DIFOTI (Digital Fiberoptic Transillumination)

This scientific work aims to determine the sensitivity and
specificity of the DIFOTI system and to assess its diagnostic
performance in detecting initial caries. Forty five sound teeth
were randomly assigned to two groups. One group consisted of
unaltered teeth (N=18). In the second group, an artificial carious
lesion was created (Damato et al., 1988) on the buccal (N=9),
proximal (N=9) and occlusal surface (N=9). To evaluate proximal
caries, a proximal contact was simulated. Images of teeth were
taken with DIFOTI. In addition to the DIFOTI standard software, the
images were analyzed with View3D, a special software, performing an
evaluation comparable to the one described by De Josselin de Jong
et al., 1995 to quantify light scattering. The specimens were then
sectioned, revealed an average depth of 62.39µm of the carious
lesions and used as gold standard for the subsequent evaluation.
Based on visual observation of the digitized images, sensitivity
and specificity values were: 1.00 and 0.88 for smooth surfaces,
0.82 and 1 for occlusal surfaces, and 0.44 and 0.83 for approximal
surfaces, respectively. Calculation of area under the ROC curve
utilizing light penetration measurements of the device revealed
values 0.69 corresponding to the 1% quantile gray level of DIFOTI
image, and 0.59 for the median. DIFOTI has shown excellent
sensitivity and specificity for buccal and occlusal superficial
artificial lesions. However, it has poor ability to detect
incipient proximal caries. Nevertheless, it still has good
specificity proximally. With the currently used software, View3D,
the gray-value evaluation fails to serve as an objective measure
for caries diagnosis and requires further investigation.