Search Results (338)

Background/Objectives: Accurate intraoperative tumour localisation remains challenging in minimally invasive colorectal surgery because tactile palpation is lost and conventional markers can migrate or provide imprecise localisation. Building on a preceding tri-frequency study that identified 13.56 MHz as the preferred RFID band for the intended application, this work develops a shaft-compatible NFC antenna–reader platform and evaluates its electromagnetic behaviour from bench-top reference media to five-layer tissue-equivalent phantoms. Methods: A Ø3 × 25 mm Fair-Rite Material 67 ferrite-rod antenna was designed from material and geometric parameters using finite-rod demagnetisation, inductance, resonance, and field calculations, followed by FEM cross-validation and experimental characterisation. The primary dataset comprised 480 detection distance measurements (2 media × 4 tag angles × 30 repetitions × 2 encapsulation variants). Phantom testing added 1440 measurements at 22 °C and 600 measurements at 37 °C across three fabrication batches, with the 37 °C non-coaxial subset limited to one batch. Results: The fabricated antenna measured 16.9 µH versus a 17.4 µH analytical estimate (−2.9%), with loaded Q = 23. The coaxial detection range was 16.45 ± 0.29 mm in air and 16.26 ± 0.21 mm in saline; angle was the dominant determinant of range (partial η² = 0.989). In the multi-layer phantom, detection was 100% at 0 and 10 mm perirectal fat thickness under coaxial alignment at 22 °C, whereas performance declined markedly with angular misalignment and no detections occurred at fat thicknesses ≥ 20 mm. Across detectable phantom configurations, FEM showed r² = 0.994, RMSE = 0.81 mm, and mean bias +0.70 mm. Bare and resin-overcoated tags showed no statistically detectable range difference. Multi-tag discrimination reached 100% for up to three tags separated by ≥20 mm under coaxial alignment, but deteriorated with angular misalignment. Conclusions: The study demonstrates a physics-informed route from antenna miniaturisation to measured system performance, and defines the present operating envelope under controlled bench and tissue-equivalent phantom conditions. The electromagnetic measurements apply to the antenna–electronics subassembly; integrated-shaft, multi-prototype, multi-operator, ex vivo, and in vivo validation remain necessary before clinical performance can be determined. Full article

Conditional imitation learning (CIL) for CARLA depends on diverse expert data spanning map, weather, traffic, and control-perturbation axes, yet it remains unclear which axes actually drive closed-loop behavior, and whether ablation conclusions survive deployment guardrails. We train a matched baseline and four equal-budget leave-one-axis-out (LOO) variants of a fixed CIL architecture (v14) and evaluate each under three nested tiers: pure policy rollout, minimal traffic-rule shields, and a fully deployed stack with route blending and recovery. The factorial design comprises $18\times 5\times 3$ scenario-variant-tier cells, each repeated under n = 5 traffic-seed replicates (1350 closed-loop episodes) to estimate NPC-seed variance on every eval stack. No single withheld axis dominates pooled outcomes. LOO effects are tag- (scenario-category) and spawn- (vehicle starting location) specific: removing perturbation-labeled recovery data costs $-285$ m on geometry_stress but can gain distance on in-distribution spawns; removing multi-town data changes held-out Town05 mobility on some routes while depressing others. Axis-importance rankings reorder across tiers; Kendall $\tau$ between pure and full rankings is $0.0$, and guardrails compress or erase pure-tier gaps (e.g., drop_perturbation pooled distance $\Delta$ from $-93$ m to 0 m). Pure-tier seed replicates show that geometry-driven lane-tracking degradation under drop_perturbation is seed-stable; traffic-axis and full-tier drop_traffic loads are more seed-sensitive. We release the evaluation ledgers, parsing scripts, and analysis tooling with the paper. Training-data ablation claims should report per-scenario or tag-stratified LOO metrics under pure evaluation; guardrailed tiers are supplementary deployment checks, not substitutes for isolating what the policy learned. Full article

Time-of-flight magnetic resonance angiography (TOF-MRA) is widely used for noninvasive visualization of arterial anatomy, but extracting hemodynamics like blood velocity typically requires supplementary phase-contrast scans, tagging or multi-TE images. This study proposes a novel, physics-informed computational framework to extract variable fluid velocity directly from standard TOF-MRA signal profiles. We analytically expand the approach-to-steady-state Bloch equations to include convective flow, establishing a mathematical relationship between the spatial decay of longitudinal magnetization and fluid velocity. The velocity derivation was further extended to pointwise estimation over a 1-D centerline, overcoming the limitations of constant-velocity assumptions. To validate and solve this problem, a MATLAB (R2025b) simulation framework was developed to model fluid flow in two variable-geometry flowing tube cases, i.e., continuous narrowing and focal stenosis, under synthetic scanner noise. A global inverse optimization approach utilizing Dual-Tikhonov regularization was applied to stably invert the ill-posed transit time integral, actively penalizing high-frequency numerical ringing while preserving structural curves. The computational simulations successfully recovered ground-truth point-wise velocities, tracking gradual hemodynamic accelerations and sharp stenotic jets. This theoretical framework and the example centerline TOF-MRA signal intensity provide a robust mathematical proof-of-concept that quantitative, localized functional hemodynamic metrics can be extracted from standard structural MRA imaging, establishing a foundation for advanced flow quantification without requiring additional scan time. Full article

Supervised fine-tuning is often discussed in terms of data volume and target quality, but the format of the supervision itself can change the reasoning strategy a model expresses in its outputs. We study this effect on a six-category reasoning benchmark spanning symbol transformation, text decryption, bit manipulation, gravitational constant estimation, numeral conversion, and unit conversion (drawn from the NVIDIA Nemotron Model Reasoning Challenge). Using NVIDIA Nemotron-3-Nano-30B-A3B with matched LoRA training settings, we compare three symbol-supervision formats: verbose English rule descriptions, compact family tags, and compact formula notation. We hypothesize that supervision renderings bias token-level reasoning priors, and that these priors transfer across task boundaries in multi-task SFT. In the canonical strict-rescore inventory, the best compact tag and formula checkpoints are statistically equivalent in aggregate within a pre-specified ±4-point margin: K8A-800 reaches 72.3% strict-scored overall accuracy and K8B-700 reaches 71.2% (TOST p = 0.003). Compact tags nevertheless provide a cleaner behavioral profile: an earlier K8A-400 checkpoint reaches 66.4% overall, 98.7% gravity accuracy, and 36.9% bit accuracy without the same contamination signatures. In contrast, verbose English rule descriptions are associated with heuristic parroting, with up to 57% of symbol failures at audited verbose checkpoints collapsing to a single remove-operator template, while formula notation is associated with cross-category contamination: numeric-looking predictions appear more often in text decryption (higher at five of six matched training steps under the canonical seed; matched-step means 15.8 vs. 11.7 numeric predictions per 157 text rows), and gravity failures at a representative K8B formula checkpoint shift toward shortcut stubs and explicit g = 9.8/9.81 fallbacks. We further show that checkpoint selection and strict evaluation auditing materially change branch decisions. Across three training seeds, neither compact format shows a consistent aggregate advantage, while the contamination signatures are partly seed-specific: the gravity-shortcut severity difference persists but is not exclusive to the formula branch, and the numeric–text signature does not reproduce under reseeding. These results support treating supervision format as a first-class hyperparameter for multi-task reasoning SFT in large language models—at least in this benchmark-and-model setting—rather than a mere rendering detail. Because such symbolic and procedural reasoning tasks recur in domains including cybersecurity, mathematics, and code generation, the same formatting choices plausibly shape the policy that any later reinforcement-learning stage would inherit, which we flag as future work. Full article

In Global Navigation Satellite System (GNSS)-denied indoor environments, ultra-wideband (UWB) localization of unmanned ground vehicles (UGVs) is challenged by position-dependent anchor visibility and mixed line-of-sight (LOS)/non-line-of-sight (NLOS) ranging. This study proposes a soft continuous confidence weighting method within an adaptive Kalman filter (AKF)-based UWB/inertial measurement unit (IMU) localization framework. The vehicle model uses motor pulse increments and IMU yaw-rate measurements as inputs and outputs vehicle position and heading estimates. Virtual forward–backward iteration converts inconsistencies between the current UWB ranges and tag–anchor geometry into terminal virtual-anchor displacements. A half-Gaussian function then maps each displacement to a continuous confidence coefficient. The resulting coefficients are incorporated into weighted least-squares (WLS) and AKF localization, while the UWB measurement-noise covariance is adaptively updated using the range innovations. The proposed method was evaluated through static calibration and dynamic localization experiments. These experiments compared soft and hard weighting schemes and assessed the contribution of AKF fusion. These results indicate that the method proposed in this study improves localization accuracy, robustness, and temporal continuity under position-dependent anchor visibility and mixed LOS/NLOS conditions. Full article

The textile and clothing value chain faces increasing pressure to improve reuse and recycling rates, particularly in post-consumer scenarios, where garments must be rapidly assessed, classified, and routed toward appropriate end-of-life pathways. Post-consumer garment sorting must preserve reusable items while directing non-reusable textiles toward appropriate recycling or inspection pathways. This article presents a two-stage image-only decision-support framework that combines YOLO-based image classification, a locally executed vision–language model (VLM), two ConvNeXt-Tiny textile classifiers, and deterministic routing rules. In Stage 1, YOLO classifiers estimate garment type and dominant color, while Qwen2.5-VL-3B-Instruct VLM assesses visible stains, holes, pilling or lint, tags, dirt or discoloration, intentional distressing, condition, and supporting evidence. The backend validates these outputs and applies explicit precedence and uncertainty rules to assign categories A (resale), B (donation/reuse), C (recycling-oriented pre-sorting), or D (critical review). Stage 2 is triggered only for C/D garments and aggregates predictions from multiple RGB crops to estimate broad material-family hints and visible fabric structure before proposing an initial route, container, color group, recycling mechanism, and validation requirement. The YOLO garment-type classifier achieved 78.6% top-1 and 99.2% top-5 accuracy on the test set. The ConvNeXt-Tiny fabric-structure classifier achieved 78.55% accuracy and 78.64% macro-F1, whereas the material-family classifier achieved 56.39% accuracy and 55.83% macro-F1. In a controlled Stage 1 pilot test, binary reuse-oriented versus additional-processing routing achieved 80.0% accuracy, 75.0% precision, 75.0% recall, and an F1-score of 0.75. A Stage 2 end-to-end pilot test achieved 66.7% correctly recommended final routes, with macro-F1 of 0.767. These results provide evidence that complementary models and explicit validation rules can support preliminary explainable garment triage. However, RGB imagery cannot confirm exact fiber composition, blend percentages, or chemical contamination. Full article

Accidental releases of radionuclides from nuclear-powered vessels or ships carrying nuclear weapons may expose marine organisms to radioactive contaminants. The consequences of such events can differ markedly between resident fish populations and migratory species that move through affected areas. While these contrasting responses have been evaluated by the authors for coastal nuclear accidents, comparable assessments for open-sea scenarios remain limited. We simulated radionuclide releases from vessels in the North Atlantic to evaluate exposure pathways and potential ecological effects on both resident and migratory fish, since several naval bases operating nuclear-powered vessels are located along the eastern North American coastline. Physical transport of radionuclides was modeled using a Lagrangian framework that incorporates advection by ocean currents, three-dimensional turbulent diffusion, radioactive decay, and dynamic sediment–water exchanges. This transport model was coupled to a four-compartment food-web bioaccumulation model representing phytoplankton, zooplankton, non-piscivorous fish, and piscivorous fish. The approach allows evaluation of contaminant uptake in resident fish populations and along the migration routes of highly mobile species. Bluefin tuna (Thunnus thynnus), a key ecological and commercially valuable species, was selected as the migratory case study. Migration paths reconstructed from electronic tagging data were integrated to estimate radionuclide exposure along individual trajectories. This combined modeling framework provides new insight into how open-sea nuclear releases may differentially affect resident fish communities and wide-ranging migratory species in the North Atlantic, with relevance for fish ecology, population risk assessment, and fisheries management. Full article

A camera-guided warehouse robot keeps its bearings by repeatedly estimating its pose against known visual references, and it must relocalize whenever that estimate is lost. What limits this process is often not identification but the availability of a usable reference along the route. The references used in practice are usually single-sided planar fiducial markers such as QR-like codes, ArUco markers, and AprilTags, which stay readable only within a limited cone about their surface normal; a spatial reference, by contrast, can in principle be recognized from any azimuth. We quantify what that difference is worth at the navigation level. The framework is built in Unity with NavMesh navigation and a purely geometric-availability model, rather than an image-based recognizer, whose single switchable property is the availability rule. In the idealized all-azimuth spatial-reference regime (the spatial regime), a reference is available from any direction; in the single-sided, angularly constrained planar-reference regime (the planar regime) it is available only within $\pm 20°$ of the surface normal. A full-factorial experiment with 54 configurations ($3\times 3\times 2\times 3$) and $n=100$ paired replications, 10,800 runs in all, was run in both regimes over four deployment factors: deployment scheme, camera field of view, recovery step, and detection range. Under this geometric model, the spatial regime reached $5.7\times$ higher reference coverage ($41.6\%$ vs. $7.3\%$) and a mission-completion rate 30 percentage points higher (86.2% vs. 55.9%). A paired Wilcoxon signed-rank test confirms the coverage difference ($p<0.001$, matched-pairs $d_z=1.84$), and McNemar’s test together with a logistic regression confirms the completion difference. In a factorial analysis of variance, the detection range dominates (partial ${\eta }^2=0.903$), and a strong deployment × range interaction concentrates the advantage in the rack aisles, where a planar reference is seen edge-on. Three further analyses point the same way: an angular-threshold sweep from $10°$ to $60°$, an equal-count deployment control, and route- and time-normalized visibility and relocalization metrics. The advantage also held across square, L-shaped, and U-shaped aisle layouts (32,400 runs in total), with a negligible regime × layout interaction. All these numbers are model-based estimates under an explicitly stated availability model: they measure the navigation-level value of azimuthal reference availability and do not validate any particular physical object, decoding algorithm, or AR device. Full article

Background/Objectives: Early childhood caries (ECC) is a common, preventable, and socially patterned disease, yet the literature on ECC is vulnerable not only to limitations in evidence generation but also to errors in evidence translation. This conceptual framework paper focuses on evidence translation rather than estimating a new treatment effect. Methods: Literature and framework development were informed by targeted narrative searches of PubMed/MEDLINE, Scopus, Web of Science, and Google Scholar, with an emphasis on ECC literature from 2010 onward and foundational methodological sources. Sources were purposively retained if they clarified an evidentiary domain, a cross-domain bridge, or a competing interpretation; priority was given to systematic reviews, trials, longitudinal and causal studies, natural experiments, clinical guidance, and implementation evaluations. No pooled effect estimates were produced. Results: Evidence drift is the movement of a finding into a stronger or different claim without adequate bridging evidence. The proposed framework comprises association, mechanism, causation, consequence, disease control, and policy implementation, with commercial and structural determinants operating as a cross-cutting upstream layer. It classifies the inference advanced rather than study design alone and permits primary and secondary domain tags. A four-question test asks whether the research question and conclusion occupy the same domain, what bridge supports the movement, and whether uncertainty is retained. Five trajectories illustrate the framework: vitamin D, dental rehabilitation under general anesthesia, oral microbiome research, silver diamine fluoride (SDF), and sugar taxation. In the SDF case, lesion-arrest evidence is explicitly separated from the still-unproven implementation hypothesis that endpoint-only pathways may create differential standards of care. Conclusions: The original contribution is an integrated vocabulary, decision procedure, and validation agenda for ECC evidence appraisal. The framework generates three prespecified, testable hypotheses: (1) trained raters will classify claims with reproducible inter-rater agreement (kappa ≥ 0.70); (2) claims judged to contain evidence drift will be significantly less likely than domain-concordant claims to include an explicit bridge; and (3) using the structured framework will improve the transparency and proportionality of reviews compared with usual appraisal. Prospective testing of content and construct validity, as well as practical utility, in peer review, guideline development, education, and policy evaluation is required before standardized implementation. Full article

We study a structured QND central-spin testbed under a hybrid finite-resolution protocol. The central-spin state is temporally integrated without retaining the internal readout-time label, whereas environmental fragment information is quantified by a time-tagged scalar-averaged Holevo benchmark. On a converged numerical grid, we identify finite time regions in which the integrated central-spin state has low off-pointer visibility while model-resolved late-time coherence bursts remain present and the typical-fragment Holevo benchmark exceeds a prescribed threshold. A detector-window variance decomposition supplies a model-assisted diagnostic; it requires a fine-grained trajectory or a microscopic model. The Holevo quantity is an upper bound on accessible classical information, and no fragment measurement attaining it is constructed. Grid refinement, an independent trapezoidal quadrature, and a half-cell grid shift preserve the qualitative four-region baseline structure. A sampled-subset calculation benchmarks the typical-fragment classification, but we do not establish constructively disjoint or state-integrated fragment redundancy. A single-realization detector-width/disorder map and a 100-realization ensemble delimit the structured mesoscopic scope. The dimensional conversion is an NV-like scale estimate for an engineered quasi-homogeneous testbed and not an experimental protocol for an arbitrary natural bath. Full article

Several of the North Carolina blue crab’s (Callinectes sapidus) life history parameters such as growth rate, natural mortality, and length–weight relationships would benefit from being updated or localized. To this end, we designed a mark–recapture study of blue crabs using coded wire tags to estimate carapace width (CW)-to-weight relationships, growth, and natural mortality in a North Carolina blue crab population. We used data from captured blue crabs to estimate their CW-to-weight relationships and used recapture data to estimate growth rates via the Fabens method using both frequentist and Bayesian approaches. Additionally, we built a Cormack–Jolly–Seber (CJS) model to estimate apparent survival. Our CW-to-weight estimates fell largely between the CW-to-weight estimates from the Chesapeake Bay and Florida stocks, suggesting possible regional variation. Our frequentist and Bayesian growth rate estimates had higher k values than other estimates based on a limited number of growing recaptures. There was no single CJS model that was strongly supported by data, but we identified important covariates that inform the precision and accuracy of the CJS model for future studies. Our results suggest that coded wire tags can be useful for mark–recapture studies of species without permanent hard parts, but considerable effort may be necessary to produce viable results. Full article

Tomato yellow leaf curl virus (TYLCV) is a major threat to tomato production, creating a need for sensitive, low-cost detection methods that can be applied to early symptomatic or low-viral-load samples. Recombinant antigen configuration may influence serological assay development, although the specific contribution of multiple-cloning-site (MCS)-derived intermediate sequences remains uncertain. In this study, a recombinant Trx-His-coat protein (CP) fusion antigen was produced using an MCS-free direct-fusion construct that retained the Trx-His tag while removing the MCS-derived intermediate sequence, followed by gradient refolding. No direct comparison with linker-containing, tag-cleaved, or tag-free antigen constructs was performed. The purified antigen was used to immunize rabbits and generate a high-titre polyclonal antibody (pAb). The resulting indirect enzyme-linked immunosorbent assay (ELISA) achieved a theoretical limit of detection of 1.8 ng/mL and an estimated pre-dilution equivalent the limit of detection (LOD) of 72 ng/mL after sample dilution. The assay showed favourable tolerance to crude tomato leaf matrices, with spike-recovery rates of 95.45–100.40%. In a preliminary evaluation using a balanced panel of 32 field-collected samples, ELISA absorbance correlated with droplet digital PCR quantification (R² = 0.9819) and plant disease index values (R² = 0.9774). Liquid chromatography–tandem mass spectrometry (LC-MS/MS) peptide mapping and AlphaFold2-based modelling were used only to provide preliminary computational context for antigen interpretation. The assay showed cross-recognition toward Tobacco curly shoot virus (TbCSV), indicating that it should not be considered strictly TYLCV species-specific. Therefore, this assay may support preliminary serological screening under the tested conditions, whereas molecular confirmation remains necessary when species-level identification is required. Full article

Ambient backscatter communication is a promising low-power technology for the Internet of Things (IoT), yet direction of arrival (DOA) estimation and detection in multi-backscatter device (BD) scenarios remain challenging. This paper thus investigates joint DOA estimation and signal detection in multi-BD IoT systems. By exploiting the Toeplitz structure of the covariance matrix of the received signals, an atomic norm minimization (ANM)-based optimization framework is first employed for DOA estimation in varying signal-to-noise ratio (SNR) conditions. Specifically, the alternating direction method of multipliers (ADMM) in combination with the estimation of signal parameters via the rotational invariance techniques (ESPRIT) algorithm is used to iteratively solve the ANM optimization issue and obtain the DOA angles. Then, the muti-BD signal detection is considered by iteratively implementing the minimum variance distortionless response beamforming with successive interference cancellation, to guarantee successive detection from the strongest to the weakest signal. Simulation results demonstrate that the proposed ANM-based scheme achieves high DOA estimation accuracy and low bit error rate (BER) of muti-BD detection across the whole SNR range of [0–20] dB. For example, the root mean square error of DOA estimation under three-tag conditions can achieve ${10}^{-2}$ at 10 dB, with the BER of BD detection being $2.9\times {10}^{-3}$, validating the effectiveness of our method in multi-BD IoT scenarios. Full article

Indoor passage monitoring requires people-count estimation, behavior recognition and identity association while avoiding privacy-invasive sensing. This paper reports an RFID-only corridor-style passage-monitoring framework that combines wearable EPC identity tags, a wall-mounted passive tag array and distance-aware RSSI temporal modeling. Identity-tag readings and multi-channel tag-array RSSI sequences are aligned on a unified reader-timestamp timeline. An effective behavior duration (EBD) mechanism is used as a rule-based valid-window screening step before model inference. Target count, behavior state and identity association are then estimated using distance-aware LSTM temporal fusion. The RSSI attenuation relationship is used as a physical motivation for feature design and distance-mode interpretation, rather than as an explicit loss constraint or separately validated physical prior. A subject-independent dataset was collected with 18 array tags and 18 volunteers across single-person, multi-person, crossing and following scenarios, yielding 11,148 EBD-valid windows. Under the EBD-valid event-window protocol, the proposed method achieves a mean absolute error (MAE) of 0.226, a root mean square error (RMSE) of 0.357, a Macro-F1 of 0.934 and an identity-association accuracy of 0.961. Relative to threshold rules, traditional machine learning and a single-branch LSTM evaluated on the same dataset, the MAE is reduced by 59.5%, 46.1% and 28.9%, respectively, and the RMSE is reduced by 55.8%, 43.5% and 28.9%, respectively. The results indicate that native RFID identity, spatial tag-array sensing, EBD screening and distance-aware temporal modeling can jointly support privacy-preserving passage monitoring under controlled corridor-like conditions. Full article

Building-integrated photovoltaics (BIPV) provide an important pathway for expanding distributed solar generation in dense urban areas, but comparable evidence on roof–facade resources across different urban morphologies remains limited. This study develops a scalable workflow to estimate the technical BIPV potential of roofs and facades within standardized 3 km × 3 km urban-core windows in 10 representative Chinese cities. Building footprints, height-related attributes, and functional tags derived mainly from OpenStreetMap were audited, cleaned, and completed through a hierarchical imputation strategy. A 2.5D urban geometry model was then used to estimate annual solar irradiation on building envelopes, with shading, orientation, and sky visibility explicitly considered. The results show that inter-city variation in BIPV potential is not governed by sunshine duration alone, but is strongly shaped by building density, height structure, envelope composition, and roof–facade contribution patterns. High total potential and high envelope-use efficiency do not necessarily occur in the same cities, indicating that total supply capacity and spatial deployment efficiency should be evaluated separately. The analysis further shows that facade-led BIPV pathways may be important in high-density urban cores, but facade-related estimates are sensitive to height-data completeness and usable-facade assumptions. These findings suggest that urban BIPV planning should move beyond aggregate solar-resource ranking and adopt morphology-aware, surface-specific, and data-quality-conscious assessment frameworks. The proposed workflow is intended for early-stage screening and cross-city comparison and provides a basis for identifying differentiated deployment priorities for roofs and facades in urban solar energy systems. Full article