Purpose: To assess the influence of liquid attached on the tooth surfaces on the accuracy (trueness and precision) of intraoral scanners and the effectiveness of the drying method (using compression air) to exclude the influence of liquid on the scanning results. Materials and methods: A mandibular jaw model was scanned using an industrial computed tomography scanner to obtain a reference model. A scanning platform was designed to simulate three specific tooth surface states (dry, wet, blow‐dry). Two kinds of liquids (ultra‐pure water and artificial saliva) were used for the test. Two intraoral scanners (Trios 3 and Primescan) were used to scan the mandibular jaw model 10 times under each condition. All scanning data were processed and analyzed using dedicated software (Geomagic Control 2015). Trueness and precision comparison were conducted within the 12 groups of 3D models divided based on different intraoral scanners and liquids used under each condition. The root mean square (RMS) value was used to indicate the difference between the aligned virtual models. The color maps were used to evaluate and observe the deviation distribution patterns. The 3‐way ANOVA (condition, intraoral scanner, liquid) followed by the Tukey test were used to assess precision and trueness. The level of significance was set at 0.05. Results: The mean RMS values obtained from wet condition were significantly higher than those of the dry and blow‐dry condition (p < 0.001, F = 64.033 for trueness and F = 54.866 for precision), which indicates less accurate trueness and precision for wet condition. For two different types of liquids, the mean RMS value was not significantly different on trueness and precision. The deviations caused by liquid were positive and mainly distributed in the pits and fissures of the occlusal surface of posterior teeth, the interproximal area of the teeth, and the margin of the abutments. Conclusions: Liquid on the tooth surface could affect intraoral scanning accuracy. Blow‐drying with a three‐way syringe can reduce scanning errors.

Artificial intelligence (AI) can predict an output from unknown input data by pre-learning the relationship between two known datasets. Deep learning is a type of machine learning approach and is J Prosthodont Res. 2022; **(**):

This study aimed to propose an evaluation method for testing the mechanical strength of film-formed self-adhesive resin cements (SARCs) while reflecting cement layer thickness. Three commercially available dual-cure type SARCs were used for tensile and shear tests using specimens with varying thicknesses (0.05, 0.2, and 0.4 mm). There were no significant differences in tensile strengths among the various specimen thicknesses. In the shear test, there was a significant decrease in the strength with a reduction in specimen thickness. Stress distribution and fracture patterns were analyzed using in silico nonlinear dynamic finite element analysis. Finite element analysis demonstrated that stress distribution on the specimen surface was homogeneous even with different thicknesses in the tensile test, whereas it was inhomogeneous and induced different fracture patterns on the 0.05-mm-thick specimen in the shear test. These results suggest that the tensile test is useful for testing the mechanical strength of film-formed SARCs.

No abstract

No abstract