Assessment of AI-Driven Software Accuracy in Diagnosing Oral Lesions Using Radiographic Imaging

A BSTRACT Background: The application of artificial intelligence (AI) in dentistry has gained significant attention, particularly in diagnosing oral lesions using radiographic imaging. Materials and Methods: A total of 150 digital radiographic images of patients with suspected oral lesions were analyzed. The dataset was split into training (70%) and testing (30%) subsets for the AI model. The AI software was trained to identify and classify oral lesions into three categories: benign, malignant, and precancerous. The results were compared with diagnoses made by three experienced oral radiologists, which served as the reference standard. Diagnostic accuracy, sensitivity, specificity, and precision were calculated to evaluate the performance of the AI system. Results: The AI-driven software demonstrated an accuracy of 92.5%, with a sensitivity of 94% and specificity of 91% in diagnosing oral lesions. Precision values for benign, malignant, and precancerous lesions were 89%, 95%, and 93%, respectively. Interobserver agreement between the AI and human radiologists was found to be substantial (kappa = 0.84). The AI software showed faster diagnostic processing, with an average time of 3 seconds per image compared to 2 minutes by human experts. Conclusion: AI-driven software exhibits high accuracy and efficiency in diagnosing oral lesions using radiographic imaging, providing a valuable adjunct to traditional diagnostic methods. Its rapid processing capability and consistency in performance make it a promising tool for enhancing diagnostic workflows in dental practice.