Key words
hand - CT - radiation safety - cone beam computed tomography - CBCT - low dose
Introduction
Cone Beam CT Scanners (CBCT) are compact flat detector computed tomography systems
that have become well established in craniofacial diagnostics over the last 15 years
[1]
[2]
[3]. For the past few years these systems have been sold commercially for imaging of
the extremities [4]
[5]. The primary difference between the CBCT and modern multidetector computed tomography
systems (MDCT) is the detector [6]. The flat detector enables a higher spatial resolution [7], but has a lower dynamic range and is more susceptible to scatter radiation; in
contrast to multi-slice detectors in MDCT it does not have a scatter radiation grid
[8]. For this reason, as well as the significantly higher generator power, a MDCT provides
better soft tissue contrast [9].
There is differing information in the literature with respect to the dose associated
with craniofacial examinations. Many studies present a significantly lower dose in
CBCT protocols compared to MDCT protocols [10]
[11]
[12].
When the two modalities are compared at the same dose however, MDCT provides better
contrast resolution [6]
[9], and to some extent better spatial resolution as well [6].
Due to the smaller volume of the examination object, examinations of the hand produce
less scatter radiation compared to cranial examinations. As this could represent a
benefit when using the flat detector, we asked whether the CBCT was capable of performing
high-quality examinations of the hand at a lower dose than the MDCT.
Materials and Methods
Phantom
The examinations were performed on the cadaver hand from a body donor. This meant
that an additional vote by the Ethics Commission was unnecessary.
Dose calculations
Calculating the effective dose for examinations on the extremities based on equivalent
doses is difficult, because there are no conversion factors for CT examinations of
the extremities, and, in particular none for CBCT examinations. The concept for fluoroscopic
examinations [13] may be applied, but in this case there is only information regarding standard projections
(a. p., p. a., lat.). Thus, the concept may be insufficient for CBCT dose calculations.
However, to make a reliable assertion regarding the applied dose for examinations
of the hand, we determined radiation exposure using Monte Carlo simulation. At the
beginning of this study, models of the systems to be observed in this regard were
created, validated, and calibrated. The uncertainties of the simulation with respect
to validation measurements in a Standard Computed Tomography Dose Index (CTDI) head
phantom were less than 5 %.
The volumes examined contained no organs at risk that had to be included in the calculation
of effective dose. For this reason, the absorbed energy dose in individual anatomical
structures of the hand was used for the comparison. This results in dose values in
mGy instead of mSv, which are not comparable with specifications of effective dose.
They apply only to the comparison of the two scanners examined. The absorbed dose
was calculated in the arm of the ICRP male phantom [14], which was extracted from the data set to simplify matters. The simulations for
the respective KV and mAs combination were performed using GMCTdospp software [9].
CT examinations
MDCT
A 320 detector row MDCT (Aquilion One, Toshiba, Otawara-shi, Japan) was used. Images
were acquired with a 0.5 second 180º rotation and without table movement. The examinations
were performed in standard dose (120 KV und 40 mAs) and every possible combination
of lower KV and mAs; that is, in all combinations of 80 to 120 KV (80 KV, 100 KV,
120 KV) and 5 to 40 mAs (5 mAs, 7 mAs, 10 mAs, 15 mAs, 20 mAs, 30 mAs, 40 mAs). In
total, examinations were performed using 21 different dose settings. Axial slices
were reconstructed with a slice thickness and slice spacing of 0.2 mm each. The image
matrix was 512 × 512 pixels.
CBCT
A CBCT scanner designed for imaging of the extremities (Verity, Planmed, Helsinki,
Finland) was used. Images were acquired through 300 projections during an 18-second
210º rotation without table movement. The examinations were performed in standard
dose (90 KV und 36 mAs) and every possible combination of lower KV and mAs; that is,
in all combinations of 80 to 90 KV (80 KV, 84 KV, 88 KV, 90 KV) and 12 to 36 mAs (12 mAs,
18 mAs, 24 mAs, 30 mAs, 36 mAs). In total, examinations were performed using 20 different
dose settings. Axial slices were reconstructed with a slice thickness and slice spacing
of 0.2 mm each. The image matrix was 801 × 801 pixels.
Semiquantitative image quality determination
The images were sent to the PACS (AGFA Impax 6, Agfa, Mortsel Belgium). The images
were assessed under standard conditions in accordance with DICOM 14 [15]. Five raters, unaware of the modality (three radiologists with 1, 3, and 5 years
experience, and two trauma surgeons with 10 and 18 years experience), performed the
evaluations independently of one another. We used numerous image quality descriptors
for the semiquantitative image quality determination: cortical bone, cancellous bone,
articular surfaces, soft tissue, and artifacts. The raters evaluated the depiction
of the cortical bone, cancellous bone, articular surfaces, and soft tissue in the
examinations using a Likert scale of 1 (very good), 2 (good), 3 (acceptable), 4 (poor),
and 5 (very poor), as suggested by Dehmeri [16]. In addition, the extent of artifacts was evaluated using a Likert scale of 1 (none),
2 (slight, with no impact on diagnosis), 3 (moderate, with no impact on diagnosis),
4 (significant, with impact on diagnosis), and 5 (very significant, with impact on
diagnosis). The sum of evaluations for the depiction of cortical bone, cancellous
bone, articular surfaces, and soft tissue were added up individually for each examination,
and plotted against the dose. If the sum for an individual structure gave a value
< 10, good image quality was assumed regarding depiction of this structure. The sum
of ratings for the artifacts were added up individually for each examination, and
plotted against the dose. A sum of these ratings by the 5 raters of < 10 each was
assumed to denote non-relevant artifacts.
To analyze overall image quality, all 25 evaluations of an examination (5 raters,
each with 5 quality descriptors) were summed and plotted against the dose. A sum < 50
was assumed to denote good overall image quality.
Statistics
Pearson Correlation was used to analyze the correlation between the ratings and the
experience level of the raters [17]. The correlation between the ratings of all 5 raters was analyzed using Kendall’s
W, with correction for ties [18]. P = 0.05 was accepted as the significance level. A Bonferroni P-value correction
was performed to prevent familywise error rates [19]. The statistical analysis took place in R (version 3.0.3).
Results
The calculated average dose of the hand for the various settings ranged from 1.14
(CI 1.11 – 1.17) mGy to 34.36 (CI 33.50 – 35.22) mGy in MDCT, and from 1.76 (CI 1.73 – 1.79)
mGy to 7.15 (CI 6.99 – 7.30) mGy in CBCT. The dose for the standard examination in
MDCT was higher, at 13.21 (CI 12.96 – 13.46) mGy, than in CBCT, which was 7.15 (CI
6.99 – 7.31) mGy ([Fig. 1A]).
Fig. 1 Examples of axial reconstructions at the level of the distal forearm, the distal
carpus and metacarpus in MDCT and CBCT at A default settings and B lowest dose with good overall image quality (level/window, 1000/3300).
MDCT achieved good depiction of the cancellous bone and imaging without relevant artifacts
at a lower dose than was the case with CBCT ([Table 1, ]
[Fig. 2], [3]). In contrast, CBCT achieved good depiction of the cortical bone and articular surfaces
at a lower dose than was the case with MDCT ([Table 1, ]
[Fig. 2]). Neither modality achieved good depiction of soft tissue ([Fig. 2]).
Table 1
Minimal dose of MDCT and CBCT resulting in a good image quality regarding depiction
of cortical bone, cancellous bone, articular surfaces, soft tissues and non-relevant
artifacts.
|
MDCT
|
CBCT
|
|
dose in mGy (confidence interval)
|
dose in mGy (confidence interval)
|
|
cortical bone
|
4.95
(4.86 – 5.04)
|
4.1
(4.03 – 4.19)
|
|
cancellous bone
|
4.54
(4.44 – 4.64)
|
5.42
(5.30 – 5.54)
|
|
articular surfaces
|
2.31
(2.27 – 2.35)
|
2.17
(2.12 – 2.22)
|
|
soft tissue
|
none
|
none
|
|
non-relevant artifacts
|
1.14
(1.11 – 1.17)
|
3.53
(3.47 – 3.59)
|
Fig. 2 Sum of all ratings for A cortical bone, B cancellous bone, C articular surfaces and D soft tissues plotted against the dose for each examination.
Fig. 3 Sum of all ratings for artifacts plotted against the dose for each examination.
Among all examinations with good overall image quality (sum of all evaluations < 50),
MDCT achieved the lowest dose, 4.54 (CI 4.43 – 4.64) mGy with an application of 100
KV and 20 mAs, and CBCT achieved a minimum dose of 5.72 (CI 5.59 – 5.85) mGy with
an application of 90 KV and 30 mAs ([Fig. 1B], [4]).
Fig. 4 Sum of all ratings plotted against the dose for each examination.
The experience of the raters correlated significantly with a poorer assessment of
the MDCT (P = 0.004). In contrast, with CBCT there was no correlation between the
experience level and the assessments by the raters (P = 0.59). The correlation among
all raters equaled 0.43 (P< 0.001).
Conclusion
Although the dose of the standard protocols in the CBCT is lower than in the MDCT,
the MDCT can realize a good overall image quality at a lower dose than the CBCT. Dose
optimization of CT examination protocols for the hand is useful in both modalities,
the MDCT has an even greater potential for optimization.
Discussion
In this study on the comparison of the dose between MDCT and CBCT for examinations
of the hand, we were able to show that MDCT requires less dose to perform an examination
with good overall image quality. CBCT has a lower dose when comparing standard settings.
Good overall image quality using low-dose protocols was possible on both units. This
shows that dose optimization for the protocols makes sense for both modalities, although
there is greater potential for optimization in MDCT. An advantage for the MDCT can
be seen only with reduced-dose imaging protocols.
We used semiquantitative evaluation by raters to evaluate image quality, as this enabled
us to analyze the image quality relevant in the clinical routine. Overall image quality
was determined using a number of descriptors, most of which referred to bones. This
correlates with the clinical requirements regarding non-contrast CT examination of
the wrist [16]
[20].
We selected raters from various disciplines and different levels of experience in
order to better represent the various user groups. This could also be the most important
reason why the correlation between the raters was only moderate. This suspicion is
confirmed by the fact that the experienced evaluators judged MDCT significantly worse.
Apparently the experienced evaluators were used to the standard image quality of MDCT
and evaluated the low dose protocols accordingly worse. That this did not happen with
CBCT as well was not a surprise, as all of the evaluators had limited experience,
approximately 6 months, with CBCT.
For the application of CBCT in the head area and extremities, multiple studies indicate
a lower radiation exposure than with MDCT [11]
[12]
[21]
[22]. However, these studies failed to apply the dedicated dose optimization in MDCT
protocols, which limits the comparability of both modalities. In our study we were
able to show that dose optimization of the examination protocol is both possible and
crucial, particularly for MDCT. We were able to show that it is possible to realize
good overall image quality in MDCT at a lower dose than in CBCT. These data are supported
by previously published studies that determined, when comparing MDCT and CBCT at the
same dose, better contrast resolution in MDCT and an equivalent display of fractures
[6]
[9]
[23].
Similar to our results, in examinations of the midface Hoffmann et. al were able to
show that diagnostically usable scans in MDCT were possible at a dose lower than CBCT
in some cases, depending on the device used [24]. The diverging results between the different studies can be explained by the fact
that CBCT scanners in some cases differ significantly with respect to image quality
and radiation dose. In addition, different dose settings were used in these studies
[6]
[24]
[25].
Looking individually at the image quality descriptors in our study, it is clear that
the greatest differences in quality for both modalities were in the area of artifacts;
MDCT presented significantly fewer artifacts. The susceptibility of CBCT to artifacts
has already been documented and is due in large part to technical reasons [26]. The two modalities are relatively close to one another for the other analyzed structures,
although the CBCT enables a good representation of cortical bone and articular surfaces
at a lower dose than MDCT. This is because CBCT has better spatial resolution than
MDCT, enabling high-contrast structures such as bone to be better presented. For cancellous
bone, MDCT enables a good depiction at a lower dose than CBCT. This is surprising
given that the literature describes a superior depiction of cancellous bone using
CBCT. However, we analyzed low-dose protocols in our study where, due to the reduction
in radiation dose, image noise rises quadratically to dose reduction. It can be assumed
that the iterative reconstruction technique in MDCT better compensates for the increase
in image noise. There is no iterative reconstruction technique currently available
for the model of CBCT scanner used.
It was notable that the dose values of the 80 KV MDCT protocols in some cases were
significantly higher than those of the other examination protocols. The dose increase
at a reduction in ray tube voltage to 80KV in the MDCT can be explained by the corresponding
low-energy spectrum of the X-ray beams, which deposit more dose in the examination
object. The significantly stronger prefiltering of the CBCT scanner explains why the
same effect is not observed with CBCT.
Although a study with good overall image quality
was conducted in lowest dose by MDCT in our
experimental study, a good depiction of cortical
bone and joint surfaces was possible with lower
dose in the CBCT. CBCT also has other advantages, such as higher spatial resolution.
The CBCT has a simpler technical design [1]
[27], which may mean lower production and maintenance costs [28]. In addition, because of its limited examination spectrum operation of the CBCT
scanners is generally easier and faster to learn. However, additional studies are
necessary to confirm this. In our opinion, both modalities are suited to imaging the
hand, but dose optimization in the protocols for both modalities is strongly recommended.
Limitations
The results of our experimental study are limited to the two scanners used. Cadaver
hands were used as examination objects, as we could not justify multiple examinations
on patients. An analysis of sensitivity and specificity of the modalities for specific
pathophysiological changes (such as fractures or erosive changes) and for motion artifacts
therefore was not possible.
Clinical relevance of the study
Both MDCT and CBCT achieve good image quality in dose-optimized protocols. Although
the CBCT applies a lower dose in the standard setting, the lowest dose in an examination
with good overall image quality could be realized with the MDCT. Adapting the examination
protocols in both modalities therefore appears to be absolutely necessary, although
there is greater potential for optimization with MDCT.
