Key words
spine - CT - surgery
Introduction
Degenerative spine disease often causes multifocal symptoms which may lead to frequent
imaging. Magnetic resonance imaging may not be a reasonable alternative in patients
with prior osteosynthesis due to the susceptibility to artifact disturbance [1]. This can lead to repeated CT imaging with resulting radiation exposure or myelography
examinations with many projections.
Dual energy CT (DE-CT) allows either a reduction of the radiation dose with a comparable
signal-to-noise ratio (SNR) or an increase of the SNR with a similar dose in comparison
to single-energy (SE) acquisitions [2]
[3]. In DE-CT two X-ray tubes with different tube voltages rotate simultaneously. This
technique allows an almost simultaneous and fast image acquisition with the opportunity
to reduce movement artifacts [4]. Additionally a reduction of beam hardening artifacts was described [5]
[6]
[7]
[8]
[9], which leads to a higher image quality especially in patients with prior osteosynthesis.
The different absorptions of X-rays in DE-CT [10]
[11] makes discrimination between different tissues possible [12]. Using appropriate software, tissue which is not of interest can be removed [11]
[13]. Therefore, it is possible to generate a “virtual” myelography from the original
dataset using the volume rendering technique (VRT).
The aim of the current study was to evaluate the different DE-CT datasets with respect
to artifact disturbance and to compare VRT myelography with conventional myelography
regarding nerve root delineation.
Materials and Methods:
Study population
This study was approved by the local ethics committee. Imaging data of 30 patients
(19 female, 11 male) age 32 – 81 years (mean 63 years) who had received prior lumbar
pedicle screw fixation or disc replacements was included and evaluated retrospectively.
Depending on the previous surgery and existing osteosynthesis material, different
numbers of lumbar levels were assessed. In all cases the indication for imaging was
persistent or new radicular complaints with or without back pain.
Myelography and dual energy CT
After intrathecal injection of iodine contrast agent (10 ml Solutrast 250 M®, Bracco Altana Pharma, Konstanz, Germany), the patients underwent X-ray imaging including
lateral, anterior-posterior, and left/right oblique views. Lumbar spine CT imaging
was performed with a DE scanner (Somatom Definition, Siemens, Erlangen, Germany) with
tube voltages of 140 kV and 80 kV (kV / mAS = 140 / 86 and 80 / 468, FOV = 167, collimation = 14 × 1.2,
pitch = 0.55). The computed tomography dose index (CTDI) of the entire scan was 17.49 mGy.
Post-processing
In addition, a virtual monochromatic 120 kV series with a ratio of 0.3 (70 % of the
information from the 140 kV dataset and 30 % from the 80 kV dataset) was generated.
The data was reconstructed with a slice thickness of 1.5 mm, an increment of 1.0 mm,
a medium sharp kernel (D50f) at a level of 400 HE and a center of 70 HE. In post-processing
a 3 D dataset of each tube voltage was archived and virtual myelography VRT datasets
were generated using the software component “dual energy, body bone removal” (syngo
MMWP, VE 32B, Siemens, Erlangen, Germany) in lateral, anterior-posterior, and two
oblique views (45°) comparable to the conventional data.
Artifact evaluation and statistical analysis
The evaluation of artifacts according to the visualization of the spinal canal and
the foramina was performed with a four-point evaluation scale in consensus by two
experienced radiologists. 77 lumbar levels were included for the evaluation of the
spinal canal and 66 for the evaluation of the spinal foramina. Each dataset was evaluated
in a three-dimensional view parallel to the intervertebral disc. For statistical evaluation
a Friedman test was applied (StatView, SAS Institute Inc.©, Version 5.0.1). Statistical
differences with an error probability of p < 0.05 were accepted as significant.
Comparison of VRT and myelography
Data of 202 lumbar spinal nerve roots in anterior-posterior views and 197 roots in
oblique views was included. The VRT dataset was compared to the conventional myelography
with respect to the visualization of the roots in healthy and in fused levels by counting
the number of visible roots for each modality.
Results:
Artifact evaluation
The results of the four-point artifact evaluations are given in numeric data in [Table 1]. With tube voltages of 140 kV, the beam hardening artifacts were least pronounced.
With this voltage most evaluated levels were only slightly disturbed by artifacts
concerning the spinal canal as well as the spinal foramina. In comparison to the other
tube voltages with this voltage most levels with “no artifact disturbance” could be
found. Artifact disturbance was either slight or moderate in more than 50 % of the
evaluated levels in the spinal canal ([Fig. 1a]). The imaging quality was somewhat worse in the spinal foramina ([Fig. 2a]). In the 80 kV series artifacts limited the evaluation severely in most levels of
the spinal canal ([Fig. 1b]) and the spinal foramina ([Fig. 2b]), for both areas in more than 50 % of the evaluated levels. In the evaluation of
the virtual monochromatic 120 kV series ([Fig. 1c], [2c]), a moderately disturbed imaging quality for most levels was found in the evaluation
of the spinal canal. By evaluating the spinal foramina, almost the same number of
levels was slightly, moderately or severely disturbed by artifacts. All these differences
were statistically significant (p < 0.0001).
Fig. 1 Axial views of the spinal canal with 140 kV a, 80 kV b and the virtual 120 kV series c. In this example beam hardening artifacts lead to a slight disturbance in the 140 kV
series, a severe disturbance in the 80 kV series and a moderate disturbance in the
virtual series.
Abb. 1 Axiale Darstellung des Spinalkanals mit 140 kV a, 80 kV b und der virtuellen 120 kV c Serie. In diesem Beispiel führen Aufhärtungsartefakte zu einer geringen Einschränkung
in der 140 kV Serie, zu einer deutlichen in der 80 kV Serie und einer moderaten in
der virtuellen Serie.
Fig. 2 Axial views of the spinal foramina with 140 kV a, 80 kV b and the virtual 120 kV series c in two different patients. Arrows are pointing to the nerve roots. In this example
beam hardening artifacts lead to no disturbance in the 140 kV series, a severe disturbance
in the 80 kV series and a slight disturbance in the virtual series.
Abb. 2 Axiale Darstellung der Neuroforamina mit 140 kV a, 80 kV b und der virtuellen 120 kV c Serie bei zwei verschiedenen Patienten. Die Pfeile zeigen auf die Spinalnerven. In
diesem Beispiel führen die Aufhärtungsartefakte zu keiner Einschränkung in der 140 kV
Serie, zu einer deutlichen in der 80 kV Serie und einer geringen in der virtuellen
Serie.
Comparison of VRT and conventional myelography
In an anterior-posterior view, 162 of 202 (80 %) roots were seen with VRT and 110
of 202 (55 %) with conventional myelography ([Fig. 3] a–c). In oblique views 161 of 197 (82 %) roots were displayed with VRT and 96 of 197
(49 %) with conventional myelography. Due to the opportunity of material removal in
VRT, no overlay problems were present, whereas in conventional myelography overlay
from intervertebral material in the anterior-posterior view and from the fixateur
interne in the oblique views was seen.
Fig. 3 Conventional myelography and DE-VRT in anterior posterior a, lateral b and oblique c views.
Abb. 3 Konventionelle Myelografie und DE-VRT in anterior-posterior a, lateraler b und schräger c Darstellung.
Discussion
In the present setting a low CTDI of 17.49 mGy could be achieved with DE-CT lumbar
spine imaging. In phantom studies and in other organ systems than the spine, a reduction
of dose while maintaining a comparable SNR was described by comparing the 120 kV virtual
monochromatic DE series and 120 kV SE images [2]
[3]
[6]. This was explained by the fact that with an increasing tube voltage, the SNR increases
and the tissue contrast decreases, whereas lowering the tube voltage leads to an increased
tissue contrast and a decreased SNR. DE imaging is supposed to combine these advantages
[2]
[14]. It can either be used for dose reduction while maintaining a similar SNR compared
to SE imaging or to increase the SNR with a comparable dose to SE imaging [2]
[3]. A limitation of the present study is that DE and SE CT myelography studies have
not been compared concerning dose radiation, which could represent a topic for another
study. As the existing studies in several other organ systems comparing both techniques
show a similar or even better SNR [13]
[15]
[16]
[17], this was transferred to the spine.
In the present study the default ratio of 0.3 (12 kV) for the virtual monochromatic
series was applied, which was shown to exhibit a similar SNR compared to SE in the
central SNR [3]. In several other studies other ratios were described to provide a further increase
of imaging quality [18].
In the present study it was shown that beam hardening artifacts are less with tube
voltages of 140 kV than with 80 kV or the virtual monochromatic 120 kV series. This
is in accordance with earlier studies about the evaluation of intracranial aneurysm
clips [5]
[19] or metal implants in different body regions [20]. In the last mentioned study additional high voltage interpolation post-processing
from DE data lead to a further decrease in artifacts. Post-processing modifications
were not the focus in the present study. With monoenergetic extrapolation of DE data,
a reduction of beam hardening artifacts from implants could be achieved, in comparison
to single energy acquisition [7]
[8]. However, in the present study the monochromatic evaluation of 140 kV led to evaluation
of the spinal canal and the spinal foramina with minimal artifact disturbance. The
80 kV series could display advantages in areas where artifact disturbance does not
play a role but a high contrast might be helpful such as in searching for CSF leakages.
Modification of the monochromatic series was not the topic of the present study but
could represent a topic for further investigation.
Virtual myelography showed marked advantages in comparison to conventional myelography
as overlay problems of the fixateur interne in the oblique view and of intervertebral
material in the anterior-posterior view are not present. This post-processing step
reveals the opportunity to shorten conventional imaging with shortened radiation exposure
for patients and fewer painful movements. In addition, the radiation exposure for
the radiologist or technician, who is in the room during data acquisition, is reduced
as well. However, DE VRT cannot replace dynamic conventional studies in patients with
adjacent level instability. In the present study functional conventional imaging was
only performed in some individuals in whom instability was assumed. In patients who
do not need functional conventional imaging, conventional imaging could be dispensed
with.
There are some limitations concerning the imaging evaluation in this study: the applied
four-level scale for the consensus evaluation of artifacts is a common method for
comparing imaging data [5]
[20] but it is a subjective approach. In counting spinal roots in conventional imaging
and VRT myelography, no reviewer blinding concerning the methods was possible since
the images revealed the technique.
In conclusion, DE-CT myelography can be an opportunity to reduce dose, depending on
the designated imaging quality. The minimization of artifact disturbance due to a
high tube voltage and the possibility of “virtual” myelography are significant advantages
in patients with prior osteosynthesis. The presented new methodology may partly replace
conventional imaging because no overlay disturbance from osteosynthesis material is
present.
Clinical relevance:
-
DE CT myelography can be a tool for dose reduction.
-
DE CT myelography combines high contrast imaging and minimal artifact disturbance.
-
DE “virtual myelography” is advantageous after osteosynthesis in comparison to conventional
myelography for the evaluation of spinal nerve roots.
-
DE “virtual myelography” can partly replace extensive X-ray examinations in conventional
myelography.
Table 1
Consensus evaluation of artifact disturbance in the spinal canal and the spinal foramina
with different tube voltages and the virtual series with number of levels assigned
for each scale point.
|
Tube voltage
|
140 kV
|
80 kV
|
120 kV
|
|
4-point scale
|
0
|
1
|
2
|
3
|
0
|
1
|
2
|
3
|
0
|
1
|
2
|
3
|
|
spinal canal
|
number of levels (n = 77)
|
6
|
28
|
23
|
20
|
0
|
6
|
17
|
54
|
0
|
18
|
33
|
26
|
|
spinal foramina
|
number of levels (n = 66)
|
11
|
23
|
16
|
16
|
1
|
10
|
14
|
41
|
7
|
20
|
19
|
20
|
Criteria: no artifacts = 0, artifacts with slight disturbance = 1, artifacts with
moderate disturbance = 2, artifacts with severe disturbance = 3.
