Search Results (5,946)

Background/Objectives: Deep inferior epigastric perforator (DIEP) flap reconstruction represents the gold standard for autologous breast reconstruction, while profunda artery perforator (PAP) flap reconstruction has developed as a reliable alternative, particularly in patients with low body mass index or inadequate abdominal tissue. Even though several comparative studies have evaluated surgical and patient-reported outcomes between PAP and DIEP flaps, evidence regarding reconstructive timing, oncologic safety, and interactions with adjuvant therapies remains scarce, especially for PAP reconstruction. Methods: A narrative review of the literature was conducted using PubMed. Studies assessing PAP and DIEP flap breast reconstruction were included, with particular focus on surgical outcomes, patient-reported outcomes, reconstructive timing (immediate or delayed reconstruction), oncologic safety, recurrence, and the effects of radiotherapy and chemotherapy. Comparative studies, cohort studies, systematic reviews, and meta-analyses were synthesized through a narrative review. Results: Twenty studies were included. Comparative evidence showed similar flap survival rates and overall patient satisfaction between the two methods, with flap success rates approaching 98–100%. PAP reconstruction was associated with increased donor-site wound complications and, in some studies, increased fat necrosis rates, while long-term patient-reported and aesthetic outcomes remained equivalent between techniques. In contrast to the relatively limited PAP literature, DIEP reconstruction has been widely studied in terms of reconstructive timing and oncologic safety. Current evidence indicates that immediate DIEP reconstruction does not increase the risk of flap loss, major complications, or recurrence in comparison with delayed reconstruction and might optimize early postoperative recovery and patient-reported outcomes. Nevertheless, none of the identified studies directly compared PAP and DIEP reconstruction with respect to immediate versus delayed timing, exposure to radiotherapy or chemotherapy, or long-term oncologic outcomes. Conclusions: PAP flap appears to represent a reliable alternative to DIEP flap reconstruction. However, major gaps in the literature persist involving PAP reconstruction in oncologic and timing-related settings. Future prospective multicenter studies that directly compare PAP and DIEP flaps according to reconstructive timing, exposure to adjuvant therapy, recurrence, and patient-reported outcomes are warranted to establish evidence-based reconstructive strategies for oncologic breast reconstruction. Full article

Introduction: Three-dimensional deep learning offers promise for automated accurate brain tumor segmentation and survival prediction but requires robust validation across multiple MRI modalities to be effectively implemented in clinical practice. Methods: This study presents a comprehensive 3D deep learning framework using 369 cases from the BraTS2020 dataset. A 3D U-Net architecture was developed for tumor segmentation utilizing combined imaging data and optimized for computational efficiency and memory. The final 3D U-Net model segmentations were used to build machine learning 6-month and 12-month survival classifiers. Segmentation models were evaluated using multiple metrics, including the Dice Similarity Coefficient, Hausdorff Distance, and Cohen’s d. The classification models were evaluated using AUC-ROC and balanced accuracy. Results: Segmentation achieved a modest, but promising, performance across 30 epochs and with 295 training patients, achieving the best mean validation Dice = 0.8388 and a final-epoch mean Dice of 0.8263. Survival classification with a hybrid clinical and imaging logistic regression showed promising results, with 12-month prediction achieving AUC = 0.746 and 69% accuracy. The top contributing features for the 12-month prediction classifier were extent of resection, T1 contrast-enhanced tumor median, and FLAIR tumor median. Conclusions: This comprehensive framework demonstrates that a multi-modal approach provides meaningful performance gains, while segmentation-derived features show a promising ability to enable survival prediction. Full article

Serum and plasma are the most widely used matrices in metabolomics and human biomonitoring studies; however, the optimal matrix for integrated non-targeted analysis (NTA) workflows combining metabolomics and exposomics has not been systematically evaluated. This pilot study applied parallel NTA workflows to paired serum and plasma samples from five individuals to characterize matrix-dependent differences and provide an empirical basis for matrix selection in integrated studies. Three analytical methods were employed: one metabolomic method (Method 1) using Hydrophilic Interaction Liquid Chromatography (HILIC) and Reversed-Phase Liquid Chromatography (RPLC) columns and one exposomics (Method 2) method using an RPLC column, each analyzed in both electrospray ionization (ESI) positive and negative modes. Overall, serum and plasma showed broad similarity, with substantial overlap in detected features and strong linear correlations between paired samples (R² = 0.70–0.87). However, PCA revealed systematic differences between the two matrices along PC1 and PC2, likely attributable to matrix effects arising from coagulation-related compositional changes in serum. For metabolomics, glycerophospholipids, sphingolipids, and acylcarnitines were consistently enriched in serum, attributable to platelet activation and phospholipase release during blood coagulation, consistent with prior reports. In contrast, oxidized fatty acid species were predominantly elevated in plasma, warranting caution in oxylipin-focused studies using serum. For exogenous chemical profiling, the two matrices performed comparably, with 32 out of 36 annotated features showing no significant matrix-dependent differences (p > 0.05), including PFAS, pharmaceuticals, and diverse xenobiotics. These findings support the interchangeability of serum and plasma for broad exposomics studies. Overall, while both matrices provided broadly comparable global coverage, plasma may represent a more appropriate matrix for integrated NTA workflows, as it better preserves in vivo metabolite composition and minimizes coagulation-induced confounding, though validation in larger cohorts is needed. Full article

In large-scale state-owned enterprise audit and compliance tasks, ensuring that similar violations receive consistent disciplinary decisions is essential for procedural fairness and institutional credibility. However, existing retrieval methods face three major challenges: lexical matching methods fail to recognize semantically equivalent violation descriptions, general-purpose semantic encoders lack knowledge of inspection-specific terminology and regulatory distinctions, and retrieved precedents are often not directly transformed into actionable disciplinary references. To address these problems, this paper proposes InspectCL, a domain-enhanced contrastive learning and Retrieval-Augmented Generation framework for similar case retrieval, validated on audit data from a provincial power grid company. First, to provide task-specific supervision that is unavailable in existing benchmarks, we construct InspectCase, a de-identified dataset of 4200 audit and compliance cases across 12 violation categories, with expert-validated positive pairs and hard negative pairs. Second, to overcome the weak domain awareness of generic encoders, we design a domain-enhanced contrastive learning model. Specifically, terminology-masking augmentation improves robustness to specialized inspection expressions, regulatory semantic injection incorporates disciplinary rules to distinguish factually similar but legally different cases, and hierarchical contrastive optimization strengthens both case-level similarity learning and category-level boundary separation. Third, to convert retrieved precedents into practical decision support, the Top-K similar cases are used as evidence for a large language model to generate structured disciplinary recommendation summaries, including violation classification, penalty references, applicable regulations, and rectification measures. Experimental results on InspectCase show that InspectCL substantially outperforms BM25, BERT-base, SimCSE, and Legal-BERT baselines, achieving 56.9% ± 0.7% Recall@5 and an 87.6% ± 0.4% Penalty Consistency Score (PCS). These results demonstrate that the proposed problem-driven modules jointly improve semantic retrieval accuracy and disciplinary decision consistency, offering a practical reference for similar power-grid audit scenarios, with broader applicability to be validated in future cross-domain studies. Full article

Despite the vast biodiversity of Mexican vipers, venom of endemic species has been barely studied. Here we analyzed the venom composition of three endemic species of rattlesnakes: Crotalus aquilus, C. triseriatus, and C. ravus. We used quantitative chromato-mass-spectrometry and compared venoms with C. molossus, a species commonly found in North America, in a comparative and phylogenetic framework. In total, we identified 165 proteins grouped in 19 main protein families, consistent with previous reports for viperid venoms. In C. aquilus and C. triseriatus, the most predominant protein-family type was Serine Proteases, and in C. triseriatus and C. molossus it was Snake Venom Metalloproteases. The Label-free quantification revealed a high proportion of Snake Venom Metalloproteases in C. aquilus, C. triseriatus, and C. molossus, reaching 28–47% of the total venom. In contrast, in C. ravus 47% of the venom was composed of Phospholipases A₂. Among the four species analyzed, C. triseriatus and C. aquilus were most similar in compositional profiles and their profiles are highly correlated. Venom composition in terminal clades and taxa were better explained by protein losses than evolution of new proteins. The triseriatus group share seven proteins, while the clade C. aquilus + C. triseriatus share seven derived protein features, of which six are protein losses. Full article

Automated defect detection in precision agriculture serves as a critical technology for enhancing the quality of agricultural products. Although supervised-only semantic segmentation has demonstrated remarkable performance in citrus surface defect detection, it relies heavily on training with large-scale labeled data, which results in prohibitive acquisition costs. Semi-supervised learning mitigates reliance on labeled data by generating pseudo-labels. However, existing semi-supervised segmentation methods still face challenges. On the one hand, the instability of pseudo-labels and the propagation of noise can mislead the training of semi-supervised models. On the other hand, due to the lack of semantic constraints in feature learning, models often suffer from insufficient feature discriminability when handling complex samples, such as citrus surface defects characterized by similar textures and blurred boundaries. Therefore, this study proposes UP-ETS, a dual-guided semi-supervised semantic segmentation model based on the Mean Teacher–Student framework, specifically designed for the segmentation of complex citrus surface defects. UP-ETS employs Uncertainty Estimation (UE) based on Kullback–Leibler (KL) divergence to quantify the prediction discrepancy between the teacher and student models on blurred and ambiguous pixels. This mechanism guides the model to dynamically adjust weights, thereby reducing noise propagation and enhancing pseudo-label stability under complex citrus surface textures. Prototype Contrastive Learning (PCL) is utilized to align pixel-level features of difficult samples with class prototypes, optimizing the feature discriminability for complex citrus surfaces. Experimental results demonstrate that the UP-ETS model exhibits superior semi-supervised segmentation performance. Notably, at a labeled data ratio of only 1/16, the dice improved from 85.57% to 87.76% compared to the supervised-only baseline. Furthermore, the model shows significant performance enhancements in segmenting difficult samples, such as small targets, complex boundaries, and blurred regions. The results of ablation studies and t-SNE visualization prove the effectiveness of the proposed UE and PCL. These two methods synergistically guide the model to construct a feature space that is better structured and highly discriminative. Furthermore, UP-ETS outperforms various representative semi-supervised segmentation models in terms of segmentation performance, parameters, and inference speed. In cross-dataset validation, the model exhibits robust generalization capabilities, achieving performance comparable to supervised-only methods trained on the full augmented dataset. Consequently, the framework introduced in this study effectively mitigates the heavy dependency on annotated datasets, providing significant practical value for agricultural deployment. Full article

This paper examines changes in the macroeconomic indicators of the member countries of the Comprehensive and Progressive Agreement for Trans-Pacific Partnership (CPTPP) following their accession to the agreement. This study aims to identify shifts in the structural comparability of national economies and to assess the processes of macroeconomic convergence in the context of sustainable development. To achieve this objective, reference pools of CPTPP member countries are constructed, and their digital profiles are developed based on key macroeconomic indicators and grouped into three blocks: (1) indicators of economic growth and the state of the real sector, including GDP (constant 2015 US$), GDP growth, annual %, gross capital formation, % of GDP, unemployment, total % of total labor force, and national estimate; (2) indicators of foreign economic activity and trade openness, including exports of goods and services, % of GDP, imports of goods and services, % of GDP, external balance on goods and services (% of GDP), foreign direct investment, net inflows, % of GDP, and trade, and % of GDP; (3) indicators of financial and macroeconomic stability including inflation, consumer prices, annual %, central government debt, % of GDP, and gross savings, and % of GDP. Based on the digital profiles, similarities/differences in the economies were examined by applying linear discriminant analysis (LDA). The empirical framework covers two periods: (1) 2013–2017 (pre-accession) years and (2) 2019–2023 (post-accession) years. The results indicate that the economies of member countries in 2013–2017 exhibited a high degree of heterogeneity. In contrast, the 2019–2023 period demonstrates a tendency toward partial convergence of macroeconomic parameters, as evidenced by a reduction in distances between country profiles in the discriminant space. While interpreting the results, it is acknowledged that the 2019–2023 period coincided with the effects of the global crisis caused by the COVID-19 pandemic, which significantly impacted international trade dynamics. For most countries, this period was characterized by a decline in several macroeconomic indicators and investment activity, an increase in debt burdens, and enhanced heterogeneity in economic dynamics, which was taken into account when interpreting macroeconomic convergence processes within the CPTPP. The scientific novelty of the study lies in its application of an approach based on the analysis of the structural similarity of the macroeconomic profiles of CPTPP countries, which complements traditional assessments of the effects of economic and trade integration. The practical significance of the findings is associated with their potential use in evaluating the prospects for CPTPP expansion and in modeling alternative scenarios of participation and sustainable development within international trade agreements under conditions of global economic transformation. Full article

Background/Objectives: Contrast-enhanced (CE) breast MRI is highly sensitive for evaluating breast cancer extent and response to neoadjuvant therapy (NAT) but requires intravenous administration of gadolinium-based contrast agents (GBCA), increasing cost, time, patient discomfort, and health concerns. This study explored the feasibility of reducing GBCA use in treatment monitoring using a deep learning (DL) model to synthesize CE-MRI from non-contrast MRI. Methods: This IRB-approved retrospective pilot study evaluated women with breast cancer enrolled in an ongoing trial using serial MRI to monitor NAT prior to surgery. A pre-trained DL model was used to synthesize CE-MRI from T1-, T2-, and diffusion-weighted MRI. Changes in tumor volume at early (post-1-cycle NAT) and mid-treatment were measured on synthetic and acquired CE-MRI. Performance for predicting residual cancer burden (RCB) class 0/1 was evaluated using AUC and compared with DeLong’s test. Results: 27 women were included in the study (median age, 47 years [range = 28–75]); 14 (52%) achieved RCB class 0 and six (22%) achieved class 1. Synthetic CE-MRI-derived tumor volumes showed strong correlation with those from acquired CE-MRI at pre-treatment ($\rho$ = 0.92, p < 0.001) and early treatment ($\rho$ = 0.83, p < 0.001), but lower agreement at mid-treatment ($\rho$ = 0.57, p = 0.002). Change in tumor volume on synthetic CE-MRI was numerically similar to acquired CE-MRI for predicting RCB class 0/1 vs. 2/3 at both early (AUC = 0.84 vs. 0.86, p = 0.83) and mid-treatment (AUC = 0.73 vs. 0.75, p = 0.80). Conclusions: Synthetic CE-MRI demonstrates preliminary feasibility as a non-contrast surrogate for predicting favorable outcomes (RCB class 0/1) in this pilot study, but inconsistencies in tumor volume measurement vs. acquired CE-MRI warrant further model refinement and validation. Full article

Soil moisture (SM) is a key variable for assessing plant water availability, especially in rain-fed systems where imbalances strongly affect crop development. Satellite missions such as SMAP provide global SM estimates, though representing vertical SM variability remains challenging. This study evaluates the performance of SMAP Level 4 Global 3-hourly 9 km grid EASE-Grid Surface and Root-Zone Soil Moisture Geophysical Data (SPL4SMGP, version 7 and the new and scarcely evaluated version 8) using field observations from the Argentine Pampas, a region dominated by Typic Argiudolls soils (~16 million ha). The analysis covered normal-wet and dry conditions across several crop seasons. Surface (SSM, ~5 cm) and root zone (RZSM, 0–100 cm) soil moisture were compared against field data using Pearson’s correlation (r), bias, and unbiased root mean square deviation (ubRMSD). Both SSM and RZSM achieved ubRMSD values close to the SMAP accuracy target (≈0.04 m³/m³). SSM correlated moderately with observations (r = 0.57–0.72) and showed a consistent negative bias (−0.08 ± 0.05 m³/m³). In contrast, RZSM exhibited low sensitivity to soil profile variability and a narrow dynamic range. Version 8 showed similar performance to version 7, with a tendency toward overestimation, mainly during dry periods. Overall, SPL4SMGP products effectively capture SSM dynamics but show limited skill in representing root zone variability in Typic Argiudolls. Full article

Mixedwood forests emerge from nonlinear interactions among disturbance, aspen recruitment, and conifer succession, yet ecological processes alone may not explain the contrasting long-term trajectories observed in managed landscapes. This paper proposes a coupling of classical aspen–conifer dynamics with a rarity-sensitive mechanism of human perception and intervention to examine how social feedback reshapes mixedwood stability. The ecological subsystem alone produced transitions among pure aspen, mixed-species, and pure conifer equilibria, but the introduction of social feedback fundamentally reorganized these stability regimes. In particular, social feedback generated bistability between mixed-species and pure conifer equilibria, indicating that similar ecological conditions may support contrasting long-term forest outcomes depending on intervention history and societal response. Under low social sensitivity, weak intervention destabilized coexistence and expanded bistability, whereas high social sensitivity stabilized coexistence and compressed bistable regions. Strong intervention eliminated bistability entirely and produced unstable or oscillatory dynamics, including persistent cycles under some parameter combinations. These findings suggest that mixedwood resilience depends not only on ecological recovery and succession processes, but also on how societies perceive species decline and respond through management intervention. Full article

Severe cutaneous adverse reactions (SCARs) are rare, life-threatening drug hypersensitivity syndromes. Although pharmacovigilance can identify drugs disproportionately reported with SCARs, it does not reveal which local chemistries recur among them. To address this, we assessed whether drugs with FAERS-derived SCAR disproportionality signals share interpretable chemical motifs. We screened FAERS data from 2004Q1 to 2024Q3, identified 5523 drugs with available Simplified Molecular-Input Line-Entry System (SMILES) representations, and constructed a signal-enriched dataset of 1676 compounds with nominally significant broad-SCAR associations after excluding predefined therapeutic/supportive confounders. Compounds were assigned to positive-signal [natural logarithm of reporting odds ratio (lnROR) > 0, n = 1219] or non-positive-signal (lnROR ≤ 0, n = 457) classes and encoded with 9753 explicitly mappable atom-centered local substructure descriptors. A LightGBM signal-classification model evaluated using random repeated nested cross-validation (six-fold outer × 50 repeats) achieved moderate internal discrimination (mean area under the receiver operating characteristic curve = 0.7041 ± 0.0337). Descriptor-space cluster-based repeated nested cross-validation, designed to reduce train–test structural leakage, yielded lower but still above-chance performance (mean ROC AUC = 0.6409; permutation p = 0.001), indicating that random-split estimates should be interpreted as optimistic for structurally novel compounds. Sensitivity analyses using minimum SCAR case-count thresholds and retention of predefined therapeutic/supportive drugs showed broadly similar performance and motif rankings. SHapley Additive exPlanations (SHAP) analysis revealed a fragment-level contrast: allylamine-like, ethanolamine-related, and diaminopropane-related motifs were associated with higher positive-signal class probability, whereas phenol and pyrimidine motifs were associated with lower positive-signal class probability. These findings suggest that FAERS-derived broad-SCAR signal direction is not chemically random within the selected dataset. Overall, the proposed framework should be viewed not as a direct predictor of absolute clinical SCAR risk but as an exploratory, pharmacovigilance-driven cheminformatics approach for prioritizing compounds and motif families for further SCAR-focused evaluation. Full article

Reliable evaluation resources for semantic deduplication, semantic textual similarity (STS), and retrieval remain limited for low-resource and morphologically rich languages. This study introduces KazakhTextDuplicates, a Kazakh-language benchmark for controlled evaluation of semantic duplication under varying levels of semantic preservation and surface-form distortion. The benchmark includes two complementary versions. KazakhTextDuplicates v1.0 is a diagnostic dataset derived from naturally occurring duplicate and near-duplicate pairs for analyzing relationships between duplication labels and lexical overlap. KazakhTextDuplicates v2.0 is a controlled benchmark created using deterministic transformations that define seven semantic duplication regimes: exact duplication, contextual reformulation, paraphrasing, partial semantic overlap, and three levels of character-level noise. Each pair is assigned both a regime label and a predefined similarity score, enabling evaluation across duplicate classification, STS, and retrieval tasks. Five embedding models (BGE-M3, LaBSE, Multilingual-E5-Large, KazEmbed-v5, and OpenAI text-embedding-3-large) were evaluated in a zero-shot setting. Results show that v1.0 is strongly affected by surface-form similarity and is therefore more suitable for diagnostic analysis. In contrast, v2.0 provides a more challenging and informative evaluation environment. OpenAI text-embedding-3-large achieved the strongest STS performance (Pearson = 0.510, Spearman = 0.652, MSE = 0.075) and the best duplicate regime classification results (Accuracy = 0.155, Macro-F1 = 0.066). Retrieval performance remained strong at higher cutoffs despite lower first-rank stability. The results demonstrate that benchmark design substantially affects semantic similarity evaluation and emphasize the need for controlled assessment in low-resource agglutinative languages. Full article

Temperature stress strongly affects plant metabolism, and recurrent heat exposure can modify physiological responses depending on developmental stage. This study examined the biochemical and physiological adjustments of Senecio madagascariensis subjected to single (naïve) or repeated (primed) heat stress at 40 °C during vegetative and reproductive stages. Sampling was conducted after the second heat stress and after the subsequent recovery period. Principal component analyses (PCAs) revealed marked stage-specific contrasts. In the vegetative stage, PCA1 and PCA2 explained 75.7% of total variance, clearly separating treatments: naïve plants were associated with elevated proline, soluble sugars, phenolics, glycine betaine, hydrogen peroxide (H₂O₂) and lipid peroxidation, whereas primed plants were linked to enhanced superoxide dismutase (SOD) and ascorbate peroxidase (APX) activities and reduced oxidative markers. Under stress, naïve plants showed substantial increases in soluble sugars (+198%) and proline (+66.9%), whereas primed plants exhibited attenuated oxidative responses and reduce phenolic accumulation. After recovery, primed plants exhibited markedly reduced H₂O₂ levels (−57.5%) and lipid peroxidation, alongside higher SOD activity. In the reproductive stage, PCA indicated more subtle priming effects, with overlapping clusters among treatments. Primed plants accumulated the highest soluble sugar levels under stress (+276.5%), while naïve plants showed higher proline and glycine betaine levels. Following recovery, osmolyte levels were similar among groups. Senecionine remained unchanged during the vegetative stage but increased in both naïve (+21.4%) and primed (+19.1%) plants during the reproductive stage after recovery. Oxidative markers revealed contrasting patterns, with primed reproductive plants showed the lowest superoxide under stress but the highest H₂O₂ and lipid peroxidation at both time points. Overall, the findings demonstrate that heat-stress responses in S. madagascariensis are developmentally regulated, with stronger priming effects during vegetative growth and phenology-dependent metabolic adjustments during reproduction. All results are directly supported by the measured biochemical and physiological data. Full article

Silver-doped diamond-like carbon (Ag–DLC) coatings were investigated with respect to their tribological behavior under ambient and physiologically relevant conditions. Gradient Ag–DLC coatings deposited on AISI 316L stainless steel were tested in air, simulated body fluid (SBF), and an albumin-containing solution using a pin-on-disk configuration. Increasing silver content resulted in a systematic reduction in the H³/E² ratio, leading to increased coating wear irrespective of the environment. In contrast, friction behavior was strongly controlled by the surrounding medium. Under dry sliding in air, all coatings exhibited similar steady-state friction governed by the DLC matrix. The lowest steady-state friction coefficients were obtained in SBF, indicating that the aqueous ionic environment provided the most favorable friction conditions among the tested media. In the albumin-containing medium, friction also remained low, indicating that protein adsorption and interfacial layer formation modified the sliding conditions, although the CoF did not fall below that observed in SBF. Wear was highest in air and generally lowest in SBF, while tests in albumin promoted surface layer formation. Surface analyses indicated silver redistribution, transfer-layer formation, and the presence of protein-related surface agglomerates, with higher apparent surface coverage on coatings containing more Ag. Overall, the results show that Ag-doped DLC coatings exhibit environment-dependent tribological behavior under physiologically relevant conditions. The present work should be regarded as a tribological study rather than a direct validation of antibacterial performance. Future studies should combine tribological assessment with dedicated antibacterial and cytocompatibility experiments. Full article

High-precision environmental perception is essential for deep-sea exploration and autonomous underwater vehicle operations. However, physical factors such as light scattering and selective absorption, along with high target–background similarity, cause existing detection methods to suffer from low recall and inaccurate localization on targets with blurred edges, low contrast, or category ambiguity. To address these challenges, we propose YOLO-CAB, a YOLOv13-based underwater object detector designed to handle these complex scenes by optimizing feature extraction, boundary perception, and the loss function. First, we introduce the Context-Aware Large Selective Kernel (CALSK) module into the shallow backbone layers to expand the receptive field and adaptively enhance spatial features via multi-scale depthwise convolutions. Furthermore, the Spatial Boundary Attention Module (SBAM) is applied before the feature pyramid enters the detection head to refine multi-scale features and enhance sensitivity to target boundaries. To address the variance in detection difficulty across categories, we also develop the Momentum-based Category-Aware Weighted Intersection over Union (MCAWIoU) loss. Consequently, the proposed weighting mechanism improves localization accuracy and confidence distribution for challenging samples. Evaluated on the RUOD benchmark dataset, YOLO-CAB improves mean average precision ($mA{P}_{50}$) by 4.67%, the stricter localization metric ($mA{P}_{50-95}$) by 3.0%, and $F_1$ score by 3.7% over the vanilla YOLOv13n baseline. Ablation studies confirm the individual and synergistic contributions of these components. With 8.85 GFLOPs and an inference time of 5.21 ms under the tested hardware setting, YOLO-CAB improves detection accuracy while maintaining real-time inference. Full article