Artificial intelligence demonstrates potential to enhance orthopaedic imaging across multiple modalities: A...

Area of Science:

• Orthopaedic Surgery
• Medical Imaging
• Artificial Intelligence (AI)
• Machine Learning (ML)

Background:

• Numerous AI-assisted medical imaging applications exist in orthopaedics.
• A lack of comparative studies on the clinical efficacy and utility of these AI applications is evident.
• This systematic review addresses the need to evaluate AI's role in orthopaedic imaging.

Purpose of the Study:

• To systematically review and evaluate the effectiveness and reliability of AI applications in orthopaedic imaging.
• To assess the impact of AI on diagnostic accuracy, image segmentation, and operational efficiency.
• To analyze AI performance across various imaging modalities in orthopaedics.

Main Methods:

• A comprehensive literature search was conducted following PRISMA guidelines (PubMed, Cochrane, Scopus) from inception to March 2024.
• Studies published between September 2018 and February 2024 evaluating ML model effectiveness in orthopaedic imaging were included.
• Bias assessment was performed using the Joanna Briggs Institute (JBI) Critical Appraisal tool and ROBINS-I tool.

Main Results:

• 53 studies analyzed 11,990,643 images; 39 reported Dice Similarity Coefficient (DSC), 15 reported accuracy and sensitivity.
• Convolutional Neural Networks (CNNs) were the most frequent ML models, appearing in 17 studies (32%).
• AI models demonstrated high performance metrics across various imaging modalities for segmentation and analysis.

Conclusions:

• AI applications show diverse capabilities in accurately segmenting and analyzing orthopaedic images.
• AI models achieve high performance metrics in orthopaedic imaging tasks.
• Further research, including randomized controlled trials, is necessary to validate AI findings in clinical settings.