AI vs. Human: The Accuracy of Deep Learning in Detecting MB2 Canals

Abstract

Complex root canal anatomy and limitations of diagnostic tools and clinicians’ ability to accurately detect the MB2 canal in maxillary molars make successful root canal treatment challenging. Artificial intelligence (AI) has been used in various areas of dentistry and represents a potential tool to aid in detecting MB2 canals. This review aimed to evaluate the accuracy of AI in detecting the MB2 canal in maxillary molars and compare it with human performance. A literature review was conducted using PubMed and Google Scholar databases for studies published between 2010 and 2026, retrieving articles investigating AI-based detection of MB2 canals using CBCT imaging. AI models demonstrated variable sensitivity (53.8%–100%) with overall good diagnostic performance (AUC 0.80–0.90). AI models, image quality, and resolution had a significant impact on AI diagnostic accuracy. AI shows promising potential in enhancing MB2 canal detection; however, it should be used as an adjunct to support, not replace, clinical expertise.