Performance of 5 different displays in the detection of artificial incipient and recurrent caries-like lesions

Abstract

The aim of this study was to assess whether auto-calibrating medical-grade monitors perform better than off-the-shelf monitors and tablet computers in detecting artificial incipient and recurrent caries-like lesions. Sixty extracted teeth (30 premolars and 30 molars) were selected. All molars received class II amalgam and composite restorations. A 7-mm2 area on the crowns of half of the teeth was demineralized. Phantoms consisting of 4 teeth were created. Three observers using a 5-point scale evaluated digital periapical radiographs for the presence of caries on 5 displays: 2 autocalibrating medical-grade monitors, 2 tablets, and 1 off-the-shelf monitor. Sensitivity, specificity, accuracy, and receiver operating curve data were calculated and verified through analysis of variance and Tukey tests. Observer agreements were assessed using Cohen's κ test. Intraobserver agreement ranged from 0.347 to 0.612 (molars) and 0.617 to 0.811 (premolars). Interobserver agreement ranged from 0.239 to 0.559 (molars) and 0.657 to 0.858 (premolars). The performances of tablets and the off-the-shelf monitor were similar to medical monitors when the same tooth groups were compared. Medical-grade monitors presented fewer statistically significant differences when different lesions where compared within the same display and restorative material. Evaluations of similar lesions were not significantly different among the 3 types of displays. However, the autocalibrating medical-grade monitors performed better when incipient and recurrent lesions were compared.