Artificial Intelligence–Assisted Interpretation of CBCT Scans for Early Periapical Pathology Detection: A Comparative Endodontic Study

Abstract

Early detection of periapical pathologies is critical for successful endodontic treatment and preservation of dental structures. Cone-beam computed tomography (CBCT) provides detailed three-dimensional imaging, yet interpretation can be time-consuming and subject to observer variability. This study evaluates the efficacy of artificial intelligence (AI) in assisting the interpretation of CBCT scans for early periapical lesion detection, comparing its performance with that of experienced endodontists. A set of CBCT scans from patients with suspected periapical pathology was analyzed using a trained AI model and independently by clinicians. Diagnostic accuracy, sensitivity, specificity, and area under the receiver operating characteristic curve (AUC) were assessed. Results demonstrate that AI-assisted interpretation achieves comparable or superior accuracy to human evaluation, with potential to reduce diagnostic time and improve early intervention outcomes. Integration of AI into endodontic imaging workflows may enhance diagnostic consistency and support clinical decision-making.