Methods: 24 subjects were recruited from a mixed population of low, moderate, and high caries risk patients, categorized using the American Dental Association certified caries risk assessment tools. Both visual examination (3 calibrated caries detection experts) and the investigational device (used by one trained clinician) were used to detect potential incipient lesions on buccal, occlusal, and lingual surfaces of teeth. Ground truth was generated from consensus visual examination results plus image information from white light and FIRE images. 17 subjects satisfied the data analysis criteria. 410 surfaces with potential incipient lesions were used in evaluating performance of detection of incipient caries + hypomineralization combined. Among these, incipient caries were found on 233 surfaces; these surfaces were used to separately evaluate incipient caries detection performance. A total of 277 incipient lesions (Buccal 163, occlussal 55, lingual 59) and 367 hypomineralizations (Buccal 123, occlussal 167, lingual 77) reached visual examination consensus. The performance of Visual examination alone, device detection in Still Image Mode, and device detection in Video Mode, were evaluated against ground truth using these data set. Sensitivity performances of device and visual examination were determined and compared descriptively.
Results:
|
Overall Sensitivity (all surfaces) evaluated against ground truth |
||
Visual examination |
Device detection in Still Image mode |
Device detection in Video mode |
|
Incipient Caries |
0.94 |
0.884 |
0.848 |
Incipient Caries + Hypomineralization |
0.95 |
0.916 |
0.883 |
Conclusions: The present study demonstrated (1) the clinical effectiveness of the investigational device, and (2) that the device is at least as good as expert visual examination in the detection of potential incipient lesions.
Keywords: Caries, Caries detection, Diagnosis and Digital image analysis
See more of: Cariology Research - Detection, Risk Assessment and Others