Deep learning in oral lichen planus diagnosis: a systematic review of clinical image-based detection approaches.

OBJECTIVES

To systematically evaluate the diagnostic performance of deep learning models in detecting oral lichen planus (OLP) using clinical photographs.

STUDY

DESIGN

This systematic review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and included studies utilizing deep learning architectures (e.g., Convolutional Neural Networks (CNNs), and Vision Transformers) for OLP diagnosis. Performance metrics such as accuracy, sensitivity, specificity, and Area Under the Receiver Operating Characteristic Curve (AUC) were extracted. Study quality was assessed using the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) tool.

RESULTS

All included models showed high diagnostic accuracy, with some exceeding 95%. Architectures such as InceptionResNetV2 and Xception have achieved notable sensitivity and specificity. However, limitations include small, homogeneous datasets, inconsistent image preprocessing, and limited external validation.

CONCLUSIONS

Deep learning shows strong potential for OLP diagnosis via clinical images, but the real-world application remains limited. Broader datasets, robust validation, and integration of explainable artificial intelligence (AI) are essential for clinical adoption.