Search Results (2,581)

Objective evaluation of calligraphic handwriting similarity is essential for digital calligraphy analysis, yet it remains difficult because brush writing differences are expressed through coupled variations in spatial layout, stroke contour, stroke width, topology, and global visual appearance. Existing human assessment is informative but difficult to reproduce at scale, whereas conventional image similarity or classification models do not directly measure how closely a handwritten copy preserves the visible evidence of a reference stele. Here, we propose a stele-guided deep visual and structural framework for assessing calligraphic handwriting similarity using the Yanta Shengjiao Xu stele as the reference. The framework combines non-generative image preprocessing, quality-controlled glyph segmentation, automatic same-character pairing, and a paired neural similarity model that integrates self-supervised visual features, structural image channels, explicit geometric-topological descriptors, and cross-image attention. The model outputs five dimension-specific scores covering spatial layout, stroke geometry, stroke width and ink appearance, topology, and deep visual similarity, together with an overall score, a heuristic uncertainty index, and local difference visualisations. Without relying on expert similarity labels, the model is trained and validated using controlled perturbations of original stele glyphs, enabling objective tests of monotonic degradation under known structural, contour, width, topology, and combined distortions. On the locked controlled test set, the full model achieved 96.65% within-one-level ordinal accuracy, 0.9990 damage-discrimination AUC, and 1.0000 different-character AUC, outperforming DINO-only, ResNet18, and handcrafted structural baselines. Application to validated handwritten copybook pairs further demonstrated the feasibility of producing quantitative, dimension-specific similarity profiles accompanied by local difference visualisations for individual calligraphic samples. This study establishes a reproducible reference-based framework for evaluating visual handwriting restoration to classical stele evidence, while its conclusions are limited to objective image-level similarity rather than expert aesthetic judgement. Full article

Compromising electromagnetic emanations (CEs) from digital video interfaces represent a significant risk to information confidentiality in systems processing sensitive visual data. Previous studies have demonstrated the feasibility of reconstructing video content from electromagnetic leakage of Transition-Minimized Differential Signaling (TMDS)-based interfaces such as HDMI and showed that reconstruction quality depends on displayed image characteristics. However, predictive approaches for evaluating content-dependent leakage susceptibility remain insufficiently explored. This paper proposes a content-dependent multi-domain framework for modeling and evaluating compromising emanations from HDMIs. The methodology combines near-field electric, near-field magnetic, and far-field measurements to establish relationships between electromagnetic leakage behavior and displayed image properties. Based on experimentally reconstructed TMDS-related signals, a measurement-informed predictive model is developed to estimate the reconstructability of different display configurations without exhaustive measurement campaigns. Experimentally reconstructed and simulated images are quantitatively compared using image-processing similarity metrics under varying display and noise conditions. The obtained results demonstrate consistency between measured and simulated reconstruction characteristics. The proposed approach can significantly reduce the complexity and duration of compromising emanation assessment for HDMI-based systems. Full article

Early damage detection in External Thermal Insulation Composite Systems (ETICSs) is essential for building safety, yet conventional non-destructive testing methods are often hindered by the wave-absorbing properties of the insulation layer. To overcome this limitation, this study proposes a field monitoring approach based on Digital Image Correlation (DIC) that enables quantitative, multi-point deformation measurement and early warning of ETICS degradation. A 60-day continuous monitoring campaign was conducted on a real building facade under combined environmental loads, during which surface displacements and four environmental variables—temperature, relative humidity, wind speed, and wind direction—were synchronously recorded. By integrating environmental normalization with a dual-criteria anomaly detection rule, the proposed framework separates reversible environmental deformation from irreversible structural damage. A major structural anomaly on 2 October was successfully identified and confirmed by field inspection, demonstrating the system’s early-warning capability. The approach provides a practical, non-invasive solution for ETICS condition assessment without requiring material parameters or destructive testing. Further full-area detection tests on buildings with diverse structural configurations are in progress to strengthen the generalisability of the findings. Full article

Maintaining radiometric consistency throughout the mission lifetime is essential for the quantitative application of high-resolution optical satellite imagery, particularly for sensors without onboard absolute radiometric calibration systems. This study presents an operational absolute radiometric calibration framework for the Compact Advanced Satellite 500-1 (CAS500-1) based on repeated field campaigns conducted during 2023–2024. Reflectance-based vicarious calibration was performed independently for the summer (TDI 1) and winter (TDI 2) operating modes using ground-measured surface reflectance, quality-controlled atmospheric observations, sensor viewing geometry, and MODTRAN6 radiative transfer simulations. Annual gain and offset coefficients were derived through band-wise linear regression between simulated top-of-atmosphere (TOA) radiance and CAS500-1 digital numbers (DNs). The reliability of the calibration was evaluated through campaign-specific uncertainty assessment, independent cross-calibration with Sentinel-2A/B over Libya-4 and RadCalNet sites, and operational image-based validation. The 2024 calibration showed strong linear relationships between simulated TOA radiance and CAS500-1 DN, with campaign-specific combined uncertainties ranging from 4.59% to 7.77%. Sentinel-2 cross-calibration provided independent support for the field-derived coefficients in the visible bands, although the agreement varied with spectral band, site, and operating mode. Image-based validation showed that negative TOA radiance occurred at low proportions for most images and was primarily concentrated over low-radiance surfaces when elevated values occurred. Repeated observations over a stable asphalt reference site showed generally consistent temporal behavior in the visible bands, whereas the NIR band exhibited greater absolute variability that was also present in the original DN measurements. The results demonstrate that operational radiometric calibration should be managed through an integrated framework combining field calibration, uncertainty quantification, independent cross-calibration, and product-level validation. Continued PICS and reference-site monitoring is required to further distinguish long-term sensor-response changes from calibration-related variability. Full article

Magnetic Resonance Imaging (MRI) is an imaging technique that provides detailed structural and functional information about organs and tissues. To support the validation of MRI measurement techniques and provide a tool for scanner benchmarking, we develop an MRI simulator exploiting graphics processing units (GPUs) for computational acceleration. The GPU-Accelerated MRI Simulator (GAMS) enables us to generate synthetic images with high-resolution detail for quantitative estimation of relaxation times. GAMS is designed to reproduce the entire MRI image acquisition pipeline, including the definition of the MRI pulse sequence in a standard Pulseq format, the input of the digital phantom and its voxelization, the time integration of the Bloch equation at voxel level, and the processing of the derived k-space for image synthesis. The Bloch equation solver is implemented in CUDA Fortran, using GPUs to increase computational efficiency. The solver accuracy is tested in simple cases by comparison with analytical solutions and with the open-source Jülich Extensible MRI Simulator (JEMRIS). GAMS is applied to an inverse reconstruction problem to estimate the relaxation properties of a calibration phantom, starting with synthetic data. Then, it is used to reproduce the imaging process of a high-resolution digital brain phantom, generating realistic images under diverse MRI pulse sequences, which enable differentiation of tissues based on relaxation properties and/or proton density. The results demonstrate GAMS’s ability to generate synthetic images of complex anatomical structures comparable to those acquired with real MRI scanners and to reliably quantify sample relaxation properties. The simulator is also characterized by high computational efficiency, extensibility for multi-GPU implementation, and cross-platform compatibility. These features make GAMS a valuable tool for the metrological assessment of quantitative MRI (qMRI) and the testing of new pulse sequences for qMRI technique development. Full article

Generative artificial intelligence (GenAI) can accelerate the production of digital teaching materials, but long-term use also requires provenance, reproducibility, and maintainable resource workflows. In a single-account 30-day longitudinal case study, we generated images of the moss Pleurozium schreberi with ChatGPT and Gemini under persistent-thread (P) and new-thread (N) conditions and assessed structural similarity (SSIM), perceptual distance (LPIPS), and CLIP embedding distance. The experimental contrast concerned visible persistent-thread continuity versus newly opened conversations; account-level memory, personalisation, model routing, and other provider-side state were not independently controlled. Under the primary preprocessing, within-day P-N divergence was greater in the observed ChatGPT series than in Gemini for perceptual and CLIP-embedding distance. The LPIPS contrast remained robust under padding-free preprocessing, whereas the structural SSIM cross-platform contrast was strongly preprocessing-sensitive; equivalent geometric-preprocessing sensitivity was not assessed for CLIP. Persistent-thread outputs showed greater observed day-to-day similarity for ChatGPT at structural and perceptual levels, while Gemini showed smaller but consistent P-N differences. Padding-free sensitivity analyses preserved the perceptual within-day contrast and the principal day-to-day thread-condition patterns but showed that the cross-platform within-day SSIM result depended strongly on image preprocessing. The study did not measure learning outcomes, botanical accuracy, energy use, carbon emissions, educator time, or monetary cost; the sustainability implications are therefore limited to provenance-aware digital-resource management and should not be interpreted as demonstrated environmental or economic benefits. Prompt-only documentation was insufficient for recovering a stable visual asset; AI-generated visuals intended for educational reuse should therefore be archived and versioned with their prompt, platform, date, and thread context, while content quality requires separate botanical and pedagogical validation. Full article

Geological CO₂ storage is essential to pathways to carbon neutrality, but its deployment depends on trustworthy estimates of storage capacity, injectivity, trapping, reactive evolution, and containment. Digital rock physics can provide pore-scale inputs to these estimates from X-ray and electron microscopy images, but translation to formation-scale performance requires additional geological, fluid, and operational information. Each image-derived result depends on image segmentation, which converts grayscale data into pore, mineral, fracture, and fluid phases. This review evaluates classical methods, machine learning, deep learning, transformers, and foundation models according to whether they support reliable storage decisions rather than image overlap scores alone. Evidence is synthesized from imaging of dry rocks, CO₂–brine experiments, multiscale studies of carbonates and shales, and analyses of fractured rocks. We introduce a framework with seven dimensions: class accuracy, boundary fidelity, topology, morphology, calibrated uncertainty, sensitivity of physical properties, and consequences for engineering decisions. The evidence shows that visually similar segmentations can yield substantially different predictions of permeability, connected porosity, residual trapping, reactive surface area, and leakage paths when errors occur at critical pore throats, fluid interfaces, or fractures. We therefore recommend selecting methods according to storage task and lithology, validating them on independent samples, propagating ensembles of plausible segmentations, using metrics that account for topology, and comparing predictions with laboratory measurements. The central message is simple: segmentation should be treated as both a measurement process and a form of risk control. Segmentation with auditable and quantified uncertainty has the potential to improve the inputs to site screening, injection design, and monitoring. These project-level benefits are proposed consequences requiring upscaling and project-specific validation, not outcomes demonstrated by this review. Full article

The credibility of artificial intelligence (AI) algorithm testing and evaluation directly determines the assessment of their intelligence level and application reliability, which heavily depends on the quality of input datasets. Currently, datasets for AI algorithm testing generally suffer from unclear sources, arbitrary annotations, and heterogeneous formats, while lacking quantitative evaluation specifications that integrate metrological characteristics with algorithm requirements, leading to poor comparability and low credibility of evaluation results. To address these gaps, this study proposes a dataset quality evaluation method oriented to AI algorithm testing by integrating metrological traceability and measurement uncertainty control concepts. Firstly, following the principles of scientificity, quantifiability, and traceability, a hierarchical technical framework and a full-process technical roadmap were constructed. Secondly, a five-dimensional quantitative evaluation index system was established, covering repeatability, completeness, accuracy, consistency, and traceability, with an accompanying uncertainty evaluation, where weighted scoring and measurement uncertainty evaluation methods were introduced to realize quantitative classification of dataset quality. Verification experiments on a temperature and humidity meter image dataset show that the proposed method can effectively assess and grade dataset quality levels. The evaluated dataset obtained a score of 0.9765 ± 0.010 (k = 2), reaching the “excellent” grade, demonstrating the framework’s ability to recognize high-quality datasets. This proof-of-concept demonstrates the framework’s feasibility for assessing high-quality datasets. Formal Standard Reference Data certification is beyond the scope of this study and requires further validation. This research provides a feasible path for the standardized construction of candidate reference data in the field of digital metrology and holds significant value for promoting the standardized application of AI technologies in critical domains such as measurement and control. Full article

Background/Objectives: Multimodal artificial intelligence (AI) is increasingly used in radiology to combine medical images with radiology reports, clinical narratives, structured health records, laboratory measurements, and other patient data. These systems may support more context-aware interpretation, reporting, and clinical decision-making than unimodal approaches. This narrative review examines recent technical and clinical advances in multimodal radiology AI and identifies barriers to responsible implementation. Methods: Relevant biomedical and technical literature was identified through targeted searches of PubMed, IEEE Xplore, ACM Digital Library, Web of Science, Scopus, arXiv, ScienceDirect, SpringerLink, and Google Scholar. A documented screening process identified 111 reviewed publications from 1154 records. Study selection and data extraction were conducted by one reviewer without independent verification. Original research, reviews, and commentaries, including selected preprints, were synthesized thematically. Results: The field has progressed from early and late feature fusion toward cross-modal attention, contrastive image–text pretraining, vision–language models, multimodal large language models, and general-purpose foundation models. Applications include diagnostic classification, prognostic modelling, image–text retrieval, visual question answering, clinical decision support, and automated radiology report generation. Despite promising technical results, comparison across studies remains difficult because of heterogeneous datasets, tasks, metrics, and validation designs. Clinical translation is further limited by scarce external and prospective validation, uncertain interpretability, hallucination and omission risks, privacy and fairness concerns, and limited real-world workflow evaluation. Conclusions: Multimodal AI may enable more clinically informed radiological interpretation and reporting, but progress in benchmark performance has outpaced evidence of safety, generalizability, and clinical utility. Future research should prioritize multicentre evaluation, clinically meaningful metrics, transparent reporting, robust safety assessment, and workflow-centred implementation. Full article

High-frequency full-field vibration measurement using three-dimensional digital image correlation (3D-DIC) is limited by the trade-off between camera frame rate, spatial resolution, and measurement noise. This study presents a method to overcome the frame-rate limitation of 3D-DIC vibration measurement by combining stroboscopic equivalent-time sampling with spatiotemporal noise decoupling. Short-pulse stroboscopic illumination freezes structural motion at different vibration phases, allowing a low-frame-rate stereo camera system to reconstruct high-frequency periodic responses through inter-cycle phase sampling. The measured three-dimensional displacement fields are further processed as space–time data, where target-frequency extraction and spatial-frequency filtering are combined to suppress noise and enhance small-amplitude vibration responses. The proposed method was experimentally validated using a plastic plate vibrating at 251.1 Hz. With an actual camera frame rate of approximately 8.34 fps and an equivalent temporal sampling frequency of 2511 Hz, the reconstructed mode shape achieved a Modal Assurance Criterion (MAC) value of 0.9233, comparable to that obtained using a high-speed camera (0.9225), while providing higher spatial resolution and lower hardware cost. A pulse-width experiment on an aluminum plate vibrating at 7154 Hz demonstrated the influence of stroboscopic exposure duration on measurement accuracy. The proposed approach was further integrated into the EMODE-1 full-field vibration measurement system and applied to a Lenovo ThinkPad touchpad vibrating at 1051 Hz, achieving a MAC value of 0.8878 compared with continuous-scanning laser Doppler vibrometry. The results demonstrate the potential of the proposed method for high-frequency full-field vibration sensing of periodic structures using compact and cost-effective imaging systems. Full article

Digital histology of human postmortem brain tissue is fundamental to understanding neurodegeneration, yet resolving the spatial organisation of neuroinflammatory markers remains challenging because manual scoring and scalar summaries capture only part of the image information. We present a dual-tiered computational framework combining stochastic spatial point-process analysis with explainable deep learning. Using high-resolution spinning-disk confocal immunofluorescence images from a human Parkinson’s disease cohort (Braak Stage 3/4), we analysed spatial associations between IBA1-positive microglia and phosphorylated alpha-synuclein (pSyn) aggregates across the nigrostriatal pathway. Second-order spatial point-process statistics did not detect robust group differences in microglial spatial clustering, while a voxel-level Overlap Index showed a nominal elevation of microglial–pSyn co-occupancy in the substantia nigra that did not survive multiple-testing correction ($d=0.74$, nominal $p=0.045$, BH-adjusted $p=0.550$; diagnosis-by-region interaction $p=0.464$). To test whether pixel-level image information provided donor-level discrimination, we trained 2.5D multi-slice ResNet-18 models under repeated donor-contained cross-validation across all five outer folds, with predictions aggregated to the donor level. In the substantia nigra, the pSyn-only model provided the most consistent discrimination (mean AUROC $0.694\pm 0.062$ across seeds; seed-averaged donor out-of-fold AUROC $0.669$, 95% CI $0.405$–$0.901$), whereas adding IBA1 did not yield measurable incremental discrimination ($\Delta$AUROC $-0.074$, 95% CI $-0.248$ to $0.083$; $p=0.395$). Putamen performance was near chance and unstable across seeds. Cohort-level quantitative attribution analysis indicated that relevance maps were associated with structures present in the input images, but no disease-specific attribution differences survived multiple-testing correction. The framework provides a donor-aware protocol for comparing interpretable spatial summaries, imaging modalities, and held-out explanations; external validation in larger cohorts is required before generalisation. Full article

Background/Objectives: Multidimensional lifestyle interventions that combine healthy diet, physical activity, and psychosocial support are central to the effective self-management of type 1 and type 2 diabetes. Although digital health tools can improve lifestyle monitoring, existing AI-based dietary assessment methods often struggle to estimate food quantity accurately because the food size and scale are difficult to determine from images. This can lead to inconsistent estimates of food mass and energy content. This paper aimed to develop and technically evaluate PC²FoodNet, an AI-based framework for food recognition and physics-constrained physical quantity estimation, and to integrate dietary assessment and exercise support within the proposed AI-assisted Diabetes Care (AIDCare) mHealth platform for multidisciplinary lifestyle support. Methods: The AIDCare platform incorporates AI-assisted nutrition and professionally guided exercise modules to support personalized lifestyle management for individuals with diabetes. For dietary assessment, PC²FoodNet uses an EfficientNetV2-S backbone with multi-task regression and food-density priors to jointly model food volume, mass, and energy. A nutrition professional-in-the-loop approach supports the development and review of individualized diet plans and the assessment of patients’ daily key nutrient intake according to individual nutritional requirements. The exercise module includes professionally designed exercises using Unreal Engine’s MetaHuman plugin. The model was evaluated using four-fold stratified cross-validation on a dataset of 22,070 images covering 40 Turkish food categories. For physical quantity modeling, volume, mass, and energy targets were derived from standardized category-level references anchored to a 100 g reference portion rather than from image-specific ground-truth measurements. Results: PC²FoodNet achieved a mean Top-1 classification accuracy of 94.39% ± 0.47% and Top-5 accuracy of 98.90% ± 0.18% across the four folds. The corresponding Macro-$F_1$ and Weighted-$F_1$ scores were 93.14% ± 0.59% and 94.16% ± 0.36%, respectively. For the standardized category-level reference targets, the physics-constrained regression framework achieved mean MAE and RMSE values of 2.14 mL and 3.01 mL for reference volume, 21.65 g and 30.17 g for reference mass, and 41.20 kcal and 57.55 kcal for reference energy across all 22,070 pooled out-of-fold predictions. The confidence-based referral mechanism identified uncertain cases for clinical review, resulting in a clinical referral rate of 23.81%. Conclusions: AIDCare combines physics-constrained AI-based dietary assessment with personalized physical activity and continuous psychosocial support. A nutrition professional remains involved when AI predictions are uncertain, reducing the risks associated with fully automated lifestyle recommendations. The proposed approach provides a practical foundation for safer, more personalized, and evidence-based diabetes lifestyle management. Full article

Surface subsidence in mining areas is characterized by extensive spatial coverage, complex evolutionary processes, and diverse influencing factors. Conventional monitoring methods and prediction models can hardly meet the demands of large-area continuous monitoring and accurate prediction. Based on time-series InSAR observations and deep learning models, this study investigates the temporal evolution of surface subsidence in mining areas and evaluates the predictive accuracy of the proposed deep learning model. A total of 71 ascending-track Sentinel-1A single-look complex (SLC) images acquired over an open-pit mining area from 9 June 2021 to 14 January 2026 were selected. The PS-InSAR inversion is first implemented to extract highly coherent stable scatterers with small deformation values. Taken as ground control points (GCPs), these points are utilized for orbital correction and interferogram re-flattening in SBAS-InSAR processing to achieve PS-point-constrained deformation monitoring. Taking time-series InSAR deformation datasets, local spatial deformation features, and environmental factors including rainfall, digital elevation model (DEM), slope, and aspect as joint model inputs, a multi-source spatiotemporal fusion prediction model GeoSTF-Former is established to achieve pixel-scale prediction of surface subsidence in mining areas. Experimental results show that the spatial continuity and temporal stability of SBAS-InSAR deformation results are effectively improved after adopting PS stable-point constraints. Compared with the unconstrained scheme, the average RMSE and MAE of monitoring results decrease by 31.9% and 35.3%, respectively, while the coefficient of determination R² increases by 0.10. On the test dataset, the GeoSTF-Former model yields an RMSE of 3.08 mm, an MAE of 2.42 mm, and a coefficient of determination R² of 0.96 at the pixel scale. These results verify that the combination of PS-stable-point-constrained SBAS-InSAR measurements and multi-source spatiotemporal deep learning models can deliver reliable technical support for surface subsidence monitoring and trend forecasting of mining areas. Full article

This paper proposes Sensor-Initialized Gaussian Splatting (SIGS), a method that uses panoramic depth data acquired by the LiDAR sensor in a smartphone for three-dimensional (3D) scene reconstruction and VR space generation. Traditionally, 3D scene generation has required specialized knowledge and significant time, posing challenges for its application in VR. In particular, point clouds estimated by Structure from Motion (SfM), which are used for initializing 3D Gaussian Splatting (3DGS), have had limitations in density and accuracy. SIGS addresses these challenges by utilizing high-accuracy point clouds directly acquired from an iPhone’s LiDAR sensor for 3DGS initialization. For this research, a dedicated smartphone application called Panoramic Depth Recorder (PDR) was developed to simultaneously capture RGB images, depth images, point cloud data, and camera position and rotation information while the iPhone is rotated. These point clouds are then integrated into a common world coordinate system. For outdoor scenes, geometric information over a wider range is supplemented by combining near-field LiDAR data with far-field point clouds estimated by SfM. Experiments demonstrated that SIGS improved rendering accuracy (Structural Similarity Index Measure and Learned Perceptual Image Patch Similarity) and visual quality compared to conventional methods initialized solely with SfM. The generated scenes exhibited fewer artifacts and reproduced more faithful geometric shapes, with improved floor surfaces and ceiling irregularities. The mesh data of the generated 3D scenes is designed for use in VR environments. After conversion from PLY to the FBX format using Blender, they can be imported into Unity, enabling collision detection and user movement control within the VR space. This opens up possibilities for applications such as creating digital twins of robot training environments to reproduce real-world spaces in VR applications. Full article

Background and Objectives: Watch-and-wait has become an accepted option for selected patients with rectal cancer who achieve a clinical complete response after total neoadjuvant therapy, sparing them total mesorectal excision and the stoma, bowel, urinary and sexual dysfunction that follow it. The decision to omit surgery, however, still rests on anatomical evidence—digital rectal examination, endoscopy and magnetic resonance imaging—which describes what remains visible rather than what remains viable, and local regrowth in about one in four such patients measures that gap. We examined whether, and how, molecular evidence of response should enter that decision. Materials and Methods: Narrative review of PubMed/MEDLINE, Europe PMC and ClinicalTrials.gov (final searches 20 September 2026), prioritizing randomized trials, prospective cohorts, registries and meta-analyses; evidence was classified by its relation to the watch-and-wait decision and reported accuracy figures were recalculated with their denominators. Results: Direct evidence comes from six small cohorts, four of them retrospective, and accuracy depended on assay, timepoint and outcome. In a 31-patient ultrasensitive sub-study of a randomized trial, detection at restaging predicted neither sustained response nor relapse, and a positive surveillance result was followed by recurrence in fewer than one in four patients (specificity 27.8%, positive predictive value 23.5%); tumor-informed polymerase chain reaction assays were highly specific but detected only 23% to 74% of residual or regrowing tumors. Mismatch repair testing already redirects treatment before surgery is planned. No molecular assay is validated to decide that surgery may be omitted, and none has been compared with carcinoembryonic antigen, endoscopy and magnetic resonance imaging. Conclusions: Molecular response should complement, not replace, clinical and radiological assessment. We propose an integrated complete response—concordant anatomical and molecular response, interpreted against baseline tumor biology—as a research framework, whose most defensible first application is to relax surveillance rather than to escalate treatment. Full article