Artificial intelligence in traumatology

Area of Science:

• Radiology
• Artificial Intelligence
• Orthopedic Surgery

Background:

• Distal radius fractures (DRFs) are common orthopedic injuries.
• Accurate detection of DRFs is crucial for effective patient management.
• Physician diagnostic accuracy can be influenced by various factors, necessitating supportive tools.

Purpose of the Study:

• To develop artificial intelligence (AI) software to aid physicians in detecting distal radius fractures (DRFs).
• To evaluate the impact of AI software support on the diagnostic accuracy of physicians for DRFs.
• To compare the performance of physicians with and without AI assistance in identifying wrist fractures.

Main Methods:

• A convolutional neural network (CNN) model was trained on a large dataset of wrist radiographs (26,121 images).
• Eleven physicians (six trainees, five hand surgeons) assessed 200 wrist radiograph pairs twice: once unaided and once with AI support.
• Diagnostic accuracy metrics, including sensitivity, specificity, and error rates, were compared between unaided and AI-assisted assessments.

Main Results:

• The CNN model achieved an area under the receiver operating curve of 0.97.
• AI assistance improved physician sensitivity for DRF detection from 80% to 87%.
• AI support increased specificity from 91% to 95%, reducing the overall error rate from 14% to 9%.

Conclusions:

• AI software, used as a second opinion, can enhance the diagnostic accuracy of distal radius fracture detection.
• The integration of AI tools shows promise in improving orthopedic diagnostic workflows.
• AI-assisted diagnosis may lead to more accurate and efficient fracture identification in clinical settings.