Received: 11 June 2019 Revised: 1 November 2019 Accepted: 11 November 2019

DOI: 10.1111/jerd.12553

RESEARCH ARTICLE

Evaluating the relationship between tooth color and enamel

thickness, using twin flash photography, cross-polarization
photography, and spectrophotometer

Wei-Hung He DDS., MS1 | Cheryl J. Park DDS., FACP2 | Sangho Byun DDS., FACP2 |
Daniel Tan BDS3 | Chih Y. Lin DDS4 | Winston Chee DDS, FACP5
1
Department of Dentistry, Division of
Prosthodontics and Operative Dentistry, Abstract
Chang Gung Memorial Hospital Kaohsiung Objectives: To understand (a) the effects of labial enamel on tooth color
Branch, Kaohsiung City, Taiwan, ROC
2 (b) relationship of color data taken by nonpolarized (NP), cross-polarization photogra-
Division of Restorative Sciences, Herman
Ostrow School of Dentistry, University of phy (CP), and spectrophotometry (SP).
Southern California, Los Angeles, California
Materials and methods: Fifty extracted human maxillary incisors were coated with
3
Advanced Prosthodontics Program, Herman
Ostrow School of Dentistry, University of resin on their palatal surfaces. Their color was measured with NP, CP, and SP and
Southern California, Los Angeles, California their dimensions were scanned by an intraoral scanner. The labial enamel was
4
Division of Endodontics, Department of
removed using a modified selective enamel demineralization technique. Tooth dimen-
Dentistry, Kaohsiung Chang Gung Memorial
Hospital, Kaohsiung City, Taiwan, ROC sions and color were recorded again. The differences in the labial enamel thickness
5
Division of Restorative Sciences, Herman (ΔT) and color (ΔE*00 and ΔL*, Δa*, Δb*) were statistically analyzed with the Pear-
Ostrow School of Dentistry, University of
son correlation coefficient and simple linear regression.
Southern California, Los Angeles, California,
Ralph and Jean Bleak Professor of Restorative Results: In CP and SP methods, ΔT and ΔE*00 were weakly to moderately positively
Dentistry, Director Advanced Prosthodontics
correlated (r = .38 and .27). In NP, CP, and SP methods, ΔT and Δb* are weakly posi-
Correspondence tively correlated (r = .27, .27 and .29). The color data of three measuring methods
Winston Chee, Ralph & Jean Bleak Professor
were highly positively correlated (r > .8). A linear relationship between ΔE*00 and ΔT
of Restorative Dentistry, Herman Ostrow
School of Dentistry, University of Southern were found (CP and SP groups).
California, Rm 4374, 925 W 34th St, Los
Conclusions: (a) Thicker labial enamel has a greater impact on tooth color.
Angeles, CA 90089.
Email: wchee@usc.edu (b) Reducing labial enamel thickness shifts the tooth color toward yellow. (c) Tooth
color measured from the three methods were highly correlated.
Clinical Significance: Knowing the relationship between enamel thickness and tooth
color, a clinician can better predict the stump shade before tooth preparation. Due to
the highly correlated measuring outcomes, it is reasonable to combine these three
methods during shade matching.

KEYWORDS
cross-polarization, enamel thickness, intraoral scanner, shade matching, spectrophotometer,
tooth color

1 | I N T RO D UC T I O N studies that describe and measure tooth color do not differentiate

1.1 | Tooth color
Coronal tooth structure is mainly composed of dentin and enamel,
and both components contribute to the overall tooth color. Many
between these two components. Instead, these investigations con-
sider tooth color as a whole.1-4
The color of the dentin layer dominates the fundamental tone of
the tooth, while the enamel plays a role in the refraction and

J Esthet Restor Dent. 2020;32:91–101. wileyonlinelibrary.com/journal/jerd © 2019 Wiley Periodicals, Inc. 91

92
scattering of light.5-7 Coops and ten Bosch5 compared the color of
intact teeth and the color of the same teeth after removing the labial
enamel and found that the colors were strongly correlated. Their
study confirmed that tooth color is determined mainly by the dentin,
while the enamel plays a minor role through scattering wavelengths in
the blue range. Based on previous findings, there is still a need to
understand how different enamel thicknesses could affect tooth color.
Therefore, this study was designed to investigate the relationship of
the enamel thickness to the overall tooth color.
1.2 | Digital methods for clinically evaluating tooth
color
Digital shade-matching techniques have become essential methods
for clinicians. For example, digital photography and spectrophotome-
try (SP) are both simple and readily available methods. To ensure clini-
cal applicability, this study adopted these methods for measuring
tooth color.
Digital, nonpolarized flash photography (NP) has increasingly
become an integral part of clinical records. Using digital images for
shade matching has also been proven to be feasible and reproducible
in both in vitro and in vivo experiments.8-11 Some studies claim that
color matching achieved through digital photography is more precise
than visual shade matching with shade guides.11
Cross-polarization photography (CP) is a technique that can remove
specular highlights by aligning two sets of polarizers perpendicular to
each other. One polarizer is attached to the light source, and the other
polarizer is in front of the camera lens. CP is a useful technique that
removes the glare from the tooth surface and has the potential to aid
in shade matching and color analysis in dentistry.12-16
Spectrophotometry (SP), due to its accuracy and reliability, is
thought to be a clinical standard for shade-matching.17 Using a spec-
trophotometer in combination with a shade guide can provide better
shade-matching results than human vision alone with a shade guide.18
When comparing these three color analysis methods, the investi-
gator needs to be aware that the color measurement outcomes under
different illuminations are not compatible with each other.19 First,
these three methods mentioned utilize different light sources (a direct
flashlight, a polarized flashlight, and a red-green-blue [RGB] light-
emitting diode.) Furthermore, the environmental lighting conditions,
calibration targets, and color sensors in the digital devices are all dif-
ferent. A previous study has shown that spectrophotometers are not
calibrated to each other.20
It has been suggested that the different methods can be combined
for better shade-matching outcomes.18 If we can confirm that the
measurement data of these three methods are highly positively corre-
lated, this correlation will help us to estimate the actual color of the
tooth.
For example, if the target tooth is brighter in value and slightly
more reddish in hue than the closest shade tab, all three measurement
methods should record the same trend.
HE ET AL.
1.3 | Objective
This study attempts to measure the effect of altering the labial enamel
thickness on tooth color by utilizing three different color measure-
ment methods (NP, CP, and SP). The data from these three measuring
methods will also be compared to understand how well they correlate
with each other.
1.4 | Hypotheses
The null hypotheses are as follows.
1. Tooth color is not affected by the difference in the labial enamel
thickness.
2. Tooth color differences are not positively correlated with differ-
ences in labial enamel thickness.
2 | MATERIALS AND METHODS
There are four major steps in this study (Figure 1):
Sample preparation: Tooth sample mounting and modified selective
enamel demineralization were carried out.
Data collection: Three-dimensional (3D) scanning and three differ-
ent methods were utilized for the color measurements.
Data processing: The differences in the enamel thickness (ΔT) and
color (ΔE) were calculated.
Statistical analysis: The Pearson correlation coefficient and a sim-
ple linear regression were used in the analysis.
2.1 | Tooth sample preparation
1. Fifty extracted intact human maxillary incisors were obtained
and stored in water for 3 days prior to any experimental
procedures.
2. Surface stains were removed with silicon finishing tips (One Gloss,
Shofu Inc., Kyoto, Japan).
3. The palatal surfaces were coated with a thin layer of filled dental
adhesive resin (Optibond FL, Kerr Corporation, Orange, California)
to protect them during the modified selective enamel deminerali-
zation procedure. The intact palatal surfaces served as a reference
for subsequent dimensional measurements.
4. Tooth samples were mounted in custom-designed mounting
stands (Figure 2), which enabled measurements to be made in the
same 3D position.
5. A light-blocking trap (64 × 32 × 42 mm) was set up behind the
samples to provide a black background.
6. A ColorChecker extra mini chart (X-Rite Inc., Grand Rapids, Michi-
gan) was used as a reference target for exposure, white balance,
and camera profile calibration. (Figure 2).
7. Preoperative measurements were carried out, including color mea-
surements and 3D scanning.
HE ET AL.
F I G U R E 1 This flowchart shows the main procedures. 50 specimens were photo recorded and 3D shape scanned, and then proceeded acid-
dissection. After acid-dissection (48 specimens left), specimens were photo recorded and 3D shape scanned again. Color and shape measuring
data were aligned and measured, followed by statistical calculations
2.2 | Modified selective enamel demineralization
A technique to remove the enamel and preserve the dentin was modi-
fied from Bazos and Magne's method.21
1. The palatal resin-coated specimens were submerged in 10% HCl
for 15 minutes to remove the labial enamel layer.
2. Excess acid was then removed from the samples by washing under
flowing water for 30 minutes. After washing, samples were stored
in water.
3. The samples were placed back onto the mounting stand, and post-
operative measurements were carried out, including color mea-
surements with the three methods described above and 3D
scanning.
2.3 | Color testing method NP: Regular twin flash
photography, no cross polarization
1. A custom-made dark box (25 × 32 × 24 cm) including a fixed
mounting seat for the tooth samples, a camera adaptor, and two
93
twin flash cold shoes was made (Figure 3). This equipment allowed
for consistent repositioning of the samples at the same distance
on the camera's optical axis.
2. A full-frame camera (D810, Nikon Corporation, Tokyo, Japan) with
a 105 mm macro-lens (105 mm f/2.8G ED-IF AF-S VR Micro
NIKKOR Lens, Nikon Corporation, Tokyo, Japan) and a twin flash
(Twin Lite MT-24EX Speedlite Flash, Canon Inc., Tokyo, Japan)
was used to capture the digital photographs. An external flash bat-
tery pack (PROPAC PB960 Lithium-Ion Flash Power Pack,
GODOX Photo Equipment Co., Ltd., Shenzhen, China) was used
ensure a consistent twin flash output.
3. Camera settings. The camera was set in manual mode (ISO
100, shutter speed 1/100, aperture f13, flash output 1/8). The
optical viewfinder was closed to prevent light leakage, and all
images were captured in RAW format.
4. Image recording. Samples were photographed before and after
modified selective enamel demineralization. To avoid color
changes due to dehydration, photographs were taken within
20 seconds of removing the samples from water.
94 HE ET AL.

5. Exposure calibration. All image files were processed with Adobe
Lightroom CC 2015 (Adobe Systems Inc., San Jose, California) and
were calibrated in grayscale mode to obtain the same exact
exposure.
6. Color and white balance calibration. Camera color profiles and
white balance were calibrated using the X-Rite ColorChecker extra
mini Chart and Adobe Lightroom CC 2015 with the X-rite Col-
orChecker Passport plugin (Figure 2).
7. Color space and measurements. All color measurements were
recorded in International Commission on Illumination color space
(CIELAB) values in a ProPhoto RGB color space with Adobe
Photoshop CC 2017 (Adobe Systems Inc.). The white point was
set as D65.
2.4 | Color testing method CP: Cross polarization
1. Cross-polarized filters were attached in front of the camera lens
and twin flashes (Figures 3 and 4).
2. The camera was set in manual mode (ISO 100, shutter
speed 1/100, aperture f11, flash output 1/1). All other
steps were exactly the same as those described for the NP
method.

2.5 | Color testing method SP: Spectrophotometry

1. A dental spectrophotometer ShadepilotTM (DeguDent, Hanau,
Germany) was used for the color measurements (spectral data:
light output from ~410 to 680 nm, image output: calibrated data
from 400 to 720 nm, steps of 10 nm; Figure 5).

F I G U R E 2 A tooth specimen (T) was fixed on the mounting stand

(S), with a 24-colored ColorChecker chart (C) at the same sagittal
level. Behind the tooth specimen, the light-blocking trap (BT) blocked F I G U R E 4 A sets of crossed polarized filters (CP1 and CP2) was
light from all directions and provided a background (B) close to pure used while taking cross-polarization photos, attached in front of the
black camera lens (CP1), and the twin flash (CP2), respectively

F I G U R E 3 A, Photography
set-up. A custom-made dark box
(BO) with fixed adaptors for a
camera (CA) and a twinflash (F). A
flash battery pack (P) was used
for providing stable power
output. A set of crossed polarized
filters (CP1 and CP2) was used
while taking cross-polarization
photos, attached in front of the
camera lens (CP1) and the twin
flash (CP2), respectively. B, The
actual photography set-up. The
cover on top of the box was
removed for showing the internal
structure
HE ET AL. 95

2. A custom stand (Figure 5) was used to maintain the position of
the spectrophotometer in relation to the tooth specimen. The stand
also provided shading to eliminate the influence of ambient light.
3. To avoid color changes due to dehydration, color measurements
were made within 20 seconds of removing the samples from the
water.
4. Measurements were taken before and after modified selective
enamel demineralization (Figure 6).

2.6 | Digital scanning

An intraoral scanner, TRIOS 3 (3Shape Trios A/S, Copenhagen,
Denmark), was used for the 3D scanning of all the samples. The
mounting stand served as an index for the orientation of all the
3D files.

2.7 | Defining and aligning the measuring points

F I G U R E 5 'A dental spectrophotometer (S) was used for color
measurement. A custom stand (ST) was used to maintain the position
of the spectrophotometer and tooth specimen. The stand also provided
shading to eliminate the influence of ambient light. The tooth specimen
was attached on a mounting holder (H) which can help to position the
specimen inside the spectrophotometer shading stand
1. The length of the tooth was defined as the distance from the
cementoenamel junction to the incisal edge (Figure 7). Ten percent
of the incisal section was subtracted from the full tooth length,
and the sample was subsequently divided equally into the incisal,
middle, and cervical sections.
2. Three measuring points were defined on the tooth samples. These
points were fixed at the intersections of the vertical midline and

F I G U R E 6 A, The noncrossed-polarized (NP) images from one of the testing specimen before and after acid-dissection. The little squares
indicate the color measuring areas. Several specular reflections on the line angle areas can be noticed. B, The crossed-polarized (CP) images from
the same testing specimen before and after acid-dissection. There is almost no specular reflection. The overall color is slightly different from the
images of NP method even though exposure and camera profile were calibrated. C, Spectrophotometer Images from the same testing specimen
before and after acid-dissection. The overall color is also slightly different from the images of NP and CP method
96
the midpoint of each horizontal section, as previously described
(ie, incisal, middle, and cervical).
3. Screen-overlapping software (Ghost-it! 1.04, Matthew T. Pandina)
was used to locate the measurement points via the different soft-
ware packages for the digital photographs, 3D files, and spectro-
photometer images.
2.8 | 3D digital comparison
1. The front surface of the mounting stand was used to orient the
3D files so that the labial surface of the tooth sample remained
perpendicular to the camera's optical axis (Figures 6, 8, and 9).
2. The resin-coated palatal surfaces were used as references for
superimposing the 3D files obtained before and after modified
selective enamel demineralization.
F I G U R E 7 Three measuring points (I, M, C) are defined,
overlapping onto the sample photo for color measuring. (L, full crown
length; L0 , 10% of L; mid, midline)
F I G U R E 8 Three dimensional (3D) digital comparison. Superimposed two 3D files (original and after acid-dissection) the thickness of
removed structure (ΔT) was measured on those defined color measuring points
HE ET AL.
3. The dimensional differences obtained before and after modified
selective enamel demineralization were measured in the direction
of the camera's optical axis using a 3D engineering software
(Geomagic Control, 3D Systems, Rock Hill, South Carolina). This
allowed the change in enamel thickness to be quantified.
2.9 | Statistics
The symbols represent the measuring outcomes and the units of the
measurement.
1. All colors are measured and recorded in CIELAB values, and the
superscripts represent the different measuring methods, namely,
NP, CP, and SP. For example, the CIELAB values measured with
CP are L*CP, a*CP, and b*CP (Table 1).
2. ΔE represents the color difference after modified selective enamel
demineralization. The CIEDE2000 system was used as a standard
to represent the color difference, recorded as ΔE*00. In the follow-
ing data, the superscripts also denote the different measuring
methods. For example, the color differences measured by the
spectrophotometer and calculated with the CIEDE2000 system
were recorded as ΔE*00SP. For a better comparison with previous
studies, the color differences according to the older CIELAB ΔE*
definition are also attached and are recorded as ΔE*ab (Table 1).
3. ΔT represents the dimensional change measured in the direction
of the camera's optical axis after modified selective enamel demin-
eralization (recorded in millimeters).
2.10 | Statistical methods
1. The Pearson correlation coefficient (r) among the color measure-
ments (CIELAB values) of three different methods (NP, CP, and
SP) were calculated. ΔE*00 and ΔE*ab represent the acceptable
color differences.
HE ET AL. 97

2. The Pearson correlation coefficient (r) and a simple regression positive correlation, with correlation coefficients of 0.38 and 0.27,
analysis were used to calculate the differences in the enamel respectively. There was no significant correlation between the
thickness (ΔT) and color (ΔE*ab and ΔE*00). changes in the enamel thickness and color of the NP measurements
3. The Pearson correlation coefficient (r) was also used to calculate of ΔE*00NP. (Table 2).
the differences in the enamel thickness (ΔT) and color in ΔL*, Δa*, In all three testing methods, ΔT and the color differences present
and Δb*. in Δb* showed a weak positive correlation, with correlation coeffi-
4. All the statistical calculations were conducted by JMP Pro 13 soft- cients of 0.27 (NP), 0.27 (CP), and 0.29 (SM).
ware (SAS Institute, Cary, North Carolina). There was no significant correlation between the enamel thick-
ness and color differences present in ΔL* and Δa* in all three
methods (Table 3).
3 | RESULTS
T A B L E 2 Pearson correlation coefficient (r) between the enamel
During the process, two samples were damaged and removed from thickness (ΔT) and color difference (ΔE)
this study. All color differences were calculated in the CIEDE2000
Pearson
system as a standard. To allow a comparison with previous studies, correlation
the data calculated in CIELAB ΔE* are also provided. coefficient (r) P value Correlation
ΔT and ΔE*00NP .08 .3475 Very weak
ΔT and ΔE*00CP .38 <.0001* Moderate
3.1 | Correlation coefficients among all
ΔT and ΔE*00SP .27 .0011* Weak
measurements
ΔT and ΔE*ab NP
.06 .4880 Very weak
In the CP and SP methods, the enamel thickness (ΔT) changes and the ΔT and ΔE*ab CP .31 .0002* Weak
color differences ΔE*00CP and ΔE*00SP showed a weak-to-moderate ΔT and ΔE*ab SP .25 .0027* Weak

*P < .05. Values with a superscript asterisk indicates a statistically

significance.

T A B L E 3 Pearson correlation coefficient (r) between the enamel

thickness (ΔT) and color difference (ΔL, ΔA, ΔB)

Pearson
correlation
coefficient (r) P value Correlation
ΔT and ΔL*NP −.06 0.4787 Very weak
ΔT and ΔL* CP
.08 0.3376 Very weak
ΔT and ΔL*SP −.07 0.3882 Very weak
ΔT and Δa* NP
.16 0.0589 Very weak
ΔT and Δa*CP .07 0.3946 Very weak
ΔT and Δa* SP
.13 0.1161 Very weak
ΔT and Δb*NP .27 0.0012* Weak
ΔT and Δb* CP
.27 0.0011* Weak
F I G U R E 9 The palatal surface (P) was used for overlapping and
aligning the three dimensional files before and after acid-dissection. ΔT and Δb*SP .29 0.0004* Weak
The arrow shows the measuring direction, equal to the camera
*P < .05. Values with a superscript asterisk indicates a statistically
optical axis
significance.

TABLE 1 The code for the measuring data

Color measurements in Color differences in ΔL*, Color differences in Color differences in

Measuring method CIELAB Δa*, Δb* CIELAB ΔE* CIEDE2000

Nonpolarized twin flash L*NP, a*NP, b*NP ΔL*NP, Δa*NP, Δb*NP ΔE*ab NP ΔE*00NP
photography
Cross-polarization L*CP, a*CP, b*CP ΔL*CP, Δa*CP, Δb*CP ΔE*ab CP ΔE*00CP
photography
Spectrophotometry L*SP, a*SP, b*SP ΔL*SP, Δa*SP, Δb*SP ΔE*ab SP ΔE*00SP
98 HE ET AL.

T A B L E 4 Pearson correlation
CIELAB Measuring methods Pearson correlation coefficient (r) P value Correlation
coefficient (r) between the color
L* NP and CP .93 <.0001* Very strong measuring values (CIELAB) from three
NP and SP .93 <.0001* Very strong different testing methods (NP, CP, SP)
CP and SP .88 <.0001* Very strong
a* NP and CP .93 <.0001* Very strong
NP and SP .92 <.0001* Very strong
CP and SP .95 <.0001* Very strong
b* NP and CP .93 <.0001* Very strong
NP and SP .82 <.0001* Very strong
CP and SP .84 <.0001* Very strong

*P < .05. Values with a superscript asterisk indicates a statistically significance.

F I G U R E 1 0 A, The simple regression analysis shows that ΔE00CP is positively related to ΔT with an intercept (arrow), for example,
ΔE00CP = 1.2 + 2.4 ΔT (mm). B, The simple regression analysis shows that ΔE00SP is positively related to ΔT with an intercept (arrow), for example,
ΔE00SP = 1.7 + 1.7 ΔT (mm)

F I G U R E 1 1 A, The simple regression analysis shows that Δb*NP is positively related to ΔT, Δb*NP = −3.0 + 4.9 ΔT (mm). B, The simple
regression analysis shows that Δb*CP is positively related to ΔT, Δb*CP = −0.8 + 4.0 ΔT (mm). C, The simple regression analysis shows that Δb*SP
is positively related to ΔT, Δb*SP = 0.4 + 4.1 ΔT (mm)

The correlation coefficients among the three measuring methods The simple regression analysis shows the linear relationships
are higher than 0.8, showing that the color measuring outcomes of all between the differences in the enamel thicknesses (ΔT) and color:
three methods are strongly correlated (Table 4). ΔE*00CP = 1.2 + 2.4ΔT (mm) (Figure 10a).
HE ET AL.
ΔE*00SP = 1.7 + 1.7ΔT (mm) (Figure 10b).
ΔE*abCP = 2.1 + 3.1ΔT (mm).
ΔE*abSP = 2.7 + 2.7ΔT (mm).
Δb*NP = −3.0 + 4.9ΔT (mm) (Figure 11a).
Δb*CP = −0.8 + 4.0ΔT (mm) (Figure 11b).
Δb*SP = 0.4 + 4.1ΔT (mm) (Figure 11c).
The statistical results show that the color differences (ΔE*ab and
ΔE*00 and Δb*) are weakly to moderately correlated with the differ-
ent enamel thicknesses (ΔT). Linear relationships are found between
ΔE and ΔT (the CP and SP methods). Additionally, linear relationships
are found between Δb* and ΔT. This result means that changing
enamel thickness leads to a color change. For example, reducing the
enamel thickness makes the tooth color appear more yellow. Hence,
the first null hypothesis is rejected.
The statistical results also show that the color data of the three
measuring methods are highly positively correlated, indicating that all
three methods measure color in a very similar manner.
4 | DISCUSSION
4.1 | Calibration
Since the color information recorded by digital devices is device
dependent, proper calibration of all digital equipment is necessary
before measurements are made.22
The purpose of the camera calibration is to ensure the color accu-
racy and consistency of the pictures taken. This procedure includes
calibrating the camera color profile, white balance, and exposure. In
this study, the camera color profile calibration was performed with a
ColorChecker chart. The white balance and exposure calibrations
were done with the gray card portion of the same chart.23,24
The same calibration procedure is also clinically advisable. If the
camera is not correctly calibrated, the technician can only estimate
the white balance and exposure from the shade-matching pictures
based on personal experience.
This study utilized the CIEDE2000 system for calculating color dif-
ferences, which provides better indicators of the human perceptibility
and the acceptability of the color differences than does
CIELAB ΔE*.25
4.2 | Selective removal of labial enamel
To accurately observe how labial enamel affects tooth color, the pala-
tal enamel needs to be preserved. Coops and ten Bosch grounded 5
away enamel with a special tool to maintain the parallelism with the
anatomical surface to assess the effect of enamel on tooth color.
However, it is technically challenging to remove enamel without dam-
aging the dentin structure.
Selective enamel demineralization was used by Bazos and
21
Magne for observing enamel and dentin patterns. Their technique
uses acid treatment to selectively remove the enamel and maintain
the dentin. However, this method cannot preserve the palatal
enamel.
99
For the present study, the labial enamel needed to be removed
without damaging the underlying dentin or the palatal enamel. A mod-
ified selective enamel demineralization technique was developed. We
coated a thin layer of liquid resin (Optibond FL) with an enamel bond-
ing technique onto the palatal surface of each tooth specimen prior to
any color and dimensional measurements. When performing enamel
demineralization, this coating protected the palatal enamel and kept it
intact. According to our pilot study, tooth acid treatment with 10%
HCl for 15 minutes is the most suitable protocol for enamel removal
for our purposes.
4.3 | Intraoral scanner
In this study, we chose an intraoral scanner that did not require the
use of powder. This approach removed any concern regarding the
alteration of the tooth surface and reduced the measuring time; thus,
the tooth was able to remain hydrated. By comparing the 3D scanning
data before and after selective enamel demineralization, the amount
of labial enamel removed from each specific area was known. This
quantity was previously difficult to measure without the risk of dam-
aging or altering the specimens. After superimposing the two dimen-
sional color mapping data onto the 3D measurement data, we
obtained the relationship between the differences in the labial enamel
thickness and the color.
4.4 | Analysis of outcome
4.4.1 | Cross polarization
The result of this study shows that there is a positive correlation
between ΔT and ΔE in cross-polarization measurements. However,
there is no correlation between ΔT and ΔE in the NP method. This dif-
ference might be caused by the interference of the specular reflection
in the NP method. Without the specular reflection, CP can aid in the
observation of subtle color changes that are not evident in regular
flash photography.
4.4.2 | Enamel homogeneity
If the enamel layer is a perfectly homogeneous structure, then the ΔE
values should be strongly correlated to ΔT. However, in this study, ΔE
and ΔT are only weakly to moderately correlated. This outcome might
indicate that the enamel layer is not homogeneous.
4.4.3 | Effect of labial enamel on tooth color
In a previous study, Coops and ten Bosch5 measured the influence of
the enamel layer on the color. These researchers confirmed that tooth
color is determined mainly by dentin, with enamel playing only a
minor role in the blue range. In the present study, similar findings
were observed. ΔT and Δb* (the color difference on the blue-yellow
axis in the CIELAB color space) are positively correlated. When more
100
labial enamel was removed, the tooth appeared more yellowish. All
three testing methods reveal the same findings.
Coops and ten Bosch5 also found that the color of a labial enamel-
removed tooth correlated strongly with the color of the complete
tooth. By adding one more factor, the enamel thickness, the present
study shows that color differences (ΔE*ab and ΔE*00 and Δb*) are
weakly to moderately correlated with the variation in the enamel
thickness (ΔT).
4.4.4 | Applying the results in a clinical situation
To interpret the simple linear regression analysis, we used the out-
come of the cross-polarization method as an example. Simple linear
regression analysis gives two formulas: ΔE*00CP = 1.2 + 2.4ΔT(mm;
Figure 10a) and Δb*CP = −0.8 + 4.0ΔT(mm; Figure 11b).
In a clinical situation, ΔT can be the labial enamel reduction in a
veneer preparation.
If a veneer preparation removes 0.5 mm of the labial enamel
(ΔT = 0.5), then the color difference (ΔE*00CP) will be 2.4. With 1 mm
labial enamel removal (ΔT = 1), the color difference will be 3.6. With
0.5 and 1 mm labial enamel removal, the Δb* is 1.2 and 3.2, respec-
tively. This result means that when more labial enamel is removed, the
tooth color becomes more yellowish.
According to these formulas, a clinician can roughly estimate the
stump shade even before preparing the tooth.
4.4.5 | The intercept in the simple regression
analysis
Interestingly, a simple regression analysis shows that ΔE is positively
related to ΔT with an intercept; for example, ΔE*00CP = 1.2 + 2.4ΔT
(mm). This finding means that when ΔT is close to zero, ΔE*00CP is still
1.2 (Figure 10a). If only a very thin layer of enamel is removed by
modified selective enamel demineralization, then there is already a
color change ΔE*00 of ~1.2, which is very similar to the effect of the
macroabrasion procedure. Further studies are needed to explain this
finding.
4.4.6 | The relationship between the different
testing methods
The color measuring results of NP, CP, and SP testing methods are
strongly and positively correlated with respect to the CIELAB values.
This correlation means that all three measuring methods point to the
same trend in the color difference in the CIELAB color space. This
result may enable clinicians and technicians to select and combine
these three methods to perform shade matching. However, more spe-
cific studies on this topic are still needed.
While matching the shade of a tooth clinically, NP should be the
primary reference as it is the closest to normal illumination conditions
and also records the tooth's surface texture. Without interference
from the specular highlights produced by the NP flash reflection, both
CP and SP techniques are very useful for assisting in shade selection.
HE ET AL.
5 | C O N CL U S I O N S
1. With the CP and SP testing methods, there is a weak-to-moderate
positive correlation between the varying labial enamel thickness
and the tooth color difference. A thicker labial enamel has a
greater impact on the tooth color.
2. With all three testing methods, the change in the labial enamel
thickness (ΔT) and color differences present in Δb* showed a
weak positive correlation. Reducing the labial enamel thickness
shifts the tooth color appearance toward yellow.
3. The measurement results of NP, CP, and SP are highly correlated,
which suggests that clinicians and technicians can combine all
three different methods for shade matching.
6 | CLINICAL IMPLICATIONS
1. This study establishes the relationship between varying labial
enamel thickness and differences in tooth color. The greater the
thickness of the removed labial enamel is, the more yellowish the
tooth color is. This basic rule could be helpful to clinicians when
preparing veneers.
2. It is reasonable to combine NP, CP, and SP for shade matching;
however, further research is still needed.
DISCLOS URE STATEMENT
No financial interest exists. The authors do not have any financial
interest in the companies whose materials are included in this article.
OR CID
Wei-Hung He https://orcid.org/0000-0003-2580-1958
RE FE RE NCE S
1. Yuan JC, Brewer JD, Monaco EA Jr, Davis EL. Defining a natural tooth
color space based on a 3-dimensional shade system. J Prosthet Dent.
2007;98(2):110-119.
2. Douglas RD. Precision of in vivo colorimetric assessments of teeth.
J Prosthet Dent. 1997;77(5):464-470.
3. Hasegawa A, Ikeda I, Kawaguchi S. Color and translucency of in vivo
natural central incisors. J Prosthet Dent. 2000;83(4):418-423.
4. Goodkind RJ, Schwabacher WB. Use of a fiber-optic colorimeter for
in vivo color measurements of 2830 anterior teeth. J Prosthet Dent.
1987;58(5):535-542.
5. ten Bosch JJ, Coops JC. Tooth color and reflectance as related to
light scattering and enamel hardness. J Dent Res. 1995;74(1):
374-380.
6. Spitzer D, Bosch JT. The absorption and scattering of light in bovine
and human dental enamel. Calcif Tissue Res. 1975;17(2):129-137.
7. Fried D, Glena RE, Featherstone JD, Seka W. Nature of light scatter-
ing in dental enamel and dentin at visible and near-infrared wave-
lengths. Appl Optics. 1995;34(7):1278-1285.
8. Denissen H, Dozic A. Photometric assessment of tooth color using
commonly available software. Eur J Esthet Dent. 2010;5(2):204-215.
HE ET AL.
9. Smith RN, Collins LZ, Naeeni M, et al. The in vitro and in vivo validation
of a mobile non-contact camera-based digital imaging system for tooth
colour measurement. J Dent. 2008;36(Suppl 1):S15-S20.
10. Tam WK, Lee HJ. Dental shade matching using a digital camera.
J Dent. 2012;40(Suppl 2):e3-e10.
11. Jarad FD, Russell MD, Moss BW. The use of digital imaging for colour
matching and communication in restorative dentistry. Br Dent J.
2005;199(1):43-49. discussion 33.
12. Gasparik C, Grecu AG, Culic B, et al. Shade-matching performance using
a new light-correcting device. J Esthet Restor Dent. 2015;27(5):285-292.
13. Clary JA, Ontiveros JC, Cron SG, Paravina RD. Influence of light
source, polarization, education, and training on shade matching qual-
ity. J Prosthet Dent. 2016;116(1):91-97.
14. Lakhanpal S, Neelima MS. Accuracy of three shade-matching devices
in replicating the shade of metal ceramic restorations: an in vitro
study. J Contemp Dent Pract. 2016;17(12):1003-1008.
15. Gordon P, Wander P. Specialised equipment for dental photography.
Br Dent J. 1987;162(9):346-359.
16. Hein S, Tapia J, Bazos P. eLABor_aid: a new approach to digital shade
management. Int J Esthet Dent. 2017;12(2):186-202.
17. Kim-Pusateri S, Brewer JD, Davis EL, Wee AG. Reliability and accu-
racy of four dental shade-matching devices. J Prosthet Dent. 2009;
101(3):193-202.
18. Chu SJ, Trushkowsky RD, Paravina RD. Dental color matching instru-
ments and systems. Review of clinical and research aspects. J Dent.
2010;38(Suppl 2):e2-e16.
19. Agrawal VS, Kapoor S. Color and shade management in esthetic den-
tistry. Universal Res J Dent. 2013;3:120-127.
101
20. RWG H, Pointer MR. Spectroradiometers and spectrophotometers.
Measuring Color. 4th ed. West Sussex, UK: Wiley; 2011.
21. Bazos P, Magne P. Bio-emulation: biomimetically emulating nature
utilizing a histoanatomic approach; visual synthesis. Int J Esthet Dent.
2014;9(3):330-352.
22. Wee AG, Lindsey DT, Kuo S, et al. Color accuracy of commercial
digital cameras for use in dentistry. Dent Mater. 2006;22(6):
553-559.
23. Berns RS, Billmeyer S. Color Imaging. Principle of Color Technology.
4th ed. Hoboken, NJ: Wiley; 2019.
24. Hein S, Zangl M. The use of a standardized gray reference card in
dental photography to correct the effects of five commonly used dif-
fusers on the color of 40 extracted human teeth. Int J Esthet Dent.
2016;11(2):246-259.
25. Wee AG, Lindsey DT, Shroyer KM, Johnston WM. Use of a porcelain
color discrimination test to evaluate color difference formulas.
J Prosthet Dent. 2007;98(2):101-109.
How to cite this article: He W-H, Park CJ, Byun S, Tan D,
Lin CY, Chee W. Evaluating the relationship between tooth
color and enamel thickness, using twin flash photography,
cross-polarization photography, and spectrophotometer.
J Esthet Restor Dent. 2020;32:91–101. https://doi.org/10.
1111/jerd.12553