at the farm level. Methods: We compared the prevalence of fecal carriage of ESBL-E. coli isolated from dairy cattle (N = 244 animals; 25 farms), pigs (N = 261; 25), and laying hens (N = 255; 10). We also administrated questionnaires to 59 farmers regarding their socioeconomic status, husbandry practices, animal diseases, and antibiotic use. Results: All but one of the 60 farms sampled had at least one animal carrying ESBL-E. coli. A statistically higher prevalence of ESBL-E. coli was estimated in dairy cows (75%) and pigs (61%) from Lima compared to laying hens from Ica (34%). Our statistical analyses (Poisson generalized linear models) using two variable selection approaches revealed that the prevalence of ESBL-E. coli was lower in farms raising laying hens, when farmers oversaw both animal husbandry and healthcare, and in farms with a higher number of gastrointestinal outbreaks in the last semester. Socio-economic features of farmers and self-reporting antibiotic use varied across farms (i.e., highest antibiotic use over the last semester was reported among pig farmers (96%), followed by laying hen (70%) and dairy cattle farmers (50%)), but these factors were not associated with the prevalence of ESBL-E. coli. Conclusions: Despite a relatively low number of farms sampled, our findings suggest that the widespread circulation of ESBL-E. coli among livestock in Peru could be mainly associated with unknown species-specific drivers, independently of the socioeconomic status of farmers and antibiotic use. Therefore, our study calls for future research to identify the specific drivers promoting the high prevalence of ESBL-E. coli among cattle and pigs in Peru. Full article

With the development of intelligent connected vehicle and artificial intelligence technologies, mixed traffic scenarios where autonomous and human-driven vehicles coexist are becoming increasingly common. Autonomous vehicles need to accurately predict the lane change behavior of preceding vehicles to ensure safety. However, lane change behavior of human-driven vehicles is influenced by both environmental factors and driver preferences, which increases its uncertainty and makes prediction more difficult. To address this challenge, this paper focuses on the mining of driving preferences and the prediction of lane change behavior. We clarify the definition of driving preference and its relationship with driving style and construct a representation of driving operations based on vehicle dynamics parameters and statistical features. A preference feature extractor based on the SimCLR contrastive learning framework is designed to capture high-dimensional driving preference features through unsupervised learning, effectively distinguishing between aggressive, normal, and conservative driving styles. Furthermore, a dual-branch lane change prediction model is proposed, which fuses explicit temporal features of vehicle states with implicit driving preference features, enabling efficient integration of multi-source information. Experimental results on the HighD dataset show that the proposed model significantly outperforms traditional models such as Transformer and LSTM in lane change prediction accuracy, providing technical support for improving the safety and human-likeness of autonomous driving decision-making. Full article

Functional near-infrared spectroscopy (fNIRS) offers a safe and portable signal source for brain–computer interface (BCI) applications, particularly in motor imagery (MI) decoding. However, its low sampling rate and hemodynamic delay pose challenges for temporal modeling and dynamic brain network analysis. To address these limitations in temporal dynamics, static graph modeling, and feature fusion interpretability, we propose TopoTempNet, an innovative topology-enhanced temporal network for biomedical signal decoding. TopoTempNet integrates multi-level graph features with temporal modeling through three key innovations: (1) multi-level topological feature construction using local and global functional connectivity metrics (e.g., connection strength, density, global efficiency); (2) a graph-modulated attention mechanism combining Transformer and Bi-LSTM to dynamically model key connections; and (3) a multimodal fusion strategy uniting raw signals, graph structures, and temporal representations into a high-dimensional discriminative space. Evaluated on three public fNIRS datasets (MA, WG, UFFT), TopoTempNet achieves superior accuracy (up to 90.04% ± 3.53%) and Kappa scores compared to state-of-the-art models. The ROC curves and t-SNE visualizations confirm its excellent feature discrimination and structural clarity. Furthermore, the statistical analysis of graph features reveals the model’s ability to capture task-specific functional connectivity patterns, enhancing the interpretability of decoding outcomes. TopoTempNet provides a novel pathway for building interpretable and high-performance BCI systems based on fNIRS. Full article

Ship acoustic signal classification is essential for vessel identification, underwater navigation, and maritime security. Traditional methods struggle with the non-stationary nature and noise of ship acoustic signals, reducing classification accuracy. To address these challenges, we propose an automated pipeline that integrates Empirical Mode Decomposition (EMD), adaptive wavelet filtering, feature selection, and a Bayesian-optimized Random Forest classifier. The framework begins with EMD-based decomposition, where the most informative Intrinsic Mode Functions (IMFs) are selected using Signal-to-Noise Ratio (SNR) analysis. Wavelet filtering is applied to reduce noise, with optimal wavelet parameters determined via SNR and Stein’s Unbiased Risk Estimate (SURE) criteria. Features extracted from statistical, frequency domain (FFT), and time–frequency (wavelet) metrics are ranked, and the top 11 most important features are selected for classification. A Bayesian-optimized Random Forest classifier is trained using the extracted features, ensuring optimal hyperparameter selection and reducing computational complexity. The classification results are further enhanced using a majority voting strategy, improving the accuracy of the final object identification. The proposed approach demonstrates high accuracy, improved noise suppression, and robust classification performance. The methodology is scalable, computationally efficient, and suitable for real-time maritime applications. Full article

This study investigates the use of artificial neural networks (ANNs) and deep neural networks (DNNs) for estimating photovoltaic (PV) energy output, with a particular focus on hyperparameter tuning. Supervised regression for photovoltaic (PV) direct current power prediction was conducted using only sensor-based inputs (PanelTemp, Irradiance, AmbientTemp, Humidity), together with physically motivated-derived features (ΔT, IrradianceEff, IrradianceSq, Irradiance × ΔT). Samples acquired under very low irradiance (<50 W m⁻²) were excluded. Predictors were standardized with training-set statistics (z-score), and the target variable was modeled in log space to stabilize variance. A shallow artificial neural network (ANN; single hidden layer, widths {4–32}) was compared with deeper multilayer perceptrons (DNN; stacks {16 8}, {32 16}, {64 32}, {128 64}, {128 64 32}). Hyperparameters were selected with a grid search using validation mean squared error in log space with early stopping; Bayesian optimization was additionally applied to the ANN. Final models were retrained and evaluated on a held-out test set after inverse transformation to watts. Test performance was obtained as MSE, RMSE, MAE, R², and MAPE for the ANN and DNN. Hence, superiority in absolute/squared error and explained variance was exhibited by the ANN, whereas lower relative error was achieved by the DNN with a marginal MAE advantage. Ablation studies showed that moderate depth can be beneficial (e.g., two-layer variants), and a simple bootstrap ensemble improved robustness. In summary, the ANN demonstrated superior performance in terms of absolute-error accuracy, whereas the DNN exhibited better consistency with relative-error accuracy. Full article

Deep learning has significantly advanced medical image analysis by enabling accurate, automated diagnosis across diverse clinical tasks such as lesion classification and disease detection. However, the practical deployment of these systems is still hindered by two major challenges: the limited availability of expert-annotated data and substantial domain shifts caused by variations in imaging devices, acquisition protocols, and patient populations. Although recent semi-supervised domain generalization (SSDG) approaches attempt to address these challenges, they often suffer from two key limitations: (i) reliance on computationally expensive uncertainty modeling techniques such as Monte Carlo dropout, and (ii) inflexible shared-head classifiers that fail to capture domain-specific variability across heterogeneous imaging styles. To overcome these limitations, we propose MultiStyle-SSDG, a unified semi-supervised domain generalization framework designed to improve model generalization in low-label scenarios. Our method introduces a multi-style ensemble pseudo-labeling strategy guided by entropy-based filtering, incorporates prototype-based conformity and semantic alignment to regularize the feature space, and employs a domain-specific multi-head classifier fused through attention-weighted prediction. Additionally, we introduce a dual-level neural-style transfer pipeline that simulates realistic domain shifts while preserving diagnostic semantics. We validated our framework on the ISIC2019 skin lesion classification benchmark using 5% and 10% labeled data. MultiStyle-SSDG consistently outperformed recent state-of-the-art methods such as FixMatch, StyleMatch, and UPLM, achieving statistically significant improvements in classification accuracy under simulated domain shifts including style, background, and corruption. Specifically, our method achieved 78.6% accuracy with 5% labeled data and 80.3% with 10% labeled data on ISIC2019, surpassing FixMatch by 4.9–5.3 percentage points and UPLM by 2.1–2.4 points. Ablation studies further confirmed the individual contributions of each component, and t-SNE visualizations illustrate enhanced intra-class compactness and cross-domain feature consistency. These results demonstrate that our style-aware, modular framework offers a robust and scalable solution for generalizable computer-aided diagnosis in real-world medical imaging settings. Full article

With the continuous growth of network threats, intrusion detection systems need to have robustness and adaptability to effectively identify malicious behaviors. However, factors such as noise interference, class imbalance, and complex attack pattern recognition have posed significant challenges to traditional systems. To address these issues, this paper proposes a dynamic preprocessing-enhanced DyP-CNX framework. The framework designs a sliding window dynamic interquartile range (IQR) standardization mechanism to effectively suppress the temporal non-stationarity interference of network traffic. It also combines a random undersampling strategy to mitigate the class imbalance problem. The model architecture adopts a CNN-XGBoost collaborative learning framework, combining a dual-channel convolutional neural network (CNN) and two-stage extreme gradient boosting (XGBoost) to integrate the original statistical features and deep semantic features. On the UNSW-NB15 and CSE-CIC-IDS2018 datasets, the method achieved F1 values of 91.57% and 99.34%, respectively. The experimental results show that the DyP-CNX method has the potential to handle the feature drift and pattern confusion problems in complex network environments, providing a new technical solution for adaptive intrusion detection systems. Full article

Bank transaction fraud detection has emerged as an important area of research in the economic sector, driven by the developing sophistication of fraudulent activities and the considerable economic losses they entail. This paper reviews numerous methodologies and technologies employed in the real-time identification and mitigation of fraudulent transactions, including traditional statistical techniques, machine learning algorithms and advanced artificial intelligence strategies. It enhances the need to combine anomaly detection structures with behavioral analytics to enhance detection accuracy while addressing challenges like data privacy, the need to balance false positives and negatives and the need for adaptive systems. By evaluating the most recent developments and case studies, this study provides a comprehensive assessment of what is happening in bank transaction fraud detection and presents future directions for enhancing safety features. Full article

As a high-value product used as food, medicine, or cosmetics, honey is particularly susceptible to adulteration. Therefore, it must be regularly tested at various stages of its life cycle to ensure its quality and authenticity, especially its botanical origin. Sensory quality features play a huge role in creating the quality of products, but also in determining their authenticity. Sensory analysis helps determine the honey’s overall quality based on attributes like color, aroma, taste, and texture. Sensory evaluation of honey can reveal issues like crystallization, off-flavors, or off-odors that might indicate adulteration or spoilage. The aim of our work was therefore sensory quality assessment of 84 honey samples in order to create sensory profiles for the varietal classification of honeys. In order to obtain information on the differences in sensory features and their classification based on the assessment of honey quality descriptors, a discriminant analysis was carried out. Then, an assessment was carried out to check whether the compared varieties differ in terms of the value of the sensory feature parameter assessment. As a result, a statistical tool was constructed (canonical discriminant functions, distinguishing/classifying the varieties of honeys tested). These models will ensure the repeatability of results in the classification of sensory profiles of varietal honeys on the example of Polish honey varieties. The results indicate that the sensory analysis is a good analytical tool to differentiate honey types. The findings of this study can be applied by honey producers, suppliers, and customers to differentiate and determine honey varieties according to their sensorial attributes. Full article

Voice pathology diagnosis is essential for the timely detection and management of voice disorders, which can significantly impact an individual’s quality of life. This study employed logistic regression to evaluate the predictive power of variables that include age, severity, loudness, breathiness, pitch, roughness, strain, and gender on a binary diagnosis outcome (Yes/No). The analysis was performed on the Perceptual Voice Qualities Database (PVQD), a comprehensive dataset containing voice samples with perceptual ratings. Two widely used voice quality assessment tools, CAPE-V (Consensus Auditory-Perceptual Evaluation of Voice) and GRBAS (Grade, Roughness, Breathiness, Asthenia, Strain), were employed to annotate voice qualities, ensuring systematic and clinically relevant perceptual evaluations. The model revealed that age (odds ratio: 1.033, p < 0.001), loudness (odds ratio: 1.071, p = 0.005), and gender (male) (odds ratio: 1.904, p = 0.043) were statistically significant predictors of voice pathology. In contrast, severity and voice quality-related features like breathiness, pitch, roughness, and strain did not show statistical significance, suggesting their limited predictive contributions within this model. While the results provide valuable insights, the study underscores notable limitations of logistic regression. The model assumes a linear relationship between the independent variables and the log odds of the outcome, which restricts its ability to capture complex, non-linear patterns within the data. Additionally, logistic regression does not inherently account for interactions between predictors or feature dependencies, potentially limiting its performance in more intricate datasets. Furthermore, a fixed classification threshold (0.5) may lead to misclassification, particularly in datasets with imbalanced classes or skewed predictor distributions. These findings highlight that although logistic regression serves as a useful tool for identifying significant predictors, its results are dataset-dependent and cannot be generalized across diverse populations. Future research should validate these findings using heterogeneous datasets and employ advanced machine learning techniques to address the limitations of logistic regression. Integrating non-linear models or feature interaction analyses may enhance diagnostic accuracy, ensuring more reliable and robust voice pathology predictions. Full article

Coral reef ecosystems are home to diverse marine flora and fauna. However, these ecosystems are threatened by an array of environmental and anthropogenic factors. Here, we investigated coral reef diversity, structure, and health status, and identified their key environmental drivers. Coral reef data were collected from Koh Seh Island, located inside the Marine Fisheries Management Area in the Kep archipelago. We found that the reef cover largely comprised live corals (64%, mainly Porites and Tubinaria species), followed by Zoanthids (15%) and sand/rubble (15%). Based on Ward’s hierarchical cluster analysis, coral communities were grouped into three zones: East, South, and West zones. Coral diversity was slightly higher in the East zone, though not statistically significant. Zone East showed a positive association with sediment loads and water temperature. Elevated levels of salinity, dissolved oxygen, and pH were characteristic of the East and South zones, whereas the West zone was distinguished by deeper water conditions. We also found that Favites was the key indicator for coral communities in the East zone, which features shallow, high-DO, high-pH waters with more sediments, strong currents, and significant human activities like fishing and transportation. Goniastrea species were abundant in the South and East zones, making it the indicator taxon, while the West zone had no indicator, suggesting that coral species are sparse in this zone. Interestingly, only a few dead corals were found, and no signs of diseases were detected around the Koh Seh coral reefs. This may reflect the effectiveness of joint protection efforts by Marine Conservation Cambodia and the Marine Fisheries Department in Kep province. Overall, our study provides a valuable baseline for assessing future changes in benthic reefs and coral communities on Koh Seh island, throughout the Kep Archipelago and its surrounding areas. Full article

Background/Objectives: Cardiac amyloidosis (CA) and cardiac sarcoidosis (CS) are two distinct infiltrative cardiomyopathies that can present with overlapping clinical features, including heart failure and arrhythmias. However, they arise from fundamentally different pathophysiological mechanisms: amyloid protein deposition in CA versus granulomatous inflammation in CS. These differing pathophysiologies result in divergent imaging patterns, clinical trajectories, and treatment strategies. This study aims to compare the clinical presentations, imaging characteristics, and outcomes of patients with CA and CS to identify key differentiating factors that can improve diagnostic precision and guide therapy. Methods: This single-center, retrospective, cross-sectional study analyzed electronic medical records of patients diagnosed with CA (limited to transthyretin CA) or CS at Mount Sinai Morningside system from January 2017 until October 2023. Patients were identified using diagnostic codes and confirmed by histology or disease-specific imaging criteria. Clinical data, transthoracic echocardiography (TTE), cardiac magnetic resonance (CMR) imaging, pyrophosphate scintigraphy (PYP), and fluorodeoxyglucose positron emission tomography (FDG-PET) findings were collected. Statistical comparisons between groups were performed using chi-square tests and independent t-tests, with p < 0.05 considered statistically significant. Results: A total of 16,834 patients were screened and 216 patients were included in the analysis (125 CA, 92 CS). CA patients were older (78.2 vs. 62.0 years, p = 0.01), had greater interventricular septal thickness (1.57 vs. 1.10 cm, p = 0.01), and exhibited diffuse late gadolinium enhancement (LGE) and elevated extracellular volume (ECV) on CMR. CS patients had higher rates of ventricular tachycardia (53.3% vs. 10.7%, p = 0.01), increased myocardial fluorodeoxyglucose (FDG) uptake on positron emission tomography (PET) (90%), and more frequent implantable cardioverter-defibrillator (ICD) placement (66.3% vs. 13.0%, p = 0.01). Conclusions: CA and CS demonstrate distinct imaging profiles, arrhythmic risks, and treatment patterns. Early differentiation using advanced imaging is crucial for implementing disease-modifying therapies in CA and for immunosuppression and ICD implantation in CS, thereby improving patient outcomes. Full article