Radiology
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online before print March 27, 2003, 10.1148/radiol.2272011149
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
2272011149v1
227/2/585    most recent
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Baskaran, V.
Right arrow Articles by Finn, J. P.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Baskaran, V.
Right arrow Articles by Finn, J. P.
(Radiology 2003;227:585-592.)
© RSNA, 2003

Technical Developments

Myelographic MR Imaging of the Cervical Spine with a 3D True Fast Imaging with Steady-State Precession Technique: Initial Experience1

Visveshwar Baskaran, BA, F. Scott Pereles, MD, Eric J. Russell, MD, Stavros A. Georganos, MD, Ali Shaibani, MD, Kenneth A. Spero, MD, Elizabeth A. Krupinski, PhD, Al Zhang, PhD and J. Paul Finn, MD

1 From the Department of Radiology, Northwestern University Medical School, 676 N St Clair St, Suite 800, Chicago, IL 60611 (V.B., F.S.P., E.J.R., S.A.G., A.S., K.A.S., J.P.F.); Department of Radiology, University of Arizona Health Sciences Center, Tucson (E.A.K.); and Siemens Medical Systems, Chicago, Ill (A.Z.). Received July 3, 2001; revision requested August 22; final revision received August 13, 2002; accepted September 27. Address correspondence to F.S.P. (e-mail: s-pereles@northwestern.edu).


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 Materials and Methods
Results
 Discussion
 REFERENCES
 
The majority of spinal magnetic resonance (MR) imaging has been performed with spin-echo sequences and spoiled gradient-echo sequences. Advances in gradient MR imaging performance now permit imaging with coherent steady-state sequences. In this study, the authors compare a three-dimensional true fast imaging with steady-state precession sequence with a three-dimensional spoiled gradient-recalled-echo sequence for MR evaluation of the cervical spine in the transverse plane. Initial experience indicates that the steady-state sequence is superior to spoiled gradient-recalled-echo sequences for MR evaluation of cervical spine anatomy and abnormalities.

© RSNA, 2003

Index terms: Magnetic resonance (MR), pulse sequences, 31.121411, 31.121412 • Spinal cord, MR, 31.121411, 31.121412 • Spine, anatomy, 31.92 • Spine, MR, 31.121411, 31.121412


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
Materials and Methods
Results
 Discussion
 REFERENCES
 
Magnetic resonance (MR) imaging has become established as the modality of choice for noninvasive imaging of the cervical spine (1–3) and compares favorably with the reference standard of computed tomographic myelography in sensitivity and specificity (4). Sagittal T2-weighted fast spin-echo sequences generally provide excellent spinal cord and cerebrospinal fluid (CSF) contrast (5–7), but the through-plane motion of CSF renders those techniques useless for transverse high-signal-intensity MR imaging. Gradient-recalled-echo (GRE) sequences lend themselves to greater flexibility in adjusting CSF contrast and flow sensitivity, and T2*-weighted GRE techniques have been used routinely for transverse MR imaging of the cervical spine for many years. The use of three-dimensional (3D) implementations offers advantages over two-dimensional imaging, such as better through-plane resolution and shorter echo times (8,9). To achieve high CSF signal intensity relative to that of the spinal cord (myelographic effect), spoiled GRE sequences must be used with low flip angles and sufficiently long echo times to ensure adequate T2* weighting. The requirement of a low flip angle limits the signal-to-noise ratio (SNR), an effect that is offset to some extent by using narrow readout bandwidths. However, longer readout times in turn are predisposed to artifacts related to CSF pulsation, magnetic susceptibility, and motion in the tracheoesophageal structures. Artifactually narrow intervertebral foramen size can cause errors in the assessment of potential foraminal compromise (10,11).

Steady-state GRE techniques offer substantial advantages over spoiled MR imaging techniques for both SNR and contrast in tissues with long T2*, such as CSF (12,13). The most efficient of these, true fast imaging with steady-state precession (FISP) (TrueFISP; Siemens Medical Systems, Iselin, NJ) (14), is also one of the most sensitive to inhomogeneous magnetic fields and magnetic susceptibility. The degree to which these factors degrade image quality is related directly to the repetition time, and true steady-state imaging has not been practical with the repetition times achievable in the past.

The development of high-performance gradient subsystems (with slew rates approaching 200 mT/m/msec) has refocused attention on the potential of steady-state GRE MR imaging. True FISP imaging uses a gradient profile balanced in section-, read-, and phase-encoding directions, which, when combined with a short repetition time, assumes several desirable properties for imaging the heart and blood pool (15,16). With true FISP, the steady-state signal is related to the ratio of T2 to T1, which generates high contrast between CSF and the spinal cord (12,13). The fully balanced gradient waveform allows both dynamic and static spins to achieve the steady state, which confers relative motion insensitivity and diminishes troublesome effects of CSF pulsation. Previous applications of true FISP, without the benefit of rapid-gradient subsystems and recent improvements in localized shimming, have been associated with characteristic "dark-band" artifacts. These artifacts arise from radio-frequency resonance-offset angle effects due to magnetic field nonuniformities (12,13) and can be addressed by minimizing the repetition time and optimizing the local shim state (12,17).

We propose that the true FISP sequence will, as predicted, have superior contrast properties and speed relative to those of spoiled GRE imaging for transverse evaluation of the cervical spine. The purpose of this pilot study was to compare a 3D true FISP sequence, which requires a short repetition time and an alternating-phase radio-frequency excitation scheme, with a conventional 3D spoiled GRE sequence in the clinical MR evaluation of the cervical spine and spinal cord in the transverse plane.


   Materials and Methods
TOP
 ABSTRACT
 INTRODUCTION
 Materials and Methods
Results
 Discussion
 REFERENCES
 
Patients
The study group comprised 30 consecutive patients (13 men, 17 women; mean age, 44.1 years; age range, 23–85 years) who underwent MR imaging of the cervical spine. Indications for MR examination included clinically suspected radiculopathy or myelopathy associated with pain, paresthesia, or upper extremity weakness in 20 patients; clinically suspected intrinsic spinal cord lesions in five patients; clinically suspected syringomyelia in two patients; and follow-up examination in three patients (one with neurofibromatosis, one with traumatic arm avulsion, and one with cervical fluid collection). Information was available for comparison with relevant clinical and laboratory data in all patients. The institutional review board of Northwestern University approved the study, and informed patient consent was obtained.

MR Imaging Technique
MR imaging was performed with a 1.5-T system (Magnetom Sonata; Siemens Medical Systems) with a maximum gradient amplitude of 40 mT/m and a peak gradient slew rate of 200 mT/m/msec. MR images of the cervical spine were obtained with a phased-array head and neck coil. All patients underwent standard MR examination of the cervical spine as required for evaluation of clinically suspected abnormalities. In cases of suspected radiculopathy, sagittal T1-weighted (repetition time msec/echo time msec, 711/12) and T2-weighted (4,769/132) fast spin-echo sequences were performed in addition to transverse GRE sequences. In cases of clinically suspected myelopathy, additional sagittal short inversion time inversion-recovery, or STIR, (4,200/60) and contrast–enhanced sagittal and transverse spin-echo (664/15 msec) sequences were performed.

All patients were imaged in the transverse plane from the cephalad aspect of C1 to the caudad aspect of T1 by using a conventional spoiled GRE MR sequence with the following parameters: 36/15; flip angle, 5°; number of signals acquired, one; slab thickness, 130 mm; number of partitions, 76–80; effective partition thickness, 1.5–1.7 mm. The image matrix was typically 157 x 256 with a 7/8 230-mm rectangular field of view, resulting in a pixel size of 1.28 x 0.89 mm. The readout bandwidth for the conventional GRE sequence was 81 Hz per pixel, and the total imaging time for this sequence was 7 minutes 12 seconds. Flow compensation was implemented in section-selection and frequency-encoding directions. Prior to use of the true FISP sequence, an automated localized 3D shim was performed in all patients to optimize main field uniformity across the neck.

All patients subsequently underwent transverse true FISP MR imaging from C1 to T1 with a short repetition time in an alternating-phase radio-frequency excitation scheme: 4.8/2.4; flip angle, 50°; number of signals acquired, three; slab thickness, 120 mm; number of partitions, 80; effective partition thickness, 1.5 mm. The typical image matrix for this sequence was 224 x 256 for a 7/8 200-mm rectangular field of view, resulting in a pixel size of 0.78 x 0.78 mm. The readout bandwidth for the 3D true FISP sequence was 980 Hz per pixel, and the total imaging time was 2 minutes 50 seconds.

Image Analysis
For purposes of quantitative analysis, paired conventional GRE and true FISP images were evaluated at C3–4, C4–5, and C5–6 disk levels. All images had the same orientation and were chosen to have the closest correspondence in section position. Circular regions of interest were constructed in the CSF (mean area, 0.81 mm2; range, 0.6–1.0 mm2) and spinal cord (mean area, 0.81 mm2; range, 0.6–1.0 mm2) on the paired images to obtain signal intensity measurements in each region. Background noise for each image was determined by constructing a region of interest (area, 1.00 mm2) in a standardized area outside of the patient’s body. All region of interest measurements were performed by the same observer (V.B.). Contrast-to-noise ratio (CNR) between the CSF and spinal cord was calculated as follows: [(SI of CSF - SI of spinal cord)/SD of background noise], where SI is signal intensity. Because SNR scales linearly with voxel size and the square root of the imaging time, CNRs were normalized for differences in these variables (15,18) by dividing the CNR by voxel size and the square root of the imaging time. Average uncorrected CNR (dimensionless) and corrected CNR (per second per cubic millimeter) were calculated at each level and were compared for the two sequences by using analysis of variance.

For qualitative comparison, paired true FISP and conventional GRE images were evaluated at every cervical level (including the intervertebral disks and disk spaces) from C2–3 to C7–T1 by three neuroradiologists (A.L.S., K.A.S., S.A.G.). The three readers worked independently and were blinded to imaging parameters. Images were evaluated at six cervical levels for each of 30 patients, resulting in a total of 180 image pairs. Overall qualitative ratings (ie, for all spinal levels) were also reported for each image quality parameter assessed. Images were evaluated for spinal cord definition and sharpness, visualization of intradural nerve roots, gray matter–white matter contrast in the spinal cord, vertebral artery visualization, and intervertebral foramen visualization. CSF characteristics, including in-plane uniformity, contrast relative to the spinal cord, and contrast relative to intradural nerve roots, were also evaluated and scored.

Each parameter was rated with a five-point scale (1 = poor and nondiagnostic, 2 = poor but diagnostic, 3 = fair, 4 = good, and 5 = excellent). Disk signal intensity relative to that of bone and CSF was assessed with a seven-point scale (hypointensity [score of -1, -2, or -3], isointensity [score of 0], or hyperintensity [score of +1, +2, or +3]). The presence or absence of artifact was determined at each cervical level, and the severity of artifact was rated with a five-point scale (1 = no artifact, 2 = mild artifact with no compromise in diagnostic quality, 3 = mild artifact with potential compromise in diagnostic quality, 4 = moderate artifact with potential compromise in diagnostic quality, and 5 = severe artifact with compromise in diagnostic quality). Potential compromise in diagnostic quality was defined as artifact outside of the region of interest but with potential to be seen in the region of interest, whereas definite compromise in diagnostic quality referred to artifacts in the region of interest that inhibited accurate diagnosis. Readers were also asked to report the presence or absence of dark-band artifact on true FISP images at all levels.

Interobserver variability was assessed with a Kruskal-Wallis test. All qualitative ratings data were analyzed with a nonparametric Wilcoxon signed rank test. The distribution of data approaches the normal form quite rapidly, so for reasonably sized samples such as those used here, the normal approximation may be used and the normal deviate (z) can be calculated. The Wilcoxon signed rank test ranks the raw data and compares pairs in terms of how they rank (positive and negative differences, accounting for tied values as well). The unit of analysis for the qualitative analysis was 30 patients with six cervical levels rated (n = 180). Some correlation did exist, since we compared the same patients at the same levels with two different image renditions. Thus, use of the nonparametric Wilcoxon signed rank test appeared appropriate. If analysis included the 30 patients alone, it would require that we ignore the levels, thus yielding only general statistics. Means were included to graphically depict differences in the data. All statistical tests were two tailed, and P < .05 was considered to indicate a statistically significant difference.


   Results
TOP
 ABSTRACT
 INTRODUCTION
 Materials and Methods
Results
Discussion
 REFERENCES
 
All true FISP MR examinations consistently resulted in high-quality images, with high SNR and excellent contrast between the CSF and spinal cord. All abnormalities detected with conventional GRE MR imaging were confirmed with use of the true FISP sequence. Final diagnoses were spondylosis and/or degenerative disk disease (n = 20), compressive myelopathy (n = 2), neurofibromatosis type 1 (n = 1), syringomyelia (n = 1), epidural hematoma (n = 1), and arachnoid cyst (n = 1). Four patients had normal cervical MR images.

In all cases of radiculopathy and compressive myelopathy, disk material was better defined and spinal cord contour was better delineated from CSF on true FISP images than on GRE images (Figs 1, 2). In one case of neurofibromatosis type 1 (Fig 3), two neurofibromas extending bilaterally into the intervertebral foramina at C3–4 were better visualized with true FISP imaging (Fig 3). The cystic component of this intramedullary tumor at C5–6 was better visualized with true FISP because of better contrast between intramedullary fluid and the spinal cord (Fig 3). In a case of syringomyelia in a patient with a known Chiari I malformation, contrast between the syrinx and spinal cord was superior with true FISP rather than with conventional GRE imaging (Fig 4). An epidural hematoma that occurred in a patient after traumatic right arm distraction with nerve root avulsion was much better demonstrated on true FISP images as a hyperintense epidural fluid collection with higher signal intensity than that of CSF. On the comparison conventional GRE image, hematoma and CSF were similar in appearance, making the hematoma difficult to delineate fully (Fig 5).



View larger version (87K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1. MR images in a 43-year-old woman with a mild central disk protrusion at C4-5. Left: Transverse spoiled GRE image (36/15) demonstrates indistinct spinal cord contour and poorly defined thecal sac margins. Right: Transverse true FISP image (4.8/2.4) better demonstrates spinal cord contour and thecal sac margins. Note the visible intradural nerve roots as they exit the spinal cord.

 


View larger version (76K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2. MR images in a 40-year-old man with small left paracentral disk protrusion at C5-6. Left: Transverse spoiled GRE image (36/15) shows low SNR and poor contrast between CSF and the spinal cord. Right: Transverse true FISP image (4.8/2.4) shows improved SNR and superior contrast between CSF (arrow) and the spinal cord.

 


View larger version (143K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3a. MR images in a 23-year-old man with postlaminectomy neurofibromatosis type 1 from C2 to T1. (a) Sagittal T2-weighted fast spin-echo image (4,769/132) shows hyperintense fluid collection (pseudomeningocele) (short arrows) and focal hyperintense cystic changes (long arrow) in an otherwise solid mass. (b) Transverse MR images show two neurofibromas extending into the intervertebral foramina bilaterally at the C3-4 level. Left: Spoiled GRE image (36/15) demonstrates distended intervertebral foramina with tumor (arrows). Note overall low SNR and poor image contrast. Right: True FISP image (4.8/2.4) better demonstrates extension of tumor into intervertebral foramina (arrows). (c) Transverse MR images demonstrate the cystic component of the intramedullary tumor at C7 to T1 levels. Left: Spoiled GRE image (36/15) shows poor differentiation between intramedullary CSF and spinal cord. Right: True FISP image (4.8/2.4) better shows the intramedullary cystic component (arrow) due to heightened contrast between intramedullary fluid and spinal cord.

 


View larger version (89K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3b. MR images in a 23-year-old man with postlaminectomy neurofibromatosis type 1 from C2 to T1. (a) Sagittal T2-weighted fast spin-echo image (4,769/132) shows hyperintense fluid collection (pseudomeningocele) (short arrows) and focal hyperintense cystic changes (long arrow) in an otherwise solid mass. (b) Transverse MR images show two neurofibromas extending into the intervertebral foramina bilaterally at the C3-4 level. Left: Spoiled GRE image (36/15) demonstrates distended intervertebral foramina with tumor (arrows). Note overall low SNR and poor image contrast. Right: True FISP image (4.8/2.4) better demonstrates extension of tumor into intervertebral foramina (arrows). (c) Transverse MR images demonstrate the cystic component of the intramedullary tumor at C7 to T1 levels. Left: Spoiled GRE image (36/15) shows poor differentiation between intramedullary CSF and spinal cord. Right: True FISP image (4.8/2.4) better shows the intramedullary cystic component (arrow) due to heightened contrast between intramedullary fluid and spinal cord.

 


View larger version (78K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3c. MR images in a 23-year-old man with postlaminectomy neurofibromatosis type 1 from C2 to T1. (a) Sagittal T2-weighted fast spin-echo image (4,769/132) shows hyperintense fluid collection (pseudomeningocele) (short arrows) and focal hyperintense cystic changes (long arrow) in an otherwise solid mass. (b) Transverse MR images show two neurofibromas extending into the intervertebral foramina bilaterally at the C3-4 level. Left: Spoiled GRE image (36/15) demonstrates distended intervertebral foramina with tumor (arrows). Note overall low SNR and poor image contrast. Right: True FISP image (4.8/2.4) better demonstrates extension of tumor into intervertebral foramina (arrows). (c) Transverse MR images demonstrate the cystic component of the intramedullary tumor at C7 to T1 levels. Left: Spoiled GRE image (36/15) shows poor differentiation between intramedullary CSF and spinal cord. Right: True FISP image (4.8/2.4) better shows the intramedullary cystic component (arrow) due to heightened contrast between intramedullary fluid and spinal cord.

 


View larger version (139K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 4a. MR images in a 55-year-old woman with syringomyelia and known Chiari I malformation. (a) Sagittal T2-weighted fast spin-echo image (4,769/132). Note large syrinx extending from C6 to T3 levels (arrows). (b) Transverse MR images demonstrate intramedullary CSF consistent with syrinx (arrows). Left: Spoiled GRE image (36/15) at the C7-T1 level. Right: True FISP image (4.8/2.4) at the C7-T1 level better demonstrates hyperintensity of intramedullary CSF and better depicts syrinx margins.

 


View larger version (86K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 4b. MR images in a 55-year-old woman with syringomyelia and known Chiari I malformation. (a) Sagittal T2-weighted fast spin-echo image (4,769/132). Note large syrinx extending from C6 to T3 levels (arrows). (b) Transverse MR images demonstrate intramedullary CSF consistent with syrinx (arrows). Left: Spoiled GRE image (36/15) at the C7-T1 level. Right: True FISP image (4.8/2.4) at the C7-T1 level better demonstrates hyperintensity of intramedullary CSF and better depicts syrinx margins.

 


View larger version (88K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 5. MR images in a 34-year-old man with an epidural hematoma that occurred after traumatic right arm distraction with nerve root avulsion. Center: Sagittal T2-weighted fast spin-echo image (4,769/132) demonstrates a ventral epidural fluid collection centered at C6-7 that is consistent with epidural hematoma (arrow). Left: Transverse spoiled GRE image (36/15) at the C6-7 level demonstrates a hyperintense epidural hematoma dorsal to the posterior longitudinal ligament that compresses and displaces the spinal cord. Right: Transverse true FISP image (4.8/2.4) at the C6-7 level better demonstrates the epidural hematoma (arrows) because of superior contrast between the hematoma, spinal cord, and surrounding CSF.

 
Quantitative Analysis
Mean corrected and uncorrected CNRs (with and without adjustment for differences in voxel size and measurement time per section, respectively) are summarized in Figure 6 for conventional GRE and true FISP sequences at C3–4, C4–5, and C5–6. Overall, uncorrected CNR was significantly greater for true FISP than for conventional GRE imaging (t = 12.73, df = 89, P < .0001 [Bonferroni correction used]). For GRE imaging, there was a significant difference in uncorrected CNR between levels when CNR was tested with analysis of variance (F = 9.47; df = 2, 87; P = .002), with values for C5–6 being significantly lower (P = .003 with use of a Fisher protected least-squares difference post-hoc test) than those for both C4–5 and C3–4. Values for C3–4 and C4–5 did not differ significantly.



View larger version (31K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 6. Bar graph shows CNRs for true FISP and spoiled GRE MR sequences at C3-4, C4-5, and C5-6 levels with and without correction for differences in voxel size and imaging time. CNRs for true FISP imaging are significantly greater than those for spoiled GRE imaging at all levels with (P < .05) and without (P < .05) correction.

 
For true FISP, there was also a significant difference in uncorrected CNR between levels when CNR was tested with analysis of variance (F = 3.151; df = 2, 87; P = .04), with values for C3–4 being significantly higher (P = .01 with the Fisher test) than those for C5–6. Although uncorrected CNR at C4–5 was higher than that at C5–6, it did not reach statistical significance. After adjustment for differences in voxel size and section imaging time, the corrected CNRs for true FISP remained significantly higher than those for conventional GRE imaging overall (t = 14.32, df = 89, P < .0001). For true FISP, the progressive decline at lower cervical levels persisted in the corrected CNRs (F = 3.15; df = 2, 87; P = .04), with C3–4 and C5–6 differing significantly (P = .01 with the Fisher test). For conventional GRE imaging (F = 9.48; df = 2, 87; P = .0002), corrected CNRs at C3–4 and C4–5 were not statistically different, but C5–6 had a significantly lower CNR than did C3–4 and C4–5 (P = .0003 with the Fisher test). The falloff in corrected and uncorrected CNRs in the lower cervical spine in both sequences likely arose from a parallel decrease in the receiver coil sensitivity profile in the same area.

Qualitative Analysis
Paired true FISP and conventional GRE images were assessed for the image quality parameters listed previously. Since there was no significant interobserver variability (P < .05) between readers, ratings data for the three readers were pooled. Mean scores for each parameter are summarized in Figure 7. True FISP was rated as significantly better than conventional GRE at all cervical levels, as well as in overall ratings for spinal cord definition and sharpness (z = 13.44, P < .0001) and visualization of the intradural nerve roots (z = 12.93, P < .0001) and vertebral artery (z = 15.87, P < .0001). With regard to CSF characteristics, true FISP was rated as having significantly better CSF signal intensity relative to that of the spinal cord (z = 10.14, P < .0001) and intradural nerve roots (z = 12.93, P < .0001) and also in relation to CSF uniformity (z = 7.42, P < .0001). These comparisons were in favor of true FISP for all spinal levels, as well as for overall ratings for all parameters, with the exception of CSF uniformity ratings at C7–T1, which were not significantly different for the two sequences.



View larger version (43K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 7. Bar graph shows median subjective ratings for true FISP and spoiled GRE imaging for qualitative parameters. Higher scores imply better visualization. True FISP subjective ratings are significantly greater (P < .05) than those for spoiled GRE imaging for all parameters except spinal cord gray matter-white matter contrast and intervertebral foramina visualization. Pooled ratings for intervertebral foramina visualization are not significantly different, whereas pooled ratings are greater with spoiled GRE imaging for spinal cord gray matter-white matter contrast.

 
For visualization of gray matter–white matter contrast in the spinal cord, conventional GRE imaging was rated as superior to true FISP (z = 15.43, P < .0001) at all cervical levels, as well as in overall ratings. For intervertebral foramen visualization ratings, true FISP was rated as superior at C2–3, C3–4, and C4–5 levels, whereas conventional GRE imaging was rated as superior at C5–6, C6–7, and C7–T1 levels.

Signal intensity ratings of disks relative to those of bone demonstrated that with true FISP, disks were rated as isointense or minimally hyperintense, which is significantly different from findings at conventional GRE imaging, in which disks were rated as hyperintense (z = 16.382, P < .0001). Relative to CSF, disks on both true FISP and conventional GRE images were rated as hypointense; however, disks on true FISP images were rated as significantly more hypointense than those on corresponding conventional GRE images (z = 14.24, P < .0001). These findings were demonstrated in ratings for each cervical level, as well as in overall ratings.

Artifacts associated with true FISP images were reported as more severe than those associated with conventional GRE images (z = 6.79, P < .0001). Cervical levels with the greatest differences in artifact ratings were C5–6, C6–7, and C7–T1 (Fig 8). However, the average rating for true FISP artifact for all cervical levels was lower than 2 (mild artifact with no compromise in diagnostic quality). Dark-band artifacts were reported on 22% (117 of 630) of true FISP images, with most occurring at the C5–6 level or lower.



View larger version (87K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 8. MR images at the C7-T1 level in a 24-year-old woman. Left: Transverse spoiled GRE image (36/15) shows superior image quality compared with that of the true FISP image. Right: Transverse true FISP image (4.8/2.4) demonstrates image degradation due to dark-band artifact, which wraps inward from the periphery (arrows) of the field of view. This true FISP image was uniformly rated as "severe artifact with compromise in diagnostic quality" by all three readers.

 

   Discussion
TOP
 ABSTRACT
 INTRODUCTION
 Materials and Methods
Results
 Discussion
REFERENCES
 
The results of this study indicate that MR imaging with a 3D true FISP technique produces significantly better myelographic images of the cervical spine than does a 3D spoiled GRE technique and has significant advantages in SNR and CNR. An additional benefit of the ultrashort repetition times used with the true FISP sequence is the marked improvement in imaging time compared with that of conventional GRE imaging. Imaging times with true FISP were more than 2.5 times faster than those with conventional GRE, with no significant difference in in-plane or through-plane resolution. CNR measurements for the 3D spoiled GRE sequence may be affected adversely by patient motion during the prolonged acquisition time; this limitation is overcome with the 3D true FISP sequence.

The normalized CNRs with 3D true FISP are significantly greater than those with conventional GRE imaging at all cervical levels assessed. The myelographic effect is on the order of 14-fold greater with true FISP imaging than that with conventional GRE imaging. Qualitatively, true FISP images were rated as superior to conventional GRE images for spinal cord definition and sharpness, visualization of intradural nerve roots and vertebral arteries, and CSF characteristics such as uniformity and hyperintensity relative to the spinal cord and intradural nerve roots.

Disk signal intensity on true FISP images was significantly lower in comparison to the relatively hyperintense disks on conventional GRE images. While hypointense disks may be easier to distinguish from hyperintense CSF, they may be more difficult to distinguish from hypointense bone. Therefore, this difference in signal intensity properties of disks may or may not be an advantage, depending on the needs of the referring physician and the abnormality in question (19). Subjectively, disk signal intensity appears to be greater than that previously observed with 3D fast spin-echo MR sequences used in the transverse plane (1,5).

Potential limitations of true FISP observed in this study include intervertebral foramen visualization in the lower cervical spine and intramedullary spinal cord gray matter–white matter contrast. Intervertebral foramen visualization in the upper cervical spine was rated as superior with true FISP, whereas artifacts made it somewhat inferior in the lower cervical spine. However, developments in localized volume shim techniques will likely improve this. With regard to the relationship between gray matter–white matter contrast and sensitivity to intramedullary abnormalities, such as demyelinating disease, a reasonable assessment could not be made because of absence of these diagnoses in the study population. When two cases of spondylotic spinal cord compression were reviewed, abnormal T2 hyperintensity was well appreciated, but visualization required narrower window level settings. This suggests that intramedullary processes can also be detected with true FISP. Further investigation is required to delineate the capabilities and limitations of true FISP in this regard.

The requirement of low flip angle and longer echo times to achieve the myelographic effect is a known limitation of conventional GRE imaging that can result in images with poor SNR and contrast (2,19). Flow compensation methods involve application of additional gradient lobes to reduce the spin dephasing induced by CSF pulsation and thereby improve the myelographic effect. However, the requirement of longer echo times generates increased vulnerability to artifact, particularly due to motion and magnetic susceptibility differences. Efforts to improve contrast on GRE images have resulted in a variety of developments primarily involving magnetization transfer schemes (20,21). Theoretically, saturation of bound water in the spinal cord with an off-resonance magnetization transfer pulse should decrease spinal cord signal intensity and improve contrast between CSF and the spinal cord.

In addition, Melhem et al (21) described a technique with which a magnetization transfer scheme coupled with GRE acquisition allows significant reduction in echo time. This can minimize artifact due to CSF pulsation, motion, magnetic susceptibility, and chemical shift (21). However, one of the limitations of magnetization transfer schemes is the decreased tissue contrast in the paravertebral soft tissues and disk space. Additionally, there is a requisite time penalty to apply magnetization transfer prepulses, which can be somewhat offset by amortizing magnetization transfer prepulses over several acquisition periods (21). Overall, there has been limited clinical evaluation for all of the proposed magnetization transfer GRE sequences.

Steady-state free precession techniques such as true FISP have been known for some time to possess high SNR and excellent contrast characteristics due to T2 and T1 weighting (12,13). Also, the motion insensitivity conferred by the fully balanced gradient waveform partially offsets the effects of CSF pulsation. However, these sequences have been compromised by dark-band artifacts that occur at sites of inhomogeneity in the main magnetic field due to resonance offset effects. These artifacts, which we observed at the lower cervical levels in our patients, can be limited in a number of ways.

One such technique involves acquisition of two separate data sets with 0 and 2{pi} phase offset, with subsequent digital recombination on a pixel-by-pixel basis to exclude banding (12). This technique has been applied to both two-dimensional and 3D acquisitions in the form of the multiecho data image combination, or MEDIC, and 3D constructive interference in steady state, or CISS, pulse sequences. However, 3D CISS constructive interference in steady state limited through-plane resolution in practical imaging times for routine cervical spine imaging, thereby limiting effective evaluation of the intervertebral foramina (22). Primarily, constructive interference in steady state has been used intracranially to visualize structures in the cerebellopontine angle (23). The two-dimensional multiecho data image combination sequence lacks the through-plane resolution necessary for visualization of the intervertebral foramen, and findings in investigations with the 3D multiecho data image combination sequence have not yet been reported.

Shortcomings of the present study include a limited range of abnormalities imaged. However, the conditions described are representative of those encountered in routine clinical practice, and the findings support the assertion that true FISP has excellent contrast properties and image characteristics. Further studies of more diverse disease processes are warranted. A fairly frequent shortcoming of true FISP imaging was dark-band artifact seen in the most inferior portions of the cervical spine images, a region where the spoiled GRE images were noisy but acceptable. This effect is unpredictable on a case-by-case basis, and for routine purposes, we recommend supplementing 3D true FISP imaging with a limited 3D spoiled GRE sequence set to cover the C6–T1 levels. This can be completed in less than 2 minutes and serves as a backup in cases in which true FISP has unacceptable artifact in the lower cervical spine.

In conclusion, 3D MR imaging with a true FISP technique provides enhanced myelographic effect, high structural definition, and excellent tissue contrast compared with those of a conventional 3D spoiled GRE technique. Additionally, the true FISP sequence is fast and relatively insensitive to CSF pulsation.


   FOOTNOTES
 
Abbreviations: CNR = contrast-to-noise ratio, CSF = cerebrospinal fluid, FISP = fast imaging with steady-state precession, GRE = gradient recalled echo, SNR = signal-to-noise ratio, 3D = three-dimensional

Author contributions: Guarantors of integrity of entire study, F.S.P., J.P.F.; study concepts, J.P.F., A.Z., F.S.P., V.B., E.J.R.; study design, J.P.F., F.S.P., V.B., E.J.R.; literature research, F.S.P., V.B., E.J.R.; clinical studies, F.S.P., V.B., S.A.G.; data acquisition, V.B., F.S.P.; data analysis/interpretation, V.B., S.A.G., A.S., K.A.S., E.A.K.; statistical analysis, E.A.K., V.B.; manuscript preparation, V.B., F.S.P.; manuscript definition of intellectual content, editing, revision/review, and final version approval, all authors.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 Materials and Methods
Results
 Discussion
 REFERENCES
 

  1. Georgy BA, Hesselink JR. MR imaging of the spine: recent advances in pulse sequences and special techniques. AJR Am J Roentgenol 1994; 162:923-934.[Abstract/Free Full Text]
  2. Ross JS. Newer sequences for spinal MR imaging: smorgasbord or succotash of acronyms? AJNR Am J Neuroradiol 1999; 20:361-373.[Free Full Text]
  3. Boutin RD, Steinbach LS, Finnesey K. MR imaging of degenerative diseases in the cervical spine. Magn Reson Imaging Clin N Am 2000; 8:471-489.[Medline]
  4. Yousem DM, Atlas SW, Hackney DB. Cervical spine disk herniation: comparison of CT and 3DFT gradient echo MR scans. J Comput Assist Tomogr 1992; 16:345-351.[Medline]
  5. Jones KM, Mulkern RV, Schwartz RB, Oshio K, Barnes PD, Jolesz FA. Fast spin-echo MR imaging of the brain and spine: current concepts. AJR Am J Roentgenol 1992; 158:1313-1320.[Free Full Text]
  6. Melki PS, Mulkern RV, Panych LP, Jolesz FA. Comparing the FAISE method with conventional dual-echo sequences. J Magn Reson Imaging 1991; 1:319-326.[Medline]
  7. Sze G, Merriam M, Oshio K, Jolesz FA. Fast spin-echo imaging in the evaluation of intradural disease of the spine. AJNR Am J Neuroradiol 1992; 13:1383-1392.[Abstract]
  8. Russell E. Cervical disk disease. Radiology 1990; 177:313-325.[Free Full Text]
  9. Tsuruda JS, Norman D, Dillon W, Newton TH, Mills DG. Three-dimensional gradient recalled MR imaging as a screening tool for the diagnosis of cervical radiculopathy. AJNR Am J Neuroradiol 1990; 10:1263-1271.
  10. Czervionke LF, Daniels KL, Wehrli FW, et al. Magnetic susceptibility artifacts in gradient-recalled echo MR imaging. AJNR Am J Neuroradiol 1988; 9:1149-1155.[Abstract]
  11. Tsuruda JS, Remley K. Effects of magnetic susceptibility artifacts and motion in evaluating neural foramina on 3DFT gradient-echo MR imaging. AJNR Am J Neuroradiol 1991; 12:237-241.[Abstract]
  12. Haacke EM, Wielopolski PA, Tkach JA, Modic MT. Steady-state free precession imaging in the presence of motion: application for improved visualization of the cerebrospinal fluid. Radiology 1990; 175:545-552.[Abstract/Free Full Text]
  13. Zur Y, Wood ML, Neuringer LJ. Motion-insensitive, steady-state free precession imaging. Magn Reson Med 1990; 16:444-459.[Medline]
  14. Oppelt A, Graumann R, Barfuss H, Fischer H, Hartl W, Schajor W. FISP: a new fast MRI sequence. Electromedica 1986; 54:15-18.
  15. Carr JC, Simonetti O, Bundy J, Li D, Pereles FS, Finn JP. Segmented trueFISP: a new technique for cine MR angiography of the heart. Radiology 2001; 219:828-834.[Abstract/Free Full Text]
  16. Pereles FS, Kapoor V, Carr JC, et al. Usefulness of segmented trueFISP cardiac pulse sequence in evaluation of congenital and acquired adult cardiac abnormalities. AJR Am J Roentgenol 2001; 177:1155-1160.[Abstract/Free Full Text]
  17. Deimling M, Heid O. Magnetization prepared trueFISP imaging (abstr) In: Proceedings of the First Second Meeting of the Society of Magnetic Resonance Imaging in Medicine. Berkeley, Calif: Society of Magnetic Resonance Imaging in Medicine, 1994; 495.
  18. Kaufman L, Kramer DM, Crooks LE, et al. Measuring signal-to-noise ratios in MR imaging. Radiology 1989; 173:265-267.[Abstract/Free Full Text]
  19. Melhem ER. Technical challenges in MR imaging of the cervical spine and cord. Magn Reson Imaging Clin N Am 2000; 8:435-452.[Medline]
  20. Melhem ER, Benson BL, Beauchamp NJ, Lee RR. Cervical spondylosis: three-dimensional gradient-echo MR with magnetization transfer. AJNR Am J Neuroradiol 1996; 17:705-711.[Abstract]
  21. Melhem ER, Caruthers SD, Jara H. Cervical spine: three-dimensional MR imaging with magnetization transfer prepulsed turbo field echo techniques. Radiology 1998; 207:815-821.[Abstract/Free Full Text]
  22. Georgy BA, Hesselink JR, Joseph PH. 3D CISS in MR imaging of the cervical spine (abstr). Proceedings of the American Society of Neuroradiology 36th Annual Meeting Philadelphia, Pa: American Society of Neuroradiology, 1998; 198.
  23. Casselman JW, Kuhweide R, Deimling M, et al. Constructive interference in steady state-3DFT MR imaging of the inner ear and cerebellopontine angle. AJNR Am J Neuroradiol 1993; 14:47-57.[Abstract]



This article has been cited by other articles:


Home page
 Am. J. Neuroradiol.Home page
D.J. Seidenwurm and for the Expert Panel on Neurologic Imaging
Myelopathy
AJNR Am. J. Neuroradiol., May 1, 2008; 29(5): 1032 - 1034.
[Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
2272011149v1
227/2/585    most recent
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Baskaran, V.
Right arrow Articles by Finn, J. P.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Baskaran, V.
Right arrow Articles by Finn, J. P.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
RADIOLOGY RADIOGRAPHICS RSNA JOURNALS ONLINE