Search Results (239)

Background: Thrombus formation remains a major complication during extracorporeal membrane oxygenation (ECMO), potentially resulting in circuit failure and thromboembolic events. Although several techniques have been proposed for thrombus detection, the relationship between measured ultrasonic features and target size has not been systematically characterized. This study evaluated size-dependent ultrasonic signal changes using standardized cylindrical surrogate targets under controlled conditions. Methods: A through-transmission ultrasonic system incorporating paired point-focused transducers (5 and 10 MHz), a water chamber, an ultrasonic pulser/receiver, a digital storage oscilloscope, and a manual translation stage was constructed. Cylindrical targets made of acrylic, acrylonitrile butadiene styrene (ABS), and SUS304 stainless steel, with diameters ranging from 0.1 to 5.0 mm, were scanned at 0.5 mm intervals over a 30 mm range. Baseline-subtracted waveforms were used to calculate peak height, pulse area, and squared-amplitude sum. Results: Exploratory regression analyses demonstrated positive diameter-dependent associations for all evaluated features. Pulse area showed strong linear associations across the six material–frequency conditions (R² = 0.904–0.997), although no single feature consistently demonstrated the highest goodness of fit. Strong associations were observed at both frequencies; however, the different receiver gains precluded direct inference regarding relative frequency sensitivity. Under the 10 MHz condition, the 0.3 mm SUS304 wire target was the smallest tested surrogate target meeting the study-specific operational detection-index criterion (DI > 3). Conclusions: Through-transmission ultrasound demonstrated size-dependent changes in signals obtained from standardized cylindrical surrogate targets. Pulse area was identified as a practical candidate feature because of its consistently strong associations and computational simplicity, but neither statistical superiority nor a predictive sizing model was established. The 0.3 mm result applies to a high-acoustic-contrast SUS304 target under the specified experimental conditions and does not represent a biological thrombus detection limit. Full article

Flax (Linum usitatissimum) is an important member of the genus Linum. L. usitatissimum has been commonly cultivated as a fiber or oil source. Despite its economic significance, comprehensive genome analyses encompassing a wide range of genotypes remain limited. Nuclear DNA content is defined as the amount of DNA within the nuclei of a eukaryotic organism. The elucidation of the nuclear DNA content may provide valuable insights for genetic studies, biodiversity studies, conservation efforts, and breeding programs, thereby reinforcing the significance of its estimation. Nuclear DNA content variation may arise under different environmental conditions, and its potential adaptive role has attracted considerable interest. In the present study, isolated intact nuclei were stained with propidium iodide as a DNA stain and flow cytometry analyses were performed to estimate the nuclear DNA content of two varieties and 40 accessions from geographically distant locations. FCM analyses revealed that the nuclear DNA content varied from 1.19 pg/2C to 1.37 pg/2C. This corresponds to a 15.13% difference in nuclear DNA content among the genotypes, with PI 182226 having the smallest genome and PI 194998 the largest genome. A minor variation was detected among the genotypes; however, accession PI 194998 showed statistically significant differences compared with some of the analyzed genotypes. The small intraspecific variation observed among the analyzed genotypes may indicate detectable genomic diversity, which could potentially be attributed to differences in the abundance of repetitive elements among these genotypes. The chromosome number of the selected genotypes was determined to rule out ploidy-level variation as a potential explanation of differences in nuclear DNA content, and all shared the expected same ploidy level. Geographical location (longitude and latitude) showed no statistically significant correlation with nuclear DNA content. The potential causes of the intraspecific variation are discussed in the text. The results obtained in this study may constitute a valuable genetic resource for subsequent molecular and breeding research aimed at improving this economically important crop. Full article

Traffic crashes remain a major transportation safety problem, particularly for motorcycle riders, who have limited physical protection and high exposure to road hazards. Powered two-wheeler riders account for roughly a fifth of global road deaths, and the burden is concentrated in South-East Asia, where motorcycles dominate daily mobility. Although hazard perception is a trainable skill, the most recent meta-analytic synthesis of hazard perception training reports only three intervention studies with motorcyclists and the smallest pooled effect of any road-user group, indicating that rider-specific evidence is scarce. This study examined changes in motorcycle riders’ visual attention and riding performance following repeated exposure to identical Honda Riding Trainer (HRT) routes. A within-subject experiment was conducted with 20 licensed riders with 1–3 years of riding experience. Each participant completed two sessions separated by one week. In each session, participants rode five HRT routes containing 34 hazard events while their eye movements were recorded with an eye tracker. Visual attention was indexed by Time to First Fixation (TTFF), First Fixation Duration (FFD), and fixation count (FC); riding performance was indexed by driving score, total crashes, and total wrong lanes. To evaluate the effect of repeated exposure, before–after differences were compared with the Wilcoxon Signed-Rank test, and the unaggregated data were then re-analyzed with mixed-effects models carrying crossed random effects for participant and for hazard. Spearman rank correlation was applied as an exploratory analysis of the associations between visual attention and riding performance measures. All six variables differed significantly between the two sessions, with large effect sizes throughout (r = 0.564–0.868). Visual attention metrics improved, with TTFF decreasing by 32.52%, FFD by 41.47%, and FC by 29.00%. Riders therefore detected hazards faster, processed them more efficiently, and required fewer fixations after prior exposure. Riding performance variables improved in parallel, where driving score increased by 7.53%, total crashes decreased by 26.32%, and total wrong lanes decreased by 61.84%. Improvement was largest for turning or crossing vehicle hazards and smallest for blind spot hazards, consistent with evidence that riders respond more slowly to gradual-onset hazards. Exploratory correlations linked driving score to crash frequency in both sessions. Repeated HRT exposure thus appears to support implicit learning of hazard anticipation, whereas latent, blind spot hazards may require explicitly targeted training. Full article

Wind turbine main bearings are critical components in the drivetrain and are characterised by long service life, low failure rates, and limited failure-interval samples, which increases uncertainty in reliability assessment and maintenance decision-making. To improve the utilisation of limited failure-interval information in small-sample reliability modelling, this study develops a unified three-way expansion Bootstrap strategy combined with a three-parameter Weibull distribution. The principal methodological contribution lies in integrating intra-interval supplementary sampling, left-boundary expansion, and right-boundary expansion within the same sample-generation framework, thereby enabling the main distributional information and boundary information contained in the available failure-interval samples to be utilised jointly. Based on 36 equivalent failure-interval samples obtained from Romax fatigue-life simulations under different operating conditions, the proposed method is compared with traditional Bootstrap and two-way expansion Bootstrap methods. The results of the two-sample K-S test indicated that no statistically significant distributional difference was detected between the expanded samples and the original sample. Using the three-parameter Weibull fitting results obtained from the original 36-sample dataset as the reference, the proposed three-way expansion method yields the smallest relative deviation of the scale parameter η among the three expansion strategies, at 2.69%. The MTBF relative deviations of the traditional Bootstrap, two-way expansion Bootstrap, and three-way expansion Bootstrap methods were 4.81%, 7.36%, and 7.76%, respectively. Repeated simulation results further show that the three-way expansion method provides substantially lower MTBF variability than the traditional Bootstrap method, although the two-way expansion method yielded the smallest MTBF standard deviation. The results demonstrate the methodological potential of the proposed strategy for small-sample MTBF modelling of wind turbine main bearings under the investigated simulation conditions. Full article

Accurate detection of critical infrastructure in high-resolution remote sensing imagery is important for infrastructure inventory, regional monitoring, and risk assessment. However, differences in object scale, spatial density, geometric form, and background complexity make it difficult to evaluate detectors using a single aggregate metric. This study develops a diagnostic evaluation framework utilizing CI-DIOR-7, a task-oriented benchmark derived from the DIOR dataset by retaining horizontal bounding-box annotations for seven infrastructure categories: Airport, Harbor, Bridge, Dam, Train Station, Storage Tank, and Windmill. The benchmark contains 23,463 images and 44,579 instances and preserves the original training, validation, and test splits. Four representative detectors—YOLO11s, RT-DETR-L, Faster R-CNN R50-FPN v2, and RetinaNet R50-FPN v2—were evaluated under a unified COCO-style evaluation protocol, modified with a maximum detection limit of 500 (maxDets = 500) to accurately evaluate highly dense infrastructure scenes. On the official test set of 11,738 images and 33,766 instances, RT-DETR-L achieved the highest mAP@0.5:0.95 of 0.3212 and the smallest validation-to-test performance drop. YOLO11s obtained an mAP@0.5:0.95 of 0.2785 while providing the lowest inference latency and the lowest false-positive burden on zero-GT images, at 0.168 false positives per image. Class-wise and difficulty-oriented analyses showed that Bridge was the most challenging category, with a mean class AP of 0.1514 and small-bridge recall ranging from 0.105 to 0.161 across the four models. Storage-tank detection was strongest in moderately dense scenes but deteriorated when more than 20 instances occurred in one image. These results show that no single architecture simultaneously achieved the highest detection accuracy, the lowest measured inference latency, and the lowest false-positive burden under the zero-GT protocol. CI-DIOR-7 therefore provides a reproducible benchmark and diagnostic framework for evaluating critical infrastructure detectors beyond overall mAP. Full article

Background: This exploratory study provided a primarily descriptive comparison of two xenogeneic and two synthetic blocks for mandibular onlay augmentation in rabbits. Methods: Twelve rabbits received bilateral block grafts, providing 24 sites allocated to Bio-Oss^® Block, SP-Block, ReproBone^® Block, or Alos Block (n = 6/material). After 10 weeks, undecalcified sections were evaluated within standardized inferior and superior regions. Newly formed bone was the primary outcome; secondary outcomes included residual graft, IBN-like tissue, soft-tissue components, and cross-sectional augmented area. Mixed-effects models accounted for clustering within animals. Results: No statistically significant difference in newly formed bone percentage was detected among biomaterials (p = 0.475). Residual graft differed significantly (p < 0.001) and was lower for all test materials than for Bio-Oss^® Block. ReproBone^® Block consistently exhibited IBN-like tissue (17.3 ± 2.9%). SP-Block showed more intra-compartment soft tissue than Bio-Oss^® Block. Alos Block had the smallest augmented area (12.9 ± 6.1 mm²; adjusted p = 0.001 versus Bio-Oss^® Block), and extra-compartment soft tissue was present in five of six sites. Conclusions: Although newly formed bone percentages did not differ significantly, the biomaterials displayed distinct profiles of scaffold persistence, incorporation, tissue organization, and cross-sectional augmented area at 10 weeks. These features should be considered together when evaluating block biomaterials for onlay augmentation. Full article

Wildfires represent a major environmental hazard with significant impacts on ecosystems, climate, biodiversity, and human activities. The increasing frequency and intensity of wildfire events have highlighted the need for reliable and timely detection techniques based on satellite remote sensing. This study investigates the application of Proper Orthogonal Decomposition (POD) to thermal observations acquired from the Spinning Enhanced Visible and Infrared Imager (SEVIRI) onboard the Meteosat Second Generation (MSG) satellite for wildfire anomaly detection. A wildfire event that occurred on 8 August 2021 in Calabria, Southern Italy, was selected as the primary case study. To assess the consistency of the POD response beyond the primary case, the analysis was further extended to two additional wildfire events, Viggianello–Abate and Pazzano–Montestella, using the 15 × 15 pixel extraction window. Middle Infrared (MIR, 3.9 $\mathsf{\mu }$m) observations collected at 15 min intervals over a complete day were analyzed using four different spatial extraction windows (3 × 3, 15 × 15, 30 × 30, and 45 × 45 pixels). POD was employed to separate dominant background thermal variability from localized fire-induced anomalies. The analysis focused on higher-order POD modes, particularly the 6th, 7th, and 8th modes, which exhibited enhanced sensitivity to wildfire activity. Results showed that POD successfully identified thermal anomalies corresponding to wildfire occurrence times independently detected by the RST-FIRES methodology. The comparison of extraction window sizes revealed that the 15 × 15 pixel window provided the best balance between anomaly enhancement, spatial localization, and noise reduction. Larger windows introduced excessive spatial smoothing and reduced localization capability, whereas the smallest window was more affected by noise. The findings demonstrate the potential of POD as an effective complementary approach for wildfire detection and monitoring using geostationary satellite observations. Full article

We analyzed AIRS CH₄ volume mixing ratio (VMR) Standard L3 v6/v7 IR-Only Daily products and ground-based measurements from 16 stations of the Network for the Detection of Atmospheric Composition Change (NDACC) at 24 pressure levels from 1000 to 1 mbar. We assessed the dependence of maximum AIRS sensitivity on latitude. At high latitudes, the zone of maximum sensitivity is closer to the surface, at 700–500 mbar; in mid-latitudes, it is 500–250 mbar; and in tropical and subtropical regions, good initial agreement between satellite and ground-based data is observed at 400–200 mbar for both AIRS product versions. At the vast majority of pressure levels and all comparison sites, a unidirectional negative drift in the difference between satellite and ground-based measurements (i.e., discrepancy drift) was observed. Drift coefficients were calculated for each statistically supported pressure level. Two regions of maximum drift were identified: one in the lower atmosphere (925–850 mbar) and another near 50 mbar. The smallest drift was observed at 400–200 mbar. As the main result of the study, we developed and applied correction factors for all 23 AIRS v6 and v7 levels. Using these coefficients led to much better agreement between long-term methane trends from ground-based and satellite measurements and to higher correlation coefficients across all comparison sites. Full article

Retrieval-augmented fine-tuning (RAFT) can improve small language models (sLMs) on retrieval-grounded question answering, but the synthetic training data produced by commercial large language models (LLMs) vary in quality. This paper contributes a quality-aware selection protocol—rather than a new RAFT or QLoRA method—that scores LLM-generated alternatives along four embedding-based dimensions (question relevance, answer faithfulness, QA coherence, and semantic similarity) and selects one alternative per task before parameter-efficient fine-tuning. Under pre-specified paired-bootstrap contrasts with Holm correction, the parameter-free faithfulness-based selector only-AF significantly exceeds random selection on Gemma-2-9B-IT (ΔF1 = +0.106, 95% CI [+0.043, +0.174], Holm-corrected p = 0.019), and its pre-specified weighted companion af-70 (w_AF = 0.70) shows the same confirmed pattern (Holm-corrected p = 0.002). Both effects persist under a Korean character-level F1 that removes particles and punctuation (Holm-corrected p = 0.004 and p = 0.042), indicating robustness to the choice of lexical metric. Relative to training on the full 150-row augmented pool, the quality-selected 50-row sets are statistically indistinguishable while using one third of the training data, which we interpret as data efficiency rather than superiority. Across six instruction-tuned models (2B–27B), a significant selector-by-model interaction indicates that the optimal quality axis is model-dependent, and the two smallest models show no benefit from selection. The study’s confirmatory contrasts use a small controlled Korean corpus under a transductive design; two pre-registered validation experiments probe external validity. On an independent five-fold larger corpus with a passage-level train/test split, fine-tuning transfers strongly and the selected one-third subsets show no significant difference from the full pool, while the advantage over random selection is directionally positive but small and not significant; under controlled corruption of 35% of the pool, the metrics detect the damaged rows, and for the score-sum selector the selection-versus-random benefit is significantly larger than on the clean pool (difference-in-differences p = 0.0014; directionally consistent but not significant for the faithfulness selectors). Within this scope, quality-aware selection is a promising, data-efficient safeguard for synthetic RAFT data—performing comparably to full-pool training at one third of the cost, with growing value as pool quality degrades—and larger-scale external validation remains future work. Full article

Large language models are increasingly used to read résumés and judge who advances in hiring, a task once reserved for people and now handed to systems whose reasoning is hard to inspect. Whether these models carry the demographic biases that have long shaped human hiring is therefore an urgent question, and the published evidence so far is mixed and difficult to interpret, partly because studies tend to test a single condition and rarely confirm that their measurement instrument can detect bias at all. This paper audits demographic bias in résumé evaluation across three current models, one of them open-weight, and it treats robustness as a central concern rather than seeking a single verdict. Each résumé is scored through a reference-anchored comparison task in which the model rates the candidate against a fixed neutral reference for the same occupation. Effect sizes are estimated as standardised mean differences under false-discovery control, the sensitivity of the instrument is tested with an embedded seniority control, and the null findings are corroborated by formal equivalence tests against a justified smallest effect size of interest and by mixed-effects models that account for the clustered structure of repeated evaluations. The audit pairs a positive control that confirms the models read genuine differences in candidate quality with a deliberate attempt to provoke bias by weakening candidates, relaxing the prompt, and adding culture-fit language of the kind used in real hiring. Across more than thirty thousand evaluations, gender and race effects prove negligible and remain so under every one of these conditions. The one systematic preference that emerges favours candidates who appear more experienced, and closer inspection shows that most of it is an artefact of how the résumés were built rather than a bias against age, leaving only a modest effect that surfaces when the prompt is casual. A separate and quieter pattern appears in the open-weight model, which reacts to a few explicit signals of minority status. The broader lesson is that fairness measured on a clean benchmark does not by itself guarantee fairness in deployment, because how a model is prompted can decide whether bias appears. Full article

Background/Objectives: Torque teno virus (TTV) is small non-pathogenic virus, currently under investigation as a potential biomarker to monitor immunocompetence in patients with rheumatoid arthritis (RA) receiving various immunosuppressive therapies. To allow for large-scale assessment of existing pan-European RA cohorts, minimal specimen (serum) input for TTV DNA detection through quantitative PCR (qPCR) needs to be identified in relation to the type and quantity of immunosuppression. Methods: TTV qPCR analysis was performed by measuring 1:1, 1:2, 1:4 and 1:8 dilutions of 278 human serum samples derived from patients with RA under different therapies. The smallest detectable difference (SSD) as well as the intra-class correlation (ICC) were assessed. The reproducibility of results was determined by measuring an additional 19 sample duplicates. Results: RA patient sera showed a mean TTV viral load equivalent of 2.9 log10 genome copies/mL (SD ± 1.47). A total of 32 samples (11.6%) were negative for TTV; 54 (19.6%) were below 2.4 log10, the manufacturer’s lower limit of quantification; 44 (16%) were between 2.4 and 3.0 log10; and 145 (52.7%) showed a high viral load above 3.0 log10. Dilution and replication experiments showed a high stability and reproducibility of TTV measurements (overall ICC: 0.95 (95%CI: 0.951 to 0.966)). TTV differed between undiluted and 4-fold- as well as 8-fold-diluted samples (1 vs. 2: p = 0.993; 1 vs. 4: p < 0.001; 1 vs. 8: p < 0.001), aligned with decreasing ICC and less sensitivity. RA disease activity did not influence reproducibility. Conclusions: Quantitative TTV DNA detection in serum samples from RA patients yields highly reproducible results, including 2-fold-diluted serum samples. Full article

Background/Objectives: Accurate landmark identification underpins reliable cephalometric analysis. This study evaluated a ResNet50-based, single-stage regression convolutional neural network for direct automatic localization of 15 landmarks on lateral cephalograms. Methods: This retrospective study included 669 lateral cephalograms (669 patients): 619 for training and 50 randomly selected for an internal holdout test set. One of five orthodontists annotated each cephalogram, and coordinates served as the reference standard. One orthodontist independently re-annotated all test images after more than 2 weeks to assess reproducibility. Model performance was evaluated using Euclidean localization errors and success detection rates (SDR). Results: Across 750 landmark predictions, mean localization error was 1.25 ± 1.39 mm (95% confidence interval, 1.15–1.35 mm) and median error was 0.72 mm. SDRs within 1.0, 2.0, and 4.0 mm were 64.5%, 81.5%, and 94.4%, respectively. A statistically significant overall difference was observed among the 15 landmarks (Friedman χ²(14) = 28.84, p = 0.011). Point A had the numerically largest mean error (1.66 mm) and the mandibular central incisor the smallest (1.04 mm). The mean difference between original and repeated annotations was 0.52 ± 0.92 mm. Conclusions: In this single-center internal holdout test set, mean localization error was below the 2.0 mm benchmark. However, 18.5% of predictions exceeded 2.0 mm, and external validity remains unconfirmed. Artificial-intelligence-generated landmarks should be verified by orthodontists, and external validation is required before broader clinical use. Full article

Large language models (LLMs) are increasingly integrated into educational settings, yet their pedagogical reliability remains insufficiently understood. Beyond overt hallucinations, which informed users readily recognize, a subtler failure mode consists of explanations that are coherent, authoritative, and pedagogically plausible while harbouring hidden conceptual flaws, responses we term Socratic traps. This paper introduces SocraticTrap-CS, a publicly available benchmark that probes the capacity of open-weight LLMs to generate such strategic misconceptions on demand. A single structured prompt explicitly elicited three outputs per concept (a correct explanation, an overt hallucination, and a strategic misconception), yielding 735 expert-annotated response segments from seven open-weight models across 35 core concepts in algorithms and data structures, programming languages and paradigms, databases, computer networks, and operating systems. Three domain experts independently annotated each segment using a three-class schema, achieving near-perfect agreement (Fleiss’ $\kappa =0.9487$). Because models were explicitly instructed to produce the misconception, the central metric quantifies adversarial instruction-following capacity rather than the base rate of such errors in naturalistic use and should be read as a conservative upper bound on model capability. Under these conditions, compliance reached 91.7% overall (100% for three models; 57.1% for the smallest model, Mistral 7B, whose lower rate plausibly reflects weaker instruction-following rather than greater safety). Expert-judged persuasiveness was moderate to high, errors were predominantly conceptual rather than factual, models differed significantly, and no statistically significant domain-level differences were detected. The benchmark reframes the evaluation of educational LLMs around pedagogical trustworthiness rather than factual correctness alone. Full article

Cooking fumes are an important source of indoor and outdoor air pollution. Containing potentially carcinogenic particles and toxic chemical components, they pose significant health threats, making precise detection essential for risk prevention and the formulation of emission standards. This study used the manual gravimetric analysis as the benchmark reference method to systematically evaluate the performance of three alternative measurement techniques: (1) an optical particle counter (Promo 3000, Palas GmbH, Karlsruhe, Germany), (2) a photometer (DustTrak 8533, TSI, Shoreview, MN, USA), and (3) infrared spectrophotometry. A data correction model was constructed by simulating typical cooking conditions. Results indicate that the manual gravimetric analysis yielded the highest particulate concentrations. Infrared spectrophotometry measured only 75.6% of the benchmark value, mainly because of its selectivity toward oil-derived organic components and possible sampling losses. The optical particle counter, influenced by particle light-scattering properties and density differences, underestimated the mass concentration of particles larger than 0.3 μm, capturing only 46.9% of the benchmark value. Conversely, the photometer exhibited the smallest deviation, with readings closest to the gravimetric reference under the tested cooking fumes conditions. Theoretical derivation and data fitting determined correction factors of 1.32 for infrared spectrophotometry, 2.13 for the optical particle counter for particles larger than 0.3 μm, and 0.97 for the photometer. This correction system enables the standardization of data across different detection methods, allowing on-site monitoring data to be calibrated against the gravimetric reference method. Full article

To eliminate the complex nonlinear mixing relationships among spectral features in near-infrared (NIR) quantitative analysis, and to overcome the limitations of principal component analysis (PCA), which relies solely on covariance structure and linear assumptions and is therefore incapable of effectively handling nonlinear signals, this study employs kernel-independent component analysis (KICA), for nonlinear feature extraction from NIR spectra, combined with a regression model to achieve rapid detection of tannin content in sorghum grains. KICA effectively separates nonlinearly mixed source signals by mapping spectral data into a high-dimensional feature space via the kernel trick. The prediction model built on KICA-extracted features and support vector regression (SVR) consistently delivered the highest test-set prediction accuracy and exhibited the smallest training-to-test R² gap among all evaluated models across repeated random splits, confirming its superiority over PCA-based feature extraction methods and standalone SVR, and its competitive performance relative to ICA-based methods, in both predictive accuracy and generalization capability. Additionally, KICA yielded a lower reconstruction error for the original spectra, indicating its ability to more completely retain the nonlinear informative content of the spectral data. By calculating the mean absolute coefficient of each independent component, it was found that the component with the highest contribution was strongly correlated with the wavelength range near the characteristic absorption peaks of tannin, thereby enhancing the chemical interpretability of the features. On a publicly available corn NIR dataset, the proposed method also achieved superior prediction results compared with benchmark methods, validating its generalization capability across different sample types and quality attributes. This study confirms the feasibility of introducing nonlinear blind source separation via KICA into NIR quantitative analysis, offering a promising approach for spectral feature extraction in the rapid quality assessment of agricultural products with complex matrices. Full article