Optimizing Failure Mode Analysis of Dental Restorative Materials: Balancing Efficiency and Accuracy

To quantify the adhesion of ceramics to dental hard tissues, macroscopic shear bond strength (SBS) tests are commonly used. A failure mode analysis is then performed to determine the location of the failure and its correlation with the corresponding SBS values. Most analyses in the literature use an optical microscope. In this work, we introduce a new approach to perform objective and efficient failure mode analysis using a weakly supervised deep neural network. Our experimental results demonstrate that this approach significantly outperforms human analysis.