RESUMEN
ABSTRACT@#Dental caries is a commonly progressive disease that proceeds through various degrees of severity that a dentist can detect. The aims of the in vivo study were to assess the accuracy of the individual model (near-infrared light transillumination [NILT] device, visual and radiographic examinations) in detecting occlusal caries, and to evaluate the performance of visual and NILT device combination for occlusal caries detection in deciding the treatment options. Fifty-two non-cavitated occlusal surfaces from 16 patients were assessed with three different diagnostic devices in random order. Identified lesions were prepared and validated. Logistic regression analysis was performed for each method. The sensitivity and specificity values for each method and the combined models were statistically measured using RStudio version 0.97.551. At the enamel level, visual detection was the most sensitive method (0.88), while NILT was the most specific (0.93). NILT scored the highest for sensitivity (0.93) at the dentine level and visual detection scored the highest for specificity (0.88). Visual detection + NILT model was significantly better (p = 0.04) compared to visual detection or NILT alone (df = 1). The visual-NILT combination is a superior model in detecting occlusal caries on permanent teeth. The model provided surplus value in caries detection hence improving the treatment decision-making in occlusal surfaces.