Accuracy of artificial intelligence technology in detecting number of root canals in human mandibular first molars obturated and Indicated for retreatment : (Diagnostic accuracy experimental study) /

Abstract

Statement of problem: Failure to disinfect and to locate all canals is a major reason for root canal primary and secondary failures, which may lead to periapical periodontitis. Cone beam computed tomography (CBCT) is considered the gold standard in canal detection although its use must be justified beforehand due to its high radiation dose. Using artificial intelligence (AI) software may enhance canal detection and avoid human error in interpreting CBCT images. Aim of the study: To evaluate the accuracy of new AI technology for detecting root canals in mandibular first molars retreatment cases in comparison to dentist clinical access cavity and CBCT imaging. Materials and methods: Thirty-five patients with obturated lower first molar(s) referred for retreatment was participated in this study. After a pre-treatment periapical x-ray to aid practitioner in access cavity formation a CBCT was performed for all cases. Stage 1: CBCT scans performed to all participants were randomly distributed and observed by the principle investigator and supervisors and the number of canals found was recorded. Stage 2: patients were randomly distributed on 6 post graduate students* students enrolled in the endodontic master’s program at MIU* students were then performed access cavity, the number of canals found will be recorded. Stage 3: CBCT images were uploaded to AI software, and the number of canals detected was be recorded. Data collected will be compared using 3 groups: Group 1: CBCT with co supervisor interpretation the control group Group 2: clinically after performing access cavity. Group 3: CBCT with the AI technology. All cases that failed to be detected by the AI software were then evaluated for more morphological features that may influence the accuracy of AI software.