Search Results (10,533)

Medical image segmentation plays a crucial role in clinical diagnosis and treatment planning. However, existing segmentation frameworks frequently exhibit high computational complexity and often fail to retain fine-grained structural details—especially along intricate anatomical boundaries such as blood vessels and tumor margins. To overcome these limitations, we propose VMMedSAM-X, an efficient and computationally economical medical image segmentation framework that incorporates structured state space modeling into the Medical Segment Anything Model (MedSAM) architecture. The proposed method adopts a state-enhanced encoder that combines extended long short-term memory (xLSTM) with two-dimensional selective scanning (SS2D) and a dual-path cross-attention mechanism to enhance long-range dependency modeling while maintaining linear computational complexity. Experiments conducted on the $1024\times 1024$ ACDC cardiac MRI dataset show that the proposed encoder reduces floating-point operations from 369.44 G to 17.36 G and achieves a $2.4\times$ improvement in inference speed compared with the Vision Transformer (ViT)-based encoder. Additional evaluations on the SegTHOR and MSD-Lung datasets demonstrate consistent improvements in Dice Similarity Coefficient (DSC) and Intersection over Union (IoU) metrics over MedSAM and Vision Mamba U-Net (VM-UNet) baselines. These results indicate that the proposed framework provides an effective and computationally efficient solution for high-resolution medical image segmentation. Full article

In response to the subjective issues, inconsistent quality standards, high labor intensity and low sorting efficiency during the drying process of green pepper, an improved YOLOv11 algorithm was proposed for quality detection. A multi-scale edge enhancement module (MEEM) is introduced into the backbone network, replacing the original basic C3K2 module with C3K2-MEEM to enhance the extraction of detailed features in images of dried green Sichuan pepper and prevent missed detections, false detections, and boundary confusion. The LRSA module is integrated into the 10th layer of the backbone network to improve the clarity of the tumor-like texture of the Sichuan pepper and reduce the influence of impurities, automatically allocating attention based on feature similarity to preserve local information. In the neck layer, the DPCF module is added to the FPN+PAN feature fusion stage to achieve multi-scale feature collaboration, meeting the detection requirements of dried green Sichuan pepper. The results show that the accuracy recall rate, mean average precision, and model size of the improved MLD-YOLOv11 algorithm are 92.1%, 96.6%, 95.6%, and 11.06 MB, respectively. Compared with the training results of the original YOLOv11 model, the average accuracy of the improved model has increased by 2.2 percentage points, and GFLOPs have definitely decreased by 2 G, with parameter reduction of approximately 3.10%. Compared with other mainstream models, the MLD-YOLOv11 model has significant advantages in terms of mean average precision, model size, and floating point operations per second, making it more suitable for industrial applications and providing an efficient, accurate, and lightweight solution for the quality detection of dried green Sichuan pepper. Full article

Background: Magnetic Particle Imaging (MPI) is an emerging biomedical imaging modality that detects superparamagnetic iron oxide nanoparticles (SPIONs), providing high contrast, sensitivity, and quantification capabilities without ionizing radiation, making it particularly suitable for cancer diagnostics. Considerable engineering efforts are underway to translate MPI technology to clinical settings. Most of these MPI scanners feature a cylindrical bore geometry similar to that of other clinical imaging modalities, which limits their potential application primarily to head scanning. Methods: We have developed a single-sided MPI scanner designed to expand the modality’s applicability to other regions of the human body through a unique hardware design developed in our previous work. Imaging experiments were performed on an anatomical breast phantom containing implanted SPION point sources placed at anatomically plausible locations for breast tumors. These point sources served as artificial tumors for evaluating the system’s suitability for breast imaging applications. Results: The scanner successfully detected and clearly resolved the implanted SPION tumors in two orthogonal imaging planes. Tumor positioning was independently validated by ultrasound imaging, confirming MPI’s accurate localization. In addition, sensitivity measurements demonstrated a detection limit of 4.0 $\mathsf{\mu }\mathrm{g}$ of iron, below the estimated 4.8 $\mathsf{\mu }\mathrm{g}$ sensitivity threshold required for breast tumor detection with electronic depth scanning up to 3.5 cm deep. Conclusions: Together, these results demonstrate the capability of a single-sided MPI geometry for breast imaging applications. Imaging an anatomical breast-shaped volume presents significant challenges for MPI due to the size and accessibility constraints of conventional hardware. The results presented highlight the advantages of this approach and support its potential to extend MPI from small-animal imaging to clinically relevant applications. Full article

Frequency response analysis (FRA) is recognized as an effective method in power transformer winding fault diagnosis. However, the traditional numerical index methods focus on the overall features of FRA curves, making it difficult to capture subtle deformations in transformer windings. Similarly, existing digital image processing methods rely on a single feature or a simple feature combination without adaptive fusion. These methods ignore differences in the data distributions of features, leading to feature mismatch, the loss of sensitive fault information, and lower diagnostic accuracy. To solve this problem, a novel adaptive multiple-image-feature fusion method for transformer winding fault diagnosis is proposed. First, a multi-dimensional feature space combining image pixel matrix similarity, morphological features, and image texture features is built to decode the difference in fault of FRA images. Second, the multiple kernel learning (MKL) framework is used to dynamically adjust the fusion weights of different kernels to make features compatible and remove redundant information. Finally, comparative and ablation experiments show that the proposed method outperforms the traditional methods in identifying different types and levels of faults. The method achieves over 99% accuracy in fault type identification across SVM, KNN, and RF classifiers. For radial deformation (RD) severity prediction, the accuracy of the proposed model is 93.37% with SVM and 94.85% with KNN, outperforming the full-feature concatenation method. These results confirm the method’s robustness and diagnostic precision. Full article

Terahertz technology has demonstrated immense potential across a wide range of applications, particularly in the realm of THz photodetection. However, state-of-the-art detectors typically face fundamental trade-offs among sensitivity, response speed, operating temperature, and spectral bandwidth. While previous studies have shown that graphene field-effect transistors (GFETs) exhibit a broadband, room-temperature photoresponse to THz radiation—often attributed to photothermoelectric (PTE) and plasma-wave rectification effects—the similar functional dependence of these mechanisms on the gate voltage has historically made it challenging to disentangle their individual contributions. In this study, we leverage monolayer graphene as the photoactive material to overcome these limitations within a single device architecture. We present a novel THz photodetector driven predominantly by the PTE effect, facilitated by a precisely designed counterclockwise spiral antenna. The demonstrated device achieves exceptional room-temperature sensitivity, featuring a minimum noise equivalent power (NEP) of 80.7 $\mathrm{p}\mathrm{W}/\sqrt{\mathrm{H}\mathrm{z}}$ alongside a rapid response time of less than 11 μs. Furthermore, by systematically analyzing the temporal response dynamics, we unambiguously identify the PTE effect as the dominant operating mechanism. These results provide a robust strategy for the development of high-performance, room-temperature THz optoelectronics, paving the way for advanced practical applications in high-capacity wireless communications and real-time THz imaging. Full article

Syngas (synthesis gas) is a promising gaseous biofuel for small-scale power generation, but its highly variable composition, which depends on the biomass source and gasification process, poses challenges for engine optimization. This study investigates syngas–diesel dual-fuel combustion in an optically accessible engine using chemiluminescence imaging of OH*, CH*, and CH₂O* to characterize ignition and flame development. Three representative syngas compositions—Downdraft, Updraft, and Fluidbed—were examined. The Fluidbed composition exhibited the weakest OH* signal, approximately one-third of that observed for the other two, primarily due to its higher CO₂ dilution and lower H₂ content. Ignition delay trends were strongly correlated with dilution level: Downdraft and Updraft showed similar delays despite different H₂/CO ratios, while larger CO₂ shares led to longer delays and flattened heat-release rates. CH* and CH₂O* chemiluminescence showed better agreement with combustion timing than OH*. Methane enrichment enhanced flame propagation and reduced ignition delay, partially offsetting CO₂ dilution effects. Full article

Hydrogels derived from biopolymers have attracted considerable interest in biomedical applications because of their biocompatibility and structural similarity to the extracellular matrix (ECM). Bacterial Cellulose (BC), despite being a promising biopolymer for hydrogel preparation, lacks antimicrobial properties itself. To address this drawback, we prepared Probiotic Bacterial Cellulose (PBC) in our laboratory, which has intrinsic antibacterial properties. No research was found on the preparation of a hydrogel using PBC and κ-carrageenan, which motivated us to develop a PBC/κ-carrageenan-based hydrogel. In the study, a novel biocomposite hydrogel system has been developed by integrating PBC with κ-carrageenan, yielding a multifunctional hydrogel with enhanced antibacterial properties and biocompatibility. The novel hydrogel has been evaluated for its structural, physicochemical, antibacterial, and biocompatible properties. Fourier transform infrared spectroscopy (FTIR) analysis confirmed the formation of intermolecular interactions between PBC and κ-carrageenan. Scanning electron microscopy (SEM) images revealed a porous internal morphology and the presence of probiotic bacteria within the hydrogel networks. Porosity analysis and swelling behaviour indicated an elevated water uptake capacity and structural stability. The composite hydrogel demonstrated promising antibacterial properties against pathogenic bacteria Escherichia coli (Gram-negative) and Staphylococcus aureus (Gram-positive) and exhibited favourable in vitro biocompatibility. The developed PBC/κ-carrageenan hydrogel exhibits a synergistic combination of porosity, swelling capacity, biocompatibility, and antibacterial activity, making it a potential candidate for healthcare applications viz. wound healing and other tissue engineering applications. Full article

The development of realistic breast phantoms is critical for the evaluation of imaging systems and quantitative image analysis methods. In this work, breast samples derived from the same digital model were produced using 3D printing technology and evaluated for structural similarity and reproducibility. Four independently manufactured phantoms were imaged using mammography and breast tomosynthesis. Radiomic features were extracted from regions of interest in order to assess inter-phantom variability. The results showed very good agreement between the four printed phantoms. Most first-order and GLCM radiomic features exhibited very low inter-phantom variability, indicating consistent structural and intensity characteristics. Neighborhood-based texture features showed slightly higher variability, reflecting their sensitivity to local structural differences. Fractal and power spectrum analyses also confirmed the high structural similarity of the phantoms. These results indicate that the proposed manufacturing approach can produce reproducible breast imaging phantoms suitable for mammography and tomosynthesis imaging studies, with potential applications in imaging system evaluation and radiomic research. Full article

Physics-guided computational imaging typically aggregates data fidelity, geometric reconstruction, and sensor consistency into a single scalar loss. In low signal-to-noise ratio (low-SNR) non-line-of-sight imaging, this centralized approach creates asymmetric gradient conflicts where the dominant constraints suppress physically meaningful updates. We propose treating multi-constraint training as a gradient coordination problem rather than scalar balancing. Our framework coordinates heterogeneous objectives through branch-wise gradient routing: soft conflict projection (PCGrad), hard physical constraint enforcement (PhysGuard), learnable sensor calibration, and a staged training protocol that decouples representation learning from nuisance parameter estimation. On held-out test scenes, the fully staged model improved the peak signal-to-noise ratio (PSNR) from 19.09 dB to 20.49 dB and the structural similarity index (SSIM) from 0.67 to 0.71 over the baseline, with consistent gains across the 48, 28, and 25 dB SNR levels. Qualitative evaluation on seven real-world scenes indicates sharper structure recovery and fewer artifacts. In this NLOS setting, gradient-level coordination is more reliable than scalar aggregation under heterogeneous constraints. Full article

Model-Agnostic Meta-Learning (MAML) and Prototypical Networks (ProtoNet) establish the foundational paradigms for few-shot classification. However, MAML suffers from optimization instability caused by reconstructing classification boundaries for every new task. Conversely, ProtoNet lacks the internal mathematical capacity necessary for task-specific parameter adaptation under domain shifts. To reconcile these structural limitations, we introduce Metric-based Model-Agnostic Meta-Learning (M²AML). By completely excising the parameterized classification layer from the episodic adaptation sequence, our framework replaces traditional inner-loop classification with a dynamic self-exclusive geometric similarity metric. Substituting functional mappings with spatial distance optimizations efficiently resolves evaluation conflicts, thereby establishing perfectly synchronized inner and outer learning rates alongside substantially accelerated adaptation steps. Extensive experiments across mini-ImageNet, tiered-ImageNet, and CIFAR-FS validate our approach against a comprehensive array of established algorithms. To ensure strictly fair comparative evaluations, we meticulously reproduce the MAML, ProtoNet, and Proto-MAML baselines. Empirical results demonstrate that M²AML achieves state-of-the-art performance across most evaluation settings, delivering absolute accuracy improvements ranging from 0.1% to 2.1% over existing leading models. Full article

Although renal disease in Syrian hamsters (SHs) has been reported, imaging studies of normal kidneys in this commonly used pet and laboratory species are lacking, despite the key role of imaging in diagnosis. This study aimed to examine kidneys of Syrian hamsters using radiographic and anatomical methods, focusing on kidney location, visibility, size, and its ratio to the second lumbar vertebra, along with the effects of sex, body size, side, and recumbency. Abdominal radiographs were obtained from 29 clinically healthy adult Syrian hamsters of both sexes to assess kidney visibility, position, and size as well as the length of second lumbar vertebral body on lateral and ventrodorsal (VD) views, followed by an in situ anatomical study for comparative analysis. The kidneys were typically located opposite the first to third lumbar vertebrae. On VD views, the left kidney was generally visible, whereas the right was identified in only 28%. The mean values of radiographic RKL, LKL, and 2LVL were 15.2 mm, 12.44 mm, and 14.27 mm, respectively, and the KL/2LVL ratio ranged from 2.66 to 4.00. No significant sex differences were observed in KL or the KL/2LVL ratio in either anatomical or radiographic measurements (p > 0.05). Sex had a significant effect on both radiographic and anatomical 2LVL measurements, with females generally showing higher values than males. Unlike the anatomical measurements, no significant differences between sides were found in radiographic KL and the KL/2LVL ratio. The radiographic RKL, LKL, and 2LVL were significantly larger than those obtained from anatomical measurements. No significant correlation was found between KL, 2LVL, or the KL/2LVL ratio and body length or body weight in either radiographic or anatomical measurements, except for a correlation between body weight and anatomical KL. Right and left kidneys were symmetrically placed, as in rats, but differed from rabbits and guinea pigs. Kidney visibility on VD views was similar to that reported in rabbits. Radiographic RKL, LKL, and 2LVL values differed from those of rodents and rabbits. The radiographic ratio was larger than the values reported in rats, chinchillas, guinea pigs, and rabbits. A single KL-to-2LVL ratio reference range applies to both kidneys and sexes, simplifying clinical assessment. Full article

Permian shale is the largest and most promising shale gas exploration target in southern China after Silurian shale. The fine evaluation of shale reservoirs is a prerequisite for large-scale exploration and development. Based on the fractal method, this study, through the combined technology of nuclear magnetic resonance cryoporometry (NMRC) and Image recognition software (Image-J), clarifies the pore size distribution of Permian shale in the HGG Basin. The purpose of this study is to characterize the macroscopic parameters of Permian shale and reveal the level of reservoir space development in Permian shale. The controlling factors of porosity and pore structure are demonstrated. It is suggested that Permian shales in the HGG Basin have organic carbon contents similar to marine shales. In the favorable interval of the Dalong Formation, the average organic carbon content is comparable to that of the LMX pay zone. The lower Longtan shales have the highest organic carbon and the greatest gas generation potential, followed by the Dalong shales. TOC is the primary control on porosity in the lower Longtan and Dalong formations, whereas clay minerals dominate the control in the upper Longtan. Abundant pores between grains and between layers within clay minerals account for most of the porosity in Upper Longtan shale. In the lower Longtan and Dalong formations, organic pores are pervasive, explaining the difference in the dominant controls on porosity between these intervals. Clay minerals are a key control on the development of Permian shale reservoirs. The fractal dimension of adsorption pores (D_A) has no clear relationship with the total clay content, is negatively correlated with the illite content, and shows no clear relationship with the chlorite content. In contrast, the fractal dimension of flow pores (D_S) shows a weak positive correlation with the total clay content, a clear positive correlation with the illite content, and a negative correlation with the chlorite content. When illite interacts with water, it tends to break down and plug pores, an effect that is especially pronounced in the smallest pores hosted by organic matter; this accounts for the negative correlation between D_A and the illite content. In larger, flow-bearing pores, disintegrated illite roughens otherwise smooth walls between and within grains, increasing structural complexity and raising D_S. By contrast, reactions between chlorite and pore fluids tend to smooth the walls of flow pores, reducing structural complexity and lowering D_S. Full article

The rapid evolution of malware families poses significant challenges for cybersecurity systems, particularly when newly emerging threats lack sufficient labeled data. Although image-based deep learning approaches have achieved strong performance under fully supervised conditions, their dependence on retraining limits adaptability in dynamic environments. To address this issue, we propose a Retrieval-Augmented Few-Shot Malware Detection Framework that integrates binary-to-image visualization, multimodal embedding using a frozen Vision–Language Model (Qwen2.5-VL), and similarity-based external memory retrieval. Malware binaries are converted into grayscale images and embedded into a semantic vector space without task-specific fine-tuning. During inference, query samples retrieve similar support embeddings from a vector database, and predictions are generated through similarity-weighted aggregation, enabling adaptation without parameter updates. Evaluated on the MalImg dataset with 25 malware families under 1-shot to 10-shot settings, the framework achieves 0.886 accuracy in the 10-shot configuration. Ablation results demonstrate that combining VLM embeddings with retrieval mechanisms provides consistent improvements over individual components. These findings highlight the effectiveness of decoupling representation learning from adaptation for scalable few-shot malware detection. Full article

Ti alloys are widely used in aerospace and biomedical fields due to their high mechanical properties under severe loading. Interest in additively manufactured Ti6Al4V has increased, but further research is needed to fully characterize their properties. This work compares the effects of surface properties, internal defects, microstructure, hardness, and Hot Isostatic Pressing (HIP) or Vacuum Heat Treatment (VHT) on the fatigue behavior of Ti6Al4V produced by Selective Laser Melting (SLM) and Electron Beam Melting (EBM). Printing parameters and post-processing were optimized to achieve high density and minimal porosity, providing a solid basis for realistic fatigue comparisons. Samples were characterized in terms of microstructure (optical microscopy and SEM), mechanical properties (hardness mapping), surface texture (confocal microscopy), and internal defects (image-based analysis). Uniaxial fatigue limits were determined by a Dixon-Mood staircase method, and failed specimens were analyzed for fracture surfaces and defect areas. Applied load on flaws was evaluated to identify root causes of fatigue failure. Results showed that fatigue of as-printed samples is governed by surface roughness, while machined specimens are controlled by internal defect size. Machining increased the fatigue limit roughly threefold, and HIP further improved it by 10–20% by reducing internal porosity. In conclusion, with properly optimized melting parameters, both EBM and SLM produce similar mechanical performance at comparable roughness, supporting their use for structural components. Full article

With the continuous advancement of polarization spectral sensing technology, multi-band polarization image fusion has emerged as a novel approach to image fusion. By integrating spectral and polarization information, this method overcomes the limitations of relying on a single information source and significantly improves overall image quality. To address this, this paper proposes a new polarization spectral fusion algorithm. First, feature matching is employed to achieve pixel-level spatial alignment of multi-band polarization images. Then, a fusion strategy based on multi-scale decomposition and singular value decomposition is adopted to preserve structural information and fine details. Subsequently, fractional-order processing and guided filtering are applied to enhance details and suppress noise. Finally, a progressive reconstruction from low to high scales is performed to ensure hierarchical consistency and information integrity throughout the fusion process. In addition, spectral information is utilized for color restoration, enabling the final image to achieve high spatial resolution while maintaining natural and rich color representation.Experimental results demonstrate that the proposed method effectively integrates features from different spectral bands and polarization information while preserving maximum similarity, leading to significant improvements in both image quality and detail representation. Full article