Validation of Intraoral Scanner for Caries Diagnosis in Epidemiological Studies

Objectives: The objective of this study was to validate the use of 3D images obtained through an intraoral scanner (IOS) for caries diagnosis in a population-based cohort study.
Methods: A random sample of individuals from the Pelotas 1982 Birth Cohort was assessed at 40 years of age. Calibrated dentists clinically examined participants using merged ICDAS criteria for caries lesion detection at the tooth surface level: 0 – sound; A – initial lesion; B – moderate lesion; C – severe lesion. 3D images were acquired using an IOS (TRIOS 3 – 3Shape®). A calibrated evaluator assessed dental caries through 3D images using the same ICDAS criteria. Diagnostic properties of using IOS 3D images to detect caries were calculated, considering the clinical examination as the gold standard, with different cut-off points based on lesion stage at individual, tooth, and surface levels.
Results: A total of 99 individuals, 2,664 teeth, and 11,519 surfaces were assessed. At the individual level, sensitivity values for initial, moderate, and severe lesions were 86.1%, 77.8%, and 80.6%, respectively. Specificity was 40% for initial, 55.6% for moderate, and 93.7% for severe lesions. At the tooth level, sensitivity was 38.6% for initial, 56.8% for moderate, and 72.5% for severe lesions, with specificity values ranging from 91.5% to 99.4%. At the surface level, sensitivity values were 50.2% for initial, 71.7% for moderate, and 82% for severe lesions. Specificity values at the surface level were all higher than 97%. Areas under the ROC curve varied from 0.63 to 0.91 considering all levels and cut-off points.
Conclusions: Diagnostic properties improved as the severity of lesions increased for all analysis levels. The results demonstrated that 3D images obtained through an intraoral scanner are valid for assessing dental caries in an epidemiologic setting, with excellent properties for moderate to severe lesions.