Color Matching in Dentistry

Nehal Bhandari, Krithi KN

Introduction

Dental Composite Shades

Dentists use composites to fill cavities in teeth, and use porcelain implants to replace teeth. These composites and porcelain fillings are available in several shades, and the dentist has to select the shade that best matches the color of the tooth being replaced or filled. Dentists rely on their visual judgment to make this color matching, and this is done under the lighting conditions of their office. We have learnt in PSYCH 221 that two materials that appear to have same color under a given light, may not match when observed under a different light. So the composite chosen by the dentist may not match the color of the tooth when observed under daylight.

Motivation

The first thing we notice about a person when we meet them is his/her smile. So we believe it is very important to avoid imperfections in tooth fillings/implants. Our goal is to study the characteristics of the reflectance of teeth and composite materials under different artificial illuminants, and compare them to that under daylight. Through this study, we quantify the mismatches between the composite and the color of the tooth when the composite is chosen under an artificial light but observed under daylight. We conduct this study across several people and several lights, to arrive at conclusions on which lights are the closest to daylight when it comes to color matching, and which lights result in bad matches when observed under daylight.

Methodology

This project has three components:

1. Measure the reflectance of teeth in-situ in different individuals and the reflectance of different composite and porcelain materials through spectroradiometric measurements.
2. Second, we use this data and the CIELAB color difference metric to predict whether people will be able to detect the difference between teeth and composite material under different lighting conditions.
3. Third, we analyze the processed data to determine what factors play significant role in getting good matches for different individuals.

Data Collection

Measurement Setup

Using a spectrographic camera, we measured the spectral power distribution of the following:

1. The reflectance of teeth of eleven different people,
2. The reflectance of sixteen different shades of composite materials (A1 to A4, B1 to B4, C1 to C4, and D2 to D4),
3. The light that falls on teeth during each measurement. (Represents the white point under the lighting conditions)

Data Analysis

For each individual, we calculate the CIELAB color difference (delta-E) between his teeth and the different composites available, under daylight and various artificial illuminants, using ISET functions. We then find the best-match composite, i.e. the composite that gives the least delta-E when compared against his teeth. We then repeated this process for all people. So we now know which composites would give the best match for every individual under a given light.

Consider the following notation: $D{E}_{A-B}$. For all further discussion this notation means deltaE computed between person X's teeth and composite Y when observed under Illuminant B such that, Y is best match for X under Illuminant A. So for example, $D{E}_{F1-daylight}$ is deltaE for a composite which is best match in F1 but is being observed under daylight.

Now we know that the dentist performs color matching under an artificial illuminant. We want to quantify the mismatch when this composite chosen under an artificial light is viewed under daylight. To be able to do this analysis, we calculate the following delta-E values under each illuminant L for all subjects:

• $D{E}_{L-L}.$ : Delta-E of the composite picked under L, when viewed under L.
• $D{E}_{F1-daylight}.$ : Delta-E of the composite picked under illuminant F1, when viewed under daylight.
• $D{E}_{daylight-daylight}.$: Delta-E of the composite picked under daylight, when viewed under daylight. Here, the composite may be different from that chosen under L and we will always have $D{E}_{daylight-daylight}\le D{E}_{F1-daylight}.$

File:Error metrics.png

Error Metrics

Another interesting question would be, when a composite that is chosen such that it matches the best under daylight, how off will the teeth and composite look, when viewed under other interior lights. To answer this, we calculated the following deltaE value:

• $D{E}_{daylight-F1}$ : Delta-E of the composite picked under daylight, when viewed under illuminant F1.
• $D{E}_{daylight-daylight}$ : Delta-E of the composite picked under daylight, when viewed under daylight.

Results

Figure 1 shows spectral power distribution for one of the persons in our experiment, under daylight. We also show the SPD of composite materials which were the best match (Composite d3) and worst match (Composite c4) under daylight. The best case deltaE value is 1.3225 and worst case deltaE value is 6.8611.

File:Figure1.png

Figure 1: Best and worst composite SPDs for an individual

Next, we want to see how many times do the best matches in all the studied lights agree with the best match in daylight for each person. In Figure 2, we show percentage match in best-composite between different lights and daylight over all people. Best composite in Fluorescent 8 agrees with composites selected under daylight for all 11 people, and hence the 100% match.

File:Figure 2.png

Figure 2: Percentage match in best-composite prediction with daylight

Figure 3 shows variation in $Percentage\mathrm{\_}Error\mathrm{\_}D{E}_{F1-D}$ across different people, for all artificial lights. This plot basically helps us see that once a match has been made in some given light, how good/bad will it look when viewed under daylight for different people. We see that minimum error for all lights is zero. This is because there is always at least one person, for whom that illuminant gives same best match composite as under daylight. The highest percentage error values are also very similar across different lights, because many lights which differ from daylight in choice of best composite, agree amongst themselves. In fact for all people that we studied, there were never more than two different kind of composites chosen, as best matches across all different lights. Fluorescent 8 gives same composite as daylight for all people as discussed above. Hence zero error and zero variability.

File:Figure3.png

Figure 3: Variance in $Percentage\mathrm{\_}Error\mathrm{\_}D{E}_{Fx-D}$ where Fx are the various fluorescent lights

Figure 4 shows variation in $Percentage\mathrm{\_}Error\mathrm{\_}D{E}_{D-L}$ across different people, where L is the light in which we are viewing. This plot helps us analyze that, assuming we want best matches under daylight, how good/bad will it look when viewed in interiors with light L. Again minimum error for all lights is zero because there is always at least one person, for whom best match in daylight also translates to best match in that light. Here maximum percentage errors are quite different across different lights because deltaE values for same composite will be quite different for different lights. Here we see that a good match under daylight will translate to poor match under interiors with Fluorescent 4 light (both in terms of mean error and worst case).

File:Figure4.png

Figure 4: Variance in $Percentage\mathrm{\_}Error\mathrm{\_}D{E}_{D-Fx}$ where Fx are the various fluorescent lights

Conclusions

Through spectrophotometric measurements, we determine that making a match with Fluorescent 8 as illuminant, gives best results when viewed in daylight. Therefore, of all the artificial lights analyzed, we recommend Fluorescent8. On the other hand if we want the chosen composite to be a good match under daylight, it will not be a good match in interiors with illuminant Fluorescent 4. However, all these results are subject to experimental errors and inaccuracies. Measurements for tooth and composite spectral reflectance and the lighting conditions need to be made very carefully. Differences in angle/distance from light source may lead to erroneous data. In future one can explore more types of composite and porcelain materials under a wiser range of indoor lights.

References

1. Ng, DuYong, et al. "Non-contact imaging colorimeter for human tooth color assessment using a digital camera." Journal of Imaging Science and Technology 47.6 (2003): 531-542[1]
2. CIE Fluorescent series: http://www.cis.rit.edu/research/mcsl2/online/cie.php

Appendix I

Code

Appendix II

The experimental readings for spectrophotometric data were taken together in the lab. Both team members worked together on data processing and evaluation. There is no easy way to demarcate the contributions from each team member. Both have gained equally from this project.