Search Results (3,118)

Breast cancer remains one of the leading causes of cancer-related mortality among women worldwide, where early and accurate diagnosis is critical for effective treatment. Although recent advances in deep learning have enabled automated radiology report generation from breast ultrasound images, most existing approaches function as black-box systems, limiting clinical trust and interpretability. This study proposes a trustworthy and explainable framework for automated breast ultrasound report generation that combines Vision-Language Modelling (VLM) with multi-level Explainable Artificial Intelligence (XAI). The proposed architecture integrates a Swin Transformer for visual feature extraction, BioBERT/ClinicalBERT for clinical text representation, and a GPT-2-based decoder for report generation through a dual cross-attention fusion mechanism. The framework is evaluated on benchmark breast ultrasound datasets paired with expert-annotated radiology reports using standard natural language generation metrics, including BLEU, ROUGE-L, METEOR, and CIDEr. Experimental results demonstrate that the multimodal architecture significantly improves report quality, clinical consistency, and semantic accuracy compared with conventional image-only and single-modal baselines. To address transparency and trustworthiness, the framework provides dual-level explanations through Grad-CAM visual heatmaps and LIME/SHAP-based token attribution analysis, enabling clinicians to understand both image regions and textual features influencing generated reports. Qualitative assessment further indicates strong alignment between model explanations and radiologist-identified diagnostic findings. Full article

In the context of the large-scale integration of high-proportion renewable energy into power grids, Power Conversion Systems (PCS) and Static Var Generator (SVG), as the core power electronic devices for ensuring grid frequency stability and power quality, demand high-level safety and reliability. To tackle challenges such as the wide range of defect sizes in PCS and SVG, the low recognition accuracy for microscopic fuzzy defects, and complex background interference, this paper presents a lightweight and high-precision defect recognition model (RD-YOLO) based on the latest YOLOv11 benchmark. First, an improved Mosaic algorithm is introduced. This algorithm utilizes conflict relationship tables to preserve physical context semantics during traditional non-discriminative data augmentation. Second, to surmount the limitations of scale-aware feature extraction, the YOLOv11 is re-engineered within the backbone network by integrating a Res2Net multi-scale cascaded mechanism. This enhances the network’s capacity to capture both fine-grained defect features and large-scale defect boundaries. Third, Focal Loss is employed for difficult sample detection. Nonlinear gradient modulation is utilized to guide the model to focus on ambiguous defect edges. Finally, the Soft-NMS post-processing strategy significantly enhances the regression accuracy in densely corroded regions. Experimental validation on a self-developed dataset consisting of 8500 high-resolution PCS and SVG defect images reveals that the enhanced RD-YOLO attains an average precision of 89.6% and a frame inference rate of 98 FPS (in RTX 3090), offering robust technical support for intelligent visual maintenance in renewable energy facilities. Full article

Quantitative operational data mining and control strategies are essential for energy-saving regulation in intelligent District Heating Systems (DHSs). However, due to complex industrial environments and heterogeneous sensor networks, operational heating data often face a “zero ground-truth labels” bottleneck, rendering conventional residual-based unsupervised quality control algorithms ineffective without calibration baselines. To resolve this, this paper proposes an unsupervised multivariate data quality assessment framework constrained jointly by physical and statistical topologies. The framework evaluates time-series data streams across five dimensions: completeness, typicality, consistency, uniqueness, and timeliness. At the statistical topology level, the Mahalanobis distance identifies the spatiotemporal distribution center of multivariate variables, replacing traditional accuracy metrics with statistical typicality to enable self-consistent quantification without ground truth. At the physical topology level, coupled logical relations between primary and secondary heating networks are extracted as rigid first-principles constraints. An information entropy weight method then adaptively determines indicator weights to eliminate subjective biases. Full-sample validation was conducted using real-world SCADA data across a complete heating season from a regional network zone (46 heat exchange stations). The network-wide average data quality score reached 0.905, confirming overall control-loop input readiness, though specific stations exhibited cascading degradation from localized physical faults. This framework systematically reveals data quality heterogeneity in complex DHS and provides a generalizable theoretical baseline for Industrial Internet of Things (IIoT) data cleansing under zero-ground-truth conditions. Full article

Emissivity is a critical parameter in infrared temperature measurement and varies significantly among different materials. Infrared thermography has been widely used for the inspection of substation equipment. However, substations contain a large number of devices with complex structures, making it impractical to assign a separate emissivity value to each device or component. This limitation can significantly affect temperature measurement accuracy. To address this issue, this paper proposes an intelligent multi-emissivity temperature correction method for infrared images of substation equipment. First, a temperature–emissivity correction function is established. Then, a total of 2189 infrared images of substation equipment are collected, and the main equipment components are annotated at the pixel level. Subsequently, an equipment component segmentation model based on DeepLabv3+ is trained. Finally, different emissivity values are assigned to different component regions for temperature correction, and corrected infrared pseudo-color images are regenerated. In the experiment, the temperature values before and after correction are compared with thermocouple measurements. In the present validation experiment, the average deviation between the corrected infrared temperature and the thermocouple measurement was reduced by 79.2% compared with that before correction. Full article

Cyanobacterial blooms occur frequently in lakes worldwide, disrupting aquatic ecosystem balance and directly threatening drinking water safety and fisheries production. Establishing a reliable bloom early-warning system has therefore become an urgent priority for lake water management. This study adopts a structured narrative review approach to synthesize the major early-warning methods, including indicator threshold methods, statistical and empirical models, mechanistic models, machine learning, and remote sensing monitoring. These methods are compared in terms of their fundamental principles, data requirements, predictive capabilities, applicability, interpretability, and computational and maintenance requirements. Emerging trends in multi-source data fusion, multi-model integration, and the development of integrated early-warning systems are also summarized. The findings indicate that each method has distinct strengths and limitations with respect to forecasting lead time, spatial coverage, process interpretation, and operational costs, and that no single method can simultaneously meet the requirements of multiscale bloom monitoring and forecasting. Integrating multi-source data from in situ monitoring, remote sensing observations, and meteorological and hydrological measurements, while coordinating statistical models, mechanistic models, and artificial intelligence algorithms according to specific forecasting objectives, represents an important pathway for improving the robustness and operational applicability of early-warning systems. Given the pronounced seasonal ice cover, substantial hydrological variability, limited monitoring data, and marked regional heterogeneity of some cold and arid region lakes, future research should strengthen high-frequency monitoring during critical periods, promote coordination between remote sensing and in situ observations, and conduct local calibration of early-warning thresholds and model parameters. Season-specific models should also be developed to account for environmental differences among ice-covered, ice-off transition, and open-water periods. Overall, early warning of cyanobacterial blooms in lakes is evolving from the application of individual methods toward the integration of multi-source monitoring, multi-model integration, and decision support, thereby providing a reference for bloom risk prevention and water environment management across different types of lakes. Full article

Landslides and collapses seriously threaten the safety of residents in Xiangtan County, Hunan Province, southern China. Local human engineering activities, dominated by slope cutting for housing construction, create steep artificial free faces, greatly weakening slope stability and becoming the key anthropogenic driver aggravating geological hazard risks, while targeted quantitative susceptibility assessment for residential slope units is still lacking for precise disaster prevention. To fill this gap and support proactive geohazard mitigation, this study selects Xiangtan County as the research object. A total of 166 landslide and collapse hazard points and 869 moderately and highly susceptible residential slope units were collected, and 12 conditioning factors, such as relative height difference, average slope and engineering rock mass group, were subsequently screened. Three hybrid intelligence models, namely PSO-BP, PSO-RF and PSO-SVM, were established to map residential slope unit susceptibility across the whole study area. The ROC-AUC and Kappa coefficient were adopted to quantify and compare the predictive performance of each model, and the Jenks natural breakpoint method combined with field survey data was used to classify all 7257 residential slope units into three susceptibility grades. The evaluation results show that the PSO-RF model performs best with an AUC of 0.913 and a Kappa coefficient of 0.64, representing strong predictive reliability. Under this optimal model, moderate-susceptibility units account for 12.94% (939 units) and high-susceptibility units account for 0.69% (50 units), both of which are concentrated in the southwest, southeast and partially northern zones of the county where intensive human slope-cutting activities prevail. This research provides a feasible technical framework for identifying high-risk residential slopes and delivers clear data support for local geohazard risk control and disaster reduction. In summary, the PSO-RF hybrid model is proven suitable for fine-scale susceptibility assessment of residential slopes in hilly regions with frequent small-sized slope failures. Full article

Background/Objectives: Acute coronary syndrome (ACS) remains one of the leading causes of cardiovascular mortality worldwide. Rapid diagnosis and timely initiation of reperfusion therapy are crucial for improving patient outcomes. The aim of this narrative review was to summarize current evidence on prehospital 12-lead ECG teletransmission in the diagnosis and management of ACS and to evaluate its impact on systems of care and clinical outcomes. Methods: A narrative literature review was conducted using the PubMed database. The literature search was completed on 15 May 2026 and included publications from 1998 onwards. The review included studies evaluating prehospital ECG teletransmission, ST elevation myocardial infarction (STEMI) networks, interpretation models, treatment delays, and organizational and logistics aspects of ECG teletransmission systems. Current guidelines from the European Society of Cardiology and the 2025 ACC/AHA/ACEP/NAEMSP/SCAI guidelines were also included. Results: Available evidence indicates that prehospital ECG teletransmission enables earlier STEMI diagnosis, direct referral to primary coronary intervention (PCI)-capable centers, and earlier activation of catheterization laboratories. Multiple observational studies and meta-analyses have demonstrated reductions in first medical contact-to-device and door-to-device times associated with teletransmission systems. Limitations include interpretation variability, false-positive catheterization laboratory activations, and infrastructure-related barriers. Conclusions: ECG teletransmission is an essential component of contemporary regional STEMI care systems and contributes to improved organization of care and shorter reperfusion delays. Further development and prospective validation of artificial intelligence-assisted ECG interpretation and integrated telemedicine systems may enhance diagnostic accuracy and the effectiveness of prehospital cardiac care. Full article

Sustainable Internet of Things (IoT) deployments require network planning strategies that explicitly account for environmental impact and not only traditional performance and energy metrics. This work analyzes how input representations affect a learning-based framework for carbon footprint (CF)-aware gateway placement in LoRa multi-hop IoT networks. Building on a previous CF model and an integer linear programming dataset, a multilayer perceptron is retrained using different input encodings: end-device coordinates, traffic-based weights, spatial sampling regions (SSRs), and a global CF estimate. Their contributions are evaluated through Shapley additive explanations (SHAP)-based explainability analysis, ablation studies, and sensitivity analysis. Results show that the CF estimate is the most influential input, acting as a global guidance signal that drives large gateway relocations. The combination of raw coordinates and SSR-based spatial summaries achieves the best performance by capturing both fine spatial detail and collective relay opportunities, while traffic-based weights mainly contribute through aggregate effects. These findings provide practical guidelines for designing CF-aware learning pipelines and offer insights to support future research on environmentally aware artificial intelligence for IoT network planning. Full article

Introduction: A pathological complete response (pCR) to neoadjuvant therapy is what makes organ preservation possible in rectal cancer, so a reliable pre-treatment estimate of the likelihood of pCR would influence management. Prediction research has focused on imaging and artificial intelligence, leaving the clinical models that any unit should be able to apply largely untested beyond their original cohorts. We validated two such models eiyh our own patients, examined whether local refitting helped, and measured how often pCR occurs across the regimens we use. Methods: This was a single-centre cohort of consecutive rectal cancer patients treated with neoadjuvant therapy and resection between 2016 and 2025, drawn from a prospectively maintained registry. pCR was defined as ypT0N0 (no residual invasive tumour in the rectal wall or in the regional lymph nodes on the resection specimen). Two published clinical models (Wang and colleagues; Tan and colleagues) were applied using their original coefficients and assessed for discrimination and calibration; the better-performing model was then adjusted to the local data and its clinical utility was examined via decision-curve analysis. A local six-predictor model was developed and internally validated via bootstrappaing. Reporting followed TRIPOD and STROBE guidelines. Results: Of 425 treated and resected patients, 85 achieved pCR (20.0%). The pCR yield differed markedly by regimen, from 27.9% with total neoadjuvant therapy and 22.8% with chemoradiotherapy to 10.3% with short-course radiotherapy. The model of Wang and colleagues separated responders from non-responders only modestly (c-statistic 0.62, 95% confidence interval 0.54–0.69), and its predictors’ effects proved nearly twice as strong in our patients (slope 0.57); the model of Tan and colleagues performed worse (c-statistic 0.58). A locally refitted model performed no better (optimism-corrected c-statistic 0.60). Adjusting the best model to the local data brought its risk estimates closer to observed responses but did not sharpen discrimination, and decision-curve analysis showed only a small net benefit over default strategies. Conclusions: Two established clinical models for pCR could only be modestly applied to a contemporary European cohort; the risks they predicted overshot what we observed, and local refitting changed nothing—a low ceiling that lies in the predictors, not their coefficients. A clinical model alone should not be used to identify who is offered organ preservation. Full article

Tea is a high-value agricultural sector whose sustainable development is closely tied to the livelihoods of numerous smallholders. Persistent rural labor outmigration has made it increasingly difficult for smallholder-dominated tea production to sustain a balance between quantity expansion and quality improvement. While experimental studies show that digital–intelligent technologies can improve tea yield and quality, whether their real-world diffusion contributes to a sustainable quantity–quality balance remains underexplored. This study proposes the quantity–quality equilibrium level of tea production (QQEL_TP) to measure this balance and constructs a regional digital–intelligent transformation (RDIT) index to capture the penetration of digital–intelligent technologies into local industries. Using panel data from China’s major tea-producing provinces from 2012 to 2023, this study applies a two-way fixed-effects model to examine the RDIT–QQEL_TP relationship. Results show that RDIT significantly improves QQEL_TP. Further analysis suggests that e-commerce development is a potential channel, while heterogeneity analysis indicates that the positive effect is more likely to materialize in provinces with a stronger digital–intelligent foundation, an existing Taobao village foundation, and a smaller tea industry scale. This study provides industry-level evidence on the role of digital–intelligent transformation in promoting agricultural sustainability, and offers policy implications for addressing production-side quantity–quality imbalances in other smallholder-dominated industries. Full article

This study aimed to develop and evaluate a deep learning diagnostic model integrating a convolutional autoencoder (CAE) and ResNet18 for the early and accurate diagnosis of developmental dysplasia of the hip (DDH) in children, addressing the subjectivity of traditional methods. This study began with the construction of a CAE network. By introducing a multi-layer compression structure and a Dropout layer, the network was forced to learn the low-dimensional and robust feature representations of the input X-ray images (frog-leg lateral view). The CAE was used to reconstruct the input images and generate enhanced images that highlight abnormal regions. Subsequently, the original and enhanced images were concatenated along the channel dimension to form an information-rich enhanced input. Finally, the pretrained ResNet18 was adopted as the backbone classification network, and its input layer was modified to adapt to multi-channel input to conduct training and classification for the stitched images. Compared to those achieved by the four benchmark models (ResNet18, DarkNet19, AlexNet, and MobileNetV2), the CAE-ResNet18 model achieves excellent performance on the test set. The accuracy, recall rate, and F1-score of the Normal class are 0.9890, 1.0000, and 0.9945, respectively. The accuracy, recall rate, and F1-score of the DDH class were 1.0000, 0.9912, and 0.9956, respectively. Visual analysis shows that the t-SNE visualization of the fully connected layer feature of this model presents a more obvious inter-class separation. The CAE-ResNet18 model effectively leverages both original image information and abnormal features, significantly improving the accuracy and reliability of DDH diagnosis. It provides a potential intelligent tool for clinical auxiliary diagnosis, which may enhance patient prognosis. Full article

Offshore wind farms are rapidly expanding into deeper and more remote ocean regions. Their structural safety and operational reliability in harsh marine environments have garnered widespread global attention. Sensing technologies capture structural and environmental conditions and are indispensable to structural monitoring. Accordingly, this review examines the applications of environmental monitoring, supervisory control and data acquisition, condition monitoring, and structural health monitoring systems covering both the horizontal-axis and vertical-axis types of fixed and floating offshore wind turbines. It also summarizes key technologies for data transmission and optimal sensor placement. However, uncertainty in sensing data can significantly affect monitoring results, yet existing studies lack an adequate summary and in-depth discussion. We therefore focus on sources of sensing uncertainty, including the marine environment, the host platform, variations in environmental and operational conditions, and sparse sensing. By analyzing their effects on monitoring data, we explore key methods for overcoming data uncertainties and improving sensing accuracy. This paper also evaluates the application potential of cutting-edge artificial intelligence and digital twin technologies. Furthermore, the study points out that fusing multi-source signal data to establish a highly reliable intelligent decision-making and early warning framework is likely to become an important development direction for offshore wind power monitoring. This review aims to provide valuable support for the safe development of offshore wind farms towards deep-sea regions over the coming decades. Full article

Artificial intelligence (AI) is transforming oncologic imaging by enabling automated, accurate, and scalable diagnostic decision support. However, successful clinical translation requires not only strong predictive performance but also interpretability, transparency, and clinician confidence. We present an attention-guided deep learning framework for automated slice-level classification of parathyroid adenoma (PTA) from arterial-phase computed tomography (CT) images. The framework incorporates lightweight hierarchical attention mechanisms within residual neural networks to enhance feature representation and contextual understanding while maintaining computational efficiency for real-world deployment. Three novel architectures were developed: the Residual Block Light Attention Network (Res-BLANet), Residual Stage Light Attention Network (Res-SLANet), and Residual Layer Light Attention Network (Res-LLANet). Models were trained and evaluated on a rigorously curated, expert-annotated dataset of 63 patients, including 39 pathologically confirmed PTA cases, with 350–450 slices per patient. Training employed Hounsfield unit normalization, extensive data augmentation, and 10-fold cross-validation to ensure robust performance assessment. A key innovation is the integration of hierarchical attention modules that generate attention maps at multiple network levels, enabling qualitative visualization of diagnostically relevant regions and providing exploratory insight into the model’s decision-making process. Experimental evaluation showed strong diagnostic performance. Res-SLANet achieved the highest overall accuracy (87.37%), precision (92.56%), and F1-score (86.69%), while Res-LLANet attained the highest sensitivity (96.0%) on independent testing. These results should be interpreted as preliminary, given the limited size of the independent patient-level test cohort (n = 3). Res-BLANet delivered substantially faster inference with minimal computational overhead. These findings demonstrate that hierarchical attention-guided deep learning can achieve accurate and computationally efficient PTA detection from CT imaging while providing qualitative visual insights into the model’s decision-making process. The proposed framework represents a promising approach toward more interpretable AI-assisted diagnostic systems for oncologic imaging. Full article

This study investigated the implications of integrating artificial intelligence (AI) technologies into fashion design education and their relationship with supporting sustainable design practices. A descriptive–analytical approach was adopted, and an online questionnaire was administered to a sample of 404 female students and graduates specializing in fashion design in Saudi Arabia. The study examined the perceived usefulness of AI technologies in supporting sustainable design, the level of sustainable design self-efficacy, the relationship between these two variables, and differences associated with selected demographic variables. In addition, two optional open-ended questions provided complementary qualitative insights into participants’ perceptions of AI in sustainable fashion design. The findings revealed a high level of perceived usefulness of AI technologies in supporting sustainable design practices, as well as a high level of sustainable design self-efficacy among the participants. Statistically significant differences were found in perceived usefulness according to participants’ level of experience with AI tools and frequency of use, in favor of those with greater experience and more frequent use. The results also showed a strong positive relationship between perceived usefulness and sustainable design self-efficacy, along with statistically significant differences according to academic status, educational level, and geographic region. The qualitative findings further highlighted the role of AI in enhancing creativity, improving design efficiency, and reducing waste, while also pointing to challenges related to AI literacy, intellectual property, and ethical use. The study contributes empirical evidence from fashion design education in Saudi Arabia, a context that remains underrepresented in previous research, and emphasizes the importance of integrating AI technologies into fashion design education through educational and ethical frameworks that promote responsible AI use and sustainability-oriented design competencies. Full article

Experimental macromolecular structures are foundational for elucidating molecular mechanisms and guiding drug design and protein engineering. However, inaccurate structural models are occasionally deposited in the Protein Data Bank (PDB), potentially confounding structural analyses and misleading subsequent studies. Although the integration of artificial intelligence (AI)-predicted models to improve experimentally determined structures has been proposed, the impact of AI-guided rebuilding on functional interpretation remains largely uncharacterized. Here, AI-predicted models of δ-aminolevulinic acid dehydratase (ALAD) were analyzed, and a misinterpreted region in its original crystal structure was rebuilt using an AlphaFold3 (AF3) model as a template. The incorrectly modeled region between Cys122 and Leu142 was corrected utilizing the AF3 main-chain conformation. This rebuilding decreased the R_free value and improved structural geometry compared to the experimental structure. Notably, the initially deposited ALAD structure exhibited an inactive conformation characterized by a disrupted substrate-binding A-site. In contrast, the AI-rebuilt ALAD model restored a biologically relevant active-site conformation, featuring a coordinated zinc-binding pocket, mediated by Cys122, Cys124, and Cys132, and an intact substrate-binding configuration at the A-site. These results demonstrate the feasibility of AI-predicted model-guided rebuilding in the present ALAD case and may provide a basis for future studies evaluating its applicability to other protein systems. Full article