Detection of Proximal Caries in Conventional and Digital Radiographs

Objectives: The aim of this study was to compare the digital images to the conventional radiograph in the proximal caries diagnosis.

Methods: Fifty-one molars and 24 bicuspids were studied and divided into groups with three teeth. They were fixed in silicone and radiographed using the film InSight (Kodak) and the digital systems Digora (Soredex), DenOptix (Gendex) and CygnusRay MPS (Progeny). Twenty-five radiographs were obtained in each modality and four proximal faces were analysed in each radiograph. The images were interpreted individually by one observer at three different points in each imaging modality. The caries lesions were classified according to their depth: (0) absent; (1) restricted to enamel; (2) reaching the enamel dentinary junction; and (3) reaching the dentin. The teeth were sectioned and ground in order to obtain the gold standard and were examined histologically through stereoscopic microscopy.

Results: The Kendall test was employed to verify the intra observer agreement. A good agreement was observed (0.831) among the three evaluations implemented. The accuracy, the sensitivity, the specificity, the positive predictive value (PPV), the negative predictive value (PPN) and the Kappa range were all calculated and there were no significant differences observed among the four studied imaging modalities (ANOVA, p=0.05). The radiographic methods were also compared using the non-parametric Friedman test followed by multiple comparison tests, and no significant differences were observed. Besides, the ROC curve was calculated to each method and no statistically significant differences were observed when comparing the areas under the ROC curve at the significance level of 5%.

Conclusion: The results suggest that the diagnostic accuracy of the digital images and the conventional radiographs used to detect proximal caries is similar.