Neural network approach to evaluate the physical properties of dentin.

This study intended to evaluate the effects of inorganic trace elements such as magnesium (Mg), strontium (Sr), and zinc (Zn) on root canal dentin using an Artificial Neural Network (ANN). The authors obtained three hundred extracted human premolars from type II diabetic individuals and divided them into three groups according to the solutions used (Mg, Sr, or Zn). The authors subdivided the specimens for each experimental group into five subgroups according to the duration for which the authors soaked the teeth in the solution: 0 (control group), 1, 2, 5, and 10 min (n = 20). The authors then tested the specimens for root fracture resistance (RFR), surface microhardness (SμH), and tubular density (TD). The authors used the data obtained from half of the specimens in each subgroup (10 specimens) for the training of ANN. The authors then used the trained ANN to evaluate the remaining data. The authors analyzed the data by Kolmogorov-Smirnov, one-way ANOVA, post hoc Tukey, and linear regression analysis (P < 0.05). Treatment with Mg, Sr, and Zn significantly increased the values of RFR and SμH (P < 0.05), and decreased the values of TD in dentin specimens (P < 0.05). The authors did not notice any significant differences between evaluations by manual or ANN methods (P > 0.05). The authors concluded that Mg, Sr, and Zn may improve the RFR and SμH, and decrease the TD of root canal dentin in diabetic individuals. ANN may be used as a reliable method to evaluate the physical properties of dentin.