Effect of exposure time on in vitro caries diagnosis using the Digora system

The aim of this study was to determine the in vitro diagnostic performance for approximal caries and occlusal dentine caries diagnosis of the Digora system for digital radiography in relation to the exposure time. Radiographs were made of 220 extracted human posterior teeth and cuspids at exposure time levels of 3%, 6%, 10%, 20%, and 78% of the exposure time needed for E-speed film (image sets I, II, III, IV, and V, respectively). The image sets were put in random order (III, I, IV, II, V) and assessed independently by 3 observers using a 5-point confidence scale. True caries status was determined histologically after sectioning of the teeth. Evaluated parameters were: interexaminer kappa, sensitivity, specificity, Dz3/4 (the distance of one operating point to the diagonal in the ROC domain) and area under the ROC curve. The Spearman correlation coefficients between the parameters and both exposure time and order of viewing of the sets were calculated. For approximal surfaces, the ranges of mean parameter values were: sensitivity 26-33%, specificity 93-95%, Dz3/4 0.63-0.84, area 0.61-0.69. For occlusal surfaces, the ranges were: sensitivity 52-60%, specificity 91-95%, Dz3/4 1.03-1.34, area 0.79-0.87. Out of 24 pairwise comparisons between sets V and I, 5 were significant. Out of 24 comparisons between set V and II, only 2 were significant. It was concluded that diagnostic performance is unlikely to be impaired for an exposure time as short as 6% of E-speed film exposure. When reducing the exposure time to 3% of E-speed film exposure, caries diagnosis may be impaired.