In this section
@article{Musbahi:2025,
author = {Musbahi, O and Ahmed, A and Hall, T and Siddique, M and Katyula, K and Cobb, J and van, Arkel R and Jones, G},
journal = {Journal of Experimental Orthopaedics},
title = {Deep learning classification models demonstrate high accuracy and clinical potential in radiograph interpretation in the arthroplasty clinical pathway: a systematic review and meta-analysis},
year = {2025}
}
TY - JOUR
AB - Artificial intelligence (AI) is set to transform medical imaging, streamlining, andimproving the delivery of care. Used to diagnose disease, plan surgery, and monitorpatients post-operatively, imaging is a cornerstone of the osteoarthritis-arthroplastyclinical pathway. To date, no systematic review has examined AI’s diagnostic andprognostic role in interpreting radiographs and cross-sectional imaging in thearthroplasty pathway. With growing interest from the orthopaedic community, thismeta-analysis broadly evaluates the performance of deep learning (DL) algorithms inthese imaging tasks. Ovid Medline, Ovid Embase, Scopus, and Web of Science weresystematically searched for studies published between January 1, 2012, and April 1,2024, evaluating DL algorithms for diagnostic and prognostic tasks along theosteoarthritis-arthroplasty pathway. Eligible studies included those that usedestablished diagnostic or surgical candidacy assessments as ground truth. Studyquality was assessed using the Quality Assessment of Diagnostic Accuracy Studies 2tool, and pooled sensitivity and specificity were determined. Hierarchical summaryreceiver operating characteristic curves assessed diagnostic performance. Of the2355 studies identified, 138 studies were included. Of these, 66 studies were used inthe meta-analysis for the results of AI-only interpretation and 11 studies for the resultsof human-only interpretation. The AI studies had a pooled sensitivity of 0.88 (95% CI:0.81 to 0.92) and a pooled specificity of 0.91 (95% CI: 0.87 to 0.94). In comparison, theclinician interpretation studies had a pooled sensitivity of 0.76 (95% CI: 0.64 to 0.85)and a pooled specificity of 0.79 (95% CI: 0.59 to 0.90). This meta-analysis highlightsthe potential of DL algorithms to improve efficiency in osteoarthritis classification andprognosis in the arthroplasty pathway based on low-to-moderate quality evidence.Although the results are not generalizable, the findings suggest DL models have th
AU - Musbahi,O
AU - Ahmed,A
AU - Hall,T
AU - Siddique,M
AU - Katyula,K
AU - Cobb,J
AU - van,Arkel R
AU - Jones,G
PY - 2025///
SN - 2197-1153
TI - Deep learning classification models demonstrate high accuracy and clinical potential in radiograph interpretation in the arthroplasty clinical pathway: a systematic review and meta-analysis
T2 - Journal of Experimental Orthopaedics
ER -