Search Results (1,952)

Automated assessment of coronary artery (CA) lesions via Coronary Computed Tomography Angiography (CCTA) is essential for the diagnosis of coronary artery disease (CAD). However, current deep learning approaches confront several challenges, primarily regarding the modeling of long-range anatomical dependencies, the effective decoupling of plaque texture from stenosis geometry, and the utilization of clinically prevalent mixed-grained annotations. To address these challenges, we propose a novel mixed-grained hierarchical geometric-semantic learning network (MG-HGLNet). Specifically, we introduce a topology-aware dual-stream encoding (TDE) module, which incorporates a bidirectional vessel Mamba (BiV-Mamba) encoder to capture global hemodynamic contexts and rectify spatial distortions inherent in curved planar reformation (CPR). Furthermore, a synergistic spectral–morphological decoupling (SSD) module is designed to disentangle task-specific features; it utilizes frequency-domain analysis to extract plaque spectral fingerprints while employing a texture-guided deformable attention mechanism to refine luminal boundary. To mitigate the scarcity of fine-grained labels, we implement a mixed-grained supervision optimization (MSO) strategy, utilizing anatomy-aware dynamic prototypes and logical consistency constraints to effectively leverage coarse branch-level labels. Extensive experiments on an in-house dataset demonstrate that MG-HGLNet achieves a stenosis grading accuracy of 92.4% and a plaque classification accuracy of 91.5%. The results suggest that our framework not only outperforms state-of-the-art methods but also maintains robust performance under weakly supervised settings, offering a promising solution for label-efficient CAD diagnosis. Full article

To achieve precise and safe laser acupuncture treatment, a computational model of the skin acupoint was constructed utilizing COMSOL Multiphysics (Version 6.1). This model incorporates its multilayer anatomical structure: the epidermis, papillary dermis, reticular dermis, hypodermis, and muscle layer. A coupled multiphysics approach integrating geometric optics, radiation beams, and bioheat transfer was employed to investigate the effects of light source parameters and cooling layers on the photothermal response and thermal damage of acupoints. Under optimized parameters (808 nm, 3 mm beam waist, 50 mW) with a 0.5 mm glycerol layer, 600 s irradiation achieved a stable dermal temperature (40.86–42.04 °C) and a negligible epidermal thermal damage factor (0.0063), significantly below the subclinical injury threshold of 0.15; under identical parameters, the dermal temperature for the Gaussian periodic pulsed source was maintained between 38.85 and 40.35 °C, with a corresponding epidermal thermal damage factor of merely 0.0010. The model exhibited good robustness, tolerating variations of ±5% in laser power and ±40% in glycerol layer thickness. The resultant temperature deviations in the epidermis and dermis were well within the safe range, and the thermal damage factor remained below the injury threshold. This work serves as a guideline for selecting laser acupuncture parameters according to acupoint depth. Full article

Age-related macular degeneration (AMD) is a leading cause of irreversible vision loss; anti-vascular endothelial growth factor (anti-VEGF) therapy is standard care for neovascular AMD (nAMD), yet treatment response varies. We systematically reviewed and meta-analyzed artificial intelligence (AI) and machine learning (ML) models using optical coherence tomography (OCT)-derived information to predict anti-VEGF treatment response in nAMD. PubMed, Embase, Web of Science, and IEEE Xplore were searched from inception to 18 December 2025 for eligible studies reporting threshold-based performance. Two reviewers screened studies, extracted data, and assessed risk of bias using PROBAST+AI; pooled sensitivity and specificity were estimated with a bivariate random-effects model. Seven studies met inclusion criteria, and six were synthesized quantitatively. Pooled sensitivity was 0.79 (95% CI 0.68–0.87), and pooled specificity was 0.83 (95% CI 0.62–0.94), with substantial heterogeneity. Specificity tended to be higher for long-term and functional outcomes than for short-term and anatomical outcomes. Most studies had a high risk of bias, mainly due to limited external validation and incomplete reporting. OCT-based AI models may help stratify treatment response in nAMD, but prospective, multicenter validation and standardized outcome definitions are needed before routine use; current evidence shows no consistent advantage of deep learning over engineered radiomic features. Full article

Background: Accurate microbiological diagnosis of bone and joint infections (BJIs) is frequently hampered by low bacterial load, biofilm formation, and suboptimal tissue processing. This study evaluated the diagnostic performance of mechanical bead-milling using the Ultra-Turrax^® Tube Drive system compared with standard vortex homogenization. Methods: In a prospective cohort of 116 patients undergoing surgery for suspected BJIs, 540 intraoperative samples were processed using both methods. Culture and 16S rRNA PCR results were analyzed using classical and Bayesian statistical approaches. Diagnostic performance was assessed globally and across specimen types and anatomical sites. Results: Ultra-Turrax^® significantly improved sensitivity across all sample types (87.1% vs. 75.2%, p < 0.0001), while maintaining comparable specificity (>99%). Culture positivity increased by 17%, with the greatest gains observed in bone samples and hip prosthesis infections. Quantitative cultures demonstrated a 1.5–2 log₁₀ CFU/mL increase in bacterial yield. In culture-negative specimens, 16S rRNA PCR detection doubled with Ultra-Turrax^® processing (26% vs. 13%, p = 0.04). No increase in contamination was observed. Time to positivity was similar between methods, although Ultra-Turrax^® provided earlier results in 17% of cases. Bayesian modeling confirmed superior sensitivity (posterior probability > 0.995). Conclusions: Ultra-Turrax^® bead-milling markedly enhances microbiological detection in BJIs, particularly in low-biomass and bone-derived specimens. Its simplicity, reproducibility, and compatibility with routine workflows support its integration into diagnostic pathways. This pre-analytical optimization may improve etiological identification and guide more targeted antimicrobial therapy. Full article

Periapical lesions (PALs) are a common sequela of pulpal pathology, and accurate radiographic detection is essential for successful endodontic diagnosis and treatment outcome. With recent advancements in Artificial Intelligence (AI), deep learning systems have shown remarkable potential to enhance the diagnostic accuracy of PALs. This study highlights recent evidence on the use of AI-based systems in detecting PALs across various imaging modalities. These include intraoral periapical radiographs (IOPAs), panoramic radiographs (OPGs), and cone-beam computed tomography (CBCT). A literature search was conducted for peer-reviewed studies published from January 2021 to July 2025 evaluating artificial intelligence for detecting periapical lesions on IOPA, OPGs, or CBCT. PubMed/MEDLINE and Google Scholar were searched using relevant MeSH terms, and reference lists were hand screened. Data were extracted on imaging modality, AI model type, sample size, subgroup characteristics, ground truth, and outcomes, and then qualitatively synthesized by imaging modality and clinically relevant moderators (i.e., lesion size, tooth type and anatomical surroundings, root-filling status and effect on clinician’s performance). Thirty-four studies investigating AI models for detecting periapical lesions on IOPA, OPG, and CBCT images were summarized. Reported diagnostic performance was generally high across radiographic modalities. The study results indicated that AI assistance improved clinicians’ performance and reduced interpretation time. Performance varied by clinical context: it was higher for larger lesions and lower around complex surrounding anatomy, such as posterior maxilla. Heterogeneity in datasets, reference standards, and metrics limited pooling and underscores the need for external validation and standardized reporting. Current evidence supports the use of AI as a valuable diagnostic platform adjunct for detecting periapical lesions. However, well-designed, high-quality randomized clinical trials are required to assess the potential implementation of AI in the routine practice of periapical lesion diagnosis. Full article

Background: Chest X-ray (CXR) is widely used for the assessment of thoracic diseases, yet automated multi-label interpretation remains challenging due to subtle visual patterns, overlapping anatomical structures, and frequent co-occurrence of abnormalities. While recent deep learning models have shown strong performance, limitations in interpretability, anatomical awareness, and robustness continue to hinder their clinical adoption. Methods: The proposed framework employs a hybrid ConvNeXtV2–Vision Transformer (ViT) architecture that combines convolutional feature extraction for capturing fine-grained local patterns with transformer-based global reasoning to model long-range contextual dependencies. The model is trained exclusively using image-level annotations. In addition to classification, three complementary post hoc components are integrated to enhance model trust and interpretability. A segmentation-aware Gradient-weighted class activation mapping (Grad-CAM) module leverages CheXmask lung and heart segmentations to highlight anatomically relevant regions and quantify predictive evidence inside and outside the lungs. An ontology-driven neuro-symbolic reasoning layer translates Grad-CAM activations into structured, rule-based explanations aligned with clinical concepts such as “basal effusion” and “enlarged cardiac silhouette”. Furthermore, a lightweight out-of-distribution (OOD) detection module based on confidence scores, energy scores, and Mahalanobis distance scores is employed to identify inputs that deviate from the training distribution. Results: On the VinBigData test set, the model achieved a macro-AUROC of 0.9525 and a Micro AUROC of 0.9777 when trained solely with image-level annotations. External evaluation further demonstrated strong generalisation, yielding macro-AUROC scores of 0.9106 on NIH ChestXray14 and 0.8487 on CheXpert (frontal views). Both Grad-CAM visualisations and ontology-based reasoning remained coherent on unseen data, while the OOD module successfully flagged non-thoracic images. Conclusions: Overall, the proposed approach demonstrates that hybrid convolutional neural network (CNN)–vision transformer (ViT) architectures, combined with anatomy-aware explainability and symbolic reasoning, can support automated chest X-ray diagnosis in a manner that is accurate, transparent, and safety-aware. Full article

Ultrasound-guided femoral vascular access (UFVA) is a crucial step in cardiovascular intervention, yet training models remain costly, anatomically limited, or insufficiently durable. This research aimed to develop and evaluate a reusable, self-healing vascularised leg phantom in collaboration with clinicians. This bifurcating vascular model was cast in Plastisol using a customisable silicone mould design. The material exhibited a Shore OO hardness of 18.0 ± 2.2, a speed of sound of 1454 ± 15 m/s, an acoustic attenuation of 1.66 ± 0.02 × 106 kg/m²s, and healed 18G needle lesions within 3 h. Training capabilities were evaluated in a workshop involving 18 medical students: FVA times improved by more than 60% after 5 min of free practice. Qualitative feedback was collected from 31 medical educators via a seven-question Likert survey, with most reporting they would adopt the phantom for teaching. Phantoms cost £7.87 for materials, yet educators valued the models at £100–£500, underscoring its perceived utility. Compared to commercial alternatives, this in-house production approach reduced costs by 10–60 times, achieving comparable durability and anatomical fidelity. This study establishes a scalable, ultra-low-cost method for producing anatomically realistic, self-healing vascular phantoms, validated for effective skill acquisition in both educational and research settings. Full article

This study employs numerical simulations to analyse current densities in modelled human organs originating from extremely low frequency (ELF) electromagnetic fields emanating from a 110 kV single-circuit high-voltage transmission line. Exposure to these ELF fields gives rise to both conduction and displacement currents within the human body, potentially perturbing endogenous bioelectric currents and raising concerns of health risks. Using CST Studio Suite 2018 software, a three-dimensional multipart ellipsoidal anatomical model is developed to analyse these phenomena. Although displacement currents have lower magnitudes than conduction currents, they contribute significantly to the total current density and must therefore be included in rigorous safety assessments. Simulation results indicate that the current density values remain below the basic restrictions of the International Commission on Non-Ionizing Radiation Protection. Full article

Accurate and interpretable staging of Alzheimer’s disease (AD) remains challenging due to the heterogeneous progression of neurodegeneration and the complementary nature of imaging and clinical biomarkers. This study implements and evaluates an optimized Hierarchical Multi-Modal Fusion Framework (HMFF) that systematically integrates 3D structural MRI with clinical assessment scales for robust three-class classification of cognitively normal (CN), mild cognitive impairment (MCI), and AD subjects. A standardized preprocessing pipeline, including N4 bias field correction, nonlinear registration to MNI space, ANTsNet-based skull stripping, voxel normalization, and spatial resampling, was employed to ensure anatomically consistent and high-quality MRI inputs. Within the proposed framework, volumetric imaging features were extracted using a 3D DenseNet-121 architecture, while structured clinical information was modeled via an XGBoost classifier to capture nonlinear clinical priors. These heterogeneous representations were hierarchically fused through a lightweight multilayer perceptron, enabling effective cross-modal interaction. To further enhance discriminative capability and model efficiency, a hierarchical feature selection strategy was incorporated to progressively refine high-dimensional imaging features. Experimental results demonstrated that performance consistently improved with feature refinement and reached an optimal balance at approximately 90 selected features. Under this configuration, the proposed HMFF achieved an accuracy of 0.94 (95% Confidence Interval: [0.918, 0.951]), a recall of 0.91, a precision of 0.94, and an F1-score of 0.92, outperforming unimodal and conventional multimodal baselines under comparable settings. Moreover, Grad-CAM visualization confirmed that the model focused on clinically relevant neuroanatomical regions, including the hippocampus and medial temporal lobe, enhancing interpretability and clinical plausibility. These findings indicate that hierarchical multimodal fusion with interpretable feature refinement offers a promising and extensible solution for reliable and explainable automated AD staging. Full article

Objective: To evaluate treatment patterns, predictors of treatment selection, and overall survival (OS) in patients with noninvasive (Ta–Tis) urothelial carcinoma of the urethra (UUC). Patients and Methods: We conducted a retrospective cohort study of adults diagnosed with noninvasive UUC (stage Ta or Tis, N0, M0) between 2018 and 2021 using the National Cancer Database. Patients were categorized into prostatic and non-prostatic urethral cohorts. Treatment groups included endoluminal ablation alone, ablation combined with topical intraluminal therapy, urethrectomy, and no subsequent treatment. Multinomial logistic regression was used to identify predictors of treatment selection. The OS was assessed using Kaplan–Meier and multivariable Cox regression, with separate models for each anatomical cohort. Results: A total of 436 patients were included (185 non-prostatic, 251 prostatic); 91.9% (n = 401) were male. Ablation alone was the most common treatment in both cohorts (non-prostatic: 57.3%; prostatic: 62.6%), followed by urethrectomy (non-prostatic: 21.1%) and ablation plus topical therapy (prostatic: 20.3%). In the non-prostatic cohort, high-grade histology (OR 15.15; 95% CI, 3.82–60.04) and Tis stage (OR 3.27; 95% CI, 1.10–9.69) were associated with increased odds of urethrectomy. In the prostatic cohort, high-grade histology was associated with increased use of urethrectomy (OR 59.29; 95% CI, 4.61–763.17) and ablation plus topical therapy (OR 3.09; 95% CI, 1.21–7.90). Tis stage was also associated with ablation plus topical therapy (OR 2.53; 95% CI, 1.14–5.62). This treatment approach was associated with improved OS compared with ablation alone (HR 0.18; 95% CI, 0.05–0.60; p = 0.005). Conclusions: Treatment selection differed substantially by tumor location, stage, and grade, reflecting both treatment selection in noninvasive UUC varied by tumor location, grade, and stage. In prostatic tumors, ablation plus topical therapy was associated with improved survival, supporting its role as a risk-adapted, organ-sparing strategy in selected patients. Full article

Background: Simulation-based training may offer a useful approach to support skill acquisition in neurointerventional stroke treatment without exposing patients to procedural risks. As the global demand for thrombectomy rises, training strategies that ensure procedural competence while addressing workforce constraints are increasingly important. With this pilot study, we aim to generate a hypothesis as to whether additional exposure of trainees to mechanical thrombectomy could benefit from simulator training on top of the standard training carried out on flow models. This study was designed as an exploratory pilot investigation and was not able to provide inferential or confirmatory statistical conclusions. Methods: Six novice participants (advanced clinical-year medical students with completed anatomical and preclinical training, but without previous exposure to catheter-based interventions) performed two neurointerventional tasks, vascular access and mechanical thrombectomy (MTE), on flow models. After a baseline assessment, three participants received standard model-based training (control group), and three received additional simulator training using a high-fidelity angiography simulator (Mentice VIST G5). Performance was reassessed after four weeks using technical and clinical surrogate metrics, which were ranked and descriptively analyzed. Results: No relevant differences were observed between groups for the vascular access task. In contrast, the simulator group demonstrated a trend toward improved performance in the MTE task, with greater gains in efficiency, autonomy, and procedural safety. Conclusions: Our findings indicate a possible benefit of even brief simulator exposure for skill acquisition for complex endovascular procedures such as MTE. While conventional training may suffice for basic skills, simulation may be particularly helpful in supporting learning in more advanced tasks. Full article

Segment regeneration in earthworms is a remarkable example of postembryonic morphogenesis, yet its fidelity and cellular mechanisms remain incompletely understood. The present study investigated posterior segment regeneration in adult specimens of the earthworm model Eisenia andrei from wound closure to the 5th postoperative week using anatomical, histological, and ultrastructural approaches. Rapid wound closure occurred through fusion of the cut edges of the body wall and midgut without direct involvement of coelomocytes. The regeneration blastema consisted of dedifferentiated epithelial and muscle cells, innervated by fibers from the last intact ventral nerve cord ganglion. Coelomocytes accumulated in the last intact segments and were primarily involved in debris clearance. Notably, early regenerating tissues lacked collagen fibers, which appeared only after the third postoperative week and remained sparse until the fifth week, whereas original segments exhibited intense, region-specific collagen deposition. Transmission electron microscopy revealed characteristic cytological changes in distinct stages of body wall regeneration, including muscle dedifferentiation and the emergence of collagen-producing fibroblasts. These findings indicate that early cell migration, proliferation, and orientation in the blastema proceed independently of collagen and that collagen functions as a delayed structural scaffold, supporting tissue integrity without impeding regeneration. Importantly, no scar formation was observed between old and new tissues, resembling scarless fetal wound healing. Overall, we clarified previously controversial cellular mechanisms and propose a new, comprehensive model for the early stages of segment regeneration. Our results highlight that coordinated dedifferentiation, spatiotemporal extracellular remodeling, and delayed collagen deposition underlie effective, scar-free regeneration in earthworms, offering insights into conserved mechanisms of regenerative repair across metazoans and potential strategies for enhancing tissue regeneration in mammals. Full article

Atlantoaxial instability (AAI) in toy- and small-breed dogs remains a significant clinical challenge, as the restricted anatomical space and risk of complications complicate the selection of implants. This study aimed to compare three patient-specific Ti-6Al-4V stabilizers for the C1–C2 region: a clinically used ventral C1–C3 plate, a shortened ventral C1–C2 plate, and a dorsal C1–C2 implant. Computed tomography, segmentation, virtual reduction, CAD/CAM design, and finite element analysis were employed to evaluate the linear-static mechanical behavior of each construct under loading ranging from 5 to 25 N, with a focus on displacements, von Mises stresses, and peri-screw bone strains. Additionally, cadaver procedures were performed in nine small-breed dogs using custom drill guides and additively manufactured implants to evaluate procedural feasibility and implantation time. Finite element models demonstrated that all stabilizers operated within material and biological safety limits. The C1–C3 plate exhibited the highest implant stresses, while the C1–C2 plate demonstrated an intermediate response, and the dorsal implant minimized implant stresses, albeit by increasing bone stresses. Cadaver experiments revealed that dorsal fixation required less implantation time than ventral fixation. Collectively, the findings indicate that all evaluated constructs represent safe stabilization options, and the choice of implant should reflect the preferred load-transfer pathway as well as anatomical or surgical constraints that may limit ventral access. Full article

Fontan patients experience anatomical remodeling over time, yet the mechanisms driving these changes remain unclear. This study aimed to characterize full-route Fontan remodeling and evaluate whether observed morphological changes arise from somatic growth alone or from the combined influence of conduit properties, surgical design, thoracic anatomy, and mechanical forces. Five Fontan patients (four extracardiac, one lateral tunnel) underwent analysis using two MRI-derived 3D models obtained between 1 and 4 years apart. Directional displacement was assessed using 3D shape overlays, surface geometry was quantified using the Koenderink Shape Index (KSI), and patient-specific growth mapping estimated localized tissue dynamics. Statistical analyses included a one-sample t-test for mean anterior displacement, the Grubbs’ test for outlier detection, and the Wilcoxon signed-rank test for KSI comparisons across time points. All patients exhibited anterior displacement of the Fontan route, with a mean shift of 0.29″ ± 0.33″ and one significant outlier (lateral tunnel, 0.87″). Four of five patients showed increased convexity over time. Growth mapping revealed minimal, heterogeneous native-tissue expansion, with localized growth up to 0.2 mm/year. Individual remodeling trajectories varied and did not consistently align with localized anterior growth, indicating that Fontan route remodeling is highly individualized and cannot be explained by somatic growth alone. This retrospective longitudinal case series study highlights the value of quantitative 3D geometric tools for assessing subtle Fontan route remodeling and supports the feasibility of growth-aware, patient-specific modeling frameworks in single-ventricle physiology. Full article

Medical image segmentation acts as a foundational element of medical image analysis. Yet its accuracy is frequently limited by the scale fluctuations of anatomical targets and the intricate contextual traits inherent in medical images—including vaguely defined structural boundaries and irregular shape distributions. To tackle these constraints, we design a multi-scale feature attention network (MFA-Net), customized specifically for thyroid nodule, skin lesion, and breast lesion segmentation tasks. This network framework integrates three core components: a Bidirectional Feature Pyramid Network (Bi-FPN), a Slim-neck structure, and the Convolutional Block Attention Module (CBAM). CBAM steers the model to prioritize boundary regions while filtering out irrelevant information, which in turn enhances segmentation precision. Bi-FPN facilitates more robust fusion of multi-scale features via iterative integration of top-down and bottom-up feature maps, supported by lateral and vertical connection pathways. The Slim-neck design is constructed to simplify the network’s architecture while effectively merging multi-scale representations of both target and background areas, thus enhancing the model’s overall performance. Validation across four public datasets covering thyroid ultrasound (TNUI-2021, TN-SCUI 2020), dermoscopy (ISIC 2016), and breast ultrasound (BUSI) shows that our method outperforms state-of-the-art segmentation approaches, achieving Dice similarity coefficients of 0.955, 0.971, 0.976, and 0.846, respectively. Additionally, the model maintains a compact parameter count of just 3.05 million and delivers an extremely fast inference latency of 1.9 milliseconds—metrics that significantly outperform those of current leading segmentation techniques. In summary, the proposed framework demonstrates strong performance in thyroid, skin, and breast lesion segmentation, delivering an optimal trade-off between high accuracy and computational efficiency. Full article