Browse Articles

YOLOv11-MFF: A multi-scale frequency-adaptive fusion network for enhanced CXR anomaly detection

Chest X-ray (CXR) represents one of the most widely utilized clinical diagnostic tools for thoracic diseases. Nevertheless, computer-aided diagnosis based on chest radiographs still faces considerable challenges in anomaly detection. Certain lesions in CXRs exhibit subtle radiographic characteristics with ambiguous boundaries, low pixel occupancy, and weak contrast. While existing studies primarily focus on improving multi-scale feature fusion, they frequently overlook complications arising from background noise and varied lesion morphology. This study introduces YOLOv11-MFF, an enhanced YOLOv11 network with three key innovations. Specifically, a novel Frequency-Adaptive Hybrid Gate (FAHG) is developed to improve contrast differentiation between lesions and background. A Multi Scale Parallel Large Convolution (MSPLC) block is designed and integrated with the original C3k2 module to expand receptive fields and enhance long-range modeling capacity. Furthermore, a Feature Fusion module (FF) is introduced to reinforce target-relevant feature representation through channel-wise modulation via weight recalibration mechanisms. Benefiting from these advancements, the network achieves significant improvements in detecting multi-scale and overlapping lesions. Experimental results on the public VinDr-CXR dataset demonstrate that YOLOv11-MFF outperforms state-of-the-art models, achieving a precision of 48.2%, recall of 42.5%, mAP@0.5 of 41.5%, and mAP@0.5:0.95 of 22.6%.

The potential of TDP-43 PET ligands for a biological diagnosis of TDP-43 proteinopathies

The impact of CwlM depletion on the susceptibility of Mycobacterium smegmatis to anti-tuberculosis drugs

CwlM, identified as an N-acetylmuramoyl-l-alanine amidase, plays crucial roles in the synthesis and remodeling of peptidoglycan in mycobacteria. This protein also appears to participate in both drug susceptibility and tolerance mechanisms within these organisms. In our study, we employed CRISPR interference (CRISPRi) to deplete CwlM in Mycobacterium smegmatis ( M. smegmatis ) and examined the resulting effects on the susceptibility of mycobacteria to first-line anti-tuberculosis drugs, including isoniazid (INH), rifampicin (RIF), pyrazinamide (PZA), and ethambutol (EMB), as well as the β-lactams cefoxitin and imipenem. Our findings revealed that CwlM depletion increased the susceptibility of the bacterium to RIF, EMB, cefoxitin, and imipenem, while tolerance was heightened against INH and PZA. The enhanced antibiotic susceptibility can primarily be attributed to increased permeability of the bacterial cell wall. Conversely, the observed tolerance to INH might be ascribed to elevated expression of the amidase known as hydrazidase along with its LuxR-type regulator. Furthermore, several genes associated with peptidoglycan synthesis appeared to correlate with increased expression levels of either hydrazidase or its LuxR-type regulator. Collectively, these findings indicate that CwlM depletion significantly influences the susceptibility of M. smegmatis towards certain anti-tuberculosis drugs and may be implicated in drug susceptibility and tolerance mechanisms in M. smegmatis .

EWS::FLI1 expression in human embryonic mesenchymal stem cells leads to transcriptional reprograming, defective DNA damage repair and Ewing sarcoma

Uncovering functional INDELs responsible for splice sites in sheep with different immune profiles naturally exposed to gastrointestinal nematode infection

Modeling macroscopic brain dynamics with brain-inspired computing architecture

A deep learning based framework for music-synchronized dance choreography with pose quantization and motion prediction for activity recognition

Matrix-based imaging through dynamic scattering

Abstract Noninvasive optical imaging through complex scattering media presents a major challenge across multiple fields. State-of-the-art techniques, such as reflection matrix decomposition and neural networks, rely on multiple measurements with varying illumination within the sample decorrelation time, making their application challenging in rapidly varying dynamic media. Here, we show that due to commutativity property of the convolution operation, dynamic scattering in isoplanatic imaging is mathematically analogous to varying illumination in static media. This insight allows leveraging matrix-based approaches developed for static scattering to rapidly varying dynamic media. Specifically, we show that the covariance matrix of a set of scattered light camera frames captured through a dynamic scattering sample has the same mathematical form as the reflection matrix of a static medium, with the target object playing the scattering medium’s role. We demonstrate this concept by high-resolution diffraction-limited imaging through dynamic scattering across multiple modalities, from incoherent fluorescence microscopy to coherence-gated holographic reflection imaging.

Radon gas mapping for environmental assessment in Dessie, Ethiopia

Abstract The geological formations of Dessie town contain rocks that release radon gas, the second leading cause of lung cancer after smoking. To date, there have been no investigations of geology-based radon health risks in Dessie town. The main objective of this study was to evaluate and map the temporal and spatial distribution of radon gas concentrations within Dessie town. The evaluation and mapping were done using advanced geological and geospatial techniques, employing Quantum Geographic Information Systems (QGIS) software to integrate the research area’s weather conditions and geological data using shape files for the study area and masking with its geological map. Geological analysis helps to know whether the geological formation of Dessie town is composed of basaltic rocks, as shown on the geological map of Dessie town, which was obtained from the Ethiopian Geological Institute. Based on these geological formations and geospatial analysis, radon gas activity concentrations were estimated in two different ranges: from 3 $$\:581.5{\:\text{B}\text{q}\text{m}}^{-3}\:\text{t}\text{o}\:\text{10,744.5}{\:\text{B}\text{q}\text{m}}^{-3}$$ for the emanation coefficient 0.3 and from $$\:\:35.815\:{\:\text{B}\text{q}\text{m}}^{-3}\:\text{t}\text{o}\:107.\:445\:{\text{B}\text{q}\text{m}}^{-3}\:\:$$ for the emanation coefficient 0.1 for this study area. The radon gas distribution map indicates radon levels in Dessie town fall into two ranges, each with its own potential hotspots, as shown by circles of different colors on the map. These results indicate that there exist health-related problems at nine stations due to prolonged exposure to radon gas, and for further investigation of radon gas activity concentrations and their health impact, we recommend experimental measurements provide more accurate results.

27%-efficiency silicon heterojunction cell with 98.6% cell-to-module ratio driving new momentum towards the 29.4% limit

Effects of mindfulness breathing meditation on stress and cognitive functions: a heart rate variability and eye-tracking study

Homeostasis of glucose and lipid metabolism during physiological responses to a simulated hypoxic high altitude environment

Multi-institutional validation of AI models for classifying urothelial neoplasms in digital pathology

Abstract This study proposes a deep learning approach for classifying normal, noninvasive, and invasive urothelial neoplasms via digitized histopathologicalimages. Despite many artificial intelligence (AI) models for cancer diagnosis, few focus on bladder lesions or differentiate between these critical categories. We developed convolutional neural networks (CNNs) and transformer-based models, which were trained on 12,500 whole-slide images (WSIs) from five institutions, with preprocessing steps including stain normalization and patch extraction. Fivefold cross-validation was used for evaluation against expert-annotated labels. Among tested models, EfficientNet-B6 achieved the highest performance, with an accuracy of 0.913 (95% confidence interval (CI), 0.907–0.920), sensitivity of 0.909 (95% CI, 0.904–0.914), specificity of 0.956 (95% CI, 0.953–0.960), F1-score of 0.906 (95% CI, 0.901–0.911), and an area under the receiver operating characteristic curve (AUC) of 0.983 (95% CI, 0.982–0.984). These results demonstrate the effectiveness and generalizability of AI-based bladder cancer classification.

Explore brain-inspired machine intelligence for connecting dots on graphs through holographic blueprint of oscillatory synchronization

A novel electrode design using Cu-TCPP MOF-modified MWCNT for efficient electro-organic cross coupling synthesis of biphenyl derivatives in a Urea/Chol-Cl DES system as a green and sustainable electrolyte

Insight into the effects of gamma radiation on MLCCs: from in situ capacitance experiments to physical mechanisms

Performance comparison of post-earthquake disaster susceptibility assessment models based on GIS: a case study of the Lushan County in Ya’an City, China

Retraction Note: DNA-PK inhibition synergizes with oncolytic virus M1 by inhibiting antiviral response and potentiating DNA damage

Monitoring juvenile sicklefin lemon shark Negaprion acutidens in remote marine nurseries using unmanned aerial vehicles (UAVs)

Abstract Understanding spatial and demographic patterns in threatened coastal sharks is essential for effective conservation, yet remote reef systems remain understudied due to logistical constraints. We used dual unmanned aerial vehicles (UAVs) to monitor juvenile Negaprion acutidens around Dongsha Atoll, a no-take marine reserve in the northern South China Sea. Thirteen synchronized UAV surveys were conducted during summer and winter, covering 20 locations categorized into three zones representing different levels of human impact. We quantified seasonal variation in shark abundance, body size, spatial distribution, and environmental drivers using Generalized Linear Mixed Models (GLMMs). Results revealed stable overall abundance but strong fine-scale shifts between summer and winter. Northeastern sites showed sharp declines in shark sightings during winter, likely due to monsoonal exposure, while sheltered southern zones supported increased winter presence. Neonates were concentrated in lagoon habitats, whereas larger individuals occurred farther offshore and in seagrass areas, indicating ontogenetic habitat expansion. Human impact shaped demographic structure: low-impact areas hosted more and larger sharks, while smaller individuals and nearshore aggregation dominated high-impact zones. These findings confirm Dongsha’s role as a critical nursery habitat for N. acutidens and highlight the utility of UAV surveys for capturing spatiotemporal ecological patterns in remote ecosystems. By integrating UAV monitoring with conservation planning, managers can prioritize seasonal protection of nursery zones and respond adaptively to climate and anthropogenic pressures.

Unsteady flows uncover the limits of the stress-optic law