Sie sind auf Seite 1von 14

The British Journal of Radiology, 80 (2007), 307–320

3D MRI in multiple sclerosis: a study of three sequences at 3 T


1 1
R J MILLS, MRCP, C A YOUNG, FRCP and 2E T S SMITH, FRCP

Departments of 1Neurology and 2Neuroradiology, The Walton Centre for Neurology and Neurosurgery, Liverpool, UK

ABSTRACT. The objective of this study was to assess the feasibility of using 3D
acquisition at 3 T for imaging patients with multiple sclerosis (MS). Feasibility was
assessed by three criteria based on acquisition time, specific absorption rate (SAR) and
image quality. 47 patients with clinically definite MS underwent imaging in a Siemens
3T Trio MR scanner. Patient safety data were obtained following the scan sessions. The
study had local ethics approval. The following three-dimensional (3D) sequences, all
acquired coronally, were used: T2 fluid attenuated inversion recovery (FLAIR)
(repetition time (TR) 6000 ms, echo time (TE) 353 ms, inversion time (TI) 2200 ms),
0.560.561 mm voxels, acquisition time 10 min 38 s; T2 turbo spin echo (TSE) (TR
3000 ms, TE 354 ms), 16161 mm voxels, acquisition time 8 min 29 s; T1 inversion
recovery (IR) (TR 2040 ms, TE 5.56 ms, TI 1100 ms), matrix 5126448 (0.560.5 mm
pixels), 0.560.561 mm voxels, acquisition time 7 min 38 s. Total acquisition time was
26 min 45 s. Example images are presented. 3D scanning at 3 T provides highly detailed,
high quality images with acquisition times tolerated by MS patients, even by those with Received 21 July 2006
severe disability. The volumetric data are suitable for a wide variety of post-processing Accepted 17 August 2006
techniques; the authors suggest that 3D studies at 3 T should be considered as the
DOI: 10.1259/bjr/52670770
possible brain imaging protocol for either cross-sectional or longitudinal studies in MS
and that the 3D T2 FLAIR sequence should be considered for the purposes of ’ 2007 The British Institute of
radiological diagnosis. Radiology

The objective of this study was to assess the feasibility perspective, a total scan time which was easily tolerable
of using three-dimensional (3D) volume acquisition with a low incidence of head movement; (2) specific
sequences at 3 T for whole brain imaging in a cross- absorption rate of each sequence less than 3.2 W kg21; 3)
section of patients with multiple sclerosis (MS). a minimum image quality such that lesion conspicuity
was equal to or greater than ‘‘standard’’ [7] diagnostic
two-dimensional (2D) sequences at 1.5 T.
Background 51 patients with clinically definite MS [8] were chosen
at random from a patient database. All subjects gave
The study was set in the context of a diencephalic, MS written, informed consent. Scanning was performed on a
lesion distribution study whose protocol demanded high Siemens 3T Trio machine (Siemens, Erlangen, Germany)
resolution, contiguous slice images of no greater than using a body coil as the transmitting coil and an 8
1 mm thickness. It was surmised that a volume acquisi- channel head coil as the receiving coil. T2 weighted, 3D
tion would be desirable since this would yield a high sequences, largely based on the manufacturer’s recom-
spatial resolution and images could be reconstructed for mended settings, were trialled on the first four patients,
viewing in different planes (particularly useful for adjustments being made in order to overcome such
identifying diencephalic and mesencephalic anatomy). problems as wrap. For the purposes of the study, their
Also, by using a 3 T machine, the high field strength per data were discarded. Both a 3D T2 turbo spin echo (TSE)
se would improve the imaging of deep brain structures and a 3D T2 fluid attenuated inversion recovery (FLAIR)
[1–6]. However, such an imaging protocol would have to sequence were applied to the remaining 47 patients. In
satisfy predefined feasibility criteria (see below) and it order to optimize anatomical understanding of the
was realized that these criteria would also be relevant for diencephalon, a T1 weighted, 3D sequence was added
future studies, and even routine diagnostic imaging. to the protocol and was applied to 40 patients. Two
patients had additional sequences with contrast (0.1 ml
kg21, i.e. half the manufacturer’s recommended dose [9],
Methods and Materials dimeglumine gadopentetate 469.01 mg ml21); 11 also
had a standard 2D T2 TSE, for comparison, on the 3 T
The feasibility criteria were defined as follows: (1) machine and then went on immediately to have another
reasonable acquisition time, defined as a duration of the 2D T2 TSE and a 3D T2 TSE on a Siemens 1.5 T Symphony
total protocol of less than 1 h and, from the patient’s scanner which was located in the same department.
Images from nine normal subjects were also obtained
Address correspondence to: Dr Roger J Mills, Neurology, Walton
using the 3 T system. Post-scan safety data (heart rate,
Centre for Neurology and Neurosurgery, Lower Lane, Fazakerley, respiratory rate and blood pressure) were obtained on all
Liverpool L9 7LJ, UK. E-mail: rjm@crazydiamond.co.uk subjects and any adverse events during the imaging

The British Journal of Radiology, May 2007 307


R J Mills, C A Young and E T S Smith

recorded. For 2 of the 47 patients this was their first Example images
experience of any type of neuroimaging. The study had
both local ethics committee (04/Q1501/154) and research Figure 1 shows an example image, in the plane of
governance committee (Walton Centre for Neurology) acquisition, from each of the three study sequences
approval. (subject is a 57-year-old female, duration of MS 10 years,
Viewing and post-processing of images were per- EDSS 7.5, primary progressive disease).
formed on OsiriX (v1.7.1; http://homepage.mac.com/ There was clear demonstration of both white matter
rossetantoine/osirix/) on a Macintosh computer (http:// disease and high anatomical detail across all sequences.
www.apple.com/) by both a neurologist (RJM) and an On the T2 weighted images, the white matter not
experienced neuroradiologist (ETSS). Selected images containing obvious lesions, which might ordinarily be
were also read from hard-copy. Image quality was classed as NAWM, appeared to have a coarse texture
assessed quantitatively by calculation of signal-to-noise and was not, in fact, thought to be normal (see below).
ratio (SNR) and contrast-to-noise ratio (CNR) on a Figures 2–4 show both acquisition and reconstructed
random subset of the patients and all patients who had plane images for each sequence (subject is a 35-year-old
undergone the 1.5 T imaging. Identical, circular regions female, duration of MS 12 years, EDSS 3.5, relapsing
of interest (ROIs), of 0.05 cm2 area, were placed over the remitting disease).
brightest (or darkest on T1 inversion recovery (IR) The reconstructed images retain high anatomical detail
images) MS lesion, over cerebrospinal fluid (CSF) in and excellent depiction of lesions at the corticomedullary
the lateral ventricle, over normal appearing white matter junction, in the posterior fossa, in the corpus callosum
(NAWM), over normal appearing grey matter (NAGM) and adjacent to CSF spaces.
and outside the body to measure noise. The SNR was The post-contrast images (Figure 5) clearly demon-
calculated as mean signal of tissue divided by standard strate the major vasculature as well as small cortical
deviation (SD) of noise. The CNR was calculated as mean vessels and intraparenchymal vessels in addition to the
signal of lesion minus mean signal of tissue divided by meningeal enhancement normally seen at 3 T. There
SD of noise [10]. Values were taken from one slice and were no examples of an enhancing lesion in the patients
one lesion in each patient. scanned.

Image quality
Results
The SNR and CNR values for all the sequences
used are shown in Table 2. Comparisons were made,
Subject characteristics
by t-test, across the T2 TSE sequences only (grey
The MS patients had the following characteristics: 34 columns).
females, 13 males; mean age 47.0 years (SD 8.1, range 34– There were no differences in either the SNR or CNR
63 years); 5 had primary progressive, 26 had relapsing values between the 3 T 3D and the 1.5 T 2D T2 TSE
remitting and 16 had secondary progressive disease; sequences. The 3 T 2D sequence gave the highest signal
mean duration of MS was 16.9 years (SD 9.4, range 2– and greatest CNRs across all tissues; however, the
40 years); range of disability by Expanded Disability increase in lesion to NAWM CNR was not significantly
Status Scale (EDSS) [11] was 1.5–8.5 (modal values 4.0 better on 3 T 2D than the 3 T 3D sequence (p50.11). The
and 6.0). The normal subjects had the following 1.5 T 3D sequence was poor with SNRs significantly
characteristics: 5 females, 4 males; mean age 41.1 years worse than both the 1.5 T 2D and 3 T 3D for all tissues;
(SD 6.8, range 28–52 years). the lesion to NAWM CNR was approximately 56% less
on 1.5 T 3D than the 3 T 3D sequence.

Scan protocols
Image quality examples: 2D vs 3D
Three 3D sequences were used. All were acquired in
the coronal plane (aligned to the posterior surface of Figure 6 shows a comparison between 2D axial T2 TSE
brain stem): T2 FLAIR (TR 6000 ms, TE 353 ms, TI images at 1.5 T (Figure 6a) and 3 T (Figure 6b) and a
2200 ms), matrix 5126440 (0.560.5 mm pixels), 160 reconstructed image in the same plane from the 3D 3 T T2
contiguous slices of 1 mm thickness, acquisition time TSE sequence, all are from the same subject (a 35-year-
10 min 38 s; T2 TSE (TR 3000 ms, TE 354 ms), old female, duration of MS 13 years, EDSS 4.5, relapsing
matrix 2566192 (161 mm pixels), 160 contiguous slices remitting disease). Both the 2D images are 5 mm thick,
of 1 mm thickness, acquisition time 8 min 29 s; T1 IR the reconstructed slice is 1 mm thick, hence the
(TR 2040 ms, TE 5.56 ms, TI 1100 ms), matrix 5126448 boundaries of the lateral ventricles are much more
(0.560.5 mm pixels), 176 contiguous slices of 1 mm distinct on the thinner slice. Lesions are most conspic-
thickness, acquisition time 7 min 38 s. Total acquisition uous and demarcated on the 3D reconstructed image;
time for the three sequences was 26 min 45 s. simply comparing these three images alone, there are 7
Table 1 shows the setup parameters in detail; included plaques on the 1.5 T 2D slice, 5 plaques on the 3 T 2D slice
for comparison are those for the ‘‘standard’’ 2D axial T2 and 13 plaques on the 3 T 3D reconstructed slice. Again,
TSE on both the 3 T and 1.5 T machines and an the coarse texture of the white matter can be seen on the 3 T
abbreviated 3D sequence (only 96 slices) on the 1.5 T 3D image, especially when compared with either of the 2D
machine. images.

308 The British Journal of Radiology, May 2007


3D MRI in multiple sclerosis

Table 1. Scan parameters, the first four columns (grey) relate to the 3 T three-dimensional (3D) sequences used
Acquisition type 3 T T2 FLAIR 3 T T2 TSE 3 T T1 IR 3 T 2D T2 TSE 1.5 T 2D T2 TSE 1.5 T 3D T2 TSE

3D coronal 3D coronal 3D coronal 2D axial 2D axial 3D coronal

Repetition time, TR (ms) 6000 3000 2040 4000 4070 3000


Echo time, TE (ms) 353 354 5.57 102 99 386
Inversion time, TI (ms) 2200 0 1100 0 0 0
Echo train length 221 197 1 9 11 129
Matrix size 5126440 2566192 5126448 5126456 5126448 2566256
In plane resolution (mm) 0.560.5 1.061.0 0.560.5 0.4360.43 0.4560.45 1.061.0
Slice thickness (mm) 1 1 1 5 5 1
Interslice gap (mm) 0 0 0 1.5 1.5 0
Pixel bandwidth 930 750 130 100 100 780
Percentage phase FOV 85.93 75 87.5 89.1 87.5 100
Number of phase encoding steps 221 197 224 234 231 258
Imaging frequency 123.25 123.25 123.25 123.25 63.68 63.68
Flip angle ( ˚ ) 180 180 8 150 150 150
Number of slices 160 160 176 21 21 96
Acquisition time 10 min 38 s 8 min 29 s 7 min 38 s 1 min 50 s 2 min 51 s 12 min 36 s
SAR (W kg21) 0.1586 0.3191 0.04533 0.5577 0.0036 0.0042

FLAIR, fluid attenuated inversion recovery; TSE, turbo spin echo; IR, inversion recovery; FOV, field of view; SAR, specific
absorption rate.

Image quality examples: 1.5 T vs 3 T 3D an MS lesion. Figure 8 shows a T2 TSE image of a large
area of DAWM off the left occipital horn of the lateral
Figure 7 shows a comparison between a 3D volume ventricle. Also seen within this are bright, punctate
acquisition at 1.5 T (Figure 7a) and 3 T (Figure 7b) Virchow–Robin spaces.
(subject is a 50-year-old male, duration of MS 20 years, Figure 9 shows an example of white matter, magnified
EDSS 6.5, secondary progressive disease). The quality of from the left frontal lobe, of a normal subject (Figure 9a)
the 3 T image greatly exceeds that of the 1.5 T image, and an MS subject (Figure 9b,c) of the same sex and a
largely due to superior CNR (148 vs 65) and SNR (252 vs similar age. There are subtle irregular changes in the
91) (Table 2). There are 6 plaques visible on the 1.5 T brightness of the signal in the white matter of the MS
image and 13 plaques on the 3 T image. patient, which are different from DAWM and which do
The resolving power of the 3 T 3D examination was not appear to be in keeping with vascular structures
such that small structures, e.g. Virchov–Robin spaces or since there are no corresponding low signal changes on
the intraparenchymal vessels seen in the post-contrast the T1 image.
images, in the order of 600–700 mm in width, could be
distinguished. Similarly, some of the smallest MS lesions
detectable were between 0.9 mm and 1 mm in diameter.
The coarseness of the texture on the T2 weighted Normal subjects at 3 T
images of apparently unaffected white matter was either As part of the lesion distribution study, nine normal
due to artefact from the high field strength, or was a real control subjects underwent the same imaging protocol.
phenomenon representing abnormality in the white As mentioned above, the resolution of the images
matter in MS. If it was purely a result of high field allowed the highly vascular nature of the brain par-
strength, then no difference in the texture of white matter enchyma to become apparent with innumerable
would be expected between the MS patients and the Virchow–Robin spaces visible on the T2 TSE with
normal subjects. In order to investigate this, identical corresponding low signal on the T1 IR sequence
circular ROIs of 0.45 cm2 area were placed over lesion (Figure 10). This perivascular high signal was generally
free white matter for each patient and over white matter suppressed with FLAIR. Such small perivascular spaces
for each normal subject and the SD of the signal, on the are not generally visible at 1.5 T, certainly with slice
T2 FLAIR images, recorded. There was a small but thickness greater than 1 mm.
significant increase, by t-test, in the SD of the signal from
the white matter of the MS patients compared with the
controls: mean difference 0.81 (95%CI 0.39–1.23, p,0.01),
Tolerability, safety and adverse events
suggesting that the visible heterogeneity of signal was
possibly related to the MS disease process rather than The procedure was well tolerated by almost all subjects,
artefact. This altered texture was not thought to be even those with severe disability; the most disabled
caused by normal vascular structures since the signal patients participating were of EDSS 8.5, i.e. wheelchair
changes were more subtle and less well defined than the bound with limited upper limb use. Table 3 shows the
discrete punctate appearances typical of Virchow–Robin number of sequences degraded by movement artefact, as
spaces (see below). It was also thought to be different determined by subjective visual assessment of the images.
from what is sometimes known as ‘‘dirty-appearing’’ The sequences were applied in the same order to each
white matter (DAWM) [12], i.e. larger confluent areas of patient, starting with the T2 FLAIR. The overall proportion
increased signal not discrete enough to be described as of sequences which displayed movement artefact was

The British Journal of Radiology, May 2007 309


R J Mills, C A Young and E T S Smith

(a) (b)

(c)

Figure 1. Coronal three-dimensional 3 T (a) T2 fluid attenuated inversion recovery (FLAIR), (b) T2 turbo spin echo (TSE) and (c) T1
inversion recovery (IR) images.

about 7% and in each case, the degradation was classed as complained of transient facial flushing; and one patient
slight. Only one patient had artefact in all three sequences. developed vertiginous symptoms lasting 5 min following
There were three minor adverse events: one patient completion of the scan. Post-scan observations in all
became anxious and aborted the scan just prior to the patients were within acceptable limits, with none of the
termination of the T2 FLAIR sequence, she was subse- subjects exhibiting tachycardia or tachypnoea: mean heart
quently rescanned without complication; one patient rate (HR) 73 beats per min (bpm) (range 61–90 bpm),

310 The British Journal of Radiology, May 2007


3D MRI in multiple sclerosis

(a) (b)

(c)

Figure 2. Three-dimensional 3 T T2 fluid attenuated inversion recovery (FLAIR) orthogonal multiplanar reconstruction (MPR).

mean respiratory rate (RR) 13 min21 (range 10–20 min21), research protocols are often developed on normal
mean blood pressure (BP) 129/80 mmHg (systolic range volunteers who may be able to undergo prolonged
100–169 mmHg, diastolic range 60–106 mmHg). image acquisition without difficulty. This may be
unsuitable for patients with severe neurological disabil-
ity, a consideration which is likely to become more
Discussion
important as the attention of clinical trials in MS turns to
It is important in patient based studies that imaging patients with progressive disease and higher disability.
should be tolerated easily by the subjects. Sequences for In the present study, selection and adjustment of the

The British Journal of Radiology, May 2007 311


R J Mills, C A Young and E T S Smith

(a) (b)

(c)

Figure 3. Three-dimensional 3 T T2 turbo spin echo (TSE) orthogonal multiplanar reconstruction (MPR).

312 The British Journal of Radiology, May 2007


3D MRI in multiple sclerosis

(a) (b)

(c)

Figure 4. Three-dimensional 3 T T1 inversion recovery (IR) orthogonal multiplanar reconstruction (MPR).

The British Journal of Radiology, May 2007 313


R J Mills, C A Young and E T S Smith

(a) (b)

(c)

Figure 5. Three-dimensional 3 T T1 inversion recovery (IR), post-contrast, orthogonal multiplanar reconstruction (MPR) (63-year-
old female, duration of MS 27 years, EDSS 5.0, relapsing remitting disease).

314 The British Journal of Radiology, May 2007


3D MRI in multiple sclerosis

Table 2. Mean SNR and CNR values for each sequence. The T2 TSE (grey) sequences were compared by t-test
3 T 3D T2 FLAIR 3 T 3D T2 TSE 3 T 3D T1 IR 3 T 2D T2 TSE 1.5 T 2D T2 TSE 1.5 T 3D T2 TSE

n 27 27 27 11 7 11

SNR Lesion 172.94 (37.89) 251.51 (65.41) 157.28 (54.29) 337.90 (181.22) 266.49 (96.18) 90.74 (91.88)*{
mean CSF 66.42 (24.09) 359.47 (79.18) 80.11 (23.84) 625.26 (136.10)*{ 371.01 (111.99) 181.32 (116.94)*{
(SD) NAWM 88.76 (24.08) 103.11 (27.86) 215.96 (62.22) 134.56 (24.57){ 121.41 (58.48) 30.37 (43.82)*{
NAGM 111.78 (28.86) 147.32 (33.03) 136.90 (36.28) 192.35 (38.43){ 153.30 (63.49) 38.53 (56.81)*{
CNR CSF 108.78 (31.51) 107.96 (48.44) 77.17 (38.45) 287.36 (179.59)*{ 104.52 (53.61) 90.58 (49.07)
lesion NAWM 84.18 (26.17) 148.40 (52.34) 62.54 (38.15) 224.67 (141.54) 145.07 (55.96) 65.37 (48.86)*{
to:
mean NAGM 61.16 (23.24) 104.19 (51.61) 31.27 (24.12) 181.13 (122.20) 113.18 (46.18) 55.28 (41.43)*{
(SD)

*Difference from 1.5 T two-dimensional (2D) T2 TSE p(0.01.


{Difference from 3 T three-dimensional (3D) T2 TSE p(0.01.
SNR, signal to noise ratio; CNR, contrast to noise ratio; CSF, cerebrospinal fluid; NAWM, normal appearing white matter; NAGM,
normal appearing grey matter; SD, standard deviation; TSE, turbo spin echo.

sequences was deliberately performed on patients from abnormality in the spinal cord of MS patients [19], but
the outset. This provided practical guidance on which not in the brain [20]. Further work is necessary to
scan times might be reasonable for use on subsequent determine whether this texture abnormality has any
patients. The chosen protocol could be completed in clinical or other radiological correlation.
under half an hour; some institutions recommend safety The principal advantage of the 3 T 3D sequences was
guidelines that no patient should be in the bore of a 3 T that a true volumetric data set, of high spatial resolution,
magnet for more than 1 h [13] and so adequate time was acquired providing the opportunity of applying
would have been available for further sequences, if many post-processing techniques. For example, the
required. The protocol was tolerated by all the subjects in ability to both reconstruct and simultaneously view
the study with only a small risk of movement artefact; it images in three orthogonal planes greatly enhances their
was also notable that the noise from the scanner did not diagnostic power, not only in localizing lesions in three
cause distress. The first feasibility criterion of reasonable dimensions but also by resolving questions of partial
acquisition time was satisfied. voluming. For longitudinal studies in MS, great precision
At high field strength and with intensive acquisition, it is required in scan alignment at each examination if
is important that the specific absorption rates of the using 2D sequences, usually prolonging the time spent in
examinations are within safe limits. All sequences were the scanner for the patient, in order to allow co-
within the Medicines and Healthcare products registration of lesion sites; 3D data can be aligned by
Regulatory Agency (MHRA) guideline limits of 4 W post-processing techniques thus reducing the precision
kg21 [14] and also fell below the manufacturer’s guide- required for scan alignment [21]. A typical MRI outcome
line limit of 3.2 W kg21, thus the second feasibility measure in trials for MS therapy is rate of cerebral
criterion of acceptable SAR was satisfied. atrophy [22]; a 3D data set would be an ideal substrate
It was found that, when using volume acquisition with for whole brain volume estimation. Another advantage
voxels of 1.061.061.0 mm or less, both acquisition time of using high field strength MRI is that smaller doses of
and SNR were largely dependent on field strength, as contrast medium are required in order to enhance
expected. At 1.5 T, the 3D sequence image quality was images [23].
relatively poor and the acquisition time for even an Many studies have failed to correlate MS lesion load
abbreviated sequence exceeded that on the 3 T by about with various clinical or biological markers of disease [17,
50%. It was shown that the SNR and CNR of the volume 24, 25]. This may be due to underestimation of lesions
acquisition at 3 T were comparable with the thicker slice, at lower field strengths; there is emerging evidence that
1.5 T 2D axial images, at least for the T2 TSE sequences. 2D imaging at 3 T, compared with 1.5 T, increases
However, many more lesions were visible in the 3 T 3D significantly the detection of MS lesions [26–30], or it
images. By these objective measures, the third feasibility may be due to suboptimal scan geometry [18, 31, 32].
criterion of image quality being equal to or better than Thin, contiguous slice imaging at 3 T may therefore have
1.5 T 2D images was satisfied; certainly, the opinion of considerable impact on studies of lesion load or
both the investigators and other neuroradiologists [15] distribution.
was that the images, subjectively, were of high quality. Some disadvantages of the 3 T 3D sequences were,
Very small structures, i.e. submillimetre, of both anato- first, the data required large amounts of storage space
mical and pathological nature could be discerned. Subtle (typically 190 Mb, including localizer, per patient).
visual abnormalities in lesion free white matter were Second, the 3 T 3D acquisition times were about three
noticeable on the T2 weighted images, something which times longer than individual 2D sequences, of standard
is only normally made apparent by diffusion tensor slice thickness, at 1.5 T, which may be used for diagnostic
imaging or calculation of magnetization transfer [16–18]. purposes. Although not problematic for the patient, this
To the authors’ knowledge, such visually detectable may have an implication for departmental throughput.
texture abnormality in the brain of MS patients has not However, in a research setting, it should be considered
been reported previously. Others have shown texture that the apparent time penalty may be offset by the time

The British Journal of Radiology, May 2007 315


R J Mills, C A Young and E T S Smith

(a) (b)

(c)

Figure 6. (a) 1.5 T two-dimensional T2 turbo spin echo (TSE) axial (5 mm thick, ,7 lesions), (b) 3 T two-dimensional T2 TSE axial
(5 mm thick, ,5 lesions) and (c) reconstructed 3 T three-dimensional T2 TSE (1 mm thick, ,13 lesions) images of the same
patient.

316 The British Journal of Radiology, May 2007


3D MRI in multiple sclerosis

(a) (b)

Figure 7. (a) 1.5 T three-dimensional T2 turbo spin echo (TSE) (best window level shown) (1 mm thick, 6 lesions) and (b) 3 T
three-dimensional T2 TSE, right (1 mm thick, 13 lesions) images of the same patient.

of performing extra 2D sequences for different planes of patient with MS. However, the most common reason for
acquisition and the time required for scan alignment and imaging in MS is to reveal the extent of white matter disease
possible re-scan time following subject movement. and, to this end, the authors felt that the T2 FLAIR sequence
Notwithstanding this, some 2D sequences used in MS gave the greatest information. Not only were lesions well
research are long, e.g. 12 min 18 s for 1.5 T axial T2 demonstrated, but also other abnormality in apparently
FLAIR of 3 mm slice thickness [27]. lesion free white matter was detectable. It was easier to
Each of the three sequences employed in the study distinguish lesions from normal vascular appearances using
provided different pathological and structural information this sequence, compared with the T2 TSE sequence.
which may all have been useful in the assessment of a Therefore, it would be the sequence of choice if necessity
dictated that only one sequence could be used, perhaps for
example in a time constrained, diagnostic setting.
It was not possible to comment on the sensitivity of the
3D 3 T sequences to lesion enhancement. This would
require a separate study and a patient group purposively
chosen for more active disease. However, there is little
reason to assume that it would be significantly worse
than standard 2D imaging [27, 28].

Conclusion
Whole brain volume acquisition at 3 T can produce
very high quality images in a relatively short time and is
entirely feasible for application to MS patients, even
those with severe disability, something which was not
readily achievable at 1.5 T because of excessive acquisi-
tion time and poor image quality. The high spatial
resolution reveals visual abnormalities in the texture of
lesion free white matter on T2 weighted images. 3D
acquisition carries the advantage of allowing all the post-
Figure 8. 3 T reconstructed axial T2 turbo spin echo (TSE)
processing techniques available for volumetric data and
showing a confluent area of diffuse increased signal typical
of dirty appearing white matter (DAWM). Small punctate the authors recommend such sequences for research
areas of higher signal can be seen within this, representing protocols, and suggest that the 3D T2 FLAIR alone is
Virchow–Robin spaces. suitable for the purposes of radiological diagnosis.

The British Journal of Radiology, May 2007 317


R J Mills, C A Young and E T S Smith

(a)

(b) (c)

Figure 9. 3 T reconstructed axial T2 turbo spin echo (TSE) images showing white matter in (a) a normal subject and (b) a multiple
sclerosis (MS) patient. (c) Also shown is the corresponding T1 image for the same MS patient. There appears to be a coarser signal
texture, on the T2 image, in the MS patient which is not thought to be due simply to Virchow–Robin spaces.

318 The British Journal of Radiology, May 2007


3D MRI in multiple sclerosis

(a) (d)

(b) (e)

(c) (f)

Figure 10. Coronal, axial and sagittal (a–c) T2 turbo spin echo (TSE) and (d–f) T1 inversion recovery (IR) images of a normal
subject showing innumerable Virchow–Robin spaces in the white matter.

Table 3. Number of 3 T 3D sequences with movement artefact. 10 out of 134 individual sequences displayed movement artefact
n T2 FLAIR T2 TSE T1 IR

47 47 40

Number of sequences with movement artefact 3 (6.3%) 4 (8.5%) 3 (7.5%)


Cumulative total 3 (6.3%) 7 (7.3%) 10 (7.3%)
FLAIR, fluid attenuated inversion recovery; TSE, turbo spin echo; IR, inversion recovery.

The British Journal of Radiology, May 2007 319


R J Mills, C A Young and E T S Smith

Acknowledgments 17. Miller DH, Thompson AJ, Filippi M. Magnetic resonance


studies of abnormalities in the normal appearing white
The authors would like to thank the patient and matter and grey matter in multiple sclerosis. J Neurol
control subjects for their willing participation in this 2003;250:1407–19.
study. Also, thanks go to the staff at the Magnetic 18. Wayne Moore GR. MRI-clinical correlations: more than
Resonance and Image Analysis Research Centre, inflammation alone–what can MRI contribute to improve
Liverpool University, for their assistance with this work. the understanding of pathological processes in MS? J
Neurol Sci 2003;206:175–9.
19. Mathias JM, Tofts PS, Losseff NA. Texture analysis of spinal
References cord pathology in multiple sclerosis. Magn Reson Med
1. Schwindt W, Kugel H, Bachmann R, Kloska S, Allkemper T, 1999;42:929–35.
Maintz D, et al. Magnetic resonance imaging protocols for 20. Gasperini C, Horsfield MA, Thorpe JW, Kidd D, Barker GJ,
examination of the neurocranium at 3 T. Eur Radiol Tofts PS, et al. Macroscopic and microscopic assessments of
2003;13:2170–9. disease burden by MRI in multiple sclerosis: relationship to
2. Pinker K, Ba-Ssalamah A, Wolfsberger S, Mlynarik V, clinical parameters. J Magn Reson Imaging 1996;6:580–4.
Knosp E, Trattnig S. The value of high-field MRI (3T) in the 21. Prima S, Ourselin S, Ayache N. Computation of the mid-
assessment of sellar lesions. Eur J Radiol 2005;54:327–34. sagittal plane in 3-D brain images. IEEE Trans Med Imaging
3. Vinogradov E, Degenhardt A, Smith D, Marquis R, 2002;21:122–38.
Vartanian TK, Kinkel P, et al. High-resolution anatomic, 22. Bermel RA, Bakshi R. The measurement and clinical
diffusion tensor, and magnetization transfer magnetic relevance of brain atrophy in multiple sclerosis. Lancet
resonance imaging of the optic chiasm at 3T. J Magn Neurol 2006;5:158–70.
Reson Imaging 2005;22:302–6. 23. Trattnig S, Pinker K, Ba-Ssalamah A, Nobauer-Huhmann
4. Bink A, Schmitt M, Gaa J, Mugler JP 3rd, Lanfermann H, IM. The optimal use of contrast agents at high field MRI.
Zanella FE. Detection of lesions in multiple sclerosis by 2D Eur Radiol 2006;16:1280–7.
FLAIR and single-slab 3D FLAIR sequences at 3.0 T: initial 24. Mainero C, Faroni J, Gasperini C, Filippi M, Giugni E,
results. Eur Radiol 2006;16:1104–10. Ciccarelli O, et al. Fatigue and magnetic resonance imaging
5. Mugler JP 3rd, Bao S, Mulkern RV, Guttmann CR, activity in multiple sclerosis. J Neurol 1999;246:454–8.
Robertson RL, Jolesz FA, et al. Optimized single-slab 25. van der Werf SP, Jongen PJ, Lycklama a Nijeholt GJ,
three-dimensional spin-echo MR imaging of the brain. Barkhof F, Hommes OR, Bleijenberg G. Fatigue in multiple
Radiology 2000;216:891–9. sclerosis: interrelations between fatigue complaints, cere-
6. Tubridy N, Barker GJ, Macmanus DG, Moseley IF, Miller bral MRI abnormalities and neurological disability. J Neurol
DH. Three-dimensional fast fluid attenuated inversion Sci 1998;160:164–70.
recovery (3D fast FLAIR): a new MRI sequence which 26. Bachmann R, Reilmann R, Schwindt W, Kugel H, Heindel
increases the detectable cerebral lesion load in multiple
W, Kramer S. FLAIR imaging for multiple sclerosis: a
sclerosis. Br J Radiol 1998;71:840–5.
comparative MR study at 1.5 and 3.0 Tesla. Eur Radiol
7. Simon JH, Li D, Traboulsee A, Coyle PK, Arnold DL,
2006;16:915–21.
Barkhof F, et al. Standardized MR imaging protocol for
27. Nielsen K, Rostrup E, Frederiksen JL, Knudsen S,
multiple sclerosis: consortium of MS Centers consensus
Mathiesen HK, Hanson LG, et al. Magnetic resonance
guidelines. AJNR Am J Neuroradiol 2006;27:455–61.
imaging at 3.0 tesla detects more lesions in acute optic
8. Poser CM, Paty DW, Scheinberg L, McDonald WI, Davis
neuritis than at 1.5 tesla. Invest Radiol 2006;41:76–82.
FA, Ebers GC, et al. New diagnostic criteria for multiple
sclerosis: guidelines for research protocols. Ann Neurol 28. Sicotte NL, Voskuhl RR, Bouvier S, Klutch R, Cohen MS,
1983;13:227–31. Mazziotta JC. Comparison of multiple sclerosis lesions at
9. Data sheet for MagnevistH. Schering, 2006. 1.5 and 3.0 Tesla. Invest Radiol 2003;38:423–7.
10. Kaufman L, Kramer DM, Crooks LE, Ortendahl DA. 29. Bachmann R, Reilmann R, Schwindt W, Kugel H, Heindel
Measuring signal-to-noise ratios in MR imaging. W, Kramer S. FLAIR imaging for multiple sclerosis: a
Radiology 1989;173:265–7. comparative MR study at 1.5 and 3.0 Tesla. Eur Radiol
11. Kurtzke JF. Rating neurologic impairment in multiple 2006;16:915–21.
sclerosis: an expanded disability status scale (EDSS). 30. Wattjes MP, Lutterbey GG, Harzheim M, Gieseke J, Traber
Neurology 1983;33:1444–52. F, Klotz L, et al. Higher sensitivity in the detection of
12. Ge Y, Grossman RI, Babb JS, He J, Mannon LJ. Dirty- inflammatory brain lesions in patients with clinically
appearing white matter in multiple sclerosis: volumetric isolated syndromes suggestive of multiple sclerosis using
MR imaging and magnetization transfer ratio histogram high field MRI: an intraindividual comparison of 1.5 T with
analysis. AJNR Am J Neuroradiol 2003;24:1935–40. 3.0 T. Eur Radiol 2006;16:2067–73.
13. Smith ETS. Guidelines for use of MRI at The Walton Centre 31. Tubridy N, Barker GJ, MacManus DG, Moseley IF, Miller
for Neurology and Neurosurgery. 2005. DH. Optimisation of unenhanced MRI for detection of
14. Medical Devices Agency. Guidelines for magnetic reso- lesions in multiple sclerosis: a comparison of five pulse
nance equipment in clinical use. 2002. sequences with variable slice thickness. Neuroradiology
15. Mills RJ, Young CA, Smith ET. 3D 3T MRI in multiple 1998;40:293–7.
sclerosis. British Society of Neuroradiology, 2005. 32. Hashemi RH, Bradley WG Jr, Chen DY, Jordan JE, Queralt
16. Laule C, Vavasour IM, Moore GR, Oger J, Li DK, Paty DW, JA, Cheng AE, et al. Suspected multiple sclerosis: MR
et al. Water content and myelin water fraction in multiple imaging with a thin-section fast FLAIR pulse sequence.
sclerosis. A T2 relaxation study. J Neurol 2004;251:284–93. Radiology 1995;196:505–10.

320 The British Journal of Radiology, May 2007

Das könnte Ihnen auch gefallen