Breast cancer is one of the leading causes of mortality among women worldwide, and early and accurate mammographic assessment is important for timely diagnosis and improved patient survival. However, accurate breast lesion detection requires overcoming the issues of low lesion contrast, dense breast tissue, image noise, and variation in mammographic acquisition. In this study, we present a YOLOv8x-based computer-aided detection framework for automated breast lesion localization in full-field digital mammography (FFDM) and digital breast tomosynthesis (DBT) images acquired under heterogeneous clinical conditions. The framework was developed and evaluated using 501 clinical mammographic images obtained from multiple mammography systems, namely GE Digital Mammography for 2D FFDM and GE Senographe Pristina for DBT/3D mammography. The proposed approach integrated expert-guided radiological annotation, dedicated preprocessing, and high-resolution mammographic analysis at 1024 × 1024 pixels to improve lesion localization performance. Lesion regions were manually annotated by a specialist radiologist through clinically guided bounding-box procedures compatible with YOLO-based detection. Experimental results demonstrated accurate lesion localization performance, with precision of 98.87%, recall of 100%, mAP@0.5 of 99.50%, and mAP@0.5:0.95 of 99.16%. External validation further demonstrated promising cross-domain performance under heterogeneous imaging conditions. The findings demonstrate the potential value of clinically heterogeneous data, expert-guided annotation, and external validation for developing deployment-oriented computer-assisted breast lesion localization systems.
In Internet of Multimedia Things (IoMT) environments, video capture and frame transmission consume substantial sensor energy and impose considerable network overhead because of redundant data transmission. Traditional approaches often rely on repeated pixel-level operations and frequent execution of deep segmentation models, thereby increasing computational and transmission costs. This paper proposes an Object-Level Semantic Video Transmission Framework (OSVT) for resource-constrained IoMT environments in which high-definition visual data must be transmitted over low-bandwidth wireless multimedia sensor networks (WMSNs). The proposed OSVT incorporates a dual-stage wake-up mechanism that remains in sleep mode during static scenes, thereby reducing transmission load. The high-power YOLOv11-seg module is activated only when motion exceeds a predefined pixel threshold. Upon detecting an object of interest, the system adopts a hybrid transmission strategy. If the object is recognized from a prior frame, only spatiotemporal metadata are transmitted; otherwise, pixel data for the new object are sent. Rather than performing segmentation for every frame, the method tracks the centroids of detected bounding boxes across consecutive frames to estimate object motion. This mechanism reduces latency and bandwidth consumption. The framework was evaluated through simulations using video sequences representing diverse streaming scenarios. Performance was assessed using peak signal-to-noise ratio (PSNR), structural similarity index measure (SSIM), Learned Perceptual Image Patch Similarity (LPIPS), and the Semantic Information Index (SII). The results indicate that OSVT can support energy-efficient semantic video transmission in resource-constrained IoMT systems while maintaining reconstruction quality.
This study introduces a numerical approach for estimating alpha-decay half-lives of alpha-emitting nuclei using a modified five-parameter Woods-Saxon potential. The radial Schrödinger equation is solved with the Numerov method to calculate the wave function and the barrier transmission probability, without relying exclusively on the WKB approximation. The calculations were performed for 398 nuclides taken from the NUBASE2020 database. Of these, 264 nuclides were used for the MWS5 parameter-adjustment stage, whereas the remaining 134 nuclides were reserved for an independent evaluation of the model. The results were compared with those obtained from two reference models: the modified Gamow-Like Model (MGLM) and the Deformed Woods-Saxon Potential Model (DWS). At the global level, the proposed MWS5 approach gives physically consistent half-life estimates with deviations comparable to those of the reference models. Improved performance is observed in selected heavy and superheavy regions, particularly for nuclei with proton numbers in the range Z = 108–118. The results suggest that the proposed methodology provides a simple and reproducible computational alternative for studying alpha-decay processes and may be useful as a complementary tool in nuclear physics research and in the numerical simulation of quantum tunnelling phenomena.
Stingless bees incorporate propolis into their nest as a structural and protective material. Information remains limited on how solvent fractionation affects the antioxidant and enzyme-inhibitory activities of propolis produced by Indo nesian Tetragonula laeviceps. This study examined a crude extract prepared with 70% ethanol together with the n- hexane, ethyl acetate, and water fractions obtained from it. The analysis comprised total phenolic content (TPC), total flavonoid content (TFC), three antioxidant assays (ABTS, CUPRAC, TBARS), and inhibition of tyrosinase, elastase, and collagenase. The largest TPC and TFC values were obtained from the ethyl acetate fraction, at 85.93 ± 1.28 mg GAE/g and 71.92 ± 0.97 mg QE/g, respectively. The sample fraction ranked first in every activity assay. Its IC values were 214.68 ± 0.93 μg/mL in the ABTS assay, 19.12 ± 2.14 μg/mL in the CUPRAC assay, and 29.19 ± 1.33 μg/mL in the TBARS assay. The IC 50 50 values against tyrosinase, elastase, and collagenase were 151.18 ± 1.52, 203.23 ± 1.98, and 98.19 ± 1.56 μg/ mL, respectively. Pearson correlation analysis showed associations between TPC, TFC, three antioxidant activities and three enzyme-inhibition responses. Principal component analysis also separated the ethyl acetate fraction from the remaining samples. Under the conditions examined, this fraction showed the strongest overall activity and can therefore be prioritized for chemical characterization and further evaluation in skin-related cosmeceutical research.
Phishing attacks continue to evolve in sophistication, rendering static detection methods increasingly ineffective. Existing URL-based approaches suffer from limited adaptability to emerging phishing patterns, mislabeled training data, and insufficient validation protocols. This paper proposes a hybrid phishing URL detection system that integrates Fuzzy C-Means (FCM) clustering with XGBoost classification, enhanced by a novel Micro Adaptive Feature Extractor (MAFE). The system employs a multi-stage pipeline: feature engineering generating 36 statistical and interaction features, MAFE producing 15 adaptive features through class-aware dynamic weighting, micro-pattern detection, and entropy analysis, and FCM with K=2 clusters providing soft membership features to XGBoost. A two-pass confidence-based mislabel detection protocol identifies and removes 2.66% suspected labeling errors from the training data. The system is evaluated on the large-scale DEPHIDES dataset of 5,202,841 URLs using a proper three-way split: 60% training, 10% validation, and 30% test. The classification threshold is optimized exclusively on the validation set, ensuring unbiased test evaluation. The proposed system achieves 97.86% accuracy and 99.84% AUC on the raw test set, improving to 98.98% accuracy after verified mislabel removal. Comparative evaluation demonstrates that the system outperforms Random Forest 95.91%, LightGBM 96.31%, CatBoost 95.75%, and standalone XGBoost 96.59% trained on identical data with the same evaluation protocol. The system processes URLs at 6,528 URLs/second, with 95% confidence intervals of 97.77%–97.96% for accuracy. A sensitivity analysis confirms robustness to the MAFE adaptation rate parameter, with accuracy varying by only 0.06% across α ∈ [0.05, 0.30].
