Detection of caries with local CT

An in vitro observer study of the detection of interproximal caries by local CT.

Detection of caries by means of conventional two-dimensional (2D) digital radiography and by two CT modalities was compared. Ground truth was obtained from histological examination of the sectioned teeth. Twenty-three extracted teeth (46 surfaces) were placed in groups of six in two dry mandibles. The observers (n=10) scored the proximal surfaces for the presence of caries on a 1-5 confidence scale. Data analysis used analysis of variance (a General Linear Model). Observer and method were entered into the model as within-subject variables and lesion depth was entered as a between-subjects variable.

The analysis showed observer, method and lesion depth effects as well as several interaction effects to be significant. Observers performed significantly better with the vertically reformatted CT slices than with conventional radiographs (P=0.025). Furthermore, there were significant differences between observers, and several interactions were found to be significant. This means that although the reformatted slices performed best overall, this performance differed significantly depending on observer and on lesion depth.

Vertically reformatted CT slices obtained with local CT performed significantly better than conventional 2D digital radiographs in visually detecting caries. Axial slices did not perform better than conventional radiographs. When vertically reformatted slices are used, local CT is a promising tool for the detection of interproximal caries.