Search Results (5,720)

Multimodal object detection utilizing RGB and infrared (IR) imagery has become a critical research area for unmanned aerial vehicle (UAV) surveillance applications, providing reliable perception under various lighting and environmental conditions. Nevertheless, current methods encounter three primary challenges: (1) insufficient utilization of frequency-domain properties in heterogeneous modalities, (2) restricted adaptability in crossmodal feature integration across different environmental scenarios, and (3) inadequate modeling of fine-grained spatial relationships for accurate object localization. To overcome these limitations, we introduce MFE-DETR, a novel Multimodal Feature-Enhanced Detection Transformer that achieves superior RGB-IR fusion through three complementary innovations. First, we present the Dual-Modality Enhancement Module (DMEM) with two specialized processing streams: the Haar wavelet decomposition stream (HWD-Stream) that conducts multi-resolution frequency-domain analysis to independently enhance low-frequency structural components and high-frequency textural information, and the Attention-guided Kolmogorov–Arnold Refinement Stream (AKR-Stream) that employs learnable spline-parameterized activation functions for adaptive nonlinear feature refinement. Second, we enhance the Cross-scale Channel Feature Fusion module by integrating an Adaptive Feature Fusion Module (AFAM) with complementary gating mechanisms that dynamically adjust modality contributions according to spatial informativeness. Third, we introduce the Bilinear Attention-Enhanced Detection Module (BADM) that models second-order feature interactions through factorized bilinear pooling, facilitating fine-grained crossmodal correlation analysis. Extensive experiments on the DroneVehicle benchmark show that MFE-DETR attains 78.6% mAP${}_{50}$ and 57.8% mAP${}_{50:95}$, outperforming state-of-the-art approaches by 5.3% and 3.7%, respectively. Additional evaluations on the VisDrone dataset further confirm the excellent generalization performance of our method, especially for small object detection with 18.6% AP${}_S$, achieving a 1.5% improvement over existing techniques. Comprehensive ablation studies and visualizations offer detailed insights into the effectiveness of each proposed component. Full article

Unexpected degradation in servomotor-driven multi-mode industrial systems such as CNC feed drives and robotic machining cells compromises positioning accuracy, availability and operational safety, rendering early fault diagnosis and predictive maintenance essential in smart manufacturing environments. In this study, a predictive maintenance framework based on multi-sensor data fusion was developed to support condition monitoring, fault classification, and remaining useful life estimation of robot servomotors. Time- and frequency-domain features were extracted from synchronized electrical current, vibration, acoustic, and temperature signals using fixed-length sliding windows. Feature-level fusion was applied to combine complementary information from different sensor modalities. A data-driven health assessment approach was employed in which an autoencoder model trained on healthy operating data was used to generate a scalar Servomotor Health Score representing degradation progression. Fault types were identified using a Random Forest classifier, while remaining useful life was estimated in terms of operational cycles using a Gradient Boosting regression model. Experimental evaluations were carried out under repeated reference motion profiles, and representative mechanical and electrical fault conditions were introduced in a controlled manner. The results demonstrated that the proposed health score provided a smooth and monotonic degradation trend, enabling early fault detection without false alarms under healthy conditions. High classification performance was achieved for fault identification, and remaining useful life predictions showed low estimation error on previously unseen faulty servomotors. Feature contribution analysis indicated that electrical current and temperature signals provided the most robust indicators of degradation, while vibration and acoustic measurements offered complementary diagnostic information. The proposed framework was shown to be an effective and practical solution for predictive maintenance of servomotor-driven manufacturing systems such as CNC axes and robotic machining platforms operating under low-speed and variable-load conditions. Full article

Accurate non-invasive prediction of pathological complete response (pCR) following neoadjuvant chemotherapy (NACT) in breast cancer (BC) remains challenging despite its established prognostic significance. Objective: We aimed to evaluate the prognostic utility of baseline and post-NACT magnetic resonance imaging (MRI) and ¹⁸F-fluorodeoxyglucose positron emission tomography/computed tomography (¹⁸F-FDG PET/CT) for predicting pCR and survival outcomes, focusing on molecular subtype-specific performance and post-NACT imaging discordance. Methods: In this multicenter study, we retrospectively analyzed 335 patients with BC who received NACT between 2015 and 2025. Baseline (pre-NACT) and post-NACT imaging assessments were performed using MRI and ¹⁸F-FDG PET/CT. Pathological response was graded using the Miller–Payne classification system. Multivariable logistic regression was applied to identify independent predictors of pCR, whereas survival outcomes were examined using Kaplan–Meier analysis and Cox regression. Results: The overall pCR rate was 41.2%. Post-NACT imaging demonstrated complete response in 58.7% of patients by ¹⁸F-FDG PET/CT and 43.6% by MRI, both significantly correlating with pCR (p < 0.001). Pre-NACT MRI tumor size showed predictive value exclusively in Luminal A/B HER2-negative disease (area under curve = 0.681; p = 0.013). Importantly, post-NACT discordance between MRI and ¹⁸F-FDG PET/CT-based tumor size assessments was an independent predictor of both mortality (hazard ratio, 1.03) and disease progression (hazard ratio, 1.01). Conclusions: Post-NACT MRI and ¹⁸F-FDG PET/CT findings correlate strongly with pCR achievement, whereas pre-NACT MRI tumor size predicts pCR only in hormone receptor-positive HER2-negative subtypes. Importantly, post-NACT imaging discordance independently predicted mortality and disease progression, suggesting that dual-modality imaging assessment may identify high-risk patients requiring intensified surveillance. Full article

Accurate, non-contact vital-sign monitoring promises a scalable alternative to conventional sensors, yet low signal quality and long recording times have limited real-life adoption. We present a dual-modality system that combines Eulerian video magnified remote photoplethysmography (rPPG) from facial videos with optical flow-based shoulder tracking to estimate heart rate (HR) and respiratory rate (RR) from ultra-short 15 s recordings. With 200 participants, each providing 2 videos, 387 videos passed strict usability criteria, excluding flicker, blur, occlusion, and low illumination. For 15 s recordings, the HR estimates reached 98.5% accuracy within a ±10 beats per minute tolerance (MAE = 3.25, RMSE = 4.88, r = 0.93; p < 0.05) and the RR estimates achieved 98.4% accuracy within a ±5 respirations per minute tolerance (MAE = 0.69, RMSE = 0.87, r = 0.90; p < 0.05), exceeding prior studies that required 30 to 60 s recording lengths. Computational analysis on a standard home computer confirmed feasibility, with near real-time performance achievable on optimized hardware. By integrating complementary modalities and rigorous video quality control, the system overcomes low-SNR challenges, delivering high-fidelity, clinically validated vital signs monitoring. These results establish a robust, scalable, and precise framework for clinical and home care, demonstrating that accurate, contact-free HR and RR monitoring can now be achieved in seconds, making rapid, real-life vital signs assessment practical and accessible. Full article

This article investigates how cognitive and grammatical mechanisms shape variable singular–plural agreement in Spanish perception–verb constructions, a domain where speakers alternate between agreement with the postverbal NP₂ and agreement with the infinitival complement. Building on usage-based and cognitive linguistics approaches, this study examines whether factors related to perceptual modality and conceptual salience underlie these alternations. A corpus analysis of pronominal infinitive constructions with ver and oír reveals divergent patterns across modalities, with visual perception favoring plural agreement and auditory perception favoring singular agreement. To evaluate whether these tendencies reflect deeper linguistic preferences, an acceptability-rating task systematically manipulated modality, agreement, and animacy. The results show no overall interaction between modality and agreement, but they identify a robust effect of animacy: sentences with human referents received higher ratings than those with inanimate referents. Moreover, animacy modulated the influence of modality and agreement in opposite directions, suggesting that speakers’ evaluations are sensitive to the ontological nature of the perceived stimulus. Together, the findings show that agreement variation reflects flexible conceptual construal and that corpus and experimental evidence offer complementary insights into the interface between morphosyntax, perception and salience in Spanish. Full article

Edible Bird’s Nest (EBN) is a highly prized food product, making it a frequent target for economic adulteration. Consequently, robust quality assurance is paramount to protect consumers and ensure market integrity. A significant barrier to effective quality control, however, is an incomplete understanding of the natural chemical variability within authentic EBN. This variability, influenced by factors such as geographical origin, bird species, and post-harvest processing, can confound analytical measurements and complicate the definition of a standard reference. This study provides an existence proof in a defined cohort, characterizing microscale chemical heterogeneity in authentic A. fuciphagus EBN. We employed a multi-modal Fourier Transform Infrared (FTIR) spectroscopy approach, integrating transmission, macro-attenuated total reflectance (ATR), and high-resolution micro-ATR chemical imaging. A diverse set of validated, authentic EBN samples was analyzed using unsupervised Principal Component Analysis (PCA) to explore the data structure. Our results reveal significant and previously unquantified spectral heterogeneity, particularly in protein and glycoprotein-related regions. In our cohort, the chemical signatures of authentic EBN do not collapse to a single, uniform profile but span a broad, multi-dimensional continuum. This inherent variability presents a critical challenge for conventional quality control methods that rely on simplistic, single-spectrum standards, which may lead to the misclassification of genuine products. By establishing a robust chemical baseline for the authentic class, this work provides the foundational data essential for developing next-generation authentication models capable of reliably distinguishing this natural variance from deliberate adulteration. Full article

Municipal solid waste incineration (MSWI) is a typical complex industrial process for achieving sustainable development of the global environment. It implements the “perception-prediction–control” mode based on domain experts by using multimodal information. To harness the complementary value of different modal data, prevent information conflicts or fusion failures caused by misalignment, and ensure the availability of multimodal datasets and the reliability of analytical conclusions, constructing a benchmark operational condition multimodal dataset is essential. The objective of this work was to create a multimodal reference database for the operational status of IMSW processes. Based on the description of the MSWI process and the analysis of the characteristics of the multimodal data, the process data is first preprocessed under different missing scenarios, missing value processing and outlier processing. Then, single-frame images of the flame video are captured on a minute scale, and the missing combustion lines are quantized by using machine vision technology. Finally, the alignment of combustion line quantization (CLQ) values with the minute time scale of process data is achieved through the multimodal time synchronization module. Taking an MSWI power plant in Beijing as the research object, the combustion flame video and process data under the benchmark operating conditions were collected. The hybrid missing value management strategy combining linear interpolation with the LRDT model improved data integrity, and a spatiotemporal aligned multimodal dataset was constructed. The standardized benchmark operating condition multimodal data was obtained to support combustion state analysis during the incineration process, pollutant generation prediction, and process optimization. Therefore, the objectives of ‘reduction, harmlessness, and resource utilization’ of municipal solid waste, addressing land resource shortages, protecting the ecological environment, and promoting the dual carbon goal can be supported. Additionally, data and technical support for environmental and urban sustainable development are provided. Full article

Simulation-based driver training has shown promise for improving hazard perception in novice drivers; however, how learners with autism adapt behaviorally, visually, and physiologically during such training remains poorly understood. This study examined the effects of a game-based, hazard-focused driving simulation on hazard detection accuracy, gaze behavior, and heart rate in novice drivers with autism using a single-case, multi-phase design. Five participants completed repeated trials across baseline, treatment, and withdrawal phases while behavioral performance, eye movements, and physiological response were recorded. Across outcome domains, participants demonstrated highly individualized learning trajectories with substantial variability in both the direction and magnitude of change. Improvements in hazard detection accuracy were not consistently accompanied by changes in gaze organization or physiological response. While one participant exhibited a canonical pattern of coordinated improvement across behavioral, visual, and physiological measures, others showed dissociation between modalities, including reduced physiological arousal without performance gains or modest accuracy improvements despite sustained physiological engagement. Exploratory peri-hazard analyses further revealed participant-specific heart rate responses aligned to hazard detection, with no uniform temporal signature associated with learning. These findings suggest that hazard perception learning in drivers with autism does not follow a single pathway and cannot be inferred from any single performance or physiological metric. Instead, multimodal, within-participant analysis is critical for capturing meaningful individual adaptation during simulation-based training. The results have implications for the design and evaluation of driver training systems, supporting flexible, learner-specific assessment frameworks and adaptive approaches that accommodate diverse patterns of engagement and learning. Full article

Astronomical spectra, which encode rich astrophysical and chemical information, are fundamental to understanding celestial objects and universal laws. The advent of large-scale spectroscopic surveys, generating tens of millions of spectra, presents significant challenges for efficient data processing and analysis. To address these challenges, we develop an AI-powered platform (named “SpecZoo”) for spectral visualization and analysis. This platform integrates modern information technology and machine learning to lower the barrier to spectral data utilization and enhance research efficiency. Its core functionalities include interactive visualization, automated spectral classification, physical parameter measurement, spectral annotation, and multi-band/multi-modal data fusion, all supported by flexible user and data management systems. It has become an essential tool for the National Astronomical Data Center, directly supporting spectral data processing and research for major projects including LAMOST, SDSS, DESI, and so on. Furthermore, the platform demonstrates strong potential for science-education integration, providing a novel resource for cultivating talent in astronomy and data science. Full article

Objective: Cerebral meningiomas are the most common primary intracranial tumors in adults. Treatment of symptomatic tumors is normally surgical; tumors not suitable for surgery can be irradiated. While there is good data on the effectiveness of acute therapy, little is known about the effects of long-term team-based multidisciplinary inpatient neurological rehabilitation. We analyzed the outcome of these patients undergoing neurological rehabilitation. Methods: We performed a retrospective analysis of patients with cerebral meningioma who underwent specialized rehabilitation. We analyzed routine demographic and clinical data; the outcome was measured with the Barthel Index (BI) in patients with a BI of ≤90 on admission. Results: We analyzed 151 patients. Neuropsychological deficits were evident in 93 patients, and 9% had speech disorders. BI increased from 66.8 to 75.2%. Examination of factors influencing treatment success revealed that the number of secondary diagnoses had an influence on the average increase in the BI. No correlation was found for the other independent variables, including age, sex, tumor localization, stage, resection (complete or incomplete), complications, and length of stay. Conclusions: Even given the limitations of our analysis, rehabilitation appears to be effective in these patients. However, further investigations with a matched control group would be desirable to verify our hypothesis. Furthermore, studies regarding optimal intensity, timing, long-term outcome, and modality of rehabilitation are necessary. Full article

Pancreatic cystic lesions (PCLs) include clinically challenging conditions that range from benign to malignant prognoses. Their prevalence is increasing, and they are often detected as incidental findings during cross-sectional imaging. Thus, endoscopic ultrasound (EUS) plays a pivotal role in investigating these lesions. In this review, we analyze the complete diagnostic potential of EUS. Contrast-enhanced EUS, contrast-harmonic EUS, and elastography are useful for distinguishing between benign and malignant forms, and detective flow imaging EUS and e-FLOW EUS have enhanced the diagnostic arsenal available. Fine-needle aspiration (FNA) is important for obtaining cystic fluid for biochemical analysis and cytological examinations. Confocal laser endomicroscopy and through-the-needle biopsy represent adjunctive techniques for refined and difficult diagnosis. Moreover, artificial intelligence could be a promising modality in the EUS world. EUS allows PCLs to be detected accurately and plays a relevant role in identifying malignant forms. Full article

This systematic review summarizes the application of artificial intelligence (AI) in classifying upper-limb motion using Electroencephalogram (EEG) and Electromyogram (EMG) signals, focusing on the field’s progression from Traditional Machine Learning (TML) to Deep Learning (DL) architectures. Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, a search of PubMed, IEEEXplore, and Web of Science yielded 301 eligible studies published up to June 2025. The results indicate a change from classical classifiers like Linear Discriminant Analysis (LDA) and Support Vector Machines (SVMs) toward DL approaches. While Convolutional Neural Networks (CNNs) remain the most frequently implemented, emerging architectures, including Long Short-Term Memory (LSTM) networks and Transformers, have demonstrated remarkable performance. Despite the rise of DL, classical models remain highly relevant due to their robustness and efficiency. This review also identifies a heavy reliance on EEG-only modalities (60%), with only 7% of studies utilizing hybrid EEG-EMG systems, representing a potential missed opportunity for signal fusion. Full article

Dysregulation of iron and copper homeostasis is a pivotal driver of regulated cell death through two distinct yet interconnected modalities: ferroptosis and cuproptosis. This comprehensive review evaluates the therapeutic modulation of these metal-driven pathways within a dual paradigm: their deployment as a cytotoxic weapon in oncology and their inhibition for neuroprotection. We synthesize evidence ranging from small-molecule synergy to advanced nanomedicine, examining how the interplay between iron and copper governs cellular fate in resistant malignancies and neurodegenerative diseases such as Parkinson’s disease and Multiple Sclerosis. In oncology, bimetallic nanoplatforms and CRISPR-Cas9 nano-ionophores exploit “iron addiction” and metabolic vulnerabilities to induce fatal lipid peroxidation and FDX1-mediated proteotoxic stress, often by circumventing efflux transporters like ATP7A/B. Conversely, neuroprotective strategies focus on site-specific chelation, utilizing brain-penetrant molecules like SK4 (targeting the LAT1 transporter) and radical trapping antioxidants like Cu^II(atsm). Importantly, we elucidate the “iron trap” mechanism, where copper deficiency inactivates multicopper ferroxidases—including ceruloplasmin and hephaestin—thereby triggering iron-dependent ferroptosis. Our analysis reveals a self-amplifying cycle of oxidative damage driven by metal-induced ATP depletion and glutathione exhaustion. By delineating the molecular machinery of iron and copper metabolism, this article provides a roadmap for leveraging regulated cell death to overcome apoptosis resistance in cancer and preserve neural integrity in chronic degeneration. Full article

Background/Objectives: It has been previously shown that bioimpedance assessment (BIA) systematically underestimates adiposity compared to densitometry analysis (DXA), though the methods correlate strongly. However, whether DXA outperforms BIA for physiology modeling—using vitamin D as a sentinel signal—remains uncertain. We compared DXA and BIA side-by-side to model (i) adiposity–25(OH)D associations, (ii) mediation-style links with metabolic outcomes, and the vitamin D–PTH–calcium axis. Methods: We performed a cross-sectional analysis of 165 children with simple obesity and no vitamin D prophylaxis collected between July 2022 and July 2025. We measured adiposity through DXA and BIA methods, laboratory 25(OH)D, and associated biochemical and clinical parameters: PTH, calcium, phosphate, glucose/insulin/HOMA-IR, lipids. Information on age, sex, and season was recorded and used to adjust for potential covariates. Parallel analyses included partial correlations, linear regression, mediation models, and Bland–Altman analysis for DXA–BIA agreement. Results: The cohort median age was 13 years; median 25(OH)D level was 21.9 ng/mL. DXA fat % exceeded BIA (46.6% vs. 36.7%). Univariately, 25(OH)D correlated inversely with adiposity (DXA rho = −0.16, BIA rho = −0.19), but adiposity was not a significant determinant of 25(OH)D after season/age adjustment with either modality. No mediation of vitamin D to metabolic associations via adiposity were detected. The vitamin D–PTH–calcium axis was robust across modalities. Conclusions: In children with established obesity, seasonal and age factors dominate 25(OH)D variability, while the adiposity contributes little within-group. Vitamin D shows endocrine but not metabolic associations, and within this homogenous pediatric obesity cohort, DXA does not outperform BIA for physiologic modeling. Full article

This paper investigates deflection deformation and premature bearing failure in deep-hole machining spindles with high length-to-diameter ratios under eccentric loading. A contact stiffness model for angular contact ball bearings was developed based on Hertz contact theory. Combined with the finite element method (FEM), a comprehensive mechanical analysis model of the spindle was established. The results show that spindles with high length-to-diameter ratios exhibit significant cantilever behavior, leading to considerable front-end deflection under eccentric loading. This deflection causes the inner and outer rings to incline, resulting in localized stress concentrations, which are the primary contributors to spindle fatigue failure. To improve the spindle’s stress distribution and dynamic performance, an optimized design replacing the metal housing with carbon fiber composite material is proposed. Static and modal analyses were performed using Abaqus and Romax. The analysis results demonstrate that the carbon fiber shell reduces self-weight deformation by 35.8%, decreases coupled deformation under self-weight and grinding loads by 28.6%, and increases modal fundamental frequencies by 20.88% to 47.41%. These improvements significantly enhance structural stiffness and dynamic stability. Experimental vibration monitoring during machine testing validated the accuracy of the modeling and simulation. Full article