Artificial Intelligence Helps Diagnose Oral Potentially Malignant Disorders: A Systematic Review and Meta-Analysis

Area of Science:

• Oral medicine
• Medical imaging
• Artificial intelligence

Background:

• Oral potentially malignant disorders (OPMDs) are precursors to oral cancer, a significant global health concern.
• Early detection and prevention are critical for improving patient outcomes.
• Artificial intelligence (AI) offers novel noninvasive methods for analyzing medical data, including clinical photographs, to aid in OPMD diagnosis.

Purpose of the Study:

• To systematically review and conduct a meta-analysis on the diagnostic accuracy of AI applied to intraoral photographs for OPMDs.
• To compare AI diagnostic performance against histological examination, the current gold standard.

Main Methods:

• A systematic literature search was performed across five major databases (MEDLINE, Embase, Cochrane Library, Scopus, Web of Science) up to November 10, 2023.
• Studies comparing AI diagnostic methods with histology were included.
• Quantitative analysis assessed sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), diagnostic odds ratio (DOR), and likelihood ratios (LR+, LR-) with 95% confidence intervals (CIs).

Main Results:

• Six studies, analyzing 898 images from 4,046, met the inclusion criteria.
• The best-performing AI models demonstrated high diagnostic accuracy.
• Overall performance metrics for top AI architectures included sensitivity of 0.94 (95% CI, 0.88-0.95) and specificity of 0.95 (95% CI, 0.85-0.98).

Conclusions:

• AI-based diagnostic tools exhibit a high negative predictive value for identifying OPMD lesions from intraoral images.
• While AI can potentially reduce unnecessary specialist consultations, clinical judgment remains essential.
• Further prospective research is needed to integrate AI diagnostics into routine care for enhanced oral cancer screening and early detection.