Browse Articles

Automated error localisation and correction techniques for deep-learning-based segmentation of 3D MRI sequences based on feature-derived-region aggregation

Abstract Automatic segmentation using convolutional neural networks (CNNs) has become a key tool in musculoskeletal imaging, offering substantial reductions in processing time. However, concerns about reliability often necessitate manual inspection and correction. We present a method that leverages network-derived uncertainty to automatically identify and localise segmentation errors, reducing the need for exhaustive manual review. A 3D nnU-Net was trained on delayed gadolinium-enhanced MRI of hip cartilage. Voxel-wise uncertainty scores, computed from the SoftMax outputs of ensembled sub-networks, were aggregated over feature-based supervoxels. Each region was then evaluated for its potential impact on clinically relevant metrics, generating sensitivity scores. A logistic model combined these with uncertainty data to assign risk scores, guiding attention to areas most likely to affect clinical metrics during the initial correction steps. Using these risk scores, guided supervoxel correction of just 50 supervoxels reduced the mean absolute relative error by 2.1-fold. Guided manual correction within these regions achieved a 3.5-fold reduction, an approximate 62% supervoxel correction efficiency. Correcting the top 10 regions yielded up to 88% efficiency. This approach serves as a proof-of-concept for targeted correction in hip MRI, enhancing the clinical utility of CNN-based segmentation by demonstrating that 3D feature-derived uncertainty aggregation has the potential to reduce correction burden compared to traditional 2D methods.

Mandatory lane-changing decision and control method based on game theory

This paper proposes a lane-changing decision and control framework for mandatory lane-changing scenarios based on stage game decision-making. The lane-changing process is discretized, and payoff functions are constructed for the autonomous vehicle and the following vehicles in the target lane, enabling adaptive adjustment of driving strategies and planned trajectories according to surrounding vehicle interactions. The optimal lane-changing decision is dynamically updated by incorporating environmental information and interactive feedback from surrounding vehicles. For trajectory tracking, a composite error combining lateral displacement and heading angle deviations is defined and constrained using a preset performance boundary. The constrained error is transformed into an equivalent unconstrained form, and a sliding mode controller is designed to ensure robust tracking performance. A joint simulation platform integrating traffic simulation and driver-in-the-loop experiments is established to evaluate the proposed framework. A preliminary study involving three human drivers with different driving tendencies is conducted to analyze interaction behaviors in mandatory lane-changing scenarios. The experimental results provide initial insights into the effectiveness of the proposed approach under representative conditions.

Effect of waist drum axial position on energy separation performance of Maxwell structured tubes

Abstract Combining numerical simulations and experimental validation, this study systematically investigates the influence of the axial position of the waist drum on the three-dimensional flow field structure and energy separation performance within Maxwell structured tubes. It is found that the waist drum position is a critical geometric parameter modulating internal momentum transfer and temperature stratification, exhibiting a significant non-monotonic relationship with energy separation efficiency. The cooling and heating effects of the system simultaneously reach their peaks when the waist drum is located in the middle region of the tube. Mechanism analysis reveals that this optimal position achieves the best synergy between flow field stability and secondary flow enhancement: it avoids the vortex core breakdown near the inlet caused by an upstream position, while overcoming the insufficient effective separation length associated with a downstream position. This research elucidates the kinetic-thermal energy conversion mechanism within variable cross-section vortex tubes, providing a theoretical basis for the topological optimization of Maxwell structured tubes.

HybFusion: A holistic Android malware detection framework with advanced feature fusion and ensemble learning

Android malware detection remains a critical challenge due to the rapid increase in malware variants and the growing sophistication of obfuscation techniques. To address these issues, this paper introduces HybFusion, a holistic Android malware detection framework that integrates advanced feature fusion with ensemble learning to enhance detection effectiveness and reduce false positives. HybFusion combines two complementary feature types to overcome the limitations of existing methods in capturing a comprehensive representation of malware behaviors and semantics: (1) behavioral features extracted from the function call graph and embedded using the Graph Isomorphism Network, and (2) semantic permission features obtained from the AndroidManifest.xml file by applying a normalization process to convert permission identifiers into a permission sequence, which is then embedded using a lightweight pre-trained Transformer-based language model. This strategy enables better leveraging of semantic relationships among permissions while maintaining low computational cost. In addition, HybFusion adopts a stacking-based ensemble learning strategy that leverages the strengths of multiple classifiers to further improve detection robustness. Extensive experimental results demonstrate that HybFusion outperforms existing approaches across all evaluation metrics, achieving a recall of 99.24% and an F1-score of 99.27%.

Unified multi-task learning for hydrological processes using a shared transformer framework

Is asymmetric upper trapezius muscle activation during work associated with neck pain? A cross-sectional and longitudinal analysis

Objectives Previous studies have linked activity in the upper trapezius muscle with neck pain. However, no studies have examined whether asymmetric activation of these muscles during the working day is associated with neck pain. This study aimed to investigate this relationship. Methods Seven research institutes provided data on bilateral upper trapezius muscle activity on one working day, along with corresponding questionnaire data on cross-sectional (n = 530) and longitudinal (n = 256) neck pain intensity. The asymmetry, defined as the activity difference between the two upper trapezius muscles, was calculated as an average across the entire workday and within various intensity levels in relation to maximum voluntary isometric contraction (MVIC). Unadjusted and adjusted linear regression analyses were executed to examine the association between asymmetric muscle activation and neck pain intensity. Results In cross-sectional analyses, asymmetry in the levels 0–0.05 and 0.05–2%MVIC was significantly positively associated with neck pain intensity in both unadjusted and adjusted analyses. Asymmetry in the levels of 4–6, 6–8 and 8–10%MVIC was significantly negatively associated with neck pain in unadjusted analyses. In longitudinal analyses, significant positive associations were found for asymmetry in level 0–0.05%MVIC and negative associations for asymmetry in levels > 20%MVIC. Conclusion While asymmetry in the very low levels of muscle activity may be associated with higher neck pain intensity, asymmetry in the higher levels of muscle activity was negatively associated with neck pain intensity. However, the explained variance of the models was small, and the results should therefore be interpreted with caution. The findings suggest that work conditions facilitating simultaneous relaxation during breaks and balanced activation of both muscles during static activities may be relevant for neck pain prevention, though further research is needed to establish causality.

Simulation of long-term spatio-temporal environmental dynamics using a unified benchmark of neighbor augmenting, LSTM and graph attention models

Study on the adaptability of multilayer subway network under sudden large passenger flow disturbances

This paper proposes a multilayer network-based subway network adaptability assessment framework.Adaptability is defined as the ratio of post-disturbance cumulative network performance to its nominal value, represented as 0≤ F * ≤1. In this framework, each line is treated as a separate layer, with the introduction of composite edge weight and platform congestion factor, and the establishment a performance response function. By integrating the “physical accessibility-perceived impedance” flow diversion model and bond percolation theory, the framework characterizes failure propagation, and is verified using the Shanghai Metro regional network, with fair comparisons made against the weighted single-layer network (WSN) and the multigraph model (MLG). The results show that the critical threshold of the multilayer network is 12% and 11% higher than that of WSN and MLG, respectively; under severe attack, the network performance exceeds F * by 23%−33%, and ΔK is 22%−26% lower. Monte Carlo variance analysis further indicates a significant interaction (p < 0.01) between passenger flow arrival and route choice.

Optimization of ciprofloxacin removal from aqueous solutions using the TiO2/Fe3O4/UV process and toxicity assessment via resazurin reduction test

Developing an evidence-based framework of contemporary character development at Outward Bound Schools

This paper presents an evidence‑based framework for character development across the global network of Outward Bound (OB) Schools. Using the Eisenhardt multiple‑case, cross‑case theory‑building method, the study examined 11 OB Schools operating in diverse cultural and operational contexts to understand how character development is fostered worldwide. Analysis revealed five key “levers” that were associated with character development opportunities: educational philosophy, service, authentic adventure, educational models, and instructor behaviors. These levers vary by context yet appear most potent when intentionally aligned to create coherent educational experiences. The resulting framework provides a practical tool for OB Schools seeking to strengthen character development through aligned design and implementation strategies. More broadly, this work advances the field of character education by highlighting both the distinctive strengths of outdoor experiential learning and the ways a global educational organization can balance overarching pedagogical coherence with the need for local adaptability.

Numerical simulation of deformation and hole position shift in extrusion process of four-hole propellant

Evaluating respiratory syncytial virus immunization strategies for infants in Canada: A cost-utility analysis

Background Respiratory syncytial virus (RSV) is a leading cause of lower respiratory tract infections and hospitalizations among infants in Canada. New long-acting monoclonal antibodies (mAbs) and vaccines administered during pregnancy have expanded prevention options, yet the most cost-effective immunization program remains uncertain. Methods We updated a Canadian cost-utility model to evaluate seven seasonal RSV prevention strategies over one year (with a lifetime horizon for mortality impacts), from health system and societal perspectives. Strategies included RSVpreF vaccination in late pregnancy; targeted or universal infant mAb programs using nirsevimab or clesrovimab; and combination programs in which infants could receive protection from either RSVpreF or mAbs. Sequential incremental cost-effectiveness ratios (ICERs) were estimated in 2024 Canadian dollars per quality-adjusted life year (QALY), using a $50,000/QALY threshold. The primary analysis used immunization product list prices. Findings The most cost-effective strategy was a seasonal combination program: RSVpreF vaccination for pregnancies due during the RSV season, with mAb for infants at high risk (<32 weeks’ gestation), including catch-up for infants at high risk born before the season. This strategy had an ICER of $35,408/QALY compared to seasonal mAb for infants at moderate risk (32 0/7 to 36 6/7 weeks’ gestation) or high-risk with catch-up. Expanding mAb to unimmunized non-high-risk infants born in-season increased the ICER to $132,131/QALY. Universal infant protection, with mAb alone or combined with RSVpreF in pregnancy, was not cost-effective across analyses. RSVpreF alone was dominated. Results were most sensitive to product prices, target populations, age at administration, and RSV burden. Conclusions A seasonal combination program with RSVpreF for in-season deliveries and mAb for infants at high risk of RSV offers the best value for money for protecting Canadian infants from RSV disease. Broader infant immunization programs may be cost-effective with substantial price reductions or in regions with higher disease burden and healthcare costs.

Endoglin (CD105) overexpression is associated with an immunosuppressive tumor microenvironment in murine lung cancer models and human lung adenocarcinoma

Abstract The tumor microenvironment (TME) is a major determinant of tumor response to different therapies, especially immunotherapy. Tumors with high CD8 + T-cell infiltration that respond well to immunotherapy are referred to as “hot” tumors, whereas “cold” tumors, with an immunosuppressive microenvironment, respond less effectively to this therapy. Endoglin (CD105) plays a critical role in angiogenesis, and its overexpression has been associated with a poorer prognosis in several types of cancer. This study aims to investigate whether high endoglin levels are associated with a cold microenvironment in tumors. Transgenic mice ubiquitously overexpressing human endoglin ( ENG + ) and wild-type C57BL/6J mice (WT) were used to analyze the TME in a Lewis Lung Carcinoma (LLC) subcutaneous xenograft model and in a lung cancer model. Tumors developed in ENG + mice exhibited increased hypoxia, reduced CD8 + T-cell infiltration and an increased presence of immunosuppressive cells, such as M2 TAMs and Treg, compared to WT tumors. Thus, these tumors can be categorized as cold tumors. In addition, the analysis of the TME and endoglin expression in human lung adenocarcinoma samples showed that cold tumors have higher endoglin levels than hot tumors. These findings suggest that a hypoxic and immunosuppressive microenvironment may contribute to the poorer prognosis of tumors with high levels of this protein. This study highlights the potential of endoglin as a marker to predict the response to immunotherapy and to guide personalized treatment strategies in cancer patients.

Citizen satisfaction with public services in Peru: Regional gaps in citizen perceptions of public service provision

Citizen satisfaction with public services is a fundamental indicator of government effectiveness and democratic legitimacy in Latin America. However, there is little empirical evidence on regional gaps in countries with high territorial heterogeneity such as Peru. The objective of this study was to determine the differences in levels of citizen satisfaction with public services between the country’s coastal, mountain, and jungle regions. A secondary analysis was conducted of the 2024 National Household Survey, corresponding to the Governance, Democracy, and Transparency Module (n = 33,691), evaluating 21 public services using the Citizen Satisfaction Index. Student’s t-tests were applied for comparisons between urban and rural areas, one-factor ANOVA with Bonferroni post-hoc tests for comparisons between geographic domains, and principal component analysis to explore the dimensional structure. The results revealed statistically significant differences between the eight geographic domains (F = 89.57, p < .001, η² = .022); although this effect size is small, indicating that geographic domain accounts for 2.2% of the variance in satisfaction, the patterns are consistent and policy-relevant. The North Coast (M = 3.09) and Amazon (M = 3.13) showed the lowest satisfaction rates, while the Southern Highlands (M = 3.29) even surpassed Metropolitan Lima. Social programs (51.9%) and the Public Prosecutor’s Office (56.4%) obtained the lowest national satisfaction ratings, with the largest regional gaps. Rural areas reported higher satisfaction with public transportation (+7.0%) and public education (+3.5%) than urban areas. A four-dimensional factor structure was identified that explains 46.7% of the variance. It is concluded that territorial gaps are consistent with structural differences in regional institutional capacity, requiring territorially differentiated public policies to reduce inequalities in the provision of state services.

Development and operation of a 100 W wave energy converter for persistent ocean observation

Discharge health education needs and experiences of patients with skeletal-related events from bone metastases: A qualitative study

Objective To describe the health education needs and experiences of patients with skeletal-related events due to bone metastasis from solid tumours upon discharge and to provide a reference for the formulation of discharge health education plans for this population. Methods A qualitative descriptive design and purposive sampling method were used to select patients with skeletal-related events due to bone metastasis from solid tumours who were hospitalized in the orthopaedic department of a Class III Grade A cancer hospital in Hebei Province from June to July 2024 for in-depth semistructured interviews. A Colaizzi 7-step analysis method was used to refine the themes of the patients’ needs. Results The needs of solid tumour bone metastasis patients with skeletal-related events for discharge health education provided by medical staff were refined into five themes: 1) bone health knowledge needs: patients were eager to obtain in-depth knowledge beyond the definition of the disease, such as the specific warning of pathological fracture and the early identification of bone metastasis progression; 2) bone health self-management needs: the focus was on specific operation guidance, including the home application of analgesic stepwise therapy and the protection skills for bone destruction sites; 3) coping with daily life needs: patients were concerned about how to safely go to the toilet, take a bath and other daily activities under the risk of disability; 4) coping with psychological pressure needs: manifested as the counselling needs for fear of disability and psychological support to combat the “burden feeling” during long-term pain; and 5) family and social support needs: patients need a collaborative nursing model that includes family participation, as well as a remote consultation channel for continuous access to professional medical care after discharge. Three themes were extracted from the information experience: 1) barriers to understanding medical terms: the professional words used by medical staff (such as “skeletal-related events” and “bone scan”) confused patients, leading to incorrect interpretation or neglect of health education; 2) passive acceptance: patients were unable to obtain targeted information, and the existing education is mostly one-way indoctrination, which fails to consider the differences in personalized care after bone metastasis of different primary cancers (such as lung cancer, breast cancer and prostate cancer); and 3) the large amount of information leads to poor satisfaction: the large amount of fragmented data provided before discharge leads to information overload and cognitive fatigue among patients, and the actual mastery rate and application satisfaction after discharge are low. Conclusions Patients with skeletal-related events due to bone metastasis from solid tumours face challenges such as a lack of bone health knowledge, psychological distress and a lack of self-management ability when discharged from the hospital. Medical staff should formulate a personalized discharge health education plan according to the needs of patients.

CD49a+ NK cells promote M2 polarization and are associated with poor pathological response in NSCLC

Three-dimensional dynamic response of infinite plate resting on multilayered orthotropic foundation under moving loads

In response to the dynamic response of layered foundation pavement structures caused by traffic loads, a three-dimensional coupled model of an infinite elastic plate on an orthotropic layered foundation under moving loads is established. Based on elastodynamic theory and Kirchhoff thin-plate theory, governing equations are derived by introducing a moving coordinate system and incorporating orthotropic constitutive relations. Through double Fourier transforms and matrix analysis, a single-layer foundation transfer matrix is formulated. The global stiffness matrix is assembled using interlayer continuity conditions, ultimately yielding an analytical solution for the dynamic response of the plate and foundation in the integral transform domain. Combined with ABAQUS finite element simulations, the effects of foundation stratification, soil orthotropy, load parameters, and plate parameters on the dynamic response are systematically studied. The results indicate that the layering of the foundation significantly affects the plate deflection. The orthogonal anisotropy of the first soil layer has a more significant impact on deflection compared to isotropic conditions, and this characteristic needs to be considered in practical engineering. The increase in load speed leads to an increase in peak deflection, while the increase in load frequency results in a decrease in peak deflection. Optimization of plate parameters can effectively suppress deformation, and the amplitude of the deflection curve decreases with the increase of the elastic modulus or thickness of the plate.

Adaptive collaborative feature fusion and shape-aware optimization for multi-scale chest lesion detection

Abstract Chest diseases remain a major cause of global morbidity and mortality, and accurate detection from chest X-ray images is critical for early diagnosis and clinical decision-making. However, large variations in lesion scale, morphology, and spatial distribution pose significant challenges for automated detection systems, particularly in identifying small lesions and achieving precise localization. To address these issues, we propose a multi-scale chest lesion detection method based on adaptive collaborative feature fusion and shape-aware optimization. The method enhances multi-scale feature modeling and introduces shape structural constraints to improve detection accuracy and localization robustness in complex anatomical environments. Specifically, a Context-Embedded Feature Enhancement Network is designed to jointly capture global anatomical context and local lesion characteristics, strengthening lesion representation. An Adaptive Feature Focusing Network further improves multi-scale feature representation through adaptive spatial feature aggregation, enabling more effective detection of small lesions. In addition, a shape-aware optimization strategy integrating normalized Wasserstein distance with shape-weighted constraints improves localization stability and bounding box regression accuracy for irregular lesions. Compared with state-of-the-art methods, the proposed method achieves improvements of 6.3% in mAP , 6.4% in small-lesion mAP , and 8.6% in mean recall on VinDr-CXR, as well as improvements of 4.2% in mAP and 7.8% in mean recall on ChestX-ray8, demonstrating its effectiveness and generalization capability for multi-scale chest lesion detection.

An optimization method for flexible interconnection planning based on improved CNN-LSTM prediction and tunable relative entropy-driven chaotic evolution

As power systems evolve towards greater intelligence and flexibility, flexible interconnection technology has emerged as a critical means to enhance operational reliability and economic performance. This paper presents a data-model dual-driven planning methodology for flexible interconnection systems, integrating a multi-scale spatio-temporal cross-enhanced CNN-LSTM model for load forecasting with a Chaotic Evolutionary Optimization (CEO) algorithm to optimize system design. The proposed framework first constructs an improved CNN-LSTM hybrid architecture, trained on historical load data and simulated feature sets, to predict future load profiles. A novel Tunable Relative Entropy (TRE) metric is introduced as a complementarity quantification index, forming a multi-objective function that incorporates system balance, reliability, economy, and spatio-temporal complementarity. The CEO algorithm is then employed to solve the optimization model, determining the optimal system configuration and operational parameters. Experimental evaluations demonstrate that the forecasting module achieves high accuracy, with a Mean Squared Error (MSE) of 0.000368 and a Mean Absolute Error (MAE) of 0.006334. Moreover, the TRE index improves complementarity efficiency by 3.8%. By leveraging the predictive capability of the hybrid neural network and the CEO algorithm’s optimization efficacy, the proposed approach not only reduces load fluctuation indices but also enhances planning efficiency and operational economy, offering a viable pathway for intelligent power system development.