Search Results (3,774)

Understanding how nanocomposite gels regulate pore–throat structures and flow behavior is essential for improving profile control and flow diversion in deep-water tight reservoirs. In this study, a dual-structure-regulated nanocomposite gel (DSRC-NCG) was designed, and its structure–flow coupling behavior during gel injection, curing, and degradation was systematically investigated using multiscale flow configurations, including microfluidic models, artificial cores, and sandpack systems. Microstructural evolution and pore–throat connectivity were characterized using μCT imaging, mercury intrusion porosimetry, nitrogen adsorption, and image-based flow simulations, while macroscopic flow responses were evaluated through permeability variation, dominant-channel evolution, injectivity behavior, and quantitative indices including the structure regulation index (SRI) and pore–flow matching index (HCI). The results show that increasing SiO₂ content induces a progressive optimization of pore–flow matching by refining critical throats and suppressing preferential flow channels, whereas excessive nanoparticle loading leads to aggregation and attenuation of these effects. This study proposes a multiscale structure–flow coupling framework that quantitatively connects pore–throat regulation with macroscopic flow responses during nanocomposite gel injection and degradation. These findings offer mechanistic insights and practical guidance for the design of nanocomposite gels with improved flow-regulation efficiency and reversibility in deep-water tight reservoir applications. Full article

Mango is a high-value agricultural commodity, and accurate and efficient appearance quality grading and defect inspection are critical for export-oriented markets. This study proposes a dual-pipeline deep learning framework for automated mango assessment, in which surface defect classification and quality grading are jointly implemented within a unified inspection system. For defect assessment, the task is formulated as a multi-label classification problem involving five surface defect categories, eliminating the need for costly bounding box annotations required by conventional object detection models. To address the severe class imbalance commonly encountered in agricultural datasets, a copy–paste-based image synthesis strategy is employed to augment scarce defect samples. For quality grading, mangoes are categorized into three quality levels. Unlike conventional CNN-based approaches relying solely on spatial-domain information, the proposed framework integrates decision-level fusion of spatial-domain and frequency-domain representations to enhance grading stability. In addition, image preprocessing is investigated, showing that adaptive contrast enhancement effectively emphasizes surface textures critical for quality discrimination. Experimental evaluations demonstrate that the proposed framework achieves superior performance in both defect classification and quality grading compared with existing detection-based approaches. The proposed classification-oriented system provides an efficient and practical integrated solution for automated mango assessment. Full article

Background: Emotion processing is critical in the neuropathology of major depressive disorder (MDD), while its relationship with clinical treatment remains unclear. This study aims to indicate the associations between emotion processing and treatment effects following a sequential dual-site accelerated repetitive transcranial magnetic stimulation (rTMS) protocol. Methods: MDD patients were recruited to receive rTMS treatment with four sessions per day for four consecutive days, with stimulation sequentially delivered to the left dorsolateral prefrontal cortex (dlPFC) and the dorsomedial prefrontal cortex (dmPFC). Symptoms were assessed at baseline, end of treatment, and week 4 using the Montgomery–Åsberg Depression Rating Scale (MADRS), Snaith-Hamilton Pleasure Scale (SHAPS), and Fatigue Severity Scale (FSS). Emotional valence and arousal were evaluated with the Affect Rating Task (ART). Results: A total of 51 participants completed the clinical assessments and ART, with two excluded due to missing baseline data in the SHAPS and FSS. The linear mixed-effects models revealed significant improvement in depressive (p < 0.001, d = −0.343) and fatigue symptoms (p = 0.010, d = −0.572) following rTMS treatment. Neutral valence was correlated with MADRS scores at baseline (R² = 0.096, p = 0.027). In addition, changes in arousal for positive images (p = 0.047, adjusted R² = 0.097) and neutral images (p = 0.019, adjusted R² = 0.160) at treatment end were significantly correlated with MADRS improvement at week 4. Conclusions: Our study highlights the association between changes in emotional arousal and improvement in MDD following accelerated dlPFC-dmPFC dual-site rTMS treatment. Full article

Membrane fouling reduces permeate flux and treatment efficiency, yet most diagnostic methods are destructive and require offline analysis. Optical coherence tomography (OCT) enables in situ, real-time visualization; however, quantitative image extraction of thin foulant layers is often limited by manual processing and subjective thresholding. Here, we develop a reproducible OCT image-analysis workflow that combines band-pass filtering, Gaussian smoothing, and unsharp masking with a dual-threshold subtraction strategy for automated fouling-layer segmentation. Seventeen global thresholding algorithms in ImageJ (289 threshold pairs) were benchmarked against SEM-measured cake thickness, identifying Triangle–Moments as the most robust combination. For humic-acid fouling, the OCT-derived endpoint thickness (14.23 ± 1.18 µm) closely agreed with SEM (15.29 ± 1.54 µm). The method was then applied to other microfiltration foulants, including kaolin and sodium alginate, to quantify thickness evolution alongside flux decline. OCT with the optimized image analysis captured rapid early deposition and revealed periods where flux loss continued despite minimal additional thickness growth, consistent with changes in layer permeability and compaction. The proposed framework advances OCT from qualitative visualization to quantitative, real-time fouling diagnostics and supports mechanistic interpretation and improved operational control of membrane systems. Full article

Integrated photoacoustic and ultrasound (PAUS) imaging is a promising technology for both preclinical and clinical applications, as it exploits both advantages of photoacoustic (PA) and ultrasound (US) imaging in high resolutions and acoustic penetration depth, respectively. Using a shared US transducer, data acquisition (DAQ), and signal processing framework, the PAUS system provides simultaneous functional and anatomical information. To date, numerous studies have been reported to demonstrate the capabilities and proposed innovative approaches for the development of the PAUS probes and systems. Key performance parameters, including probe resolution, extending the region of interest (ROI), and increasing the scanning speed, play critical roles in improving image quality, expanding the scanning area, and reducing the scanning time, respectively. This review aims to summarize recent advances in PAUS probes and systems designed for rapid image acquisition. The principles and signal processing are introduced as the fundamentals for designing the PAUS probes and systems. The summaries of the PAUS probe and system design are presented and compared systematically. Furthermore, new approaches in the development of PAUS probes and systems are proposed to enhance their proficiencies in preclinical and clinical applications. Full article

Recent years have witnessed performance improvements in Multimodal Large Language Models (MLLMs) on downstream natural image understanding tasks. However, when applied to the function graph reasoning task, which is highly information-dense and abundant in fine-grained structural details, these models face pronounced performance degradation. The challenges are primarily characterized by several core issues: the static projection bottleneck, inadequate cross-modal interaction, and insufficient visual context in text embeddings. To address these problems, this study proposes a multimodal feature fusion enhancement method for function graph reasoning and constructs the FuncFusion-Math model. The core innovation of this model resides in its design of a dual-path feature fusion mechanism for both image and text. Specifically, the image fusion module adopts cross-attention and self-attention mechanisms to optimize visual feature representations under the guidance of textual semantics, effectively mitigating fine-grained information loss. The text fusion module, through feature concatenation and Transformer encoding layers, deeply integrates structured mathematical information from the image into the textual embedding space, significantly reducing semantic deviation. Furthermore, this study utilizes a four-stage progressive training strategy and incorporates the LoRA technique for parameter-efficient optimization. Experimental results demonstrate that the FuncFusion-Math model, with 3B parameters, achieves an accuracy of 43.58% on the FunctionQA subset of the MathVista test set, outperforming a 7B-scale baseline model by 13.15%, which validates the feasibility and effectiveness of the proposed method. Full article

In this paper, we propose a source camera identification method based on disentangled feature modeling, aiming to achieve robust extraction of camera fingerprint features under complex imaging and post-processing conditions. To address the severe coupling between image content and camera fingerprint features in existing methods, which makes content interference difficult to suppress, we develop a dual-branch deep learning framework guided by imaging physics. By introducing physical consistency constraints, the proposed framework explicitly separates image content representations from device-related fingerprint features in the feature space, thereby enhancing the stability and robustness of source camera identification. The proposed method adopts two parallel branches: a content modeling branch and a fingerprint feature extraction branch. The content branch is built upon an improved U-Net architecture to reconstruct scene and color information, and further incorporates texture refinement and multi-scale feature fusion to reduce residual content interference in fingerprint modeling. The fingerprint branch employs ResNet-50 as the backbone network to learn discriminative global features associated with the camera imaging pipeline. Based on these branches, fingerprint information dominated by sensor noise is explicitly extracted by computing the residual between the input image and the reconstructed content, and is further encoded through noise analysis and feature fusion for joint camera model classification. Experimental results on multiple public-source camera forensics datasets demonstrate that the proposed method achieves stable and competitive identification performance in same-brand camera discrimination, complex imaging conditions, and post-processing scenarios, validating the effectiveness of the proposed disentangled modeling and physical consistency constraint strategy for source camera identification. Full article

Mixing processes in stirred tanks are widely applied across various industries, but still offer significant potential for optimization. A promising strategy is the use of double-stage impeller setups instead of conventional single impellers. While multi-impeller configurations are common in tall vessels, their benefits for standard tanks with a height-to-diameter ratio of 1 are largely unexplored. This study systematically investigates the flow fields of single, identical, and mixed double-stage configurations of a Rushton turbine, a pitched-blade turbine, and a retreat curve impeller. To ensure balanced power input in mixed configurations, a refined method for harmonizing specific power via impeller diameter adjustment is proposed. Stereo particle image velocimetry is applied to visualize flow fields, supported by refractive-index matching to enable measurements in a dished-bottom tank. The results reveal substantial flow deficiencies in single-impeller setups. In contrast, double-impeller setups generate novel and significantly improved velocity fields that offer clear advantages and demonstrate strong potential to enhance process efficiency across various mixing applications. These findings provide new experimental insights into the characteristics of dual impellers and form a valuable basis for the design and scale-up of stirred tanks, contributing to more efficient, reliable, and sustainable mixing processes. Full article

Accurate segmentation of the airway tree is crucial for the diagnosis and intervention of pulmonary disease; however, delineating small peripheral airways remains challenging. The small size and complex branching of distal airways, combined with the limitations of CT imaging (partial volume effects, noise), often lead to missed bronchial segments. To address these challenges, we propose AirwaySeekNet, a dual-decoder neural network. The model introduces a Voxel-Selective Supervision (VSS) mechanism, a dynamic reliability-aware strategy that focuses training on uncertain voxels, mitigating annotation bias, and enhancing fine-branch detection. We further incorporate a Signed Distance Field (SDF) loss to enforce tubular shape constraints, improving the boundary delineation and connectivity of the airway tree. In experiments on a pig CT dataset, AirwaySeekNet outperformed state-of-the-art models, achieving higher topological completeness and finer branch detection, and the TD metric increased by 5.55% and the BD metric increased by 8.14%. It maintained high overall segmentation accuracy (Dice), with only a minor increase in false positives from the exploration of the smallest bronchi. Overall, AirwaySeekNet markedly improves airway segmentation accuracy and topology preservation, providing a more complete and reliable mapping of the bronchial tree for clinical applications. Full article

In the field of optical inspection for unpatterned wafer surfaces, this paper presents a novel inspection system designed to meet the semiconductor industry’s growing demand for high efficiency and cost-effectiveness. The system is built around the principles of simplicity, stability, speed, and low cost. Its core is a low-speed stepping rotary line-scan architecture. This architecture is integrated with a two-step phase-shifting algorithm. The combination leverages line-scan differential interference contrast (DIC) technology. This aims to transform DIC technology—traditionally used for detailed observation—into an industrialized solution capable of rapid, accurate quantitative measurement. Experimental validation on an equivalent platform confirms strong performance. The system achieves an imaging uniformity exceeding 85% across dual channels. Its Modulation Transfer Function (MTF) value is greater than 0.55 at 71.8 lp/mm. The vertical detection clearly resolves 3 nm standard height steps. Additionally, the throughput exceeds 80 wafers per hour. The proposed line-scan DIC system achieves both high inspection accuracy and industrial-grade scanning speed, delivering robust performance and reliable operation. Full article

The hyperspectral image (HSI) is rich in spectral information and has important applications in the field of ground objects classification. However, HSI data have high dimensions and variable spatial–spectral features, which make it difficult for some models to adequately extract the effective features. Recent studies have shown that fusing spatial and spectral features can significantly improve accuracy by exploiting multi-dimensional correlations. Based on this, this article proposes a spectral integration and focused attention network (SIFANet) with a two-branch structure. SIFANet captures the local spatial features and global spectral dependencies through the parallel-designed spatial feature extractor (SFE) and spectral sequence Transformer (SST), respectively. A cross-module attention fusion (CMAF) mechanism dynamically integrates features from both branches before final classification. Experiments on the Salinas dataset and Xiong’an hyperspectral dataset show that the overall accuracy on these two datasets is 99.89% and 99.79%, which is higher than the other models compared. The proposed method also had the lowest standard deviation of category accuracy and optimal computational efficiency metrics, demonstrating robust spatial–spectral feature integration for improved classification. Full article

Tunneling nanotubes (TNTs) are thin, actin-based intercellular bridges that enable long-range communication during cellular stress; yet the molecular pathway controlling their formation remains unclear. Here, using gain- and loss-of-function approaches in Cath. a-differentiated (CAD) neuronal cells, we identified a unidirectional regulatory pathway in which myosin-X (Myo10) functions upstream of the actin-bundling protein L-(LCP1) to drive TNT formation. Using Western blotting and fluorescence microscopy, we determined that overexpression of L-plastin significantly increased the proportion of TNT-connected cells, whereas L-plastin downregulation reduced TNT formation, demonstrating that L-plastin is both sufficient and necessary for maintaining normal TNT abundance. Having previously shown that Myo10 is required for TNT formation in CAD cells, we asked whether the relationship is reciprocal. Overexpression/downregulation of L-plastin had no effect on Myo10 protein levels. Conversely, Myo10 downregulation decreased endogenous L-plastin by ~30%, and Myo10 overexpression elevated L-plastin expression and TNT number, demonstrating that Myo10 acts as an upstream regulator of L-plastin. Dual-color 3D imaging revealed co-localization of Myo10 and L-plastin along TNT shafts and filopodia-like precursors (Proto-TNTs). Together, these findings demonstrate that Myo10-dependent TNT formation requires the bundling protein L-plastin, providing a framework for how stress-induced signaling cascades couple TNT initiation to actin-core stabilization during stress and disease. Full article

Background/Objectives: To develop and validate a predictive model that combines morphological features and dual-energy CT (DECT) parameters to non-invasively distinguish metastatic from benign axillary lymph nodes in patients with breast cancer (BC). Methods: In this retrospective study, 117 patients (median age, 65 years; 111 women and 6 men) who underwent DECT followed by axillary lymphadenectomy between April 2015 and July 2023, were analyzed. A total of 375 lymph nodes (180 metastatic, 195 benign) were evaluated. Two radiologists recorded morphological criteria (adipose hilum status, cortical appearance, extranodal extension, and short-axis diameter) and placed regions of interest to measure dual-energy parameters: attenuation at 40 and 70 keV, iodine concentration, water concentration and spectral slope. Normalized iodine concentration was calculated using the aorta as reference. Univariate analysis identified variables associated with metastasis. Multivariate logistic regression with cross-validation was used to construct two models: one based solely on morphological features and one integrating water concentration. Results: On univariate testing, all DECT parameters and morphological criteria differed significantly between metastatic and benign nodes (p < 0.01). In multivariate analysis, water concentration emerged as the only independent DECT predictor (odds ratio = 0.97; p = 0.002) alongside cortical abnormality, absence of adipose hilum, extranodal extension and short-axis diameter. The morphologic model achieved an area under the receiver operating characteristic curve (AUC) of 0.871. Increasing water concentration increased the AUC to 0.883 (ΔAUC = 0.012; p = 0.63, not significant), with internal cross-validation confirming stable performance. Conclusions: A model combining standard morphologic criteria with water concentration quantification on DECT accurately differentiates metastatic from benign axillary nodes in BC patients. Although iodine-based metrics remain valuable indicators of perfusion, water concentration offers additional tissue composition information. Future multicenter prospective studies with standardized imaging protocols are warranted to refine parameter thresholds and validate this approach for routine clinical use. Full article

Under the pressing imperative of achieving “dual carbon” goals and advancing urban low-carbon transitions, understanding how neighborhood spatial environments influence carbon emissions has become a critical challenge for enabling refined governance and precise planning in urban carbon reduction. Taking the central urban area of Xining as a case study, this research establishes a high-precision estimation framework by integrating Semantic Segmentation of Street View Images and Point of Interest data. This study employs a Geographically Weighted XGBoost model to capture the spatial non-stationarity of emission drivers, achieving a median R² of 0.819. The results indicate the following: (1) Socioeconomic functional attributes, specifically POI Density and POI Mixture, exert a more dominant influence on carbon emissions than purely visual features. (2) Lane Marking General shows a strong positive correlation by reflecting traffic pressure, Sidewalks exhibit a clear negative correlation by promoting active travel, and Building features display a distinct asymmetric impact, where the driving effect of high density is notably less pronounced than the negative association observed in low-density areas. (3) The development of low-carbon neighborhoods should prioritize optimizing functional mixing and enhancing pedestrian systems to construct resilient and low-carbon urban spaces. This study reveals the non-linear relationship between street visual features and neighborhood carbon emissions, providing an empirical basis and strategic references for neighborhood planning and design oriented toward low-carbon goals, with valuable guidance for practices in urban planning, design, and management. Full article

Fused deposition modeling (FDM) suffers from unpredictable mechanical properties in nominally identical prints. Current quality assurance relies on destructive testing or expensive post-process inspection, while existing machine learning approaches focus primarily on printing parameters rather than real-time thermal environments. The objective of this proof-of-concept study is to develop a low-cost, non-destructive framework for predicting tensile strength during FDM printing by directly measuring convective thermal gradients surrounding the print. To accomplish this, we introduce thermal fingerprinting: a novel non-destructive technique that combines Background-Oriented Schlieren (BOS) imaging with machine learning to predict tensile strength during printing. We captured thermal gradient fields surrounding PLA specimens (n = 30) under six controlled cooling conditions using consumer-grade equipment (Nikon D750 camera, household hairdryers) to demonstrate low-cost implementation feasibility. BOS imaging was performed at nine critical layers during printing, generating thermal gradient data that was processed into features for analysis. Our initial dual-model ensemble system successfully classified cooling conditions (100%) and showed promising correlations with tensile strength (initial 80/20 train–test validation: R² = 0.808, MAE = 0.279 MPa). However, more rigorous cross-validation revealed the need for larger datasets to achieve robust generalization (five-fold cross-validation R² = 0.301, MAE = 0.509 MPa), highlighting typical challenges in small-sample machine learning applications. This work represents the first successful application of Schlieren imaging to polymer additive manufacturing and establishes a methodological framework for real-time quality prediction. The demonstrated framework is directly applicable to real-time, non-contact quality assurance in FDM systems, enabling on-the-fly identification of mechanically unreliable prints in laboratory, industrial, and distributed manufacturing environments without interrupting production. Full article