European Journal of Orthodontics 8 (1986) 141-148 O " 8 6 European Orthodontic Society

The effect of projection errors on cephalometric length

measurements
J. Ahlqvist*, S. Eliasson** and U. Welander*
•Department of Oral Radiology, University of Umea and "Department of Oral Radiology, Karolinska
Institutet, Stockholm

SUMMARY The magnitude of projection errors, calculated theoretically on length measurements

in cephalometry was studied. Rotation of the object by up to 5° from the proper position,

Downloaded from http://ejo.oxfordjournals.org/ at University of Brighton on July 11, 2014

resulted in errors in length measurements that were usually less than one percent. Rotations of
more than 5° may increase the error but imply misalignments of the head that should be evident
when positioning the patient. The advantage of extremely long focus film distances seems
doubtful.

Introduction The basic mathematics for calculating projec-

tion errors in cephalometry have been described
The accuracy of cephalometric measurements
previously (Eliasson et al., 1982). The authors
has generated great interest. Many studies have
presented expressions that may be used to calcu-
been published on errors associated with land-
late the position of any landmark on the film
mark identification, errors arising from the regis-
for any misalignment between the components of
tration of landmarks, and errors due to
the cephalographic system including the object.
measurement procedures (for a review see
McWilliam, 1983). The cephalographic projection is a so-called
central projection which implies that the beam
Errors due to the projection of the three- diverges from a point. With respect to the
dimensional object on to a two-dimensional film geometry of the cephalographic projection, the
have been studied less extensively. Similarly, focal spot may be considered a point source.
there has been little analysis of errors arising Ideally the cephalographic projection is ortho-
from misalignment between the different com- gonal, i.e. the film plane is perpendicular to the
ponents of the cephalographic equipment or central ray of the beam. In a straight lateral
misalignment of the patient in the cephalo- projection, such as a cephalographic profile,
graphic system. Although these errors may be the sagittal plane of the patient should be
of significance they are usually considered to be perpendicular to the central ray of the beam and
of less importance than other errors (Bjork, parallel to the film plane.
1974; Hatton and Grainger, 1958; Miller et
al., 1966; Savara et al., 1966; Carlsson, 1967; In practice, however, the alignment of the
Baumrind and Frantz, 1971; Mitgard et al., cephalographic system may differ from the
1974; Houston et al., 1986). theoretical ideal. The relations between the dif-
ferent components of the cephalographic system
In a study by van Aken (1963) projection are affected by a number of factors (Fig. 2):
errors were found to be small but might be of
significance in cephalographs of asymmetrical —the focal spot, the cephalostat, and the film
skulls or in the case of anatomical landmarks may be linearly displaced in relation to each
that do not lie in the mid-sagittal plane. Berg- other;
ersen (1980) studied magnification and distortion —the cephalostat and the film may be rotated
in cephalometric radiography and found discrep- with respect to each other;
ancies between distances measured on the film —the patient may be linearly displaced and/or
and true distances in the object. Based on these rotated in relation to the cephalographic
results compensation tables for correction of system.
linear measurements were constructed. An analysis of the influence of these factors
142 J. AHLQVIST, S. ELIASSON AND U. WELANDER

on the cephalometric image was presented by Z

Ahlqvist et al., 1983. The factors causing image
distortion and the principal distortion effect were
studied.
The purpose of the present work was to study
the magnitude of projection errors on length
measurements in cephalometry. It was assumed
that the alignment of the cephalographic equip-
ment was accurate. Thus, the study was limited
to the effects of incorrect patients positioning
on linear measurements.
Figure 1 Directions of possible misalignments of the
patient.
Method

Downloaded from http://ejo.oxfordjournals.org/ at University of Brighton on July 11, 2014

Errors in linear measurements due to misalign- The length on thefilmof the distance between
ment of the patient was studied by performing the landmarks in all modelled situations was
mathematical calculations. By this approach all calculated when the model objects were rotated
other errors are eliminated, such as those arising around their _y-axes (Fig. 1). For a real patient
from landmark identification, landmark regis- this would mean tilting the head. Rotation
tration, and measurement procedures. around the z-axis, i.e. rotation of a patient's
The projected distance on the film plane head, will result in analogous effects and was
between differently placed points in model ob- not treated separately. Since rotation around the
jects was calculated. The mathematics employed x-axis alone will only affect the position of the
have been described in a previous work (Eliasson landmarks on the film and not the distance
et al., 1982). between them, rotation around the x-axis was
A series of computer models of distances of not considered. Rotations between —10° and
different lengths and inclinations were con- + 10° were considered. In the case of the
structed. Three principally different situations modelled patient the result of simultaneous ro-
were modelled: tation around the y- and z-axes was calculated.
—the distance between cephalometric land- Here, only rotations within ± 5° were considered
marks representing structures in the sagittal since careful positioning of the patient should
plane, such as the distance between sella and eliminate more severe malpositioning in clinical
nasion; practice. A focus to object distance of 1400 mm
—the distance between one landmark represent- and a focus to film distance of 1550 mm were
ing a structure in the sagittal plane and a pair used.
of landmarks representing bilaterally sym- The computer programs were constructed to
metrical structures, such as the distance allow for translation of the model objects along
between pogonion and articulare; all three axes of the cephalostat. Calculations
—the distance between two pairs of landmarks were performed when rotation was combined
representing bilaterally symmetrical struc- with a maximum of ± 10 mm translation along
tures, such as the distance between articulare the three axes.
and gonion. The magnitude of measurement errors was
In the case of bilaterally symmetrical struc- studied by means of diagrams in which the
tures, the mean position of the two landmarks relative length of distances between modelled
on the film plane was calculated and used to landmarks were plotted as functions of rotation
define one endpoint in a calculated distance. and/or translation of the model objects.
In order to perform calculations for a realistic
situation a patient was modelled. Three-dimen- Results
sional coordinates for anatomical structures
used to define cephalometric landmarks were In one investigated case one object point co-
found by determining the mean position of the incided with the origin of the coordinate system
structures on ten dry skulls. of the object. Another point was given different
PROJECTION ERRORS IN CEPHALOMETRY 143

- : 22.5° 22.5°
: 45.0° 45.0°
1.03 1.03 67.5°
: 67.5°
90.0°
1.02 1.02

1.01 1.01

M 1.00 f> 1.00

•5 0.99 "S 0.99

0.98 10.98

0.97 0.97

Downloaded from http://ejo.oxfordjournals.org/ at University of Brighton on July 11, 2014

-10 -5 0 5 10 -10 -5 0 5 10
A Rotation around y-axis, degrees C Rotation around y-axis, degrees

: 22.5°
: 45.0° forms a small angle to the rotational axis, the
1.03 : 67.5°
• on .0°
error of its projected length on to the film
1.02 plane is insignificant. The error increases with
increased inclination. Still, the error is less than
1.01 1 per cent between —10° and 5° rotation when
the inclination of the line is at 90° to the
rotational axis. If the length of line is increased
c
— to 100 mm there is an insignificant change of
Z 0.99 the error for distances forming small angles
to the horizontal plane (Fig. 2b). For larger
10.98
inclinations the error decreases with negative
0.97
rotation. With positive rotation, the error in-
creases with increased inclination of the distance.
The magnitude of this change is trivial (compare
-10 -5 0 5 10 Figs 2a and b). When rotation was combined
B Rotation around y-axis,, degrees with translations up to ±10 mm the additional
Figure 2 Relative length on the film plane of object effect on the error was found to be negligible
distances plotted as functions of rotation around the >>-axis. for practical purposes (Fig. 2c).
(A) A distance of 50 mm with one endpoint at the origin.
The four functions represent different inclinations to the y- It should be noted that when the head is
axis of the distance as indicated in the diagram, (B) A rotated so that the forehead approaches the film,
distance of 100 mm with one endpoint at the origin, (c) the projected length of a line such as S-N is
Same as (B) with the addition of 10 mm translation along reduced, while when the rotation is in the op-
all three axes.
posite direction, the projected length is increased
(Fig. 3).
Results of calculations performed on the
positions in the y-z-plane, corresponding to modelled patient are exemplified for the follow-
the sagittal plane of a real patient. Different ing distances: sella-nasion (Fig. 4), nasion-sub-
positions of the latter point were chosen so that spinale (Fig. 5), and articulare-pogonion (Fig.
a line connecting the two object points had 6). Here all combinations of simultaneous ro-
constant length but formed different angles to tation from — 5° to + 5° around the y- and z-
the y-axis, the rotational axis. The relative axes are plotted.
projected length of such lines, 50 mm long, when Sella-nasion forms a small angle to the y-axis.
rotated around the y-axis is illustrated in Fig. When the patient is rotated around the y-axis
2a. When a line connecting the object points the length of this distance decreases as the
144 J. AHLQVIST, S. ELIASSON AND U. WELANDER

Downloaded from http://ejo.oxfordjournals.org/ at University of Brighton on July 11, 2014

Figure 3 The projected length of an object will vary with its inclination toward the focal spot or toward the film.

L:
1.03-

0.97-

-5 5 4,

L:
1.03-

0.97-

(c)
Figure 4 Relative projected length, L, of the distance sella-nasion as a function of combined rotations around the y- and z-
ajtes, respectively. The variation of the relative length of the distance is represented by a curved plane which is illustrated
by a three-dimensional diagram shown in three aspects: in a perspective plot (a) and in plots that demonstrate the effect of
rotation around the z-axis (b) and the >>-axis (c).
PROJECTION ERRORS IN CEPHALOMETRY 145

Downloaded from http://ejo.oxfordjournals.org/ at University of Brighton on July 11, 2014

L:
1.03 -

0.97-

-5 0 5 V
Figure 5 Relative projected length, L, of the distance nasion-subspinale as a function of combined rotations around the y-
and z-axes, respectively. Three-dimensional diagram shown in three aspects.

5 V
Figure 6 Relative projected length, L, of the distance articulare-pogonion as a function of combined rotation around the
y- and z-axes, respectively. Three-dimensional diagram shown in three aspects.
146 J. AHLQVIST, S. ELIASSON AND U. WELANDER

rotation angle increases from — 5° to + 5° (Fig. above, the projected length of this distance is
4a and c), i.e. when the forehead is tilted toward most sensitive to combinations of rotation and
the film. When rotated around the z-axis the tilting of the patient's head (Fig. 6). The mini-
length of the distance increases as the rotation mum projected length occurs at simultaneous
angle increases from — 5° to + 5° (Figs 4a and negative rotations of —5° around both axes.
b), i.e. when the forehead is rotated away from The maximum projected length occurs at simul-
the film. A distance such as sella-nasion which taneous positive rotations of 5° around both
forms a small angle to the y-axis, i.e. close to axes. When negative rotation around one axis
horizontal, is most sensitive to rotation of the is combined with positive rotation around the
patient's head and less sensitive to tilting of the other axis, there are only insignificant errors
head. This is evident from the fact that the plane in the projected length. All other calculated
representing the varying projected length of distances followed these patterns depending on
sella-nasion in Fig. 5 is more markedly curved the degree of inclination.
in the dimension representing rotation around The effect of a varied focus to object distance

Downloaded from http://ejo.oxfordjournals.org/ at University of Brighton on July 11, 2014

the z-axis, than in the dimension representing on errors of length measurements is limited.
rotation around the y-axis. While relatively long focus to object distances
Since the distance nasion-subspinale is essen- are favourable, extremely long distances do
tially vertical this distance demonstrates the not change the magnitude of projection errors
opposite variation during rotation and tilting as markedly. This is exemplified in Fig. 7 where
compared to sella-nasion (Fig. 5). the relative projected length of the distance
The distance articulare-pogonion is more or articulare-pogonion is plotted for four different
less diagonal. As can be expected from the focci to object distances.

Figure 7 Relative projected length, L, of the distance articulare-pogonion as a function of rotations around the y- and z-
axes. Four different focci to object distances were employed: a) 1400 mm, b) 2800 mm, c) 4200 mm, and d) 5600 mm. The
object to film distance was held constant at 150 mm. It will be noted that the projection error demonstrates limited changes
when the focus to object distance is longer than 2800 mm.
PROJECTION ERRORS IN CEPHALOMETRY 147

Discussion increases and may become significant even at

rotations of a few degrees more than +5°.
Errors caused by the cephalographic projection On the other hand, careful patient positioning
technique represent only a part of the total error should eliminate such errors. Rotation or tilting
of length measurements in cephalometry. Other of the patient's head by more than 5° is discern-
errors originate in the identification and regis- ible and should not arise in skilful clinical work.
tration of landmarks and in the measurement Application of long focus to object distances
technique. Since all these errors would influence has been suggested to minimize projection errors
any experimental study, it was decided to investi- (Nawrath, 1961, Carlsson, 1967, van Aken,
gate the importance of projection errors in 1963). Although it is true that short focus to
cephalometry by means of theoretical calcu- object distances result in greater projection
lations. As stated above the true magnitude of errors than long focus to object distances, the
projection errors is hidden in any experiment results of the present study indicate that there is
utilizing measuring techniques, van Aken (1963) a limited gain in using extremely long distances.

Downloaded from http://ejo.oxfordjournals.org/ at University of Brighton on July 11, 2014

used a mathematical approach in a study of
'geometrical errors in lateral skull x-ray projec-
tions'. Although his study is excellent and of Address for correspondence
great importance, it is limited to specific aspects
of projection errors. Dr J. Ahlqvist
Department of Oral Radiology
In the present work the aim was to analyse the University of Umea
effect of projection errors. Thus, the computer S-901 87 Umea
programs constructed to perform the calcu- Sweden
lations were written to allow for independent
variation of the following factors: translation in
space of the focal spot, the cephalostat, the References
object, and the film; rotation of the cephalostat,
the object, and the film plane around their axes; Ahlqvist J, Eliasson S, Welander U 1983 The cephalographic
projection. Part II. Principles of image distortion in ce-
focus to object and focus to film distances. This phalography. Dento Maxillo Facial Radiology 12: 101 —
list includes 22 variables. In fact, the number of 108
variables is so great that a complete systematic van Aken J 1963 Geometrical errors in lateral skull x-ray
analysis of the influence of all of them, including projections. Nederlands Tidschrift voor Tandheelkunde
all possible combinations, is impossible for prac- 70: 18-30
tical reasons. In the present work, the investi- Baumrind S, Frantz R C 1971 The reliability of head film
gation was limited to the effect of rotation and measurements. 2 Conventional angular and linear meas-
translation of the patients head and to variations ures. American Journal of Orthodontics 60: 505-517
of the focus to film distance. These limitations Bergersen E 1980 Enlargement and distortion in cephalo-
metric radiography. Compensation tables for linear
are justified by previous results indicating that measurements. Angle Orthodontist 50: 3-244
misalignment between the different components Bjork A 1947 The face in profile. Svensk Tandlakare Tidskrift
of the cephalographic system is of minor impor- 40suppl5B
tance provided that the equipment is properly Carlsson G E 1967 Errors in X-Ray cephalometry. Odontolo-
adjusted (Ahlqvist et al., 1983). Furthermore, gisk Tidskrift 75:99-129
when several cephalographs are exposed using Eliasson S, Welander U, Ahlqvist J 1982 The cephalographic
the same cephalostat, misalignment of the equip- projection. Part I General consideration. Dento Maxillo
ment will cause consistent systematic errors, Facial Radiology 11: 117-122
while the positioning of the patient causes non- Hatton M E, Grainger R M 1958 Reliability of measurements
from cephalograms at the Burlington orthodontic research
systematic and varying projection errors. centre. Journal of Dental Research 37:853-859
In general, projection errors in length Houston W J B , Maher R E, McElroy D, Sherriff M 1986
measurements are minor in cephalometry. In Sources of error in measurements from cephalometric radio-
graphs. European Journal of Orthodontics 8: 149-151
fact, rotation of ±5° from the ideal position
McWilliam J 1983 Photographic subtraction in craniofacial
usually results in errors that are less than one analysis. Dento Maxillo Facial Radiology. Suppl 4
percent. Such as error is usually insignificant Miller P A, Savara B S, Singh I J 1966 Analysis of
and will in most instances be concealed by other errors in cephalometric measurement of three-dimensional
errors. If the rotation is increased the error distances on the maxilla. Angle Orthodontist 36: 169-175
148 J. AHLQVIST, S. ELIASSON AND U. WELANDER

Mitgard J, Bjork G, Linder-Aronsson S 1974 Reproduci- Savara B S, Tray W E, Miller P A 1966 Analysis of errors
bility of cephalometric landmarks and errors of measure- in cephalometric measurements of three-dimensional dis-
ments of cephalometric cranial distances. Angle tances on the human mandible. Archives of Oral Biology
Orthodontist 44: 56-62 11: 209-217
Nawrath K 1961 Moglichkeiteh und Grenzen der ront-
genologischen Kephalometrie. Habilitationsschrift der
Johannes Gutenberg-Universitat, Main

Downloaded from http://ejo.oxfordjournals.org/ at University of Brighton on July 11, 2014