Diagnostic Accuracy of Artificial Intelligence, CBCT, and Clinical Examination in Detecting Number of Root Canals in Conventional and Retreated Maxillary and Mandibular Molars

Introduction: Accurate root canal detection is crucial for successful endodontic treatment, particularly in complex molar cases. Conventional methods, such as clinical examination and cone-beam computed tomography (CBCT), have their limitations, as high radiation exposure. Recent advancements in Artificial Intelligence (AI) have shown promise in improving diagnostic accuracy. This study aims to compare the effectiveness of AI, CBCT, and clinical examination using a dental operating microscope (DOM) in detecting root canals in upper first, upper second, and lower first molars, in both conventional and retreatment cases. Methods: CBCT scans from 210 patients requiring non-surgical root canal therapy or re-treatment were selected. The scans were analyzed using three detection methods: clinical examination via DOM, interpretation by two experienced endodontists using CBCT, and an AI convolutional neural network (CNN) software (Diagnocat). The detected number of root canals was recorded and compared across the three methods.

Inclusion criteria

• Male and female patients who were capable of providing informed consent
• Age between 18 to 40 years old.
• A restorable tooth.

Exclusion criteria

• Patients that underwent vital pulp therapies.
• Patients with calcifications in pulp space.
• Open apex/immature roots.
• Teeth restored by full coverage crowns.
• Pregnant women by taking adequate history from patient and pregnancy test that was done in the first visit