Abstract
Background:
Carpal tunnel syndrome (CTS) is the most common entrapment neuropathy affecting the extremites. Although nerve conduction velocity (NCV) testing remains the diagnostic gold standard, ultrasound (US) has merged as a valuable non-invasive alternative. However, its diagnostic performance is often limited by operator dependency. Recently, machine learning (ML)-based artificial intelligence (AI) tools have shown promising tool to standardize and automate US image interpretation. This sysmtmic review and meta-analysis aimed to evaluate the diagnostic accuracy of ML-assisted ultrasound in diagnosing CTS compared to traditional ultrasound method.
Methods:
Following PRISMA guideline, we systematically searched PubMed, Embase, and other additional relevant databasethrough May 31, 2025. Eligible studies applied ML algorithms to sonographic diagnosis of CTS and reported sufficient data to culculate diagnostic performance. A bivariate random-effects model was used to pool sensitivity, specificity, and to construct a a hierarchical summary receiver operating characteristic (HSROC) curve. Positive (LR?) and negative likelihood ratios (LR?) were calculated, and risk of bias was assessed using the QUADA-2 tool. Publication bias was evaluated with Deeks'funnel plot.
Results:
Ten studies encompassing 1937 patients were included. The pooled sensitivity was 0.89 [95% CI: 0.85-0.93] and specificity was 0.83 [95% CI: 0.75-0.90], with an estimated area under the curve (AUC) of 0.91. LR? value ranging from 2.2 to 10.3 and LR? from 0.09 to 0.28, indicating moderate to strong diagnostic utility for several models. Deeks' funnel plot showed no significant publication bias.
Conclusion:
ML-assisted ultrasound systems demonstrate high diagnostic accuracy for CTS, with performance comparable to expert interpretation. These tools offer a reliable, non-invasive, and operator-independent alternative method for CTS screening and diagnosis. While findings are promising, further large-scale, prospective studies are necessary to validate its clinical utility.
