Artificial Intelligence-Aided X-Ray in Emergency Bone Fracture Diagnosis: Accuracy, Limitations, and Clinical Implications

Assessing AI-Assisted X-Ray for Bone Fracture Diagnosis in Emergency Settings

Bone fractures require swift and precise diagnosis in emergency rooms to avoid complications. While computed tomography (CT) scans remain the gold standard for fracture detection, their availability and cost can limit use, especially in urgent contexts. Recent advances in artificial intelligence (AI) offer promising tools to aid fracture detection via standard X-rays.

Study Overview

This study evaluated the AI tool SmartUrgence®, developed by Milvue, for detecting bone fractures on X-ray images in emergency cases. Its performance was directly compared to CT scan results to measure accuracy, sensitivity, and specificity. The goal was to understand whether AI could effectively support fracture diagnosis and identify areas needing improvement.

Methodology

• Participants: Adults (19 years or older) presenting with suspected fractures after trauma, referred from orthopedic to radiology departments.
• Imaging: All underwent both AI-assisted X-ray analysis via SmartUrgence® and CT scans.
• Exclusions: Patients with contraindications to imaging, suspected axial skeleton fractures (spine/skull), or poor-quality images were excluded.
• Analysis: AI assessments were compared against CT findings to calculate diagnostic metrics.

Key Findings

• Diagnostic Performance: SmartUrgence® achieved 91.13% sensitivity and 95.45% specificity in detecting fractures.
• Predictive Values: Positive predictive value (PPV) was 93.39%, and negative predictive value (NPV) was 93.85%, indicating reliable identification of both fractures and non-fractures.
• Overall Accuracy: The tool demonstrated 93.67% accuracy, with balanced accuracy at 93.25%, reflecting consistent performance across fracture types.
• Limitations: Despite strong results, AI performance differed significantly from CT scans in some fracture categories, highlighting areas for refinement.

Clinical Relevance

AI-assisted X-ray analysis can provide rapid, accurate preliminary fracture detection in emergency settings, especially when CT access is limited or delayed. This tool can help reduce diagnostic errors and support clinicians in making timely decisions. However, it should complement rather than replace CT imaging, which remains essential for comprehensive fracture evaluation.

Study Context and Population

The study involved 300 trauma patients with a balanced gender distribution and a wide age range (19–80 years). Lower limb fractures, particularly ankle and foot, were most common, followed by upper limb injuries such as hand fractures. The AI system flagged uncertain cases in 5.7% of instances, which were further reviewed by radiologists to confirm diagnoses.

Limitations and Future Directions

• The study was limited to a single institution and excluded axial skeleton fractures due to AI model constraints.
• Direct comparison with radiologist interpretations on uncertain cases was not performed, leaving some questions about real-world clinical integration.
• Future research should focus on enhancing AI detection for complex fracture types and improving interpretability.
• Integrating AI smoothly into clinical workflows and evaluating its impact on patient outcomes and healthcare efficiency are priority areas.

Conclusion

AI tools like SmartUrgence® show high accuracy in detecting fractures on X-rays and can serve as valuable aids in emergency care. While not a replacement for CT scans, AI can support radiologists by highlighting probable fractures quickly, especially during busy or resource-limited shifts. Ongoing improvements and validation will be key to safe, effective adoption in routine practice.

For healthcare professionals interested in expanding their knowledge of AI applications in medical imaging and diagnostics, resources such as Complete AI Training's latest courses offer practical guidance on leveraging AI tools in clinical settings.