Examining the diagnostic accuracy of interproximal chemical caries using X-ray and artificial neural network (ANN) modeling

Abstract

The diagnostic accuracy of interproximal chemical caries and microchemical corrosion is significantly improved through advanced analytical techniquesthat facilitate early detection and characterization of demineralization, enabling timely interventions and effective preventive strategies in dentistry. Radiographs were used to detect proximal caries, and various metrics were measured, including AZ value, specificity, sensitivity, positive predictive value (PPV), negative predictive value (NPV), and positive and negative likelihood ratios (LR+, LR-). The results revealed that the normal size - normal shape group had the highest sensitivity, NPV, and AZ value, while the normal size - square shape group had the highest specificity and PPV. This suggests that different jaw shape options can influence the diagnostic accuracy of interproximal caries. An artificial neural network (ANN) was shown to estimate the PPV, NPV, and +LR based on varying sensitivity and specificity. The results indicated that manipulating sensitivity and specificity impacted the estimated values, with higher specificity leading to increased PPV and positive effects on the LR+. Both sensitivity and specificity contributed to the improvement of the NPV. The prediction errors of the ANN were evaluated using linear regression and exhibited an acceptable level of error compared to empirical test results. However, it is important to note that neither the positive nor the negative likelihood ratio was sufficiently large for all groups, indicating that the jaw shape option alone may not significantly enhance the detection accuracy of interproximal caries. The micromechanical analysis of customizing the jaw shape option demonstrated its potential influence on the detection of interproximal caries While certain jaw shape options showed improved metrics, additional factors beyond the jaw shape option may be necessary to achieve optimal diagnostic accuracy in dental imaging. © 2025, Iranian Chemical Society. All rights reserved.