PROJECT SUMMARY Epidemiological studies indicate that cracked teeth are the third most common cause of tooth loss in industrialized countries. Histological studies demonstrate that all cracks are colonized by bacteria, which have the potential to cause intensely painful pulpal and periapical infections. The early detection of cracks (incomplete fractures) followed by appropriate interventions to prevent crack propagation are effective strategies to prevent infections and avert tooth loss. Current tools used to diagnose cracks are inadequate and there is an imperative need to develop an objective and reliable method to detect cracks. During our Phase I project, we developed and tested a novel algorithm for crack detection on extracted human teeth. Using machine learning and imaging features extracted from three-dimensional (3D) wavelets, we demonstrated enhanced crack detection hr-CBCT. We now propose to further refine this technology and to validate it clinically. Our hypothesis is that our method increases the predictive validity of hr-CBCT in detecting cracks. This development will happen with close clinical guidance. Also, we will collaborate with CBCT hardware vendors to increase the impact of our commercialization plan. This proposal addresses the need for quantitative, reproducible, and evidence-based ways to detect cracks in teeth, that can potentially lead to improved tooth loss prevention.