Techno bytes

Considerations in the use of cone-beam

computed tomography for buccal bone
measurements
Aaron Dean Molen
Auburn, Wash

Cone-beam computed tomography (CBCT) has become a popular modality in research, but it can be mis-
used and misunderstood. Several image quality, bone biology, and statistical factors must be considered
before designing CBCT studies or interpreting their results. Studies making small measurements, such as
changes in buccal bone thickness, are especially susceptible to these factors. The spatial resolution as de-
termined by a line pair phantom, the CBCT settings used, and a statistical power analysis should be reported
in studies that investigate small bony changes. Protocols should therefore be established and followed to
minimize the misinterpretation of results and improve the quality of research in this field. (Am J Orthod Den-
tofacial Orthop 2010;137:S130-5)

C
one-beam computed tomography (CBCT) has
become a popular modality in the evaluation of
orthodontic diagnoses and outcomes.1 It offers
high diagnostic value with a relatively low radiation
dose. Its linear accuracy has also been verified.2 This ac-
curacy has justified the use of CBCT scans in studying
implant sites,3 palatal thickness,4 and cephalometrics.5
Thus far, these studies have been limited to larger mea-
surements spanning several centimeters.
Recently, there has been interest within the research
community in using CBCT to evaluate smaller maxil-
lofacial structures: in particular, buccal-bone thickness
before, during, and after treatment. Before one under-
takes or interprets a study evaluating buccal bone with
CBCT, several image quality, bone biology, and statisti-
cal factors must be considered.
Spatial resolution is the minimum distance needed
to distinguish between 2 objects and is often incorrectly
assumed to be equal to a scan’s reported resolution or
voxel size. Factors such as partial volume averaging,
noise, and artifacts make it impossible to achieve a reso-
lution equal to the voxel size. These factors and others
will be discussed in this article.
Lecturer, Section of Orthodontics, University of California at Los Angeles; pri-
vate practice, Auburn, Wash.
The author reports no commercial, proprietary, or financial interest in the prod-
ucts or companies described in this article.
Reprint requests to: Aaron D. Molen, 2229 166th Ave E, Lake Tapps, WA
98391-4903; e-mail, aarondds@gmail.com.
Submitted, June 1009; revised and accepted, January 2010.
0889-5406/$36.00
Copyright © 2010 by the American Association of Orthodontists.
doi:10.1016/j.ajodo.2010.01.015
S130
In a study by Ballrick et al,6 a Classic i-CAT (Imag-
ing Sciences International, Hatfield, Pa), set at 120 kVp
and 5 mA, was used with a line pair phantom to test the
spatial resolution of all potential field of view (FOV)
and voxel size combinations. Those authors found that
a 0.2 mm voxel scan had an average spatial resolution
of 0.4  mm. The 2 most common voxel sizes used for
orthodontic scans—0.3 and 0.4  mm—both averaged a
spatial resolution of 0.7  mm. In areas of thin bone, a
spatial resolution of 0.7 mm would not be adequate to
properly visualize the bone.
Spatial resolution is also frequently confused with
measurement accuracy. Measurements made with
CBCT have been shown to be accurate to within 0.1 to
0.2 mm.7 However, linear accuracy over long distances
is different from a scan’s ability to differentiate between
2 objects in close proximity (spatial resolution).
Because of the multi-factorial nature of spatial resolu-
tion, each machine and scan must be evaluated individu-
ally. For CBCT studies that focus on small measurements,
it would be prudent to use a line pair phantom (Fig 1) to
determine the unique spatial resolution of the machine and
the scan settings used. Caution should be exercised when
drawing conclusions based on numbers smaller than a
scan’s spatial resolution. If a study fails to report a scan’s
spatial resolution and, instead, simply reports the voxel
size, we are left without the proper tools to interpret the
results. Also, spatial resolutions determined using the ideal
conditions in a phantom will always overestimate the sub-
sequent in-vivo spatial resolution.6
An important factor influencing in-vivo spatial reso-
lution is partial volume averaging.8 Frequently, the size

S130-135_AAOPRG_3051.indd 130 3/24/10 12:23 PM

 American Journal of Orthodontics and Dentofacial Orthopedics Molen  S131
Volume 137, Number 4, Supplement 1
Fig 1. Example of a line pair phantom (Phantom Laboratory, Salem, NY) used in determining spatial
resolution.
Fig 2. Coronal slice of sphenoid sinus. Red arrow indi-
cates where partial volume averaging has created the
false appearance that there is a communication be-
tween the sinus and the anterior cranial fossa (settings:
0.42‑mm voxel, 1.26‑mm slice thickness, 12‑in FOV).
of a voxel is larger than the object or the densities it
represents. This occurs most often along the margin of
an object or at the boundary of 2 substances of differing
densities.9 The voxel can display only 1 gray value at a
time. To account for this, the voxel displays an average
of the densities present. Simply put, if a voxel repre-
sents an area of 75% lucent soft tissue and 25% opaque
cortical bone, the voxel will appear more lucent than
opaque. This process can make boundaries between
S130-135_AAOPRG_3051.indd 131
densities harder to accurately distinguish, and results in
lower spatial resolution.
Thin bone is especially susceptible to partial volume
averaging, not just buccal bone, but also with areas such
as temporal bone and the sphenoid sinus (Fig  2). The
affect of partial volume averaging on thin bone has been
well documented in conventional computed tomography
(CT) scanners.10 It was shown that the angle at which
the image plane intersects the bone wall can cause thin
bone to appear thicker or thinner than it truly is.11 This
type of partial volume averaging can cause bone walls
thinner than 1 mm to all but disappear on CT scans.12
The most effective way to decrease the influence of
partial volume averaging is to decrease the voxel size.
There is a trade-off, however, when using smaller voxel
sizes, since they require more radiation and are more
prone to noise.13
Noise is the result of unintended energy or photons
hitting the detector and clouding the resultant image.
The noise levels in scans vary greatly between ma-
chines. Some machines have cleaner or less noisy im-
ages, whereas others are more difficult to read. Each
machine’s settings, environment, and reconstruction al-
gorithms affect the image’s noise.13
A main cause of noise in a scan is scatter radiation.14
Compared with medical CT, CBCT can have up to 15
times higher scatter levels.9,15 Medical CT’s lower scat-
ter levels allow it to image some structures better than
CBCT. For example, Loubele et al16 found that medical
CT is superior to CBCT in imaging cortical bone.
In CBCT, scatter levels increase as the size of the
FOV increases. The easiest way to decrease noise from
3/24/10 12:23 PM
 S132  Molen American Journal of Orthodontics and Dentofacial Orthopedics
Fig  3. Coronal and axial slices demonstrating metal artifacts (settings: 0.36‑mm voxel, 0.36‑mm
slice thickness, 12‑in FOV).
scatter is to use the smallest FOV that encompasses the
region of interest.17 The larger the FOV and the greater
the scatter, the worse the spatial resolution becomes. For
this reason, large FOVs, such as those frequently used
in orthodontic scans, are contraindicated for clinicians
wishing to evaluate buccal bone thickness.
Smaller FOVs may decrease noise from scatter,
but decreasing the voxel size has an inverse effect. As
voxels decrease in size, they become more sensitive to
noise, resulting in poorer spatial resolution than one
might expect.10 Reconstruction algorithms can decrease
the noise in small voxel scans but require further devel-
opment. Although a voxel size of 0.125 mm is available,
because of noise and other factors, a spatial resolution
of 0.125 mm is currently unachievable.
A number of artifacts can affect the quality of a
CBCT image. The most apparent ones in orthodontics
are metal artifacts. CBCT scans taken with braces pres-
ent show streaking artifacts around the teeth (Fig  3).
These artifacts might simply be a nuisance, except that
they can affect how a scanner interprets and reconstructs
the surrounding structures.18 Katsumata et al19 showed
that the densities of surrounding structures can affect
an adjacent voxel’s perceived density. For this reason,
a practitioner should be cautious in making measure-
ments close to braces or other metal objects, since the
spatial resolution in that area will be compromised. An
analogy would be trying to spot an airplane flying close
to the sun. The closer the airplane gets to the sun, the
harder it becomes to see because of the sun’s brightness.
The airplane is still there, but the sun has made it dif-
ficult to distinguish.
Another artifact frequently encountered in ortho-
dontics is movement. CBCT is more sensitive to patient
movement than medical CT.20 The most effective way
to limit movement artifacts is to decrease the scan time.
This is especially helpful with younger orthodontic
patients. However, with a decreased scan time comes
fewer data or frame acquisitions.6 This leads to under-
sampling, which makes resolving fine details difficult.9
S130-135_AAOPRG_3051.indd 132
April 2010
This presents a dilemma when the goal of the scan
is to achieve high spatial resolution. Do you expose the
patient to a longer scan to improve spatial resolution but
increase the risk of movement artifacts? Or do you com-
promise spatial resolution to decrease the risk of move-
ment artifacts? Most orthodontic scans are taken with
a short scan time, resulting in fewer frames acquired,
fewer movement artifacts, and lower spatial resolution.
Orthodontic scans with these characteristics are ideal
for general treatment planning but should be used with
caution if the goal is to properly evaluate the fine details
of the buccal bone.
Image quality is also affected by the gray scale bit
depth of the CBCT system used. Current CBCT sys-
tems range from 12-bit to 16-bit gray scale.21 Since the
human eye cannot distinguish beyond 10-bit gray scale
and computer monitors are currently available in only
8-bit or 10-bit gray scale, some people assume that it
would not be worthwhile to exceed these thresholds.22
However, this approach fails to take into account that
the reconstruction software uses the increased bit depth
to improve its primary and secondary reconstructions,
resulting in a cleaner and more defined volume. Most
software also allows the clinician to change the values
of gray scale displayed on the screen using a process
called windowing.23 By scrolling through the various
values of gray captured in the scan, the clinician can
better visualize the volume. In essence, the software al-
lows us to use the additional shades of gray captured by
the scan despite the limitations of our eyes and com-
puter monitors. When evaluating small structures, the
highest available gray scale should be used.
Moving beyond image quality, another consideration
is the orientation and location of the image planes used
to visualize the region of interest. Since most crowded
teeth have some rotation, the thinnest portion of bone
might not correspond to the buccal surface of the root.
A rotated tooth will naturally have thicker bone along
the buccal surface of its root. However, when the tooth
is derotated, the buccal surface of the root will move
3/24/10 12:23 PM
 American Journal of Orthodontics and Dentofacial Orthopedics Molen  S133
Volume 137, Number 4, Supplement 1
Fig 4. Axial slice of rotated premolar showing the effect
of rotation on buccal bone thickness. The black arrow
(1) indicates the rotated tooth, and the blue arrow (2)
indicates the derotated tooth. Note the thinning of bone
after the tooth is derotated (settings: 0.36‑mm voxel,
1.08‑mm slice thickness, 12‑in FOV).
into thinner bone (Fig 4). This phenomenon can easily
be mistaken for bone loss, whereas in reality the buccal
surface of the root has simply rotated into a new bony
position. For this reason, buccal bone measurements
should be made at several sites around the tooth and av-
eraged for a more honest representation of bone support
regardless of the tooth’s rotation. If the measurement is
made only at the buccal surface of the root at both pre-
treatment and posttreatment times, the derotated tooth
will appear to be in thinner bone after treatment.
The removal of braces and end of orthodontic tooth
movement are important beyond the simple avoid-
ance of metal artifacts. When teeth are being moved
orthodontically, the alveolar bone in the direction of the
applied force undergoes constant bone turnover.24 This
bone turnover is driven by the activity of osteoclasts,
which decrease the density of the active bone. This in-
crease in osteoclastic activity and the reduction in bone
density are called the regional acceleratory phenom-
enon (RAP).25 This decrease in density is important,
since CBCT scanners identify an object’s presence by
its density. For example, buccal bone undergoing RAP
would appear less clearly and more lucent on a CBCT
scan. This makes the less-active bone appear more dis-
tinctly than the adjacent bone undergoing RAP26; this in
turn gives the illusion that the boundaries of the bone
lay along the margins of the less-active bone.
S130-135_AAOPRG_3051.indd 133
This phenomenon is why changes in the alveolar
bone should be assessed only after orthodontic treat-
ment and the rematuration of the bone. RAP takes 6 to
24 months to fully subside after the end of tooth move-
ment.13 The appropriate waiting time after orthodontic
treatment before taking a final scan to evaluate buccal
bone thickness has yet to be determined, but perhaps a
minimum of 1 year posttreatment should be considered
in the interim.
An important statistical consideration is the pow-
er analysis, which is used to determine the minimum
sample size needed to make a determination with a
certain degree of confidence or power. Ultimately, the
power analysis determines the statistical sensitivity of
a study more than its spatial resolution. A study dem-
onstrating a spatial resolution of 0.4 mm with a 1.5 mm
power analysis level of sensitivity will be sensitive to
only 1.5  mm. A power analysis should be considered
during the design of the study and not retrospectively to
prevent drawing conclusions that are outside the study’s
level of sensitivity.27
There is often confusion between the power analy-
sis and statistical significance. Statistical significance is
simply the comparison of 2 groups for any differences.
If the differences are small enough, they are not con-
sidered significant; if they are large enough, they are
deemed significant. Statistical significance is super-
seded by statistical power when interpreting results.
For example, if a small study with a confidence level
of 1.5  mm has a statistically significant difference of
0.5 mm between 2 groups, the finding would be clini-
cally irrelevant, since 0.5  mm is less than the study’s
power of 1.5 mm.
Putting all previously discussed factors aside, the in-
clusion of a control group might be the single most im-
portant factor when drawing conclusions from a study’s
results. A longitudinal study with the experimental
group as its own control limits its power to make com-
parisons with groups outside the study. For example,
if a comparison is to be made between 2 orthodontic
techniques, there must be 2 groups, 1 representing each
technique. If only 1 group is used, any changes could
simply be due to orthodontic movement in general and
have no relevance to the technique used. The use of 2
groups removes this variable and allows for more pow-
erful comparisons.
Conclusions
Spatial resolution and its contributing factors should
be considered during the design or interpretation of
CBCT studies. Line pair phantoms should be used to
determine a scan’s spatial resolution so that conclusions
3/24/10 12:23 PM
 S134  Molen American Journal of Orthodontics and Dentofacial Orthopedics
are not based on measurements exceeding a scan’s
resolution. The 2 most common voxel sizes used in
orthodontics—0.3 and 0.4  mm—provide lower spatial
resolution than smaller voxel sizes and should be used
with caution if the goal is to assess small variations in
bone thickness. A smaller voxel size would be more ap-
propriate for these studies and would also decrease the
influence of partial volume averaging.
Scatter noise associated with the large FOVs fre-
quently used in orthodontics decreases spatial resolu-
tion and is contraindicated when studying buccal bone
changes. The smallest FOV containing the region of
interest should be used and precludes the use of full-
arch scans. In addition to a smaller voxel size and small-
er FOV, a long scan time with more frame acquisition
should be used to prevent poor resolution caused by
undersampling. To further improve spatial resolution, a
16-bit sensor should be used, if possible, for the best
available gray scale.
Properly conducted studies should include a time
point at least 1 year posttreatment to avoid metal arti-
facts and allow the bone to remature after bone turnover
from RAP. Buccal bone measurements made during
active treatment have limited value. Measurement site
selection and slice orientation should be carefully con-
sidered to prevent erroneous measurements caused by
tooth derotation.
Researchers should prospectively take into account
the statistical power of their study during its design. A
study’s power should always be considered during the
interpretation of its results to prevent unsupported con-
clusions. Results beneath a study’s spatial resolution or
level of statistical power are ambiguous even if they are
statistically significant. Finally, studies must include an
external control group against which comparisons can
be made.
Unless proper protocols are established and fol-
lowed, CBCT will be misused and lead to poorly de-
signed studies and improper conclusions. However,
with the right design and the correct settings, studies
can be conducted to evaluate buccal bone changes by
using CBCT. Ideally, these considerations should be
used when assessing or planning any study that seeks to
quantify alveolar bone changes secondary to orthodon-
tic tooth movement.
References
1. Molen AD. The 3D orthodontist: the modern orthodontist’s
source for information of 3D technologies. Available from:
http://www.3DOrthodontist.com/. Accessed on January 14,
2010.
2. Moshiri M, Scarfe WC, Hilgers ML, Scheetz JP, Silveira AM,
Farman AG. Accuracy of linear measurements from imaging
S130-135_AAOPRG_3051.indd 134
April 2010
plate and lateral cephalometric images derived from cone-
beam computed tomography. Am J Orthod Dentofacial Orthop
2007;132:550-60.
3. Chen LC, Lundgren T, Hallström H, Cherel F. Comparison of
different methods of assessing alveolar ridge dimensions prior
to dental implant placement. J Periodontol 2008;79:401-5.
4. Gracco A, Lombardo L, Cozzani M, Siciliani G. Quantitative
cone-beam computed tomography evaluation of palatal bone
thickness for orthodontic miniscrew placement. Am J Orthod
Dentofacial Orthop 2008;134:361-9.
5. Chien P, Parks E, Eraso F, Hartsfield J, Roberts W, Ofner S.
Comparison of reliability in anatomical landmark identification
using two-dimensional digital cephalometrics and three-dimen-
sional cone beam computed tomography in vivo. Dentomaxil-
lofac Radiol 2009;38:262-73.
6. Ballrick JW, Palomo JM, Ruch E, Amberman BD, Hans MG.
Image distortion and spatial resolution of a commercially avail-
able cone-beam computed tomography machine. Am J Orthod
Dentofacial Orthop 2008;134:573-82.
7. Hilgers ML, Scarfe WC, Scheetz JP, Farman AG. Accuracy of
linear temporomandibular joint measurements with cone beam
computed tomography and digital cephalometric radiography.
Am J Orthod Dentofacial Orthop 2005;128:803-11.
8. Chakeres DW. Clinical significance of partial volume averaging
of the temporal bone. AJNR Am J Neuroradiol 1984;5:297-302.
9. Scarfe WC, Farman AG. What is cone-beam CT and how does
it work? Dent Clin North Am 2008;52:707-30.
10. Hangartner TN, Gilsanz V. Evaluation of cortical bone by com-
puted tomography. J Bone Miner Res 1996;11:1518-25.
11. Ahlqvist JB, Isberg AM. Validity of computed tomography in
imaging thin walls of the temporal bone. Dentomaxillofac Ra-
diol 1999;28:13-9.
12. Ahlqvist JB, Isberg AM. Bone demarcation of the temporoman-
dibular joint. Acta Radiol 1998;39:649-55.
13. Kwong JC, Palomo JM, Landers MA. Image quality produced
by different cone-beam computed tomography settings. Am J
Orthod Dentofacial Orthop 2008;133:317-27.
14. Endo M, Tsunoo T, Nakamori N, Yoshida K. Effect of scat-
tered radiation on image noise in cone beam CT. Med Phys
2001;28:469-74.
15. Siewerdsen JH, Jaffray DA. Cone-beam computed tomography
with a flat panel imager: magnitude and effects of x-ray scatter.
Med Phys 2001;28:220-31.
16. Loubele M, Guerrero M, Jacobs R, Suetens P, van Steenberghe
D. A comparison of jaw dimensional and quality assessments
of bone characteristics with cone-beam CT, spiral tomogra-
phy, and multi-slice spiral CT. Int J Oral Maxillofac Implants
2007;22:446-54.
17. Miracle AC, Mukherji SK. Conebeam CT of the head and
neck, part 1: physical principles. AJNR Am J Neuroradiol
2009;30:1088-95.
18. Zhang Y, Zhang L, Zhu XR, Lee AK, Chambers M, Dong L.
Reducing metal artifacts in cone-beam CT images by pre-
processing projection data. Int J Radiat Oncol Biol Phys
2007;67:924-32.
19. Katsumata A, Hirukawa A, Okumura S, Naitoh M, Fujishi-
ta M, Ariji E, et al. Effects of image artifacts on gray-value
density in limited-volume cone-beam computerized tomog-
raphy. Oral Surg Oral Med Oral Pathol Oral Radiol Endod
2007;104:829-36.
20. Holberg C, Steinhäuser S, Geis P, Rudzki-Janson I. Cone-beam
computed tomography in orthodontics: benefits and limitations.
J Orofac Orthop 2005;66:434-44.
3/24/10 12:23 PM
 American Journal of Orthodontics and Dentofacial Orthopedics Molen  S135
Volume 137, Number 4, Supplement 1
21. Hatcher DC, Aboudara CL. Diagnosis goes digital. Am J Or-
thod Dentofacial Orthop 2004;125:512-5.
22. Kimpe T, Tuytschaever T. Increasing the number of gray shades
in medical display systems—how much is enough? J Digit Im-
aging 2007;20:422-32.
23. Buzug TM. Computed tomography: from photon statistics to
modern cone-beam CT. 1st ed. Berlin, Germany: Springer;
2008.
24. Deguchi T, Takano-Yamamoto T, Yabuuchi T, Ando R, Roberts
WE, Garetto LP. Histomorphometric evaluation of alveolar
S130-135_AAOPRG_3051.indd 135
bone turnover between the maxilla and the mandible during ex-
perimental tooth movement in dogs. Am J Orthod Dentofacial
Orthop 2008;133:889-97.
25. Verna C, Zaffe D, Siciliani G. Histomorphometric study of
bone reactions during orthodontic tooth movement in rats.
Bone 1999;24:371-9.
26. Frost HM. The biology of fracture healing. An overview for
clinicians. Part I. Clin Orthop Relat Res 1989;248:283-93.
27. Thomas L. Retrospective power analysis. Conserv Biol
1997;11:276-80.
3/24/10 12:23 PM