Artificial intelligence for X-ray scaphoid fracture detection: a systematic review and diagnostic test accuracy meta-analysis

Category	Detail
Condition	Scaphoid fractures, the most common carpal bone fracture
Key Mechanisms	Early detection and immobilization prevent complications; AI using convolutional neural networks (CNNs) improves fracture detection on wrist radiographs
Target Population	Patients with suspected scaphoid fractures undergoing wrist radiographs
Care Setting	Emergency departments and radiology settings

Scaphoid fractures have a high incidence (82–89%) and require early diagnosis to avoid avascular necrosis, carpal instability, and osteoarthritis.
Conventional wrist X-rays have limited sensitivity (66–81%) with up to 50% occult fractures missed, often due to non-specialist interpretation and fatigue.
AI deep learning models, particularly CNNs, show promise in improving diagnostic accuracy and sensitivity for scaphoid fractures on plain radiographs.

Use wrist radiographs as first-line imaging for suspected scaphoid fractures due to accessibility and cost-effectiveness.
Consider AI-assisted interpretation to improve sensitivity and reduce missed fractures on radiographs.
Reserve advanced imaging (CT or MRI) for cases with high clinical suspicion and negative radiographs, acknowledging higher cost and radiation (CT).

Early immobilization upon diagnosis to prevent complications.
Prompt referral for specialist evaluation if fracture is suspected but not confirmed on radiographs.

Monitor for signs of avascular necrosis, carpal instability, or early osteoarthritis in patients with delayed or missed diagnosis.

Risk of missed fractures due to low sensitivity of radiographs and interpretation by less experienced clinicians.
Radiation exposure with CT and high cost with MRI limit their routine use.

Patients presenting with wrist trauma and suspected scaphoid fractures

AI-enhanced radiograph interpretation may reduce missed diagnoses and improve early treatment initiation, potentially lowering complication rates.

Incorporate AI-based CNN models to assist radiograph interpretation for scaphoid fractures, especially in emergency settings with limited specialist availability.
Maintain high clinical suspicion for occult fractures despite negative radiographs and consider advanced imaging accordingly.
Ensure timely immobilization and specialist referral to optimize patient outcomes.

Clinical Scorecard: Utilizing Artificial Intelligence for the Detection of Scaphoid Fractures via X-ray: A Comprehensive Review and Meta-analysis of Diagnostic Accuracy