Search Results (883)

Computed Tomography (CT) is a widely used imaging modality that employs X-rays and computational reconstruction to visualize internal anatomy. Although higher radiation doses produce higher-quality images, they also increase long-term cancer risk, motivating the use of low-dose protocols. However, low-dose CT data inherently suffer from elevated Poisson–Gaussian noise, necessitating effective denoising strategies. In myocardial CT perfusion (CTP) imaging, this challenge is compounded by residual cardiac motion, which misaligns consecutive time points and impairs accurate estimation of perfusion maps for diagnosing coronary artery disease. Traditional approaches typically treat these two problems, noise and motion, separately, denoising the reconstructed images first or applying the registration first. Such serial pipelines often degrade clinically significant features; e.g., denoising may destroy structural details essential for registration, while motion correction can distort subtle intensity cues needed for noise modelling. To overcome these limitations, we propose a unified deep learning framework that performs noise suppression and motion correction jointly for low-dose myocardial CTP. The method integrates two complementary components through a parallel ensemble strategy: (i) a modified Fast and Flexible Denoising Network (FFDNet) that incorporates noise-level maps to mitigate blended noise effectively, and (ii) a CNN-based registration model, extended with Time Enhancement Curve (TEC) correction and 4D physiological consistency constraints to estimate temporally coherent and anatomically plausible motion fields. By combining their outputs without iterative dependencies, the proposed framework produces motion-corrected and denoised CTP sequences in a single unified processing step, thereby better preserving myocardial structure and perfusion dynamics than conventional serial pipelines. The model has been evaluated using both reference-based (MSE, PSNR, SSIM, PCC, Noise Variance, TRE) and no-reference (NIQE, FID, KID, AUC) image quality metrics, supplemented by expert human assessment. Results demonstrate that jointly learning noise characteristics and motion patterns enables restoration of low-dose CTP images while minimizing feature corruption, thereby advancing the clinical utility of low-dose myocardial CTP imaging. Full article

The identification/classification of textile fibres is essential in manufacturing, forensic science, cultural heritage preservation, and recycling. Conventional methods, including solubility tests, optical microscopy, and chromatographic techniques, are often destructive, labour-intensive, and limited in scope. Vibrational spectroscopy, particularly near-infrared (NIR), Fourier-transform infrared (FTIR), and Raman spectroscopy, has emerged as a rapid, non-destructive, and accurate alternative for fibre analysis. However, multi-composition textiles, dyes, finishing agents, and ageing effects frequently cause overlapping spectral features, hampering direct interpretation. This review examines the combined use of vibrational spectroscopy and chemometrics for textile fibre discrimination. It critically evaluates the performance of different spectroscopic techniques in classifying natural, synthetic, and blended fibres. The role of multivariate analysis methods, such as PCA, PLS, LDA, SIMCA, and machine learning algorithms, in improving spectral interpretation and classification accuracy is highlighted. Key factors affecting model robustness, including spectral pre-processing, sample heterogeneity, moisture, and colour, are also discussed. The integration of spectroscopy with chemometrics provides a robust, scalable, and sustainable solution for fibre identification, supporting quality control, fraud detection, and circular economy initiatives. This approach demonstrates significant potential for both research and industrial applications. Full article

Background: Dental students need to qualify with a clear understanding of the continuum of biological development from the molecular (genetic, epigenetic and environmental interactions) to the cellular (morphogenesis and differentiation) to the emergence of the mature tissue or organ. Histogenetics provides a core component for this understanding. The aim of this study is to investigate whether a merged approach, combining traditional and recent methods, can enhance the teaching of histogenetics to dental students. Methods: This study blended traditional (lectures, drawings, microscopy) and recent approaches (flipped classroom elements, virtual microscopy, group-based poster construction, and interactive quiz-based discussion) to enhance student engagement and perceived learning in oral histogenetics. The intervention was delivered to master-level dental students across six core oral histogenetics topics. Teaching followed a structured three-phase model: Prepare (digital lectures and short microscopy-introduction videos); Engage (microscopy session and group-based poster creation); and Test and Discuss (teacher-led quizzing and discussion). Student perceptions were evaluated through an electronically distributed 17-item questionnaire at the end of the course. Items were grouped into self-evaluation, resources, and teaching method domains and rated on a five-point Likert scale. Results: A total of 45 of 51 students responded (88%). Across all domains, positive perceptions (Agree/Strongly Agree) predominated (p < 0.001). Self-evaluation items showed strong agreement for attendance and group contribution, with more variability in preparation time and motivation. Resources were rated highly, although the accessibility of physical guidance showed more mixed responses. The merged teaching method received strong endorsement, with students reporting engagement, enjoyment, ease of understanding, and clear emphasis on clinical relevance. Conclusions: The merged approach was perceived as pedagogically valuable and clinically meaningful by the students and appears to enhance perceived engagement, clarity, and relevance in oral histogenetics teaching. These findings support the adoption of blended, student-active methodologies to strengthen comprehension and promote clinically meaningful learning in oral histology. Full article

Background and Objectives: Diabetes mellitus represents one of the most prevalent chronic metabolic disorders worldwide, necessitating precise insulin dose management to prevent both acute and long-term complications. The optimization of insulin dosing remains a significant clinical challenge, as inappropriate dosing can lead to hypoglycemia or hyperglycemia, each carrying substantial morbidity risks. Machine learning approaches have emerged as promising tools for developing clinical decision support systems; however, their practical implementation requires both high predictive accuracy and model interpretability. This study aimed to develop and evaluate an explainable machine learning framework for predicting insulin dose adjustments in diabetic patients. We sought to compare multiple ensemble learning approaches and identify the optimal model configuration that balances predictive performance with clinical interpretability through comprehensive SHAP and LIME analyses. Materials and Methods: A comprehensive dataset comprising 10,000 patient records with 12 clinical and demographic features was utilized. We implemented and compared nine machine learning models, including gradient boosting variants (XGBoost, LightGBM, CatBoost, GradientBoosting), AdaBoost, and four ensemble strategies (Voting, Stacking, Blending, and Meta-Learning). Model interpretability was achieved through SHapley Additive exPlanations (SHAP) and Local Interpretable Model-agnostic Explanations (LIME) analyses. Performance was evaluated using accuracy, weighted F1-score, area under the receiver operating characteristic curve (AUC-ROC), precision-recall AUC (PR-AUC), sensitivity, specificity, and cross-entropy loss. Results: The Meta-Learning Ensemble achieved superior performance across all evaluation metrics, attaining an accuracy of 81.35%, weighted F1-score of 0.8121, macro-averaged AUC-ROC of 0.9637, and PR-AUC of 0.9317. The model demonstrated exceptional sensitivity (86.61%) and specificity (91.79%), with particularly high performance in detecting dose reduction requirements (100% sensitivity for the ‘down’ class). SHAP analysis revealed insulin sensitivity, previous medications, sleep hours, weight, and body mass index as the most influential predictors across different insulin adjustment categories. The meta-model feature importance analysis indicated that LightGBM probability estimates contributed most significantly to the ensemble predictions. Conclusions: The proposed explainable Meta-Learning Ensemble framework demonstrates robust predictive capability for insulin dose adjustment recommendations while maintaining clinical interpretability. The integration of SHAP-based explanations facilitates clinician understanding of model predictions, supporting transparent and informed decision-making in diabetes management. This approach represents a significant advancement toward the clinical implementation of artificial intelligence in personalized insulin therapy. Full article

The integration of artificial intelligence (AI) into early childhood education presents both opportunities and challenges to longstanding Froebelian pedagogies, particularly regarding child agency and nature-based play. This mixed-methods study explores this tension within the Chinese context. It examines how Chinese Froebelian practitioners perceive the alignment between AI tools and core principles and investigates the strategies they employ to navigate the integration of technology with humanistic educational values. The survey results, from 50 practitioners, revealed that AI can support autonomous and holistic learning, yet significant concerns persisted regarding the displacement of sensory and nature-based experiences. Follow-up interviews uncovered a practitioner-led “dual-track integration” approach, which strategically blends physical manipulation and nature engagement with AI-enabled personalization. Through an iterative dialogue between theory and data, this study develops and refines the “dual-track integration” framework as an empirically grounded, sensitizing model. This framework offers principled strategies for hybrid learning that uphold the developmental primacy of play. Situated within the discourse on Sustainable Development Goal 4 (quality education) and Goal 10 (reduced inequalities), the analysis highlights AI’s dual potential to advance or hinder equity. By examining China’s hybrid position, which combines advanced digital infrastructure with persistent equity gaps, this research highlights the critical role of educator agency and pedagogical design in leveraging AI to advance equitable, high-quality early childhood education. Full article

In the context of global sustainability agendas and the rapid transformation of the media industry, cultivating new media professionals equipped with language proficiency, cross-cultural communication skills, and sustainability awareness has become a crucial educational imperative. This study implemented a pedagogical framework integrating Content-Based Instruction (CBI) and Project-Based Learning (PBL) at Communication University of Shanxi, centering on authentic media projects. A mixed-methods approach (questionnaires, N = 204; semi-structured interviews, n = 50) was employed to evaluate its effectiveness. Under this model, students demonstrated positive gains in linguistic knowledge and skills, media literacy, self-directed learning, critical thinking, and teamwork. Positive outcomes were also observed in intercultural competence and innovative thinking. Comparative analysis of pre- and post-test academic performance indicated significant improvement across all participating majors. The integrated CBI-PBL model provides a promising teaching pathway for sustainability-oriented foreign language education within similar instructional contexts. It contributes to achieving United Nations Sustainable Development Goal 4 (SDG 4) and offers theoretical and practical insights for aligning media education with the evolving sustainable demands of the industry Full article

Fluoride (F⁻) contamination in deep groundwater threatens drinking water security, yet its enrichment is commonly governed by coupled nonlinear hydrogeochemical feedbacks that are difficult to resolve with linear diagnostics alone. Here, we integrate an explainable deep learning framework (HydroAttentionNet + SHAP) with thermodynamic and mass-conservative inverse modeling (PHREEQC) to quantitatively link data-driven thresholds to mineral water processes in a multi-aquifer system. Using 258 deep-well samples, we delineate a robust evolution pathway from background to ultra-high-fluoride (Ultra-High F⁻, ≥1.5 mg/L) waters. HydroAttentionNet achieves strong predictive skill (R² = 0.77) and reveals a clear mechanistic tipping behavior: alkalinity (HCO₃⁻/CO₃²⁻) is the primary trigger for F⁻ activation, while progressive Na+ enrichment and Ca²⁺ depletion act as amplifiers by suppressing a(Ca²⁺) and weakening fluorite precipitation capacity. PHREEQC simulations confirm a coupled “salinization–decalcification–fluoridation” loop in which (i) evaporite dissolution elevates ionic strength (salt effect) and supplies Na⁺ to promote Na–Ca exchange, and (ii) carbonate re-equilibration drives calcite precipitation as an efficient Ca sink, offsetting ~45.8% of Ca²⁺ inputs; together, these processes maintain fluorite undersaturation and sustain net fluorite dissolution, contributing 56.6% of newly added dissolved F⁻ in evolved end-members. Monte Carlo health risk assessment (10,000 iterations) indicates substantial intergenerational inequity: 67.9% of children exceed the non-carcinogenic risk threshold (HQ > 1), compared with 29.3% of adults. Sensitivity analysis identifies source-water fluoride concentration as the dominant driver (Spearman r = 0.93), implying that supply-side interventions (defluoridation, well-screen optimization, and blending with low-F sources) are substantially more effective than behavioral measures. Full article

Unmanned aerial vehicle (UAV) image stitching aims to generate panoramic remote sensing images beyond the field of view of a single camera. However, there are still significant challenges in constructing panoramic images of a target area quickly and accurately, especially in terms of computationally intensive feature matching extraction and feature alignment accuracy, which are particularly sensitive to high-resolution and low-texture scenes. To address this problem, this study proposes an efficient image stitching method that incorporates dense depth kernelized feature extraction and geometric constraint optimization. The learning-based kernelized feature matching framework is adopted to achieve subpixel-level dense matching, which effectively overcomes the time-consuming and sparse matching deficiencies of traditional manual features (e.g., SIFT) in high-resolution images. Second, a two-layer geometrically constrained mismatching filtering strategy is designed, which significantly improves the alignment accuracy in low-texture and large-parallax scenarios. Finally, panoramic stitching is achieved through a hybrid strategy consisting of a single-responsive transform and max-intensity pixel blending strategy to realize panoramic stitching. Experimental results obtained on multiple datasets show that the proposed method achieves similar visual quality metrics (PSNR, SSIM, and LPIPS) while reducing the stitching time to just 17.5% of that of the baseline method. This makes it a practical solution for efficiently stitching large UAV images. Full article

This study combines multi-frequency dynamic mechanical analysis (DMA) with machine learning (ML) to characterize and predict the viscoelastic properties of a commercial polycarbonate/acrylonitrile–butadiene–styrene (PC/ABS) blend. DMA temperature sweeps at four frequencies (1–10 Hz) in single cantilever mode yielded a glass transition range of 115.8–$123.2$ °C ($E^{″}$ peak), frequency sensitivity of $7.18$ °C/$\mathrm{decade}$, and an apparent activation energy of $335\pm 85 \mathrm{k}\mathrm{J} {\mathrm{mol}}^{-1}$. Time–temperature superposition master curves were parameterized with a six-term Prony series ($R^2=0.998$). Four data-driven models (RF, XGB, SVR, MLP) and a physics-informed NeuralWLF model were evaluated through a hierarchical validation framework. Temperature-blocked CV ranked MLP ($\overline{R^2}=0.989$) above RF (0.950) for interpolation; LOFO validation revealed that NeuralWLF achieved the best cross-frequency generalization ($R^2>0.92$ for all targets) with interpretable WLF parameters ($C_1\approx 12.2$, $C_2\approx 51.7 °\mathrm{C}$). A systematic block size sweep (5–30 $°\mathrm{C}$) revealed a validation inflation effect in which MLP $tan\delta$$R^2$ dropped from 0.986 to 0.592 as the gap-to-FWHM ratio increased from 0.5 to 3.1, establishing the gap/FWHM ratio as a quantitative validation stringency criterion. A physics–data crossover was identified at gap/FWHM $\approx 2$: beyond this threshold, NeuralWLF outperformed all data-driven models in $tan\delta$ prediction by up to $+0.300$ in $R^2$, while curriculum learning (freezing the WLF layer for 300 epochs) further improved the most stringent 30 $°\mathrm{C}$ validation from $R^2=0.660$ to 0.731. The integrated framework demonstrates that honest evaluation of DMA–ML models requires validation gaps exceeding the characteristic feature width and introduces a quantifiable physics-data crossover criterion for selecting between data-driven and physics-informed architectures. Full article

Accurate energy forecasting is critical for the financial and environmental sustainability of the hospitality sector, particularly in energy-intensive subtropical climates. This research addresses a significant gap by conducting a comprehensive, comparative analysis of six machine learning algorithms—Artificial Neural Networks (ANN), Random Forest (RF), XGBoost, Radial Basis Function (RBF), Autoencoder, and Decision Trees—to predict the hourly energy consumption of a hotel in Cuba. We significantly enhance predictive performance through a novel hybrid ensemble scheme, integrating voting, stacking, and blending techniques. Furthermore, this study pioneers a long-term forecasting methodology by utilizing a Long Short-Term Memory (LSTM) model to project the hotel’s energy demand over a 50-year horizon, providing the strategic insight necessary for evaluating major retrofits. Our results demonstrate that ensemble methods, particularly blending, achieve superior accuracy and stability, with correlation coefficients up to 0.975 and the lowest error metrics. The subsequent high-fidelity predictions, including an analysis revealing a minimal specific CO₂ emission of 0.025 kg from natural gas use, provide a quantitative foundation for formulating sustainable energy policies, incentivizing investment in efficient technologies, and ensuring that long-term emission reduction targets are both financially viable and technically robust. Full article

The incorporation of slag and calcium carbonate (CaCO₃) as clinker-reducing constituents offers significant potential for lowering CO₂ emissions in cement production; however, their combined influence on age-dependent compressive strength remains complex and highly coupled. In this study, a structured literature-based dataset ($N=75$ mix conditions) was compiled from two independent experimental sources to investigate compressive strength development in slag–CaCO₃ blended cementitious systems. Compressive strength at 3 and 28 days was formulated as a multi-output regression problem to explicitly capture the correlated nature of strength evolution between early-age and later-age curing stages. Dataset-level analysis revealed that CaCO₃ replacement exerts a stronger influence on early-age strength (reductions of approximately 15–25%) than on later-age strength (typically within 5–15%), indicating a transition from clinker-dominated hydration to slag-controlled later-age strength development. Compared with independent single-output models, the proposed multi-output framework improved prediction performance by increasing $R^2$ values by approximately 4–6% and reducing RMSE by up to 15–18%. Feature importance analysis identified slag replacement ratio and CaCO₃ dosage as the dominant predictors, while chemical composition descriptors modulated age-dependent sensitivity. The results demonstrate that compressive strength at different curing ages is governed by coupled yet temporally evolving physicochemical mechanisms. From an engineering perspective, CaCO₃ replacement should be evaluated within an integrated compositional design framework that considers curing-age requirements and slag reactivity. Overall, this study provides a transparent and statistically robust approach for analyzing strength evolution in blended cement systems and highlights the value of multi-output learning for age-dependent performance prediction in sustainable cementitious materials. Full article

The sustainable stabilization of clayey soils has become a critical strategy for improving their mechanical performance while reducing environmental impact. This study compares two distinct stabilization systems applied to the same low-plasticity clay (CL) from Cartagena de Indias, Colombia: (i) lime catalyzed with sodium chloride (NaCl) and (ii) xanthan gum (XG) reinforced with polypropylene fibers (PPF). A series of laboratory tests was performed to evaluate the unconfined compressive strength (q_u) and small-strain stiffness (G_o) of both systems under controlled compaction and curing conditions. The lime–NaCl system demonstrated accelerated early-age strength and stiffness development, reaching q_u values above 2.5 MPa and G_o exceeding 10 GPa after 28 days of curing, mainly attributed to enhanced pozzolanic reactions catalyzed by NaCl. Conversely, the XG–PPF blends exhibited progressive improvements in mechanical performance, achieving notable gains after 90 days due to the polymeric bonding of XG and the fiber–matrix reinforcement that enhanced ductility and post-peak behavior. When normalized through the porosity–binder index, both systems exhibited power-law trends, with the lime–NaCl mixtures displaying higher exponents indicative of cementation-controlled behavior, while the XG–PPF mixtures showed lower exponents consistent with interparticle bonding and network formation. These results highlight the complementary mechanisms of chemical and biopolymeric stabilization, providing insights into the selection of sustainable binders tailored to specific design requirements in tropical clays. This research demonstrated that the implementation of machine learning models enhanced the fitting accuracy of the two soil stabilization methods when compared with traditional mathematical regression models commonly used in geotechnical engineering. Among the tested approaches, the neural network and Gaussian process regression models exhibited the best performance, achieving R² values ranging from 0.917 to 0.980 during the validation stage. Full article

Despite the enormous amounts of waste textiles produced by the world’s textile industry’s explosive growth, resource utilization rates are still poor. Cotton/polyester blended waste fabrics make up a sizable share, and sorting them precisely is essential to increasing recycling value and promoting the circular economy in the textile industry. Traditional mechanical and human sorting techniques are ineffective and inaccurate; current spectral analysis algorithms mainly concentrate on quantitative composition prediction and are insufficiently capable of differentiating between waste fabrics with comparable content gradients. To address these challenges, this paper proposes an improved 1DCNN model (Dual-1DCNN-Residual-SE) integrated with Near-Infrared (NIR) hyperspectral imaging technology. This model takes raw spectral data and Savitzky-Golay (SG) smoothing data as dual-channel inputs, introducing residual connections to capture subtle spectral differences between similar fabric categories, and employs SE attention mechanisms to adaptively enhance key features. Comparative experiments with four traditional algorithms—KNN, RF, SVM, and PLS—demonstrate that the proposed model achieves a classification accuracy of 95.94%, surpassing the best traditional algorithm SVM (88.12%) by 7.82%. Ablation experiments confirm each enhanced module’s efficacy. This study achieves high-precision classification of cotton/polyester blended waste fabrics, providing technical support for intelligent sorting of industrial waste fabrics. Full article

Small Private Online Courses (SPOCs) have gained attention as a promising approach to blended learning in higher education, particularly within the Flipped Classroom framework. Unlike Massive Open Online Courses (MOOCs), SPOCs cater to a limited number of students, allowing for more personalized learning experiences and enhanced interaction with instructors. This case study examines the integration of a SPOC titled “Introduction to Quantum Physics” into the undergraduate course “Introduction to Modern Physics” at the University of Crete. The research employs a mixed-methods approach, combining quantitative and qualitative data collection methods. Quantitative data were obtained from a questionnaire distributed to students and an analysis of student grades, while qualitative insights were derived from interviews with the course instructors. The findings indicate that the SPOC was associated with positive student engagement and comprehension of complex physics concepts, aligning with previous research on blended learning effectiveness. However, challenges were identified, including the need for increased student–instructor interaction in the online component. Recommendations for improving the SPOC model include the development of interactive activities and enhanced instructor support. This study aims to contribute to the growing body of research on the Flipped Classroom framework in higher education, highlighting the potential utility of SPOCs to enrich learning experiences. Full article

The accuracy of forest canopy height estimation is crucial for forest resource management and ecosystem carbon sequestration. However, existing approaches often face limitations in effectively integrating multisource remote sensing data, feature representation, and model learning strategies. To enhance the prediction performance of the model in complex terrain and multisource data environments, this study comprehensively used ICESat-2/ATLAS photon point clouds, Sentinel-2/MSI multispectral imagery, and SRTM-DEM to construct a remote sensing-driven multisource feature system, which eliminated redundant interference using permutation feature importance analysis. Additionally, a self-attention (SA) mechanism was introduced to strengthen high-dimensional feature representation. Three heterogeneous models, incorporating deep neural network (DNN), extreme gradient boosting (XGBoost), and residual network (ResNet), were independently applied for forest canopy height estimation and were further used as base learners, with a random forest as the meta-learner, and an SA-Blending heterogeneous ensemble model that combines a blending technique with an SA mechanism was proposed to enhance the accuracy of forest canopy height estimation. To evaluate the SA optimization strategy and the role of multisource fusion, this study used the original features, SA-optimized features, and multisource fusion features (i.e., the concatenation and fusion of original features and self-attention mechanism features) as inputs to comprehensively compare the performance of each single model and the integrated model. The results show that: (1) The self-attention mechanism significantly improves the prediction performance of heterogeneous models. Compared with original features inputs, the R² of DNN (SA-Only) and XGBoost (SA-Only) increased to 0.706 and 0.708, respectively, and the RMSE decreased to 1.691 m and 1.613 m. Although the R² for ResNet (SA-Only) decreased slightly to 0.699 and the RMSE increased to 1.712 m, the overall impact was not significant. (2) Under the condition of multisource fusion feature input, DNN+SA, XGBoost+SA, and ResNet+SA all demonstrated higher fitting accuracy and stability, verifying the enhancing effect of the SA mechanism on the association expression of multisource information. (3) The SA-Blending model achieved the best overall performance, with R² of 0.766 and RMSE of 1.510 m. It outperformed individual models and the SA-optimized model in terms of overall accuracy, stability, and robustness. The results can provide technical support for high-precision forest canopy height mapping and are of great significance for ecological monitoring applications. Full article