Comparison of occlusal caries detection using the ICDAS criteria on extracted teeth or their photographs

Background: Using photographs of occlusal surfaces instead of extracted teeth for the detection of caries can be useful in multicenter studies or education. Using a panel of observers, ICDAS scores on teeth or photographs were evaluated against the histological gold standard. The hypothesis was that both outcomes were equivalent. Methods: Four examiners with different experience in ICDAS scored photographs of occlusal surfaces of 100 extracted teeth on a monitor using ICDAS criteria. Two of the examiners had previously scored extracted teeth prior to photography. Digital images of histological sections of the teeth were observed by all examiners and consensus scores were given for each investigation site (gold standard). Kappa statistics and Spearman correlation coefficients as well as repeated measure ANOVA were performed. ROC curves were constructed for each examiner and the areas under the ROC-curves (AUC) of both scoring techniques (extracted teeth, digital images) were compared (alpha = 0.05). Results: Intra- and inter-rater kappa for ICDAS on teeth were 0.81-0.94 and on photographs 0.54-0.88, respectively. Correlation with histology was 0.58-0.61 for the teeth and 0.50-0.62 for the photographs. AUC of ICDAS scores of extracted teeth (mean 0.89) were slightly higher than those for photographs (mean 0.84). However, both AUC values were not statistically significant (p = 0.38). Conclusion: Using photographs to assess occlusal surfaces with the ICDAS criteria was not statistically different from scoring the extracted teeth.