Deep learning has revolutionized visual computing, but generating high-quality, content-aware holograms faces challenges such as rigid unified frameworks and reliance on manual region definition. Leveraging vision foundation models, we propose PHAROS, a novel framework that adaptively identifies salient regions for task-specific hologram generation. Unlike conventional methods like HoloNet, PHAROS integrates a controllable feature refinement pipeline via ControlNet and auto-encoded compression using VAE-KL into a classical two-stage physical process. This enables efficient handling of 4K-resolution images, specifically 3840 & times;2160. Extensive experiments on both simulation and a physical optical setup demonstrate that PHAROS achieves state-of-the-art performance, with PSNR exceeding 30 dB and SSIM above 0.9. It also excels in perceptual metrics such as Learned Perceptual Image Patch Similarity (LPIPS), outperforming existing methods by over 20% in structural fidelity. Validated by high-fidelity optical reconstructions, this work paves the way for applying foundation-model-driven holography to demanding applications such as augmented and virtual reality displays.

Adverse weather poses a major challenge to ground-based electro-optical (EO) aircraft surveillance. Existing parameter-efficient fine-tuning (PEFT) methods operate mainly in the spatial domain and treat weather as a generic domain shift rather than a frequency-dependent degradation. We propose Freq-LoRA, a frequency-domain PEFT method that applies the Type II Discrete Cosine Transform (DCT-II) to frozen encoder features, decomposes them into K=4 learned Gaussian frequency bands, and modulates the bands with an image-driven spectral gate. SpectralGate contains 140 parameters and estimates band importance from the input image’s DCT statistics, removing the need for external weather metadata at inference; weather-diverse training data are still required to learn the decomposition. On a Blender-simulated aircraft dataset covering five weather conditions, Freq-LoRA achieved a test mean Intersection-over-Union (mIoU) of 0.904, with a 95% confidence interval (CI) of [0.899, 0.908], using 559 K trainable parameters. Its point estimate differed by −0.002 from that of the weather-conditioned spatial method Feature-wise Linear Modulation (FiLM; 0.906), despite requiring no external weather metadata at inference. Relative to batch-size-matched Spatial LoRA (0.873; 95% CI: [0.867, 0.879]), Freq-LoRA had a 0.031 higher point estimate. Preliminary evaluation on real unmanned aerial vehicle (UAV) imagery yielded 0.421 mIoU (+13% relative to the zero-shot Segment Anything Model (SAM); one platform), and evaluation under six unseen image corruptions showed differences of at most 0.001 mIoU from the weather-oracle variant.

CRISPR/Cas systems hold great promise for molecular diagnostics, but their amplification-free applications are hampered by weak signals and poor quantification. Here, we developed CrisprDEM, a CRISPR/Cas12a-based digital hydrogel fluorescent-enhancing microsphere system that integrates hydrogel microsphere confinement, microfluidic digital imaging, and machine learning for ultrasensitive and quantitative nucleic acid detection without amplification. Hydrogel microspheres (HMs) efficiently captured and spatially concentrated CRISPR/Cas12a reaction reporters, achieving a 1000-fold signal amplification compared with homogeneous methods. The design of the microfluidic chip arranged the microspheres into a single-layer array, making each microsphere an independent digital reporting unit. The Intelligent Bead Analysis Software enabled automatic analysis and relative quantification via a positive bead ratio (PBR). As a proof-of-concept, we selected the respiratory adenovirus as the detection target. We optimized the CRISPR/Cas12a-microsphere enrichment reaction system and characterized the morphologies and chemical properties of the microspheres before and after enrichment. The results demonstrated that CrisprDEM technology exhibited a detection sensitivity of 10 aM for respiratory adenovirus, exhibiting no cross-reactivity with other respiratory viruses, indicating a high specificity. In the validation of 20 clinical samples, the detection results were consistent with the gold-standard real-time quantitative polymerase chain reaction (qPCR), and the PBR value showed a good linear relationship with the cycle threshold (Ct) value, enabling a relative quantification. This system expands the toolbox for amplification-free CRISPR diagnostics and holds the potential for point-of-care and early infection detection.

Digital PCR (dPCR), as a high-sensitivity technology for absolute nucleic acid quantification, holds significant value in biomedical research and environmental monitoring. However, current platforms still face challenges in multiplex fluorescence detection and rapid, high-precision droplet imaging. Moreover, the detection process is time-consuming (2-3 hours) and involves high costs. In this study, an integrated micro-droplet digital PCR (ddPCR) detection and analysis system was developed, featuring a droplet-based microfluidic chip, a high-precision thermal cycling module, and a seven-color filter-wheel-based imaging system (ATTO425 to CY7) to facilitate a seamless workflow from droplet generation to multiplex imaging. To address the challenges of identifying and segmenting massive droplets in complex fluorescence backgrounds, this paper proposes a detection method based on the You Only Look Once version 5 (YOLOv5) deep learning architecture. By integrating global coordinate remapping and sliding-window detection, the system enables rapid processing of ultra-high-resolution images (2448 × 10 000 pixels). The end-to-end analysis pipeline achieved 99.8% overall accuracy in under 800 ms. Consequently, the total detection cycle for the full digital PCR process has been successfully reduced to under one hour. Furthermore, full-process validation experiments demonstrated excellent linearity across all fluorescence channels, with R2 values exceeding 0.999, and a coefficient of variation (CV) for quantitative repeatability of less than 2% across various concentrations. These results verify the system's precision, stability, and reproducibility. The developed system significantly enhances the throughput and accuracy of ddPCR detection, and the proposed algorithm further advances the practical application of deep learning in digital PCR image analysis.

Ground-based optical remote sensing of aerial targets at kilometer-scale standoff distances requires accurate keypoint localization for six-degree-of-freedom (6-DoF) pose recovery under variable illumination, motion blur, and atmospheric degradation. Many lightweight detectors use fixed-kernel convolutions, whose spatially invariant sampling may limit adaptation to heterogeneous target geometries and spatially varying image degradation. We introduce GeoAdapt, a compact keypoint detection framework that inserts deformable convolution v2 (DCNv2) modules between the feature pyramid network (FPN) neck and the detection head. GeoAdapt also replaces the standard object keypoint similarity (OKS) loss with a combination of Wing Loss and Bone Loss. The complete model contains 5.95 M parameters, 47.9% fewer than the You Only Look Once version 8 small pose model (YOLOv8s-pose). On a synthetic ground-based optical remote sensing benchmark, GeoAdapt achieved a percentage of correct keypoints (PCK) at a threshold of 0.05 times the bounding-box diagonal (PCK@0.05D) of 89.3% and a rotation error of 11.6°, improving PCK by 11.7 percentage points over YOLOv8s-pose. Zero-shot evaluation on manually annotated real ScanEagle and Matrice 200 imagery showed consistent advantages over YOLOv8s-pose and YOLO11s-pose in all six test scenarios. A factorial ablation indicated a positive interaction between DCNv2 and the Wing+Bone loss.

Background:Real-time monitoring of bacterial dynamics is essential for preventing the progression of diabetic foot infections (DFIs). This study characterizes bacterial fluorescence in infected diabetic wounds and evaluates its utility for real-time infection surveillance during antimicrobial treatment. Methods:A diabetic rat model was established, and full-thickness wounds were inoculated with three representative bacterial species. Antimicrobial interventions were subsequently administered. Bacterial fluorescence, bacterial burden, and wound status were longitudinally assessed. Results:Inoculation with 103 CFU was sufficient to establish infection, enabling in vivo visualization of bacterial fluorescence, which emerged 1-3 days prior to the onset of clinical signs. Fluorescence intensity showed a strong association with both swab- and tissue-derived bacterial burden, and persistent fluorescence was associated with sustained infection and delayed healing. In Escherichia coli and Staphylococcus aureus infections, both antibiotic therapy and debridement markedly reduced bacterial burden, decreased fluorescence intensity, attenuated inflammation, and improved wound healing with comparable efficacy. In contrast, Pseudomonas aeruginosa infections exhibited recurrent fluorescence and bacterial regrowth following antibiotic therapy, accompanied by persistent inflammation and impaired tissue repair, whereas debridement provided superior bacterial control and more favorable healing outcomes. Conclusion:These findings elucidate the relationship between bacterial fluorescence and infection dynamics and suggest that fluorescence imaging may serve as a real-time, noninvasive tool for detecting and monitoring diabetic wound infections.

Sepsis involves life-threatening immune dysregulation where impaired neutrophil chemotaxis is a critical indicator. We developed a multi-gradient neutrophil chemotaxis analysis microfluidic chip (NCA chip) capable of simultaneously generating triple-chemotactic factor gradients (fMLP, IL-8, and LTB4). Chemotaxis trajectories of neutrophils from healthy individuals (NH = 25) and patients with sepsis (NS1 = 25) were detected, and three parameters, chemotaxis speed (V), chemotaxis index (CI), and number of chemotaxis stops (S) were manually extracted and analyzed. The three parameters were integrated into a multi-gradient-based neutrophil function analysis index (NFA Index). The NFA Index strongly correlated with sepsis severity scores (Sequential Organ Failure Assessment, Acute Physiology and Chronic Health Evaluation Ⅱ), inflammatory markers (Procalcitonin and Brain Natriuretic Peptide). In the NS1 group, the NFA Index showed robust predictive performance for 8-day mortality (AUC = 0.904; cut-off = 9.5; specificity = 92.4%; sensitivity = 75%). In an independent cohort (NS2 = 10), all patients with an NFA Index >9.5 survived, confirmed its clinical prognostic value. Furthermore, the NFA Index exhibited excellent diagnostic efficacy in distinguishing patients with sepsis from healthy individuals (AUC = 0.985; cut-off = 12.5; sensitivity = 92%; specificity = 94.3%), and achieved 100% classification accuracy in a small-sample double-blind validation (NB = 10). Additionally, a deep learning method directly analyzing chemotaxis trajectories diagnosed sepsis with 90% accuracy in the NB, supporting AI-driven immune assessment. The NCA chip and NFA Index provide a precise platform for monitoring neutrophil chemotaxis, demonstrating significant potential for sepsis diagnosis, prognosis, and immune function evaluation.

Recently, 3D indoor object detection has become increasingly important in applications such as service robotics, augmented reality and smart homes. However, due to factors like densely arranged objects, severe occlusion and sparse point cloud data, existing methods often suffer from complex architectures, high computational costs or poor performance in small object detection, making it difficult to balance accuracy and efficiency. To address these challenges, we propose an efficient multi-scale feature fusion model for 3D indoor object detection (EMF3D), offering an end-to-end, lightweight solution tailored for indoor point cloud scenarios. The method employs sparse voxel convolution to build a compact feature extraction network and introduces a sparse convolution-based squeeze-and-excitation block at the feature fusion stage to adaptively learn feature channel weights. Furthermore, the enhanced fusion module (EFM) strengthens the perception of critical structures and improves feature discriminability by aggregating multi-scale representations from earlier attention stages. Extensive experiments are conducted on three indoor point cloud datasets-ScanNet V2, SUN RGB-D and S3DIS. Results show that the proposed method outperforms major detection approaches across multiple metrics. Compared with existing methods, EMF3D achieves superior performance in small object detection and offers a better trade-off between accuracy and efficiency.

Single nucleotide polymorphisms (SNPs) are important genetic markers for disease diagnosis and precision medicine. However, conventional SNP detection methods are often limited by complex workflows, lengthy processing times, and high costs. CRISPR/Cas12a-based molecular diagnostics have emerged as a promising platform for SNP detection because of their rapid response and cost-effectiveness. Nevertheless, the high tolerance of conventional CRISPR RNA (crRNA) toward single-base mismatches limits the specificity of CRISPR/Cas12a-mediated SNP discrimination. To overcome this limitation, a truncated and dual-type mismatched crRNA (TDM-crRNA) was developed by combining a 17-nt spacer, an additional mismatch at position 2 within the seed region, and a wobble base pair at position 14. The performance of TDM-crRNA was systematically evaluated using SNP sites located both within and outside the seed region. Significantly enhanced SNP discrimination specificity was achieved in an amplification-free CRISPR/Cas12a assay, enabling reliable differentiation of both homozygous and heterozygous mutant genotypes. Furthermore, kinetic analysis and free-energy modeling revealed that the improved specificity resulted from reduced R-loop stability on mismatched targets together with an increased energetic barrier to R-loop formation, thereby enhancing the ability of Cas12a to distinguish single-nucleotide variations. Overall, TDM-crRNA provides a simple and effective strategy for improving the SNP recognition specificity of CRISPR/Cas12a and demonstrates considerable potential for SNP detection, genetic disease screening, and precision diagnostics.

Healthcare-associated infection (HAI) pathogens cause severe nosocomial outbreaks, jeopardizing patient safety and straining healthcare systems. Conventional loop-mediated isothermal amplification-lateral flow immunoassay (LAMP-ICA) allows rapid pathogen detection but is constrained by low sensitivity, a high false-positive rate, and an extended detection time. To address these limitations, we present a dual-mode (colorimetric/fluorescent) microfluidic biosensing platform based on silicon-gold/quantum dot core-shell nanoprobes (Si@Au/DQD NPs). The platform incorporates two key innovations: (1) The colorimetric/fluorescent dual-signal Si@Au/DQD nanoprobe enhances detection reliability and sensitivity through dual-signal complementary verification and multilayered QD design, halving the LAMP amplification time compared to traditional colloidal gold systems, and (2) a modular microfluidic chip integrates LAMP amplification and ICA detection within a closed system, effectively preventing leakage and contamination of amplification products. Performance evaluation showed that the fluorescence detection limit of this system for Staphylococcus aureus (S. aureus), Legionella pneumophila (L. pneumophila), and Klebsiella pneumoniae (K. pneumoniae) reaches 82-140 CFU/mL, with the entire process completed within 30 min. In addition, the detection of 25 clinical environmental samples verifies the practicality of the designed integrated detection platform. With high sensitivity, strong specificity, and dual-mode capability for qualitative colorimetric screening and quantitative fluorescence analysis, this technology offers an efficient solution for point-of-care testing (POCT) of HAI pathogens, particularly in resource-limited settings and in on-site emergency diagnostics.

Vector-borne diseases, including arboviral infections and malaria, pose a major global health threat and underscore the need for rapid, accurate, and multiplex nucleic acid detection technologies for clinical diagnosis and surveillance. However, previous studies on reagent-preloaded and switching valve-based microfluidic cartridge platforms have insufficiently addressed several key engineering challenges, particularly the reliable control of multichannel switching valves and the inherently critical requirement for uniform reagent distribution across multiple reaction zones. These limitations affect system stability and repeatability, thereby constraining their practical applicability in RNA-based arbovirus detection and DNA-based Plasmodium genotyping. We developed a Microfluidic Vector-borne Pathogen Detection (MVPD) system that introduced comprehensive advancements in fluidic architecture, reagent management, and system informatization. The MVPD system utilized a six-reaction tube microfluidic cartridge that integrated a multichannel switching valve actuated by a magnetic field-based closed-loop control mechanism, along with a dual-mode fluid-handling strategy that combined a peristaltic pump with a constant positive-pressure source for precise fluidic manipulation. Coupled with a magnetic separation module, thermal cycling unit, and six-channel optical detection, the MVPD system achieved independently verified nucleic acid extraction with > 90

Impaired neutrophil migration in sepsis is associated with a poor prognosis. The potential of utilizing neutrophil chemotaxis to assess immune function, disease severity, and patient prognosis in sepsis remains underexplored. This study employed an innovative approach by integrating a multi-tip pipette with a Six-Unit microfluidic chip (SU6-chip) to establish gradients in six microchannels, thereby analyzing neutrophil chemotaxis in sepsis patients. We compared chemotactic parameters between healthy controls (NH = 20) and sepsis patients (NS1 = 25), observing significant differences in gradient perception time (GP), migration distance (MD), peak velocity (Vmax), chemotactic index (CI), reverse migration rate (RM), and stop migration number (SM). A novel composite indicator, the Sepsis Neutrophil Migration Evaluation (SNME) index, was developed by integrating these six chemotactic migration parameters. The SNME index and individual chemotaxis parameters showed significant correlations with the Sequential Organ Failure Assessment (SOFA) score, Acute Physiology and Chronic Health Evaluation (APACHE II) score, hypersensitivity C-reactive protein (hs-CRP), and heparin-binding protein (HBP). Moreover, the SNME index demonstrated potential for monitoring sepsis progression, with ROC analysis confirming its predictive accuracy (area under the curve [AUC] = 0.895, cutoff value = 31.5, specificity = 86.73%, sensitivity = 86.71%), outperforming individual neutrophil chemotactic parameters. In conclusion, the SNME index represents a promising new tool for adjunctive diagnosis and prognosis assessment in patients with sepsis.

Bacterial autofluorescence plays a vital role in photodiagnosis (PD) and antimicrobial photodynamic therapy (aPDT), yet the autofluorescence properties of wound-associated bacteria and their responses to the physicochemical microenvironment, remain underexplored. Here, we investigated the bacterial autofluorescence of Acinetobacter baumannii, Escherichia coli, Klebsiella pneumoniae, and Staphylococcus aureus under various culture conditions, including different temperatures, NaCl concentrations, and pH levels found within wounds. Fluorescence imaging was employed to quantify red fluorescence intensity, while fluorescence spectrometry was used to correlate the observed fluorescence with spectral profiles, benchmarking them against coproporphyrin and protoporphyrin IX. Our results revealed that the selected bacteria emitted red fluorescence in vitro, consistent with their known porphyrin biosynthesis capabilities. The intensity of red fluorescence was primarily dependent on the bacterial species, growth phase, and culture conditions. Elevated culture temperature accelerated the fluorescence metabolism, whereas increasing NaCl concentrations and alkaline pH levels inhibited red fluorescence in a dose-dependent manner. Linear regression revealed a strong positive correlation between red fluorescence intensity and peak fluorescence emission when excited at 405 nm. The biosynthesis of endogenous porphyrins varied both across and within bacterial species, with distinct porphyrins produced under specific conditions. The emission spectra of the gram-positive S. aureus consistently aligned with coproporphyrin, while the gram-negative A. baumannii, E. coli, and K. pneumoniae typically displayed fluorescence peaks characteristic of protoporphyrin IX. Notably, K. pneumoniae shifted to coproporphyrin with extended culture duration or more favourable conditions, and E. coli exhibited a similar transition as the pH increased to 9. We concluded that the local physicochemical conditions found within wounds could affect the autofluorescence and endogenous porphyrins of the bacteria, which may, in turn, have connotations for PD and aPDT.

Panoptic Scene Graph Generation (PSG) aims to segment objects and predict the relation triplets within an image. Despite the impressive achievements in PSG, current methods still struggle to capture fine-grained visual context, eschewing spatial and situational information in favor of visual features related to object identity. This limitation naturally impedes the model's ability to distinguish subtle visual differences between relation triplets, such as "cat-on-person" and "cat-lying on-person". To address this challenge, we propose CVCPSG, a novel DETR-based method that uncovers composite visual clues for PSG. Specifically, drawing inspiration from how humans capture visual context using diverse visual clues, we first construct a composite visual clues bank based on three key aspects: object, spatial, and situational. Then, we introduce a multi-level visual extractor to align visual features from objects, interactions, and image levels with the composite visual clues bank. Additionally, we incorporate a cross-modal learning module with a multitower architecture to seamlessly integrate visual clues into the relation decoder, thereby improving PSG detection. Extensive experiments on two PSG benchmarks confirm the effectiveness and interpretability of CVCPSG.

LiDAR-based global localization plays a crucial role in autonomous driving and robotic systems. However, real-time changes in environmental conditions present challenges for global localization. Current methods predominantly rely on sparse LiDAR point clouds with single-feature information (semantic or geometric), leading to insufficient accuracy and robustness in 6-DOF pose estimation and similarity matching. In this work, we propose a novel semantic-geometric combined 3D histogram descriptor, called the SGHD descriptor, with robust place recognition and 6-DoF pose estimation capabilities. We first extract semantic and geometric features of objects through operations such as point cloud clustering and semantic segmentation. Next, we describe the distribution of neighboring objects using topological structures. To improve object discriminability, we construct triangular descriptors for each object based on semantic information and relative distance. Subsequently, we consider the complementary nature of semantic and geometric information and propose a 3D histogram descriptor. This 3D histogram descriptor encodes semantic, spatial angle and relative distance information, enhancing the invariance of the triangle descriptor. We perform similarity discrimination and 6-DOF pose estimation through coarse-to-fine matching of local and global semantic maps. Overall, the proposed descriptor achieves high-accuracy and high-robustness 6-DOF pose estimation and place recognition for global localization. We extensively compared the proposed SGHD descriptor with state-of-the-art methods on the public SemanticKITTI dataset. Quantitative results demonstrate that SGHD exhibits higher adaptability and significant improvements compared to its counterparts. The implementation of SGHD will be released at https://github.com/wf-hahaha/SGHD

The automatic detection and analysis of ommatidia, the fundamental visual units of insect compound eyes, is vital for advancing biological research and understanding insect vision. Compound eyes, with their multi-lens structure, provide a broad field of view and high visual resolution. Researchers typically obtain ommatidia structures through CT imaging of insect compound eyes. However, processing these high-resolution 3D images requires significant computational resources due to the complexity and scale of the data. Existing methods struggle with inefficiency and poor robustness, particularly when dealing with large and complex datasets. To overcome these challenges, we introduce a groundbreaking framework, P-ODA, for fast and robust ommatidia detection by integrating parallel computing into the CT image analysis insect compound eyes. By harnessing the power of both Graphics Processing Units (GPUs) with Compute Unified Device Architecture (CUDA) and multicore Central Processing Units (CPUs), we present a novel approach to enhance the Ommatidia Detection Algorithm (ODA). Our method efficiently processes large-scale datasets and demonstrates unprecedented scalability. We offer two independent implementations: one utilizing CUDA for GPU acceleration, and another leveraging Python's multicore CPU capabilities. Experimental results show that P-ODA dramatically improves execution time and scalability over traditional ODA methods. Moreover, the system is designed for ease of use with a Linux graphical interface. This innovative framework sets the stage for the next generation of biological dataset processing and hybrid parallel computing systems.

Fluorescence lateral flow assays (FLFA) based on quantum dot probes have attracted significant attention in recent years due to their high sensitivity and quantitative detection capabilities. FLFA requires the use of a straightforward fluorescence reader for quantitative detection. Most fluorescence readers employ narrowband filters for auxiliary imaging, which facilitates the acquisition of high-contrast signals. However, during trace detection, the weak signal from FLFA can be easily lost due to optical flux loss associated with narrowband filters, thereby indirectly diminishing detection sensitivity. To address this issue, we developed a fluorescence signal reader that employs CMOS imaging without optical filters and proposed a highly sensitive signal detection algorithm based on continuous wavelet transform (CWT) to identify weak fluorescence signals with low contrast. Experimental results demonstrate that the method achieves a fluorescence detection sensitivity for quantum dots of 10-10 mol/L, with a relative standard deviation (RSD) of < 1.45%. The designed filter-free detection system and CWT analysis algorithm were applied to various FLFA systems (including the sandwich method and the competition method), with the correlation coefficient (R2) between all detection results and sample concentration exceeding 0.997. The findings of this study offer a highly sensitive signal detection method for the precise quantification of FLFA.

Background Multiplex nucleic acid detection has emerged as an indispensable tool in modern diagnostics, with multiplex PCR established as the gold standard due to its high throughput, cost-effectiveness, and rapid turnaround time. Despite its advantages, conventional multiplex PCR remains constrained by persistent technical challenges—notably primer-dimer artifacts and nonspecific amplification. These limitations not only impose stringent requirements on primer design and reaction optimization but also restrict analytical throughput, ultimately hindering the technology's full potential in critical clinical applications where reliable multi-target detection is paramount. Results Here, we present a shared-label-tagged primer-mediated booster PCR (SLP-bPCR) system featuring an innovative primer architecture that enables efficient single-tube nested amplification using a single primer. This transformative approach achieved substantial improvements in multiplex detection performance by simultaneously addressing multiple fundamental limitations of conventional multiplex PCR. Quantitative comparisons demonstrated that the SLP-bPCR system achieved a 1.95 ± 0.10-fold improvement in signal-to-background ratio and a 3.25-fold enhancement in analytical sensitivity compared with conventional PCR. These advancements enabled reliable 5-plex detection of human reference genes at ultralow template inputs (0.01 ng total DNA) with near-ideal amplification efficiency (101.6 % ± 1.2 %), while maintaining robust performance consistency (97.4 % ± 1.9 %) when scaled to more complex 10-plex assays. Moreover, the method integrated seamlessly with droplet digital PCR (ddPCR) platform, enabling sensitive (100 %), and specific (93.8 %) identification of bloodborne pathogens, alongside precise genotyping of hepatitis B virus (HBV) that showed complete concordance with Sanger sequencing. Significance The SLP-bPCR technology establishes a transformative framework for multiplex nucleic acid detection, delivering unprecedented analytical performance through its unique combination of versatility, efficiency, and scalability. By addressing critical gaps in throughput, accuracy, and practicality, this strategy is poised to advance both high-throughput research and clinical diagnostics, particularly for applications demanding multi-analyte profiling at extreme sensitivity.

Vision-based multi-robot global localization in large-scale environments is a challenging task due to real-time environmental changes, which complicate data association between viewpoints. Recently, researchers have proposed methods to enhance viewpoint invariance by organizing object semantic information using graph structures. However, previous works still face limitations in accuracy and robustness in real-world scenarios, primarily due to the limited descriptive power of feature information (semantic or geometric) and susceptibility to noise. In this paper, we propose a novel semantic-geometric triple constraint-based graph matching multi-robot global localization method, called SGT-MGL. We first extract the semantic and geometric features of objects and describe the distribution of neighboring objects using topological structures. To improve object discriminability, we construct a triangular descriptor for each object based on semantic information and relative distances. Considering the complementarity between semantic and geometric information, we introduce a 3D histogram descriptor that encodes semantic, spatial angular, and relative distance information, enhancing the invariance of the triangular descriptor. To further mitigate noise, we propose a candidate point selection strategy guided by global geometric structures and employ a combined local and global graph matching approach for 6-DOF pose estimation. We extensively evaluate SGT-MGL on three public datasets, demonstrating superior accuracy and robustness. The implementation of SGT-MGL will be available.

Colloidal gold nanoparticles used as signal tags in traditional side-flow immunochromatography are often limited in sensitivity due to their weak colorimetric signals. To address this issue, we employed graphene oxide (GO) as a substrate and fabricated a nanozyme with dual colorimetric/catalytic signal amplification. This was accomplished by adsorbing 30 nm platinum nanoparticles (Pt NPs) onto the GO surface to provide colorimetric signals, while concurrently incorporating 5 nm gold-iridium (AuIr) nanoparticles into the interstitial spaces of the Pt NPs to enhance their catalytic performance. GO-Pt30-AuIr nanozyme utilizes the large surface area, excellent stability, and dispersibility of GO to co-assemble with Pt NPs and AuIr NPs, significantly improving the catalytic performance of the system. The detection system based on GO-Pt30-AuIr-LFA can achieve highly sensitive double detection of heavy metal Cd2 + and agricultural residue IMidacloprid (IMI) within 20 min, with a detection limit in the pg/mL range. In addition, the GO-Pt30-AuIr nanozyme developed was successfully applied to the rapid and sensitive detection of two contaminants in food and environmental samples. The detection method proposed in this study shows significant application potential in the rapid screening of small and medium-sized molecules in complex substrates, and provides a new technical approach for environmental and food safety monitoring.