Abstract
OBJECTIVE: Accurate classification of salivary gland tumors is critical to guiding appropriate management. This study evaluates the diagnostic performance of artificial intelligence models in classifying salivary gland tumors on ultrasound. DATA SOURCES: A comprehensive search of CINAHL, PubMed, and Scopus was conducted through January 28, 2025. REVIEW METHODS: Two independent reviewers screened articles and extracted data following Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. Studies evaluating the diagnostic performance of artificial intelligence in classifying salivary gland tumors were included for fixed and random-effects meta-analyses. The Quality Assessment of Diagnostic Accuracy Studies-2 tool for systematic reviews of diagnostic accuracy studies was used to assess study quality and risk of bias. RESULTS: Out of 741 articles identified, 12 studies (N = 4721) met inclusion criteria. Nine studies evaluated artificial intelligence models differentiating benign from malignant tumors, and three studies assessed classification of pleomorphic adenomas versus Warthin tumors. For benign versus malignant tumors, sensitivity was 0.91 (95% CI: 0.86, 0.95), specificity was 0.86 (95% CI: 0.80, 0.92), and accuracy was 0.85 (95% CI: 0.81, 0.90). For pleomorphic adenomas versus Warthin tumors, sensitivity was 0.81 (95% CI: 0.74, 0.89), specificity was 0.88 (95% CI: 0.81, 0.95), and accuracy was 0.84 (95% CI: 0.79, 0.90). CONCLUSION: Artificial intelligence models demonstrate strong diagnostic accuracy in ultrasound-based classification of salivary gland tumors. These results highlight the potential of artificial intelligence as a diagnostic tool, though broader validation is needed before routine clinical implementation.