Application of digital imaging in tooth whitening randomized controlled trials.

PURPOSE

The development of novel peroxide-based bleaching systems during the last several years has prompted the need for robust clinical methods to evaluate whitening response. Advances in digital camera technology and image analysis software provided the basis for an instrumental method to assess tooth color closely following a technique previously used to quantify plaque on tooth surfaces. In vitro and in vivo research was conducted to determine reproducibility of color measurements using this objective, digital imaging method.

METHODS

Each of the 16 tabs in a standard shade guide system was mounted in a jig, and measurement reproducibility was assessed in vitro from paired digital images collected over a 2-day period. Separately, clinical measurement reproducibility was assessed in vivo from paired images of 14 healthy adult volunteers collected over a 2-day period. From these digital images, mean L*, a*, and b* color values were derived for each of the 16 individual shade tabs (in vitro study), or the facial surfaces of the maxillary six anterior teeth (in vivo study) of the 14 subjects. For each data set, variability was determined using ANOVA, and between-visit color measurement reliability was determined from intra-class correlation coefficients (ICCs).

RESULTS

In the in vitro study, shade tab yellowness (b*) ranged from 9.0-18.6, lightness (L*) ranged from 63.4-76.2, and redness (a*) ranged from 0.9-3.6. Overall daily means differed by 0.08 units or less, and intra-class correlations for the image pairs were 0.998 for L*, 0.996 for a* and 0.998 for b*. In the in vivo assessment, the 14 volunteers exhibited considerable range in tooth color. Yellowness (b*) ranged from 13.5-21.3, lightness (L*) ranged from 69.2-78.0, and redness (a*) ranged from 5.2-8.8. Clinical measurement of mean tooth color from digital images was highly reproducible across visits. Intra-class correlations for the image pairs were 0.989 for b*, 0.970 for L* and 0.979 for a*.