Clinical Report: AI Software Accuracy in Detecting Pediatric Upper Limb Fractures

Overview

This study evaluated the diagnostic accuracy of the AI-based BoneView® software for detecting upper limb fractures in children aged 2 to 18 years. The AI tool demonstrated promising performance as a supportive diagnostic aid, particularly in settings lacking pediatric radiology expertise.

Background

Pediatric fractures are common and predominantly affect the upper extremities, with the forearm most frequently involved. Accurate and prompt diagnosis is critical to prevent long-term complications in the growing skeleton. Conventional radiographs remain the standard diagnostic tool, but interpretation challenges arise due to complex pediatric bone anatomy and limited pediatric radiology expertise. AI-driven tools like BoneView® offer potential to improve diagnostic accuracy while adhering to radiation safety principles.

Data Highlights

The study retrospectively analyzed radiographs from children aged 2 to 18 years with suspected upper extremity fractures. BoneView® software was trained on over 300,000 radiographs, including 30% pediatric cases. The AI outputs included fracture presence classified as “yes,” “no,” or “doubt,” and binary detection of elbow joint effusions. Radiographs were evaluated by experienced pediatric radiologists to establish a reference standard for comparison.

Key Findings

• Over 75% of pediatric fractures occur in the upper extremities, with the forearm most commonly affected.
• BoneView® AI software was trained on a large dataset including pediatric radiographs, enhancing its applicability to children.
• The AI tool provides trinary fracture classification and binary detection of elbow joint effusions, aiding indirect fracture diagnosis.
• Radiographs were reviewed by two pediatric radiologists to ensure diagnostic accuracy as a reference standard.
• AI assistance could reduce unnecessary CT scans and repeated radiographs, supporting ALARA radiation safety principles.
• AI may help mitigate diagnostic challenges due to limited pediatric radiology expertise and increasing radiology workload.

Clinical Implications

The integration of AI tools like BoneView® in pediatric emergency settings can enhance fracture detection accuracy and reduce diagnostic delays. This may decrease reliance on higher-radiation imaging modalities and support adherence to radiation safety guidelines. AI assistance is particularly valuable where pediatric radiology expertise is limited, potentially improving patient outcomes and workflow efficiency.

Conclusion

BoneView® AI software shows promise as a reliable adjunct in detecting upper limb fractures in children, addressing challenges inherent to pediatric radiographic interpretation. Further prospective validation could solidify its role in routine clinical practice.

References

1. Pediatric Fracture Epidemiology and Imaging Challenges
2. Radiation Safety and ALARA Principle in Pediatric Imaging
3. BoneView® AI Software Development and Validation
4. Study Design and Ethical Considerations in Pediatric Radiology Research