Diagnostic Performance of AI-Powered Histopathology: Can Machines Match Oral Pathologists in Identifying Epithelial Dysplasia?

Background: Epithelial dysplasia is a critical precursor to oral squamous cell carcinoma. Manual histopathology, though gold standard, is subject to variability and delay. Objective: To evaluate the diagnostic accuracy of an artificial intelligence (AI)-powered histopathology system in identifying and grading oral epithelial dysplasia. Materials and Methods: = 250) to manual diagnoses by expert pathologists. Agreement, sensitivity, specificity, precision, and Cohen's kappa were calculated. Results: AI achieved >93% diagnostic accuracy across categories. Strong agreement was noted for non-dysplastic and severe lesions; moderate discrepancies occurred in mild/moderate grades. Conclusion: AI demonstrates strong potential in aiding dysplasia diagnosis, particularly for high-volume screening, though human oversight remains essential.