Search Results (2,203)

Meticulous segmentation of medical images requires obtaining both local and global spatial detailed information. The conventional U-Net model excels at local spatial feature extraction through residual convolutional blocks but struggles to capture global features. To resolve this issue, we propose the vision transformer U-NeT (ViTUNet) model framework, which combines the self-attention mechanism of the vision transformer (ViT) to capture global information while maintaining the extraction of local features via U-NeT. This proposed architecture introduces vision transformers to the existing residual convolution blocks in the U-Net encoder path, thereby capturing both local and global features. The decoder path then rebuilds this information into high-quality segmentation maps with accurately highlighted boundaries/edges. This model is utilized to segment carious lesions in bitewing dental radiographs. These images are pre-processed using augmentation, morphological operations, and segmentation to identify the boundaries/edges of the regions of interest (caries/cavity). The proposed method is evaluated on an augmented dataset containing 3000 image–watershed mask pairs. It was trained on 2400 training images and tested on 600 testing images. The experimental results exemplified significant improvements in segmentation performance, achieving $98.45\%$ validation accuracy, $97.88\%$ validation Dice coefficient, and $95.87\%$ validation intersection over union (IoU) metric scores. These results are superior compared to other conventional and state-of-the-art U-NeT models, thus highlighting the impact of transformer-based hybrid architectures in improving medical image segmentation tasks. Full article

The aim of this study was to develop an effective and fast tool to support the automatic segmentation of skin lesions, with particular emphasis on the precise differentiation between malignant and benign lesions. In response to the problem of high false positive rates in existing CAD systems, modern neural network architectures from the YOLO family (YOLOv8, YOLOv9, YOLOv11, YOLOv12, and YOLOv26) were used in this research. The models were trained and evaluated on a new, balanced dataset (7000 images) based on the ISIC archive, where the key innovation was the introduction of a dedicated background class representing healthy skin. Through a multi-stage, rigorous optimization process, it was demonstrated that the yolov11s-seg model is highly effective for this task. It achieved a strong balance between effectiveness and processing speed, obtaining an mAP@50 score of 0.840 and an overall precision of 0.852. From a clinical perspective, the model’s high sensitivity (85.9%) in detecting the most aggressive lesion, invasive melanoma (MI), is particularly noteworthy. Thanks to its extremely short inference time (only 4.8 ms), the proposed yolov11s-seg variant overcomes the limitations of heavy hybrid architecture, providing a stable and highly efficient solution showing significant potential for deployment in real-time medical mobile applications. Full article

Large-sized, room-temperature ultrafast scintillator single crystals are highly demanded for fast timing applications such as time of flight–positron emission tomography, high-speed medical imaging, and pulse heavy-ray detection. Sub-nanosecond scintillation was discovered in 16 mm sized CsPbCl₃Br_x single crystals in our previous research. In this work, the crystal size of CsPbCl₃Br_0.03 was enlarged to 2 inches (50.8 mm). Meanwhile, by precisely optimizing the vertical Bridgman growth process, we further increased the concentration of Br dopant to realize even faster scintillation decay. In this study, we conducted a series of tests on the grown crystals, including temperature-dependent photoluminescence tests, alpha particle excitation tests, X-ray imaging tests, etc. Via the strategy of the incorporation of Br₂, Br dopant introduces highly efficient fast recombination centers in perovskite CsPbCl₃Br_0.03 crystals, resulting in an unprecedently fast scintillation decay time of 303 ps under ²⁴¹Am α-particle excitation, which is significantly shorter than that of the pure CsPbCl₃ and all other perovskites by at least two orders of magnitude. Benefiting from the excellent optical transparency and high crystalline quality of the CsPbCl₃Br_0.03 crystal, an X-ray spatial resolution of up to 20 lp/mm is achieved. These results further demonstrate the great potential of large-sized CsPbCl₃Br_x single crystals for fast timing applications. Full article

Porto-sinusoidal vascular disorder (PSVD) is an umbrella term proposed by the Vascular Liver Disease Interest Group (VALDIG) in 2019. It refers to a group of non-cirrhotic vascular liver diseases that cause portal hypertension. These were previously described as idiopathic non-cirrhotic portal hypertension, hepatoportal sclerosis, nodular regenerative hyperplasia, and incomplete septal fibrosis. PSVD is characterized by injury and remodeling of portal venules and sinusoids. Immune dysregulation, prothrombotic states, infections, medications (e.g., oxaliplatin, thiopurines), toxins (e.g., arsenic), and genetic susceptibility often drive this process. Clinically, PSVD ranges from asymptomatic patients with only abnormal liver tests to severe complications of portal hypertension, such as variceal bleeding, ascites, and portal vein thrombosis. Patients typically have preserved liver synthetic function, helping distinguish PSVD from cirrhosis. Diagnosis is based on VALDIG criteria and requires an adequate liver biopsy that shows no cirrhosis. It also requires specific combinations of clinical signs of portal hypertension and characteristic histological lesions, such as obliterative portal venopathy, nodular regenerative hyperplasia, and incomplete septal fibrosis. Non-invasive tools, including imaging and liver stiffness measurement, are supportive. They often show discordance between marked portal hypertension and low liver stiffness, suggesting a non-cirrhotic cause. Management follows cirrhosis-based portal hypertension guidelines. This includes non-selective beta-blockers, endoscopic variceal ligation, TIPS, anticoagulation in selected patients, and liver transplantation for refractory or end-stage disease. Prognosis is generally better than in cirrhosis, with a 5-year transplant-free survival rate of approximately 85% compared to 60% in matched cirrhotics. However, major gaps remain in the true epidemiology, the natural history of early or subclinical PSVD, validated non-invasive biomarkers, and disease-modifying therapies. Full article

Background: The complex anatomy of the head and neck region challenges medical students. Cinematic rendering (CR) is an advanced visualization technique that enables three-dimensional (3D) representation of cross-sectional image data and is used in education at the Faculty of Medicine at Johannes Kepler University. Methods: For the first time, CR images were used to illustrate otolaryngology anatomy in medical education. The educational value of this approach was evaluated using a questionnaire assessing six core statements and dichotomous variables, including prior experience with CR and otolaryngology. Results: All six statements showed high levels of agreement based on mean evaluation scores. Evaluation results differed according to participants’ prior experience with CR. A strong correlation was exploratorily observed between prior experience with CR and improved spatial awareness of otolaryngology anatomy (ρ = 0.80, p < 0.05). Additionally, prior experience with CR correlated with improved understanding of complex disease processes (ρ = 0.76, p < 0.05) and enhanced general comprehension of the respective field (ρ = 0.74, p < 0.05). Conclusions: These findings suggest that early integration of CR into otolaryngology education may support students’ perceived spatial understanding and facilitate comprehension of complex disease processes. Full article

Background/Objectives: Vision-language models (VLMs) such as GPT (Generative Pre-Trained Transformer) have shown potential for medical image interpretation; however, challenges remain in generating reliable radiological findings in clinical practice, as exemplified by dental pathologies. This study proposes a Self-correction Loop with Structured Output (SLSO) framework as an integrated processing methodology to enhance the accuracy and reliability of AI-generated findings for jaw cysts in dental panoramic radiographs. Methods: Dental panoramic radiographs with jaw cysts were used to implement a 10-step integrated processing framework incorporating image analysis, structured data generation, tooth number extraction, consistency checking, and iterative regeneration. The framework functioned as an external validation mechanism for GPT outputs. Performance was compared against the conventional Chain-of-Thought (CoT) method across seven evaluation items: transparency, internal structure, borders, root resorption, tooth displacement, relationships with other structures, and tooth number. Results: The SLSO framework improved output accuracy for multiple items compared to the CoT method, with the most notable improvements observed in tooth number identification, tooth displacement detection, and root resorption assessment. In successful cases, consistently structured outputs were achieved after up to five regenerations. The framework enforced explicit negative finding descriptions and suppressed hallucinations, although accurate identification of extensive lesions spanning multiple teeth remained limited. Conclusions: This investigation established the feasibility of the proposed integrated processing methodology and provided a foundation for future validation studies with larger, more diverse datasets. Full article

Background/Objectives: Simulation digital twins (DT) models that integrate patient-specific imaging with artificial intelligence (AI)-based segmentation and extended reality (XR) technologies are rapidly increasing in relevance in personalized medicine. While their clinical applications are expanding, their role as reusable educational tools and the technical pipeline utilized for their development remain incompletely characterized. This narrative review examines current approaches to digital twin creation and XR integration, illustrated by a scoliosis-specific proof-of-concept educational case study. Methods: A narrative technical review was conducted by identifying relevant search keywords within the fields of AI-based image segmentation, extended reality in medicine, and medical education based on the authors’ expertise and familiarity with the subject. PubMed, Google Scholar, and Scopus were searched for English-language studies published primarily between 2015 and 2025 addressing patient-specific three-dimensional modeling, AI-driven segmentation, and XR applications in spine, orthopedic, anesthesiology, and interventional care. A de-identified case of scoliosis is used to present a proof-of-concept example of this process of creating a simulated digital twin for the purpose of medical education in a recorded XR format. Results: Prior studies demonstrated benefits of patient-specific 3D models for anatomical understanding and procedural planning, while highlighting limitations in segmentation accuracy and workflow integration. Nevertheless, while DTs have traditionally served clinical roles in surgical planning or pre-procedural rehearsal, their pedagogical potential remains under-explored. In the proof-of-concept case study, AI-assisted segmentation enabled rapid creation of an anatomically detailed scoliosis digital twin that was incorporated into XR and used to produce a reusable, spatially anchored instructional experience focused on neuraxial access. Conclusions: AI-enabled digital twin models integrated with XR represent a promising approach for personalized, anatomy-driven medical education. Further evaluation is needed to assess educational outcomes, scalability, and integration into clinical training workflows. Full article

Deep learning has become central to semantic segmentation of three-dimensional medical images. However—despite many published models—their adoption in practice remains limited, as deployment often requires advanced programming skills and familiarity with specific machine learning frameworks. Thus, technical barriers restrict its use to specialized users. To address this, we present SegMed (version 1.0), an open-source, standalone desktop application that provides an end-to-end workflow for deep learning-based medical image segmentation. SegMed supports the loading and inspection of common medical image formats, as well as array-based formats. The application integrates standard preprocessing operations often used in the field and directly supports loading of pretrained segmentation models implemented in both PyTorch (version 2.X) and Keras (version 2.X) and those created using the Medical Open Network for AI framework (version 1.X). Models are automatically inspected to infer required configurations, such as input size and post-processing steps, enabling segmentation with minimal user intervention. Results can be exported as volumetric images or 3D surface meshes for downstream analysis, visualization, or special applications such as virtual reality. SegMed was tested using multiple publicly available pretrained models, demonstrating robustness and flexibility across diverse segmentation tasks. By abstracting low-level implementation details, SegMed lowers technical barriers, promotes reproducibility, and facilitates the integration of AI-assisted segmentation into medical imaging workflows. Full article

Background: Contemporary dentistry increasingly relies on tools and methods derived from the exact sciences, particularly mathematics and physics, to better understand the complexity of biological processes. One such tool is fractal analysis (FA), which enables the characterization and quantification of irregular, complex, self-similar structures commonly observed in nature in the form of the fractal dimension (FD). In oral radiology, it has been found useful for describing structural changes in bone tissue. Objective: The aim of this review is to present the current state of knowledge regarding the application of fractal analysis in the management of patients with, or at risk for, medication-related osteonecrosis of the jaw (MRONJ), with particular emphasis on its diagnostic and prognostic potential. This paper summarizes key research findings, and discusses the principal challenges and limitations associated with the use of this method of analysis in MRONJ cases. Materials and Methods: The inclusion criteria were as follows: original papers, the presence of MRONJ, and fractal analysis. In order to find relevant studies, international databases, including PubMed and Google Scholar, were searched. The last search was performed on 29 November 2025. Six articles were included in the systematic review. Results: The majority of the review studies show lower FD values for MRONJ patients and healthy control groups. The values are the lowest for necrotic lesions and highest for perinecrotic bone tissue. Conclusions: FD values calculated from radiological images of the jaws can be used to differentiate healthy and MRONJ-affected patients and to describe necrotic lesions. Fractal analysis has potential to be used in the diagnosis and monitoring of MRONJ after further studies and standardization of methodology. Full article

Accurate classification of breast cancer histopathology images is essential for early diagnosis and effective clinical management. However, conventional deep learning models often exhibit performance degradation under limited labeled data and lack interpretability, which restricts their clinical applicability. To address these challenges, we propose GAFR-Net, a robust and interpretable Graph Attention and Fuzzy-Rule Network designed for histopathology image classification under scarce supervision (defined here as less than 10% labeled data). GAFR-Net constructs a similarity-driven graph to model inter-sample relationships and employs a multi-head graph attention mechanism to capture complex relational representations among heterogeneous tissue structures. Meanwhile, a differentiable fuzzy-rule module integrates intrinsic topological descriptors—such as node degree, clustering coefficient, and label consistency—into explicit and human-readable diagnostic rules. This architecture establishes transparent IF–THEN inference mappings that emulate the heuristic reasoning process of clinical experts, thereby enhancing model interpretability without relying on post-hoc explanation techniques. Extensive experiments conducted on three public benchmark datasets—BreakHis, Mini-DDSM, and ICIAR2018—demonstrate that GAFR-Net consistently surpasses state-of-the-art methods across multiple magnifications and classification settings. These results highlight the strong generalization capability and practical potential of GAFR-Net as a trustworthy decision-support framework for weakly supervised medical image analysis. Full article

This review paper provides a comprehensive overview of the functional materials and assembly technologies used in intracardiac echocardiography (ICE) and intravascular ultrasound (IVUS) transducers. ICE and IVUS are advanced medical imaging technologies that play significant roles in the diagnosis and treatment of cardiovascular diseases, involving material selection and fabrication processes for miniature piezoelectric ultrasonic transducers. The review begins with an introduction to the principles and applications of ICE and IVUS, highlighting their advantages over other imaging modalities, then delves into the materials and assembly processes of the transducers, presenting the mainstream trends and research progress in various directions in this field in recent years. Finally, the paper summarizes the future technological development and clinical application trends of ICE/IVUS ultrasonic transducers. Full article

This study addresses the challenge of the difficult identification of organ structures in the standard sections of fetal abdominal ultrasounds. A deep learning-based multi-task model named SEG-FAUSP was developed to segment the core anatomical structures of seven key fetal abdominal ultrasound sections. We collected fetal abdominal ultrasound images from pregnant women in the mid-pregnancy period (18–24 weeks) using various mainstream ultrasound devices, and professional physicians annotated key anatomical structures (e.g., umbilical veins, gastric bubbles, spine) in the images. Based on an improved deep learning framework, the model accurately segments and locates the target organ structures through a parallel dual-branch semantic segmentation network, which avoids the over-reliance on large-scale pre-trained data in traditional methods. Experimental results show that the model achieves excellent performance in anatomical structure segmentation, with the intersection over union of the bladder and gastric bubble both reaching above 0.84; its segmentation accuracy for complex structures such as the inferior vena cava is also significantly superior to the baseline model. As an end-to-end model, it simplifies the clinical interpretation process of fetal abdominal ultrasound, reduces the risk of missed diagnoses caused by unclear organ identification, provides an efficient auxiliary tool for prenatal screening in grassroots medical institutions, and is of great significance for improving the quality of newborns. Full article

Background: Artificial intelligence (AI) is expected to play an increasingly important role in medicine and dentistry. While its diagnostic potential has been tested in various medical fields, limited research exists on its applications within oral medicine diagnoses using clinical images. Objective: This pilot study aimed to evaluate the diagnostic accuracy of ChatGPT, Gemini, and Copilot in identifying benign, potentially malignant, and malignant oral lesions. Methods: A cross-sectional study was conducted using clinical images from three categories: benign oromucosal conditions, oral potentially malignant disorders, and malignant oral lesions. Results: ChatGPT evaluated all images and consistently outperformed Copilot—and in some cases Gemini—across multiple diagnostic questions, with statistically significant advantages particularly in the cancer subgroup. Copilot showed the weakest performance, with high rates of missing evaluations and significantly lower proportions of correct responses in several analyses. Across both full-dataset and adjusted analyses, ChatGPT demonstrated the highest diagnostic performance overall. Diagnostic accuracy metrics for malignancy suspicion was similar for ChatGPT and Gemini. Several limitations such as sample size, lack of reproducibility testing and inability of some AI models to process images must be taken into account when interpreting the results. Conclusions: AI tools show promise but cannot yet replace clinical expertise. Further research and development are needed to improve the accuracy and applicability of AI diagnostic tools. Full article

Flexible X-ray detectors have emerged as a promising technology for portable medical imaging and wearable electronics, yet their manufacturing remains constrained by the competing requirements of device performance, mechanical conformability, and production scalability. Conventional solution-based deposition methods fail to yield high-quality perovskite thick films with uniform morphology, while vacuum evaporation techniques are limited by exorbitant operational costs and low throughput. Herein, we report an optimized solution-processed vacuum evaporation strategy that enables the fabrication of high-quality perovskite films (~1 μm thick) on flexible polyethylene naphthalate (PEN) substrates at a low processing temperature of 100 °C. By incorporating tailored additives into the precursor solution and precisely modulating the vapor-phase conversion kinetics, we achieved significant improvements in film density, crystallinity, and morphological uniformity. Systematic investigations were conducted to elucidate the structure–property relationships across three material systems: pure methylammonium lead iodide (MAPbI₃), halogen-doped methylammonium lead iodide-bromide (MAPb(IBr)₃), and synergistic cation-halogen engineered cesium-methylammonium lead iodide-bromide (CsMAPb(IBr)₃). The optimized flexible PIN-type X-ray detector based on CsMAPb(IBr)₃ exhibited exceptional performance metrics, including a dark current density as low as 5.2 nA cm⁻² and an X-ray sensitivity of up to 1.43 × 10⁴ μC·Gy_air⁻¹·cm⁻². Remarkably, the device retained over 95% of its initial performance after 400 bending cycles with a bending radius of 6 mm, demonstrating outstanding mechanical robustness and operational durability. This work establishes a viable, cost-effective technical route for the scalable production of high-performance flexible X-ray detectors, addressing critical challenges in the advancement of next-generation portable imaging technologies. Full article

Ultrasound (US) imaging is a widely used diagnostic method in clinics. Real-time-generated US images are used for rapid diagnosis without harm to patients. The quality of US imaging highly depends on the skill of the physician due to the differences among physicians. Techniques for autonomous robotic ultrasound (AU-RUS) acquisitions are expected to become an effective means to improve the level of US diagnosis, reduce the workload of physicians, and improve the standardization of US imaging quality. This paper aims to summarize the current research status of techniques for AU-RUS acquisitions, and to discuss the research trends and challenges regarding related technologies. Firstly, the techniques for AU-RUS acquisitions and systems are outlined. The techniques for teleoperated or autonomous US acquisitions are briefly discussed. Representative RUS acquisition systems are introduced. Then, the current research status of AU-RUS acquisitions is reviewed from four research directions: force sensitivity and control, scanning path-planning and positioning, US treatment guidance, and US image processing technology and quality assessment optimization. This review provides a decision-oriented autonomy perspective by mapping typical methods to workflow components across the stages of perception, decision-making, and execution. We identify major deployment bottlenecks, including safety-verifiable autonomy and failure recovery, motion compensation under deformation, and the lack of standardized, clinically meaningful US image quality metrics. Finally, the shortcomings of current research are summarized and analyzed, and the research trends and challenges for AU-RUS acquisitions are prospected. Full article