ORIGINAL ARTICLE

Orthodontic application of color image addition to visualize

differences between sequential radiographs

Stavroula Papika, DDS,a Hans Ulrik Paulsen, DDS,b Xie-Qi Shi, DDS,c Ulf Welander, DDS, PhD,c and
Sten Linder-Aronson, DDS, PhDa
Stockholm, Sweden, and Copenhagen, Denmark

A new technique was developed making it possible to visualize changes that take place after premolar
transplantation. Three sequential radiographs from a series of standardized control examinations were digitized
and registered to the first radiograph in the series using subtraction technique. The digitized radiographs were
then transformed into monochromatic images: red, blue, and green. After correction of gray level and contrast,
the three radiographs were superimposed. All details that were common in the radiographs were then displayed
in black-and-white. Details that were different appeared in color. Red and blue combines into magenta, blue and
green into cyan, and green and red into yellow. A procedure was developed to create color-coded radiographs
according to the above. The procedure was applied to cases that had been treated by premolar transplantation.
The color coding offers the possibility to identify any radiographic changes that take place over time, eg, growth,
apposition or resorption of bone, and progression or regression of pathological processes. (Am J Orthod
Dentofacial Orthop 1999;115:488-93)

Differences in information between two
radiographic images may be enhanced by applying the
well-established subtraction technique.1-10 Recently,
the principles of an alternative technique were present-
ed by Welander et al,11 namely color coding of image
differences by means of image addition. In this tech-
nique three radiographs in a sequence are registered,
made into monochromatic images, and then superim-
posed. When this is performed with the aid of a com-
puter, the three additive primary colors, red, blue, and
green, that are used for the display on color monitors
are utilized. When three monochromatic radiographs
are superimposed, all details that are common to the
three images are displayed as in an ordinary radi-
ograph, ie, in black and white. Gray levels will depend
on the light intensity. Image details that are different in
the three radiographs will appear in color. The color
will depend on the weighing between gray levels in the
original black and white radiographs.
Subtraction, if successful, eliminates all informa-
tion common in the component radiographs. Color
image addition not only retains all image information
but also provides the possibility to evaluate changes
aDepartment of Orthodontics, Karolinska Institutet Stockholm, Sweden.
bDepartment
of Orthodontics, Copenhagen Municipal Dental Health Service,
Copenhagen, Denmark.
cDepartment of Oral Radiology, Karolinska Institutet, Stockholm, Sweden.
Reprints requests to: Dr. Stavroula Papika, Department of Orthodontics, Faculty
of Odontology, Karolinska Institutet, PO Box 4064, 141 04 Huddinge, Sweden.
Copyright © 1999 by the American Association of Orthodontists.
0889-5406/99/$8.00 + 0 8/1/86796
488
that have occurred between sequential radiographs in
one and the same image.
There are a number of applications in orthodontics
of the color coding technique briefly described here. It
may be used to study any changes that take place over
time during treatment. This involves development
processes as well as the development of pathologic
conditions.
The first aim of the present work was to develop a
procedure to make color-coded images by means of
image addition from original black-and-white radi-
ographs exposed on film. A second aim was to test the
applicability of this procedure on sequential intraoral
radiographs.
PROCEDURE
Digitizing
The first step in the procedure of making color
coded images from three original radiographs exposed
on film is digitizing, ie, to scan the analog information
about optical densities in radiographs and convert these
densities to numerical values. For this purpose, a
TruScan film digitizer model D-XV (Truvel ) connect-
ed to a SUN SPARC station was used. To facilitate and
speed up the relatively time-consuming scanning pro-
cedure, a computer program was constructed that made
it possible to scan several intraoral radiographs at the
same time. The radiographs were placed in a mount
that could hold a maximum of 12 radiographs. After
scanning such a block of radiographs, the computer
program separated the block into individual radio-
American Journal of Orthodontics and Dentofacial Orthopedics
Volume 115, Number 5
graphs. These were stored in separate files comprising
432 rows and 576 columns of 8 bit image data, which
means that optical densities in the original radiographs
were converted to numbers ranging from 0 to 255.
Image Registration
Digitized radiographs from the same sequence were
registered to the first radiograph in the sequence as a
reference. Thus, the second radiograph, the third, the
fourth, etc, were oriented with respect to the first. This
was performed with the SUN SPARC station.
The subtraction technique was used for image reg-
istration. A specially designed computer program
showed the first image in a sequence, the reference
image, on the monitor with its original gray levels. The
image to be oriented was displayed superimposed on
the reference image with its gray levels reversed. By
clicking the mouse on icons on the display the super-
imposed image could be moved with respect to the ref-
erence image. Movements could be made along x-axis
and y-axis in a coordinate system in positive or nega-
tive directions. In addition, movements in positive or
negative directions could be performed along two diag-
onal axes oriented 45° to the x-axis and y-axis. Rotation
could be performed in both positive and negative direc-
tions, ie, counterclockwise and clockwise. The center
of rotation could be selected. Correction for differences
in projection geometry could be achieved to a certain
extent by stretching or compressing the radiograph to
be registered.
The position and size of the image to be oriented
was changed with respect to the reference image until
maximum cancellation of structural details was
achieved. This indicates that optimum registration has
been reached. The registered image was then stored in
a new file.
Correction of Gray Levels and Contrast
Individual radiographs in a sequence may exhibit
differences in both density and contrast. This will affect
the resulting colors when the images are made mono-
chromatic and added to create color coding. The
weighing of light intensities of the individual images
will be false and not representative of true radiograph-
ic changes that have taken place over time.
Differences in exposure from one examination to
the next that only result in different average gray levels
in the digitized images are easily compensated by a
parallel shift up or down along the gray level scale.
Contrast differences may be compensated by a
computer program that takes into account the statistical
distribution of gray levels within two radiographs, a so-
called histogram analysis. For the present work, a pro-
Papika et al 489
Fig 1. When light of the three additive primary colors,
red, blue, and green, is combined the subtractive prima-
ry colors, magenta, cyan, and yellow, are created. When
the light intensity of the primary colors is identical differ-
ent gray levels appear depending on the intensity; black
corresponds to absence of light and white to full light
intensity.
gram was used that was based on algorithms according
to Ruttiman et al.12
Image Addition
A program run under WINDOWS on an IBM com-
patible PC was used for the final step of color-coding
differences over time between sequential radiographs.
The computer was provided with a 24 bit graphics card
(Stealth 2000). This card has the capability to display
light intensities in 256 steps, ie, 8 bits. These light
intensities are independent with respect to the three
additive primary colors of the computer monitor, red,
green, and blue. The first 8 bits of a 24 bit color image
are occupied by numeric values of light intensity levels
for the red part of the image, the next 8 bits for the
green, and the last 8 bits for the blue.
When the light intensity of all the three colors is
equal, a gray scale will be displayed on a computer
monitor. The shade of gray will depend on the light
intensity. Low intensity results in dark gray and high
intensity in light gray. Absence of light is black and full
intensity white. When the light intensity differs
between the primary colors, different colors will be
displayed (Fig 1). If red is given a certain light intensi-
ty and green and blue both have zero intensity, red will
be displayed on the monitor. Green and blue are creat-
ed analogously. Their complement colors, the subtrac-
tive primary colors, will be displayed when two of the
490 Papika et al American Journal of Orthodontics and Dentofacial Orthopedics
May 1999

B
Fig 2. Addition of three sequential radiographs, the first red, the second blue, and the third green. A,
Registration with respect to the first molar; eruption path of transplanted second premolar is visual-
ized. B, Registration with respect to the transplanted premolar visualizes movements with respect to
this tooth.

additive primary colors are given equal light intensity when the balance between the additive primary colors
and the third zero intensity. Green and blue combines is changed. Yellow-red, for instance, is created by the
into cyan, green and red into yellow, and blue and red combination of red and green but in this case the light
into magenta. Intermediate color hues are displayed intensity of red dominates over green. Yellow-green is
American Journal of Orthodontics and Dentofacial Orthopedics Papika et al 491
Volume 115, Number 5

Fig 3. Registration with respect to bony structures. The magenta coloring of the mesial marginal sep-
tum adjacent to the transplanted tooth indicates resorption after the operation. The yellowish color-
ing of the distal septum indicates initial bone resorption followed by healing. Slight mesial tilting may
be noticed regarding both the transplanted tooth and first molar.

created when the light intensity of green dominates
over red. Any other color hue is created according to
the same principle. A 24-bit graphics card can display
almost 17 million combinations of the three primary
colors resulting in the same number of color hues.
The aforementioned program read three radi-
ographs to be added and placed the data in proper bits
to display one single color coded radiograph.
APPLICATION
In order to test the applicability of the procedure
described here, it was tested on a material composed of
sequences of standardized radiographs of patients who
had been followed up after transplantation of teeth to
the lower premolar region.13
When the sequential radiographs are registered with
respect to the first molar, the movement of the trans-
planted tooth relative to the first molar may be studied
during growth (Fig 2A). Relative to the first molar, the
movement of the transplanted tooth during the obser-
vation period is easily distinguished. In fact, its crown
may be identified in three positions colored red and
magenta and cyan and green, respectively. The color
coding of the root indicates that it has moved in the
mesial direction, which appears green. This indicates
that the root has moved slightly in a mesial direction.
The crown, on the other hand, is colored red and has
moved in a distal direction during the growth. Togeth-
er these features indicate a certain rotation with respect
to the first molar. The first premolar shows basically
the same features, which visualizes eruption. The color
coding of the crown of the second molar demonstrates
reversed colors and its occlusal surface appears magen-
ta. The mesial side of the root appears green. These
features must be interpreted as rotation with respect to
the first molar.
Regarding the relative movements of the teeth in
the case illustrated in Fig. 2A, the first development
stage is represented by red and the last by green. Inter-
mediate stages are represented by the color sequence of
magenta, blue, and cyan.
When the same sequential radiographs are regis-
tered with respect to the transplanted tooth analogous
color coding takes place (Fig. 2B). Then movements
relative to the transplanted tooth are indicated.
It should be noted that because the registration of
the radiographs that are combined into color-coded
images in Figs 2A and 2B is with respect to teeth, the
first molar and the transplanted premolar, respectively,
bony structures demonstrate a moiré effect with irreg-
ular coloring.
Orientation with respect to bone structures in the
492 Papika et al
Fig 4. Registration with respect to the transplanted tooth. Green and cyan color in the most apical
part of the root of the second premolar is an evidence of the root development. The same color may
be noticed in the surrounding bony structures, indicating bone calcification during healing after the
transplantation.
lower part of the mandible makes it possible to study
changes over time in these structures, such as growth or
resorption, as well as successive movements of the
teeth during eruption (Fig 3). The three images that are
combined in Fig 3 cover the period of time after the
transplantation. The first image was exposed immedi-
ately after the transplantation, the second image 3
months after, and the third image 6 months after the
transplantation. The magenta coloring of the marginal
part of the septum between the first premolar and the
transplanted tooth indicates resorption after the opera-
tion. A somewhat indistinct yellowish coloring of the
tip of the septum between the transplanted tooth and
the first molar indicates initial resorption followed by
bone healing. The first (red) image and the third
(green) image dominate over the second (blue). Thus
from the time of the operation, bone has first been lost
and then regained. Although poorly reproduced on the
printed copy, the yellowish coloring is readily seen in
the original color coded radiograph when viewed on a
computer monitor. In fact, there is also a yellowish col-
oring at the periapical region of the transplanted tooth
indicating resorption after the operation followed by
healing. It will also be noted that the transplanted tooth
American Journal of Orthodontics and Dentofacial Orthopedics
May 1999
exhibited a certain mesial tilting. This is also valid for
the first molar.
When transplanted, the root development of the
tooth is not completed. By registering the sequential
radiographs with respect to the crown of the trans-
planted tooth and to the part of the root that is devel-
oped at the first examination, root development may be
studied (Fig 4). Cyan and green coloring indicates gain
of hard tissues, ie, calcification. This is so because the
second (blue) and the third (green) sequential radi-
ographs dominate over the first (red), where the
amount of hard tissues is less in certain areas. In Fig 4,
it may be noted that the most apical part of the root is
cyan or green as an evidence of growth. It will also be
noted that the surrounding bone exhibits the same col-
oring. This indicates bone healing after the operation
when the tooth was transplanted. Thus, root develop-
ment may be evaluated in one single image.
The cases that were used to test the procedure
described in this article shows that it may be applied to
create color coded images that will demonstrate
changes over time, relative movements during growth
as well as changes of the mineral content of teeth and
bone structures.
American Journal of Orthodontics and Dentofacial Orthopedics
Volume 115, Number 5
DISCUSSION
Color coding by means of image addition has two
essential advantages as compared with the subtraction
technique. First, all radiographic information in regions
where the original radiographs are equal will be
retained in black and white whereas any differences
appear colored and thus enhanced. Second, differences
between three successive radiographs in a sequence
may be studied in one single image. This should facili-
tate the interpretation of changes that take place over
time and are recorded at three successive radiographic
examinations. The subtraction technique enhances dif-
ferences between only two radiographs and results in
the elimination of radiographic information about
regions where the two radiographs are equal.
A prerequisite condition for both techniques, sub-
traction and color coding by means of image addition,
is access to standardized radiographs exposed in the
same projection and with the same exposure parame-
ters at successive examinations. The radiographs may
not be exactly identical in these respects, however.
Within certain limits it is possible to correct for differ-
ences in projection geometry and exposure. The posi-
tion of the film or digital detector at the exposure is not
critical. By different means the images may be regis-
tered with respect to each other.
In one of the examples (Fig 2A), the radiographs
were registered with respect to the first molar. This
tooth is already formed at the time of the transplanta-
tion. The result is one image that relative to the first
molar demonstrates all changes that have occurred in
association with eruption development during the fol-
low-up period, such as movements and rotations in any
direction. This is valid not only for the transplanted
premolar but also for the other adjacent teeth.
When the radiographs are registered with respect to
the transplanted tooth (Figs 2B and 4), the same basic
aforementioned information, ie, movements and rota-
tions of the adjacent structures relative to the trans-
planted tooth may be evaluated in the combined image.
In addition, it will be possible to study root develop-
ment and bone healing after the operation.
An additional possible way of registering the com-
ponent colored radiographs is with respect to bony
structures (Fig 3). This approach makes it possible to
study changes that take place within these structures,
eg, growth, modeling (bone formation), remodeling, or
signs of destructive processes. When studying develop-
ment changes and movements of teeth during normal
eruption or orthodontic treatment, this type of registra-
tion should be the most reliable. Bone structures are
relatively stable, and changes relative to these struc-
tures reflect development or treatment effects better
Papika et al 493
than registration with respect to individual teeth that
themselves are not in stable positions.
In order to apply the color coding technique, it is
essential to understand the principle. The three additive
primary colors, red, blue, and green, can be used in any
order but every such order will provide its own color
scheme. The present order was chosen when the tech-
nique was introduced because it results in red and
magenta when bone loss has occurred and in cyan and
green when bone has been gained.11 Thus, reddish col-
ors indicate progress and greenish colors indicate
regress of a destructive process that affects bone. In the
present work, the same order of coloring the sequential
radiographs has been applied. When studying move-
ments of teeth, this interpretation of the color coding
no longer holds true, however. The colors merely
reflect the position of the respective tooth at the three
sequential occasions when the radiographs were
exposed.
CONCLUSIONS
Color coding of differences between sequential
radiographs by means of image addition may be
applied to study changes over time that take place dur-
ing any development processes. It may be applied to
study different aspects of such changes during ortho-
dontic treatment.
REFERENCES
1. Reddy MS, Jeffcoat MK. Digital subtraction radiography. Dent Clin North Am
1993;37:553-65.
2. Reddy MS. Radiographic methods in the evaluation of periodontal therapy. J Peri-
odontol 1992:63(Suppl):1078-84.
3. Jeffcoat MK, Reddy MS, van den Berg HR, Bertens E. Quantitative digital subtrac-
tion radiography for the assessment of peri-implant bone change. Clin Oral Implants
Res 1992;3:22-7.
4. Kravitz LH, Tyndall DA, Bagnell CP, Dove SB. Assessment of external root resorp-
tion using digital subtraction radiography. Journal of Endod 1992;18:275-84.
5. Jeffcoat MK. Radiographic methods for the detection of progressive alveolar bone
loss. J Periodontol 1992;63(Suppl):367-72.
6. Brägger U, Bürgin W, Lang NP, Buser D. Digital subtraction radiography for the
assessment of changes in peri-implant bone density. Int J Oral Maxillofac Implants
1991;6:160-6.
7. Nicopoulou-Karayianni K, Brägger U, Bürgin W, Nielsen PM, Lange NP. Diagnosis
of alveolar bone changes with digital subtraction images and conventional radi-
ographs: an in vitro study Oral Surg Oral Med Oral Pathol 1991;72:251-6.
8. Tyndall DA, Phillips C, Malone-Thahey A, Renner J. Validity of digital subtraction of
transcranial plain films in quantification of positional changes of the mandibular
condyle. Oral Surg Oral Med Oral Pathol 1991;71:748-55.
9. Aagaard E, Donslund C, Wenzel A, Sewerin I. Performance for obtaining maximal
gain from a program for digital subtraction radiography. Scand J Dent Res 1991;99:
166-72.
10. McWilliam JS. The application of photographic subtraction in longitudinal cephalo-
metric growth studies. Eur J Orthod 1982;4:29-36.
11. Welander U, Eklund I, Tronje G, McDavid WD, Dove SB, Mahler R, et al. Color cod-
ing of radiographic changes over time by means of image addition. Oral Surg Oral
Med Oral Pathol 1994;78:531-8.
12. Ruttimann UE, Webber R, Schmidt E. A robust digital method for film contrast cor-
rection in subtraction radiography. J Periodontal Res 1986;21:486-95.
13. Paulsen HU, Andreasen JO, Schwartz O. Pulp and periodontal healing, root develop-
ment and root resorption subsequent to transplantation and orthodontic rotation: a
long term study of autotransplanted premolars. Am J Orthod Dentofacial Orthop
1995;108:630-40.