Key words
spinal cord - MR imaging < METHODS & TECHNIQUES - imaging sequences - lesions - artifacts
Introduction
Examination of the spine and spinal cord is one of the major objectives of modern
MR imaging. Despite ongoing advances in sequence technique, imaging of the neck region
and the cervical spine is still demanding due to several reasons. The anatomical structures
are complex and rather small and a variety of artifacts, such as pulsatile flow of
vessels and cerebrospinal fluid (CSF) or swallowing, as well as truncation artifacts
occur in almost all patients [1]
[2]
[3]. Furthermore, many of the patients undergoing MRI of the cervical spine are in a
clinical condition which makes it difficult for them to cooperate sufficiently and
lie still for the duration of the MR examination. Pain, age, general condition as
well as claustrophobic problems sometimes attribute to impaired image quality.
Chemical shift artifacts at the transition between the junction of CSF dura mater
and the epidural fat of the spine obscure the dura layer in conventional TSE or fast
spin echo images. Therefore, this area often harbors difficulties in the differentiation
between epidural and subdural space and impedes lesion detection and location.
Up to now, Periodically Rotated Overlapping ParallEL Lines with Enhanced Reconstruction
(PROPELLER) or its vendor-specific implementation, the BLADE technique, has been proposed
to reduce motion artifacts in uncooperative or pediatric patients mainly for brain
imaging [4]
[5]
[6]
[7]. The sequence is based on a TSE sequence which uses radial k-space coverage in a
rotating and partially overlapping way (“blades”) instead of acquiring parallel k-space
lines as in conventional Cartesian TSE imaging. In cardiac or abdominal imaging this
special way of k-space sampling enables partial compensation of heart or bowel motion
[8]
[9]
[10]
[11]. In diffusion-weighted imaging (DWI), it reduces geometric distortions, diminishes
susceptibility artifacts and increases spatial resolution compared to standard echo-planar
imaging DWI [12]
[13]
[14]
[15]
[16]
[17]
[18]. In most of these studies BLADE or PROPELLER was used in transverse orientation.
Recent studies [19]
[20] showed that by using the BLADE technique for sagittal T2-weighted imaging (T2WI)
of the cervical spine, a significant reduction of several artifacts is possible. Although
the overall image quality and depiction of the spinal cord was superior in the BLADE
sequence, the diagnostic reliability to detect spinal cord lesions could not be judged
due to the low number of spinal cord lesions included.
The aim of this study was to evaluate if in sagittal T2WI of the cervical spine BLADE
is equivalent to TSE regarding the detection of spinal cord lesions and epidural lesions.
We hypothesized that BLADE will be at least equivalent to TSE in detecting even small
and low-contrast cervical spinal cord lesions.
Methods
Patients
We evaluated 33 consecutive patients referred for MRI of the cervical spine at our
institution in this prospective study. The average age of the 19 men and 14 women
was 53 years (age range: 12 – 86 years). The study was approved by the institutional
review board and all patients were included after giving written informed consent.
The MR examinations showed altogether 42 lesions of the cervical spinal cord as well
as 4 in the epidural layer. The findings were judged on the complete clinical MR examination
performed. The patients suffered from demyelinating lesions due to ED (n = 11 with
a total of 22 cervical lesions), myelopathy (predominantly due to degenerative spine
disease) (n = 15 with 17 lesions) and funicular myelosis caused by vitamin B12 deficiency
(n = 3). 4 patients revealed epidural abscess formation due to spondylodiscitis or
parapharyngeal inflammation. The pathologies were all proven by clinical charts, symptoms
and/or follow-up examinations.
MR examination
All examinations of the cervical spine were performed on a 1.5 T MRI scanner (Magnetom
Avanto Siemens Medical, Erlangen, Germany) using a combination of head, neck and spine
array coils. The gradient system had a maximum gradient field strength of 45 mT/m,
and a slew rate of 200 T/m/s.
Sagittal T2-weighted TSE was acquired in all patients as the first sequence and sagittal
BLADE as the second. For T2WI we applied our routine TSE sequence (rectilinear k-space
coverage, head-feet phase encoding direction, flow compensation). Geometrical and
contrast parameters of the BLADE sequence were matched ([Table 1]) resulting in a sequence with identical voxel size and acquisition time for optimized
comparability. The additional motion correction algorithm of BLADE was not used, but
a BLADE-specific high echo-train length and a high readout bandwidth were applied.
A “restore” pulse was applied in both sequences to enable a shorter TE for an increased
SNR and a shorter TR to reduce acquisition time while maintaining sufficient T2 contrast.
In the BLADE sequence cranial and caudal presaturation pulses were applied.
Table 1
Measurement parameters for sagittal T2-weighted TSE and BLADE.
Tab. 1 Messparameter für die sagittale T2-gewichtete TSE und BLADE-Sequenz.
|
TSE
|
BLADE
|
|
TR [ms]/TE [ms]
|
3000/113
|
3000/112
|
|
echo train length
|
17
|
35
|
|
bandwidth [Hz/pixel]
|
140
|
296
|
|
slice thickness [mm]/slice gap [mm]
|
3/0.6
|
3/0.6
|
|
FOV [mm × mm]
|
250 × 250
|
250 × 250
|
|
matrix size
|
384 × 384
|
384 × 384
|
|
phase encoding direction
|
H-F
|
rotating
|
|
oversampling (phase encoding direction)
|
85 %
|
100 %
|
|
number of acquisitions
|
2
|
1
|
|
flow compensation
|
yes
|
no
|
|
acquisition time [min:s]
|
4:17
|
4:20
|
Besides these sagittal T2-weighted sequences, a T2-weighted sagittal short TI inversion
recovery (STIR) and T1-weighted TSE sequences without contrast enhancement as well
as T2-weighted images in transverse orientation were acquired in all patients. Depending
on the pathology, contrast-enhanced T1-weighted images (with or without fat saturation)
in sagittal and transverse orientation were measured additionally.
Image evaluation
Two readers working independently and blinded to the imaging technique as well as
to patient data, medical history or additional MR images compared sagittal BLADE vs.
TSE images in a randomized order. Reader 1 was an experienced neuroradiologist, while
reader 2 was a resident radiologist with 2 years MRI experience. Both readers evaluated
images visually. They scored each case on a scale from 1 to 5 (1: excellent, without
any impairment of image quality 2: good, with only minimal impairment of image quality
3: moderate, showing artifacts which diminished image quality but still diagnostic
4: poor, with severe impairment of image quality and limited diagnostic reliability
5: non-diagnostic, artifacts/alterations are too severe to make a diagnosis) for the
following criteria: Image sharpness, visualization of the dura, diagnostic reliability
of spinal cord depiction as well as lesion depiction within the spinal cord.
Additionally the sagittal BLADE and TSE images of every patient were shown simultaneously
to another two experienced neuroradiologists who assessed them side-by-side for each
patient and selected in consensus the sequence they would prefer for overall image
quality and lesion depiction – TSE, BLADE or no preference. Both observers were identically
blinded to imaging technique and additional data.
The remaining images and clinical data as well as follow-up examinations were included
in the final judgment and classification of the lesions.
Statistical analysis
A statistical analysis was performed using the SPSS software (version 16.0 IBM SPSS
Statistics, Armonk, New York, USA). The sign test was used to compare the results
of the visual evaluation of TSE and BLADE – for each individual reader as well as
for the mean grading of both readers. The χ2 test was used to evaluate the results
of the consensus reading. P-values < 0.05 were regarded as statistically significant
for all tests.
Results
[Table 2] shows a comparison of the grades given by both readers to TSE and the BLADE sequence
for the different criteria. The mean grading of both readers for the BLADE sequence
was significantly superior to that of TSE regarding image sharpness, visualization
of the cervical spine dura and reliability of spinal cord depiction. Concerning image
sharpness and visualization of the dura of the cervical spine, the difference was
also significantly better for each individual reader ([Table 2]). The diagnostic reliability of spinal cord depiction just missed significance for
one reader, but with an advantage for BLADE even in his judgment.
Table 2
Superior sequence or equivalent grading of BLADE and TSE for the different criteria
evaluated by both readers and p-values (sign test).
Tab. 2 Bevorzugte Sequenz bzw. äquvalente Wertung von BLADE und TSE für die verschiedenen
Bewertungskriterien, sowie p-Werte (Vorzeichentest).
|
superior sequence
|
BLADE
|
equal
|
TSE
|
P
|
|
reader
|
1
|
2
|
1
|
2
|
1
|
2
|
1
|
2
|
|
image sharpness
(n = 33)
|
17
|
17
|
12
|
14
|
4
|
2
|
0.007
|
0.002
|
|
spinal cord depiction
(n = 33)
|
14
|
14
|
15
|
14
|
4
|
5
|
0.031
|
0.064
|
|
visualization of dura
(n = 33)
|
33
|
33
|
0
|
0
|
0
|
0
|
< 0.001
|
< 0.001
|
|
lesion depiction
(all lesions, n = 46)
|
17
|
15
|
20
|
20
|
9
|
11
|
0.169
|
0.557
|
|
lesion depiction
(spinal cord, n = 42)
|
13
|
14
|
20
|
18
|
9
|
10
|
0.523
|
0.541
|
|
lesion depiction
(ED, n = 22)
|
9
|
6
|
9
|
12
|
4
|
4
|
0.267
|
0.754
|
|
lesion depiction
(myelopathy, n = 20)
|
4
|
8
|
11
|
6
|
5
|
6
|
1.000
|
0.791
|
Regarding lesion depiction, both sequences showed no significant difference, but a
positive trend towards the BLADE sequence in the individual and mean grading of both
readers for all lesion types ([Table 2]).
As a consequence of the significantly superior visualization of the dura, the detection
of epidural lesions (4/33) was considerably easier with BLADE but the overall number
of epidural pathologies in the study was very small. Therefore, a reasonable statistical
workup of this entity was not possible.
With regard to the reliability of spinal cord depiction, none of the sagittal BLADE
sequences was rated as non-diagnostic. 3 of the TSE examinations received this grading
and were therefore clinically useless ([Fig. 1]). Although the spinal cord was depicted with high reliability, the number of examinations
that were graded as non-diagnostic (grade 5) regarding lesion depiction by at least
one reader was 8 with BLADE and 9 with TSE ([Fig. 2]). Difficulties in the exact delineation of the spinal cord and pathologic signal
were in most cases due to extensive local narrowing of the spinal canal in high-grade
spinal stenosis ([Fig. 3]).
Fig. 1 Sagittal T2-weighted TSE a in this patient is rated as non-diagnostic by both readers due to distinctive artifacts.
Sagittal BLADE imaging b enables sufficient visualization of the cervical spine showing narrowing of the spinal
canal at level C5 – 7 with presumed myelopathy, which could be affirmed in axial T2-weighted
images (not shown).
Abb. 1 Das sagittale T2-gewichtete TSE Bild a dieses Patienten wird von beiden unabhängigen Auswertern aufgrund ausgeprägter Artefakte
als nicht diagnostisch gewertet. Die sagittale BLADE Bildgebung b ermöglicht eine ausreichende Visualisierung der HWS mit Darstellung einer spinalen
Enge auf Höhe HWK 5 – 7 und Verdacht auf lokale Myelopathie, die sich in axialen T2-gewichteten
Bildern bestätigte (ohne Abbildung).
Fig. 2 Sagittal T2-weighted TSE a shows significant artifacts with reduced image sharpness and spinal cord depiction.
A dorsal midline lesion at level C3/4 is misjudged in sagittal T2 TSE a as artificial but is clearly confirmed in BLADE b and axial T2 c. Furthermore, additional lesions at level C5/6 and C7 can be identified in BLADE
and on axial T2 (not shown).
Abb. 2 Sagittales T2-gewichtetes TSE-Bild a mit deutlichen Artefakten und eingeschränkter Bildschärfe sowie Beurteilbarkeit des
Myelons. Eine Läsion in der Mittellinie des Myelons auf Höhe HWK 3/4 wird in der sagittalen
TSE-Sequenz als artifiziell fehlgedeutet, kann aber in der sagittalen BLADE b gut abgegrenzt und in der axialen T2 c verifiziert werden. Zusätzliche Läsionen zeigt die sagittale BLADE auf Höhe HWK 5/6
und HWK 7 die in axialen T2-Schichten (ohne Abbildung) bestätigt werden.
Fig. 3 Sagittal T2-weighted TSE a in good image quality but reduced spinal cord depiction at the level of spinal canal
narrowing at level C3/4 and C6/7 in this patient with a Klippel-Feil segmentation
anomalia involving C5 and C6. In sagittal BLADE b lesion detection and assessment of spinal cord status corresponds to axial T2-weighted
c images showing predominately left-sided myelopathic signal at the level of the herniated
disc C6/7.
Abb. 3 Sagittales T2-gewichtetes TSE-Bild a in guter Qualität aber reduzierter Myelonbeurteilbarkeit auf Höhe der spinalen Stenose
HWK 3/4 und 6/7, bei diesem Patienten mit einer Klippel-Feil-Anomalie HWK 5 und 6.
In der sagittalen BLADE b korrespondiert die Läsionsdetektion und die Abbildung des Myelons gut mit der axialen
T2-gewichteten Sequenz c, die ein linksbetontes Myelopathiesignal auf Höhe des Diskusprolaps HWK6/7 zeigt.
The second part of the visual evaluation, the consensus reading of two experienced
neuroradiologists, showed significant (P< 0.001) advantages for the BLADE technique
concerning overall image quality. BLADE was preferred in 27 of 33 patients, whereas
TSE was favored only in 6 patients with regard to this criterion. For lesion depiction
the side-by-side comparison of both sequences led to a preference for the BLADE sequence
in 10/33 patients and for TSE in 14/33 patients. In 9/33 patients the sequences were
graded equivalent ([Fig. 4]). Statistical evaluation showed no significant difference for this criterion.
Fig. 4 Low-contrast lesion in the cervical myelon at the C4/5 level in a patient with ED.
Sagittal T2-weighted TSE a and sagittal BLADE (4b) are rated as equal by both independent readers and in consensus
reading. Axial T2-weighted images c confirm the low-contrast dorsal, midline lesion
Abb. 4 Läsion der Zervikalmarks auf Höhe HWK4/5 mit sehr geringem Kontrast bei einem Patienten
mit ED. Sagittale T2 TSE a und BLADE b wurden sowohl in der unabhängigen Auswertung, als auch in der Konsensusbewertung
als gleichwertig eingeschätzt. Axiale T2-gewichtete Bilder (4c) bestätigen das Vorliegen
der diskreten Läsion dorsal paramedian im Myelon.
Discussion
MR imaging of the cervical spine and spinal cord is a main element of the modern diagnostic
workup of numerous neurological pathologies including disc herniation, spinal stenosis,
and inflammatory or demyelinizing disease. Thereby T2-weighted sequences in sagittal
orientation produce the most essential images for anatomic overview of the region
and lesion depiction. These images should be free of artifacts and with sufficient
contrast of anatomical details to be able to show even tiny lesions with poor contrast.
Furthermore, the dural layer in the cervical spine region is often impaired by artifacts,
calling for new MR sequence designs to solve these problems. BLADE sequences proved
their value to overcome some of the above named problems in head and spine imaging
in former studies, but clear statements concerning spinal cord lesion detection with
these sequences are not available to our knowledge. To assess the value of the BLADE
technique in comparison to the regular TSE technique in the overall image quality
for the cervical spine, image sharpness as well as visualization of the dura mater
of the cervical spine were judged qualitatively by visual evaluation. Furthermore,
the quality of lesion depiction within the spinal cord and in the epidural space was
evaluated.
Geometric as well as contrast parameters were matched in TSE and BLADE to yield sufficient
comparability of both sequences ([Table 1]). The BLADE sequence was designed with a nearly identical acquisition time to make
it applicable in the daily clinical routine. Specific characteristics of TSE and BLADE
– longer ETL and higher bandwidth in BLADE, flow compensation, head-feet phase encoding
direction and long-term averaging to reduce motion artifacts in TSE – were not transferred
to the other sequence.
In our routine patient collective consisting of mainly cooperative patients and a
few patients with restricted ability to cooperate, motion artifacts were sufficiently
corrected by the rotating k-space coverage in the BLADE technique with its repeated
measurement of central k-space areas, although the dedicated motion correction algorithm
was switched off in this study. This result is in good agreement with prior studies
of the PROPELLER or BLADE technique in MRI of the brain [21]
[22]
[23] and spine [19]
[20]. In the clinical routine motion artifacts often require repeated sequence acquisitions,
thus prolonging the overall time of the examination and impairing patient comfort
and departmental workflow. As detailed in the Results section, the intrinsic motion
correction of BLADE has the potential to save time by avoiding repetition of sequences
thereby improving patient comfort and workflow.
Spinal cord depiction and visualization of spinal cord lesions are the main objective
of most neurological and neurosurgical clinical issues and demand high standards of
MR image quality. However, diagnostic reliability for the depiction of the spinal
cord and spinal cord lesions is influenced by several factors: contrast between the
spinal cord and CSF, motion artifacts including artifacts caused by swallowing and
pulsatile CSF motion, and truncation artifacts ([Fig. 1], [2]). Most of these factors have been evaluated separately in former studies [19]
[20], which showed improved spinal cord/CSF contrast in BLADE by reduced overall motion
artifacts and improved image sharpness. The better and clearer the spinal cord is
visualized, the easier spinal lesions can be identified. Although the results of BLADE
for the diagnostic reliability of spinal cord depiction ([Table 2]) were not always statistically significant for each individual reviewer, a trend
of superior scores for the BLADE sequence as well as the absence of non-diagnostic
BLADE examinations indicate an important advantage compared to TSE.
Besides motion and flow artifacts lesions within the spinal cord can be “masked or
blurred” by low contrast or size of the lesion itself as well as narrowing of the
spinal canal with partial volume effects. The BLADE sequence showed improved or at
least equivalent lesion depiction for all evaluated pathologic entities (MS, myelopathy,
epidural lesions and overall lesion grading) even in very tiny, low-contrast lesions,
which was one of the major objectives in this study.
Visualization of the dura mater at the cervical spine is difficult in TSE sequences
for several reasons. Flow phenomena of the CSF result in local spin dephasing and,
therefore, cause hypointense areas within the CSF, which often disturb proper delineation
of the small linear dural structure. Using the BLADE k-space trajectory, flow phenomena
were significantly reduced compared with the rectilinear trajectory in TSE – similar
to the reduction of flow phenomena or pulsations artifacts seen with PROPELLER or
BLADE sequences in other anatomical regions [21]
[22]
[23].
Even more relevant, chemical shift artifacts at the junction between epidural fat
layer and CSF blur the dural structures to a major extent so that they can rarely
be differentiated in T2-weighted TSE sequences. BLADE with its radial k-space coverage
and a high readout bandwidth eliminates chemical shift artifacts and enables good
visualization of the dural layer at the cervical and upper thoracic spine. In all
examinations BLADE showed the dural layer better than TSE and as a consequence the
differentiation of epidural pathologies was easier with BLADE. Although the number
of epidural pathologies in our study was rather small (n = 4), we would suggest that
BLADE is a useful tool to visualize epidural inflammatory processes and to determine
in which compartment of the cervical spine a pathology is located.
The dedicated visual assessment of BLADE and TSE in all categories was done by two
independent readers with different levels of MRI experience. While reader 1 was an
experienced neuroradiologist, reader 2 was a resident with two years of general MRI
experience. Nevertheless, their independent image evaluations gave similar results
for all criteria in favor of the BLADE technique ([Table 2]). This confirms that T2-weighted BLADE imaging at the cervical spine is a reliable
and robust technique with good agreement to clinically proven T2-weighted TSE sequences
and better overall image quality, which can be judged even with less experience at
a good level of confidence. Compared to shorter TSE sequences used in uncooperative
individuals, BLADE images result in a better overall image quality with good depiction
of spinal discs, bony structures and the spinal cord [19]. We would recommend its clinical use especially in uncooperative patients but even
see the potential to replace TSE in standard imaging protocols at least for the Magnetom
Avanto system.
Despite its prospective design, our study has some limitations: The number of patients
with epidural lesions (4 of 33) was too low to yield a reliable result concerning
the depiction of these lesions. Therefore, the diagnostic value of BLADE for epidural
lesions has to be confirmed in a larger number of patients. Furthermore, randomized
sequence acquisition with an alternating order of TSE and BLADE would be optimal to
overcome the drawbacks of increasing patient motion with prolonged examination duration.
As we evaluated our patients in a clinical setting, we used TSE for first-line imaging
to ensure standard validated clinical imaging. This resulted in the acquisition of
the BLADE sequences 4 to 5 minutes after TSE with potentially increasing patient motion.
A major limitation of this and of most clinical studies concerning intramedullary
lesions is that a true gold standard such as histological verification is missing
and is rarely achieved. Therefore, the clinical workup and history as well as follow-up
examinations are often the only way to substantiate and confirm a suspected pathology.
Finally, all results reported in this study represent the findings on a 1.5 T Siemens
Magnetom Avanto with its specific gradient system, array coils and optimized measurement
parameters. Scanners of different producers or with higher/lower field strengths as
well as vendor-specific sequences (e. g. PROPELLER, MULTIVANE) might yield different
results. Therefore, the results of this study should be validated on the specific
scanner before use.
Conclusion
Sagittal T2-weighted imaging of the cervical spine using the BLADE technique in a
routine patient collective significantly improves overall image sharpness and visualization
of the dural layer compared to TSE. A BLADE sequence with the same spatial resolution
and acquisition time as in an optimized TSE sequence is at least equivalent with respect
to spinal cord depiction and the delineation of even small or low-contrast spinal
cord lesions and more importantly the number of non-diagnostic examinations can be
reduced. We therefore could even recommend BLADE sequences as a routine application
for sagittal T2-weighted imaging of the cervical spine replacing T2-TSE.
