Artificial intelligence in emergency musculoskeletal imaging: A critical review of current applications.

Clinical Snapshot

Key Findings

• P Value: Not applicable — narrative review design

• Effect Size: No pooled effect size reported. Individual study performance metrics are discussed qualitatively. Deep learning models are described as showing promise for fracture detection and structured classification, but no aggregate sensitivity, specificity, or AUC values are synthesised.

• Primary Outcome: Narrative summary of AI applications in emergency musculoskeletal imaging across four domains: (1) workflow efficiency and triage prioritisation, (2) abnormality detection (fractures, soft tissue injuries, infectious processes), (3) paediatric-specific AI applications, and (4) large language model applications for report generation and clinician communication

• Nnt Or Sensitivity: No NNT, pooled sensitivity, or pooled specificity reported. The review references AI systems improving workflow efficiency and reducing reporting delays, and enhancing abnormality detection, but without quantified summary statistics these claims cannot be independently evaluated.

• Confidence Interval: Not applicable — no quantitative synthesis performed

Clinical Application

Feasibility of AI integration in emergency musculoskeletal imaging is context-dependent. Workflow prioritisation tools (worklist triage) are the most immediately implementable. Fracture detection AI tools are commercially available in some jurisdictions but require local validation, integration with existing PACS/RIS infrastructure, and radiologist oversight. LLM-based report generation tools are emerging but carry significant medicolegal risk without robust governance frameworks. Implementation costs, IT infrastructure requirements, and staff training represent substantial barriers in many emergency settings. In Australia, AI-assisted radiology tools are subject to TGA regulation as Software as a Medical Device (SaMD) under the TGA's Digital Health framework. Several AI fracture detection tools have received TGA clearance or are under evaluation. The RACGP and Royal Australian and New Zealand College of Radiologists (RANZCR) have published position statements on AI in radiology emphasising the need for clinical validation, transparency, and maintained radiologist accountability. The Medicare Benefits Schedule (MBS) does not currently provide specific item numbers for AI-assisted reporting, creating a funding gap for implementation. Australian emergency departments face similar high-volume imaging pressures to those described in this review, and rural and remote settings — where after-hours radiologist access is limited — represent a particularly compelling use case for AI triage tools. However, the dataset bias acknowledged by the authors (predominantly derived from Northern Hemisphere populations) raises questions about performance on Australian demographic and epidemiological profiles. Emergency department patients requiring musculoskeletal imaging, including adults with acute fractures, soft tissue injuries, and suspected musculoskeletal infections, and paediatric patients where developmental anatomy complicates interpretation. Most applicable to high-volume emergency imaging centres with access to validated AI software.

Abstract

Artificial intelligence (AI) is increasingly shaping emergency musculoskeletal imaging, where rapid and accurate diagnosis is often challenged by high imaging volumes and time pressure. These constraints increase the risk of missed injuries and underscore the need for tools that support faster and reliable assessments. AI systems show promise in improving workflow efficiency by prioritizing urgent studies, guiding modality selection, and reducing reporting delays. Deep learning models can enhance abnormality detection by identifying fractures, soft tissue injuries, and infectious processes, and can provide structured classifications that support clinical decision making. Pediatric-focused AI systems address age-specific developmental considerations and offer valuable support for clinicians with varying levels of pediatric musculoskeletal expertise. Large language models further expand the role of AI by improving report clarity, generating structured impressions, and facilitating communication with clinicians, patients, and families. Despite these advances, challenges remain, including limited external validation, dataset bias, and medicolegal considerations. This review summarizes current AI applications across these domains and highlights key strengths, limitations, and future directions for safe and effective integration into emergency musculoskeletal imaging.

References

1. 1.Tsang, B., Gong, B., Probyn, L., & Patlas, M. N. (2026). Artificial intelligence in emergency musculoskeletal imaging: A critical review of current applications. Diagnostic and Interventional Imaging. https://doi.org/10.1016/j.diii.2026.02.003