Method: After institutional review, informed consent and operator training, one standard and two naïve examiners collected replicate tooth color measures from 20 healthy dentate adult volunteers over two days. At each visit, subjects were oriented using a chin rest, and images were collected using a high resolution digital camera, zoom lens and fixed lighting conditions. Using random order assignment, 3 images were collected each day (standard & 2 trainees). Images were captured over two days to assure subject repositioning as part of the reproducibility assessment. For each image, maxillary anterior tooth pixels were counted (pixel number) and subject average L*a*b* tooth colors were derived using standard methods. Intra-class correlations (ICC) and 95% lower confidence bounds (LCB) were calculated on a 0-to-1 scale, where 0 represented no agreement and 1 represented perfect agreement.
Result: The 20 subjects ranged from 21-63 years of age, all subjects had 6 images collected, and all data were included in the analyses. For Day 1, tooth pixel count means amongst examiners differed by less than 1.8%, while tooth color means ranged from 15.91-16.07 for b* and 75.86-76.17 for L* amongst the standard and naïve examiners. Day 2 demonstrated similar inter-examiner reproducibility. Overall, the ICC (95% LCB) ranged from 0.96 (0.93) to 0.99 (0.97) for b* yellow-blue, L* lightness and a* red-green values for all comparisons between examiners.
Conclusion: Use of digital image analysis by day amongst standard and trainee examiners showed high inter-examiner consistency for clinical measurements of tooth area (pixels) and color (L*a*b*) parameters from research conducted in a clinical trials setting.
Keywords: Color and Digital image analysis