This review explores the advancements in artificial intelligence for radiograph fracture diagnosis, emphasizing technological developments and inherent limitations. Artificial intelligence improves diagnostic accuracy and manages workflow efficiency. The review categorizes artificial intelligence applications in fracture diagnosis into four primary tasks: recognition, classification, detection, and localization. The most popular performance metrics, such as diagnostic accuracy, precision, sensitivity, specificity, and area under the curve analysis, are used to compare artificial intelligence systems with traditional radiological methods and are explained as serving as a guide. Each task and performance metric is illustrated with practical examples and success stories from recent literature, offering insights into the strengths and weaknesses of various artificial intelligence approaches, such as support vector machines, convolutional neural networks, and generative adversarial networks. We also incorporate case analyses, underscoring the potential and limitations of artificial intelligence in fracture detection. In particular, challenges were posed by external factors such as casts and anatomical complexities. Future directions are explored, emphasizing human-artificial intelligence collaboration and the development of more advanced, transparent artificial intelligence systems alongside parallel evolving ethical considerations and regulatory frameworks. This review aims to equip clinicians with the knowledge to understand and utilize artificial intelligence technologies effectively in their practice.

The fibrin matrix of thrombi is intertwined with neutrophil extracellular traps (NETs) containing histones that render resistance to fibrinolysis. During NET formation, histones are citrullinated. Our study addresses the question of whether citrullination modifies the fibrin-stabilizing effects of histones. We studied the structure and viscoelastic properties of fibrin formed in the presence of native or citrullinated H1 and core histones by scanning electron microscopy, clot permeation, and oscillation rheometry. The kinetics of fibrin formation and its dissolution were followed by turbidimetry and thromboelastometry. Co-polymerizing H1 with fibrin enhanced the mechanical strength of the clots, thickened the fibrin fibers, and enlarged the gel pores. In contrast, the addition of core histones resulted in a reduction in the fiber diameter, and the pores were only slightly larger, whereas the mechanical stability was not modified. Plasmin-mediated fibrinogen degradation was delayed by native and citrullinated core histones, but not by H1, and the action of des-kringle1-4-plasmin was not affected. Plasmin-mediated fibrinolysis was inhibited by native and citrullinated core histones, and this effect was moderated when the kringle domains of plasmin were blocked or deleted. These findings suggest that in NET-containing thrombi that are rich in core histones, alternative fibrinolytic enzymes lacking kringle domains are more efficient lytic agents than the classic plasmin-dependent fibrinolysis.

Objective: In emergency trauma care, artificial intelligence (AI) may aid fracture detection on radiographs, potentially reducing radiologists’ workload. We evaluated the role of deep learning-based decision-support software in the reporting of trauma cases. Materials and methods: We retrospectively analyzed 2317 trauma radiographs acquired at a single center: 1,174 images obtained from November 1 to 16, 2023, without access to the AI tool during reporting, and 1,143 images from February 1 to 13, 2024, with discretionary use of the AI output during reporting. The AI software output was compared with final radiology reports, with ground truth established by a musculoskeletal radiologist with 9 years’ experience. Accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated at both the fracture and patient levels. Results: The dataset included 1,914 patients with 1,188 acute fractures (621 in November, 567 in February). At the fracture level, standalone AI achieved 90.7% accuracy, 87.8% sensitivity, 94.0% specificity, 94.3% PPV, and 87.2% NPV in November, 94.1%, 93.5%, 94.6%, 94.5%, and 93.6% in February, respectively. Non-AI-assisted radiologists reached 92.4%, 89.0%, 96.2%, 96.3%, and 88.7%, AI-assisted radiologists 93.4%, 90.0%, 96.7%, 96.4%, and 90.7%, respectively. At the patient level, AI’s overall performance reached up to 96.5% accuracy and 95.6% sensitivity. Discrepancies between AI and radiologists occurred in 326 cases, often related to anatomical variants such as accessory ossicles. Conclusion: Standalone AI demonstrated near-expert accuracy and sensitivity in fracture detection at both fracture and patient levels. PPV increased with AI support, indicating more accurate detection of actual fractures. Relevance statement: By examining discretionary real-world use of AI in trauma radiography, this study shows that clinical benefit is not guaranteed by algorithmic performance alone, as optional AI integration does not consistently improve radiologist sensitivity, underscoring a critical implementation gap in practice. Key Points: Standalone AI achieves near-expert fracture detection performance in trauma radiography. Discretionary AI use does not consistently improve radiologist sensitivity. AI use reduces discrepancies, suggesting improved diagnostic consistency. Clinical benefit of AI depends on real-world implementation strategy.