J Neurosurg Spine 20:220–226, 2014 ©AANS, 2014
Improved magnetic resonance myelography of the lumbar spine using image fusion and volumetry Clinical article Knut Eberhardt, M.D.,1 Oliver Ganslandt, M.D., 2 and Andreas Stadlbauer, Ph.D. 2,3 1 MRI Center of Excellence, District Hospital Castle of Werneck, Germany; 2Department of Neurosurgery, University of Erlangen-Nuremberg, Erlangen, Germany; and 3MR Physics Group, Department of Radiology, Landesklinikum St. Poelten, Austria
Object. The goal of this study was to investigate the usefulness and diagnostic potential of an improved MR myelography approach using a dedicated, heavily T2-weighted, 3D MRI sequence in combination with image fusion. A further aim was to compare dural areas and volumes of normal and pathological altered lumbar spine levels, and accuracy in the assessment of the stenoses, with those obtained by postmyelographic CT. Methods. Fifty patients underwent myelography, postmyelographic CT, and the improved MR myelography approach using a dedicated, heavily T2-weighted, 3D MRI sequence and image fusion. Dural cross-sectional areas and volumes for normal lumbar levels and levels with stenosis were calculated for postmyelographic CT and MR myelography data. The significant differences and Pearson correlations between dural cross-sectional areas and volumes from L1–2 to L5–S1 of postmyelographic CT and MR myelography were analyzed. The 99% CIs for normal and stenotic levels in patients with claudication distances less than 100 meters were calculated. Results. For both dural areas and volumes in normal lumbar levels, the authors found no significant differences and strong correlations between postmyelographic CT and MR myelography. For the lower lumbar levels (L4–5 and L5–S1) they found significantly larger dural areas on MR myelography compared with postmyelographic CT, but not for the upper levels (L2–3 and L3–4). Dural volume analysis revealed significantly larger volumes for MR myelography at all 4 lumbar levels with stenoses in the cohort (L2–3 to L5–S1). Complete separation with no overlap was found between the 99% CIs for normal levels and stenotic levels. Conclusions. Differences in dural areas and volumes in this study may have been caused by the fact that in the case of a severely compressed thecal sac, the viscosity of the intrathecally applied contrast agent is too high in the framework of myelography. The gravitationally dependent component is thus too low to achieve sufficient fluid contrast. An optimized MR myelography approach—a dedicated 3D MR myelography sequence with high spatial resolution in combination with image fusion—is required to achieve a more reliable diagnosis of lumbar spine stenoses, especially with severe compression, compared with postmyelographic CT. This MR myelography approach may be helpful in preventing overestimation of lumbar spine stenoses. The upper limits of 99% CIs for stenotic levels can be interpreted as an indication for surgical treatment. However, further studies that include postoperative outcomes are required. (http://thejns.org/doi/abs/10.3171/2013.10.SPINE13119)
Key Words • lumbar • stenosis • computed tomography • image fusion • magnetic resonance myelography • dural volume
R
compression syndromes play a central role in the pathophysiology of spinal diseases. Symptoms of pain syndromes are often atypical and the assignment to dermatomes may be difficult or deficient. Despite the substantial progress in medical imaging, discrepancies between symptoms and imaging findings may be significant. Patients with large space-occupying lesions frequently have minor symptoms, whereas others show major symptoms without morphological findings on imaging;21,24 this was confirmed by results from a study using an animal model.19 Improvements in sectional imaging techniques have to produce a reduction of the disadicular
Abbreviation used in this paper: HASTE = half-Fourier acquisition single-shot turbo spin-echo.
220
crepancies between symptoms and morphological imaging findings. Lumbar myelography was accepted as the gold standard of detecting radicular compression syndromes for a long time. However, the diagnostic performance of this invasive method is limited,9,23 and undesirable side effects are not uncommon, such as an allergic reaction to the contrast agent or infections.14 The invention of CT allowed for combining conventional myelography with a digital sectional imaging technique (postmyelographic CT). Comparative studies by Raininko22 and Kampmann et al.11 demonstrated the advantages of this technique. Magnetic resonance imaging, a method with excellent soft-tissue contrast, was initially used for diagnosis of intraspinal masses, especially for neuronal compression J Neurosurg: Spine / Volume 20 / February 2014
Improved magnetic resonance myelography symptoms and tumors.10,25 In addition, MRI has the advantage that no ionizing radiation is used for generation of the images, and the imaging orientation can be chosen in any direction. The development of heavily T2-weighted MRI sequences allowed for depicting the dural sac threedimensionally with high spatial resolution and without artifacts due to pulsatile blood flow. Magnetic resonance myelography has the ability to visualize dural structures in a manner comparable to conventional myelography and postmyelographic CT.8 Furthermore, an isotropic 3D MR myelography data set can be reconstructed in any direction of space later by using multiplanar reconstruction. A crucial and well-known disadvantage of heavily T2weighted MR myelography sequences is the associated loss of information about the anatomical vicinity of the dural sac, which is compounded by the fact that the preferred phase encoding direction is in the coronal orientation to achieve optimal image quality. The purpose of this study was to determine whether an improved MR myelography approach using a dedicated, heavily T2-weighted MRI sequence and image fusion could demonstrate similar cross-sectional images of the dural sac area and volumes comparable in size to those detected by postmyelographic CT. Study Population
Methods
Fifty patients with suspected lumbar spinal stenosis and referred for lumbar myelography and postmyelographic CT were also evaluated with our improved MR myelography protocol. There were 15 women and 35 men, ranging in age from 36 to 82 years old (mean ± SD = 61.4 ± 12.0 years). All patients suffered from lower back pain and had neurological symptoms (radicular deficits and/or neurogenic claudication [claudicatio spinalis]). The procedures were explained to the patients and written informed consent was obtained from all patients before the procedures as required by the local institutional review board.
Myelography and Postmyelographic CT
Conventional myelography was performed while patients were prone. The nonionic contrast agent (10 ml Isovist 300 and 5 ml Isovist 240; Schering AG) was applied intrathecally at the levels where the stenoses were assumed. To achieve optimal distribution of the contrast agent in patients with an incomplete allocation of the agent within the range of the stenosis, an abdominal press, the flexion (bent forward) sitting position, and waiting periods of up to 1 hour were applied. To display the lumbar region, 3 standard projections were obtained with anteroposterior, lateral, and left anterior oblique or right anterior oblique orientation. In patients with suspicion of spinal stenosis and an incomplete or complete contrast block on conventional myelography, a postmyelographic CT examination was additionally performed within the following 2 hours using a multislice spiral CT scan (Volume Zoom; Siemens). Contiguous with the stenotic level of the myelography in the postmyelographic CT examination, the subsequent upper and lower levels were examined. The examination was performed while the patient was prone, with breath-holding.
J Neurosurg: Spine / Volume 20 / February 2014
Magnetic Resonance Myelography
Magnetic resonance imaging examinations were performed within 2 days after myelography on a 1.5-T wholebody MRI machine (Sonata; Siemens) equipped with a 6-channel spine array coil, with the patient supine. For MR myelography we used a dedicated MR sequence, a 3D heavily T2-weighted half-Fourier acquisition single-shot turbo spin-echo (HASTE) sequence with the following parameters: TR 10,000 msec, TE 165 msec, slice thickness 0.5 mm, flip angle 180°, voxel size 0.5 × 0.5 × 0.5 mm3, matrix size 256 × 256, and FOV 120 × 120 mm2. Orientation of the volume of interest, as well as phase encoding direction, was coronal. The acquisition time of this dedicated MR myelography sequence was 10 minutes. The sequence scheme included frequency-selective fat suppression and coronal saturation pulses to suppress motion artifacts due to breathing. Additionally, “partial Fourier” was used (that is, 4/8 of the k-space was sampled) to suppress anatomical structures in the vicinity of the CSF and the dural sac. In so doing, the central structures remained unaltered and fully visible, whereas the contrast resolution decreased with increasing distance to the central structures. However, the advantage of an enhanced CSF signal comes with the drawback of loss of anatomical information in the HASTE data. A sagittal T2-weighted turbo spin echo sequence was used as an anatomical-basis data set, an MR data set with anatomical information that was used as the basis data set for image fusion of the MR myelography. The parameters of the sequence were TR 4000 msec, TE 120 msec, slice thickness 2 mm, FOV 250 × 250 mm2, matrix size 256 × 256, and acquisition time 2.5 minutes. Image Data Postprocessing and Evaluation
Coronal sections were calculated from the postmyelographic CT data using multiplanar reconstruction. For each lumbar vertebra the midline of the intervertebral disc was defined as a sectional plane on these images. From this axial section plane along the midline of the intervertebral disc, 4 sectional planes above and 4 sectional planes below were defined in parallel to the central plane, with a section thickness of 2 mm each. Multiplanar reconstruction resulted in 9 axial slices around the intervertebral disc. To determine exact anatomical information within the MR myelography data, the 3D HASTE data sets were fused to the corresponding sagittal T2-weighted turbo spin echo data set. This procedure was performed on a commercially available workstation (Leonardo; Siemens) using the “3D fusion” tool. Section planes along the midline of the intervertebral disc, in the same orientation as described for the postmyelographic CT and multiplanar reconstruction, were used for reformatting to obtain 9 axial slices around the intervertebral disc. The slice thickness of these axial MR myelography images was also 2 mm. The dural area was manually defined on the 9 axial images of both the postmyelographic CT and MR myelography data in an interdisciplinary manner, using an experienced neuroradiologist, an experienced neurosurgeon, and an experienced medical physicist in consensus. The values of the dural area for the central axial image plane were used for comparisons of dural cross-sectional areas 221
K. Eberhardt, O. Ganslandt, and A. Stadlbauer between postmyelographic CT and MR myelography. The area values of all 9 image planes were used for calculation of the dural sac volume. Figure 1 shows the postprocessing steps of both the postmyelographic CT (Fig. 1 upper) and MR myelography (Fig. 1 lower) data for a normal lumbar level: multiplanar reconstruction of the 9 axial slices (Fig. 1 left column), and segmentation of the dural areas (Fig. 1 right column). Statistical Methods
Data were analyzed using SPSS statistical software (version 14.0; SPSS Inc.). To compare differences in dural area and volume between postmyelographic CT and MR myelography, we used a 2-sided paired Student t-test for all levels combined, and a Wilcoxon signed-rank test for individual lumbar levels. Comparisons of dural area and volume in normal levels and in levels with stenosis were performed using a 2-sided Mann-Whitney U-test. The 99% CI was calculated for levels with no pathological changes and stenotic levels in patients with claudication distances less than 100 meters. Pearson correlation coefficients were calculated for measuring relationships between dural area and volume assessed with postmyelographic CT and MR myelography. Correlation coefficients were interpreted under consideration of the specifications given by Zou et al.27 For all tests, the level of significance was set at p < 0.05.
Results
A total of 180 lumbar levels (L1–2 to L5–S1) were evaluated in the 50 patients (mean 3.6 levels per patients, range 2–5 levels) who were suspected to suffer from a stenosis of the lumbar spinal canal. No pathological changes were found on 88 levels in 41 patients. Eleven normal areas were found at L1–2, 19 at L2–3 and L3–4, 15 at L4–5, and 24 at L5–S1. Spinal canal stenoses were detected on 92 levels in 41 patients (31 patients with a claudication distance < 100 meters). Nine patients showed no lumbar spinal canal stenoses. No stenoses were detected at L1– 2. Twelve areas with stenoses were found at L2–3, 25 at L3–4, 34 at L4–5, and 21 at L5–S1. Figure 2 shows the postprocessing results of both the postmyelographic CT (Fig. 2 upper) and MR myelography (Fig. 2 lower) data for a patient with stenosis at L3–4.
Fig. 1. The preparation for multiplanar reconstruction (left column) and segmentation of dural areas (right column) on postmyelographic CT (upper) and improved MR myelography using a fused HASTE and T2-weighted turbo spin echo data set (lower) on a normal lumbar level (L3–4) in a 43-year-old man. The volume of interest for the multiplanar reconstruction is indicated by the black frames. The 9 axial slices are represented by black lines. The red outlines indicate the segmented dural areas.
Comparison and Correlation of Dural Areas
The mean value of the 88 normal dural areas on postmyelographic CT was 169.8 ± 47.6 mm2, and on MR myelography was 170.8 ± 46.8 mm2. These mean values were not significantly different according to a 2-sided paired Student t-test. Analyzed according to lumbar level, we found no significant differences in normal dural areas between postmyelographic CT and MR myelography for all levels (L1–2 to L5–S1). The areas on MR myelography were slightly larger, ranging between 0.0 (0%, at L1–2) and 1.9 mm2 (1.2%, at L4–5) larger (Fig. 3 upper). The mean dural area of the 92 levels with stenosis on postmyelographic CT was 83.2 ± 34.0 mm2 and on MR myelography was 94.0 ± 33.1 mm2. These mean values were significantly different (p = 0.001). Analyzed accord222
Fig. 2. Multiplanar reconstruction (left column) and segmentation of dural areas (right column) on postmyelographic CT (upper) and improved MR myelography using a fused HASTE and T2-weighted turbo spin echo data set (lower) on a lumbar level (L3–4) with stenosis in an 82-year-old man. The volume of interest for the multiplanar reconstruction is indicated by the black frames. The 9 axial slices are represented by black lines. The red outlines indicate the segmented dural areas.
J Neurosurg: Spine / Volume 20 / February 2014
Improved magnetic resonance myelography values were not significantly different. Regarding different lumbar levels we found no significant difference in normal dural volumes between postmyelographic CT and MR myelography for all levels. The volumes on MR myelography were slightly larger, ranging between 3.5 mm3 (0.1%, at L1–2) and 46.4 mm3 (1.5%, at L5–S1) larger (Fig. 4 upper). The mean dural volume of the 92 levels with stenosis on postmyelographic CT was 1794.1 ± 600.5 mm3, and on MR myelography was 1968.0 ± 552.4 mm3. These mean values were also significantly different (p = 0.001). Analyzed according to lumbar level, we found significant differences between postmyelographic CT and MR myelography for all levels (Fig. 4 lower): L2–3 (difference of 115.7 mm3, 5.8%; p = 0.041), L3–4 (difference of 172.6 mm3, 8.9%; p = 0.03), L4–5 (difference of 189.9 mm3, 9.3%; p = 0.001), and L5–S1 (difference of 183.1 mm3, 9.9%; p = 0.021). As with normal areas, Pearson correlation revealed a strong correlation between normal dural volumes on postmyelographic CT and MR myelography for the entirety of levels as well as for all individual levels. For levels with stenosis we found a moderate correlation for the entirety of levels and for all individual levels (Table 1). Comparison of Normal Levels and Levels With Stenosis
As expected, both dural areas and volumes for postmyelographic CT and MR myelography were significantly different (p < 0.001 each) between normal levels and levels with stenosis using a Mann-Whitney U-test. The 99% CIs for dural areas and volumes of normal and stenotic levels (in patients with claudication distances < 100 meters) are shown in Table 2.
Discussion
Fig. 3. Comparison between postmyelographic CT and MR myelography of the dural sac cross-sectional area for normal lumbar levels (upper) and lumbar levels with stenosis (lower). Asterisk = significant differences (p < 0.05) between mean values.
ing to lumbar level, we found significant differences between postmyelographic CT and MR myelography for the levels L4–5 (difference of 14.9 mm2, 15.7%; p = 0.002) and L5–S1 (difference of 13.4 mm2, 13.4%; p = 0.033), but not for L2–3 (difference of 3.5 mm2, 4.0%) and L3–4 (difference of 6.4 mm2, 7.0%; Fig. 3 lower). Pearson correlation revealed a strong correlation between normal dural areas on postmyelographic CT and MR myelography for all levels combined, as well as for all individual levels. However, for levels with stenosis we found a moderate correlation for the entirety of levels, L4– 5, and L5–S1, a strong correlation for L3–4, and a weak correlation that did not reach significance at L2–3 (Table 1). Comparison and Correlation of Dural Volumes
The mean volume of the 88 normal dural levels for the postmyelographic CT was 3141.1 ± 866.0 mm3 and for the MR myelography was 3187.5 ± 828.2 mm3. These mean
J Neurosurg: Spine / Volume 20 / February 2014
Several studies in the literature have compared conventional myelography, postmyelographic CT, and MR myelography. Kuroki et al.15 investigated 40 patients with symptoms of lumbar nerve root compression using conventional and MR myelography. They found that both depiction of intradural structures and image quality were comparable. Studies by Eberhardt et al.5,6 even demonstrated advantages of MR myelography compared with conventional myelography in combination with postmyelographic CT, particularly in the detection of extreme lumbar spinal canal stenoses. In a cohort of 275 patients, Ferrer et al.7 found that MR myelography improved diagnostic accuracy in 16.7% of cases. Studies from Bartlett et al.,1 Birchall et al.,2 Melhem,20 and Wang et al.26 supported these findings. Selection of the appropriate MR sequence technique is essential. Gradient echo sequences were primarily used for heavily T2-weighted MR sequences thus far.25 However, these sequences are prone to susceptibility artifacts, especially in structures with varying proton density.17 This drawback can be reduced by using spin echo sequences,3,18 which additionally have the advantage of a higher signal-to-noise ratio.17 In this study, we went one step further and used an improved 3D turbo spin echo sequence (3D HASTE) in combination with “partial Fourier” (collecting only the 223
K. Eberhardt, O. Ganslandt, and A. Stadlbauer TABLE 1: Correlation of dural areas and volumes from postmyelographic CT and MR myelography* Lumbar Level Dural Area/Volume
All
L1–2
L2–3
L3–4
L4–5
L5–S1
normal areas areas w/ stenosis normal volumes volumes w/ stenosis
0.975‡ 0.582‡ 0.966‡ 0.733‡
0.992‡ NA 0.968‡ NA
0.942‡ 0.333 0.991‡ 0.676†
0.984‡ 0.772‡ 0.985‡ 0.654‡
0.987‡ 0.562‡ 0.992‡ 0.819‡
0.980‡ 0.594‡ 0.957‡ 0.796‡
* All values are correlation coefficients. NA = not applicable. † Significant correlation at p < 0.05. ‡ Significant correlation at p < 0.001.
central parts of the k-space), and frequency selective fat suppression, to improve background contrast and avoid signal overlapping from adjacent structures, respectively. Additionally, this strategy is associated with a reduction of breathing artifacts. We found no significant differences in normal lumbar levels and strong correlations for both dural areas and volumes between postmyelographic CT and MR myelography. The lower lumbar levels (L4–5 and L5–S1) showed significantly larger areas on MR myelography when compared with postmyelographic CT, but not for the upper levels (L2–3 and L3–4). Dural volume revealed significantly larger values for MR myelography at all 4 lumbar levels with stenoses in our cohort (L2–3 to L5–S1). Both dural area and volume of levels with stenoses showed no to minor correlation between postmyelographic CT and MR myelography. Furthermore, the upper limits of the 99% CIs for dural areas and volumes for stenotic levels in patients with claudication distance of less than 100 meters (Table 2) can be interpreted as limit values and an indication for surgical treatment. Our values for the cross-sectional areas of the dural sac of normal lumbar levels are in agreement with previously published findings,16 although their values are tendentially lower. The authors investigated the effect of body position and axial load on dural sac cross-sectional area using conventional MRI, with the result that horizontal MRI with the patient supine and the legs straightened was comparable to vertical MRI, whether axial compression was added or not. This finding stands in marked contrast to the findings of Kanno et al.12,13 These discrepancies might be explained by the differences in the conventional MR protocols (TR, TE, slice thickness, and others) used in the studies, and that only dural areas but not dural volumes were evaluated. Although it is more laborious, the determination of dural volumes has the advantage of allowing for compensation of even minor differences and inaccuracies in slice positioning and orientation. The usefulness and advantages of dedicated MR myelography sequences have been demonstrated previously. Dorenbeck et al.4 compared a T2*-weighted, 2D, spoiled gradient echo multiecho sequence with magnetization transfer saturation pulse, with lumbar myelography and postmyelographic CT. They found that this gradient echo sequence was as accurate as postmyelographic CT in evaluating osteophytes and narrowing of the neural foramina. Eberhardt et al.6 compared conventional myelography and postmyelographic CT with MR myelography using a fat-suppressed 224
3D fast imaging with steady precession sequence for diagnosis of the lumbar root compression syndrome. They found that compared with the conventional method, MR myelography showed comparable sensitivity in the visualization of the spinal nerve roots in the lumbar spine, and MR myelography had significant advantages, especially in
Fig. 4. Comparison between postmyelographic CT and MR myelography of the dural sac volume for normal lumbar levels (upper) and lumbar levels with stenosis (lower). Asterisk = significant differences (p < 0.05) between mean values.
J Neurosurg: Spine / Volume 20 / February 2014
Improved magnetic resonance myelography TABLE 2: Confidence intervals (99%) for dural areas and volumes from postmyelographic CT and MR myelography Area/Volume MR myelography areas (mm2) normal stenotic* postmyelographic CT areas (mm2) normal stenotic* MR myelography volumes (mm3) normal stenotic* postmyelographic CT volumes (mm3) normal stenotic*
All
L1–2
L2–3
L3–4
L4–5
L5–S1
158–184 81–100
143–201 NA
143–181 62–123
146–184 73–106
136–190 76–103
151–222 64–124
157–183 67–87
142–203 NA
143–181 55–107
144–183 64–102
133–189 57–88
149–222 40–101
2960–3415 1738–2037
2928–3891 NA
2762–3421 1630–2600
2804–3539 1676–2155
2552–3557 1668–2105
2617–3899 1190–2002
2903–3379 1484–1835
2880–3876 NA
2765–3399 1388–2299
2772–3491 1409–2077
2564–3538 1401–1913
2441–3845 915–1651
* Calculated for the 31 patients with claudication distances < 100 meters.
cases of extreme spinal canal stenosis. The differences in dural areas and volumes in this study might be explained by the fact that in a case of a severely compressed thecal sac, the viscosity of the intrathecally applied contrast agent is too high in the framework of myelography and that the gravitationally dependent component is too low to achieve sufficient fluid contrast.
Conclusions
An optimized MR myelography approach—a dedicated 3D MR myelography sequence in combination with image fusion—is required to achieve more a reliable diagnosis of lumbar spine stenoses compared with postmyelographic CT, especially in cases of severe compression. This approach may be helpful in preventing overestimation of lumbar spine stenoses. Upper limits of 99% CIs for stenotic levels can be interpreted as an indication for surgical treatment. Further studies that include postoperative outcomes are required, however. Disclosure The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper. Author contributions to the study and manuscript preparation include the following. Conception and design: Eberhardt, Stadlbauer. Acquisition of data: Eberhardt, Ganslandt. Analysis and interpretation of data: all authors. Drafting the article: all authors. Critically revising the article: all authors. Reviewed submitted version of manuscript: all authors. Approved the final version of the manuscript on behalf of all authors: Stadlbauer. Statistical analysis: Stadlbauer. Administrative/technical/material support: Eberhardt. Study supervision: Eberhardt. References 1. Bartlett RJ, Hill CA, Devlin R, Gardiner ED: Two-dimensional MRI at 1.5 and 0.5 T versus CT myelography in the diagnosis of cervical radiculopathy. Neuroradiology 38:142–147, 1996 2. Birchall D, Connelly D, Walker L, Hall K: Evaluation of mag-
J Neurosurg: Spine / Volume 20 / February 2014
netic resonance myelography in the investigation of cervical spondylotic radiculopathy. Br J Radiol 76:525–531, 2003 3. Chiavassa H, Sans N, Galy-Fourcade D, Jarlaud T, Giobbini K, Manelfe C, et al: [HASTE sequence and cine-MRI evaluation of the cervical spinal canal: evaluation in 11 healthy subjects.] J Radiol 81:611–617, 2000 (Fr) 4. Dorenbeck U, Schreyer AG, Grunwald IQ, Held P, Feuerbach S, Seitz J: Degenerative diseases of the lumbar spine. Comparison of the multiecho data image combination sequence with magnetization transfer saturation pulse versus lumbar myelography/postmyelographic computed tomography. Acta Radiol 45:866–873, 2004 5. Eberhardt KE, Hollenbach HP, Huk WJ: [3D-MR myelography in diagnosis of lumbar spinal nerve root compression syndromes. Comparative study with conventional myelography.] Aktuelle Radiol 4:313–317, 1994 (Ger) 6. Eberhardt KE, Hollenbach HP, Tomandl B, Huk WJ: Three-dimensional MR myelography of the lumbar spine: comparative case study to X-ray myelography. Eur Radiol 7:737–742, 1997 7. Ferrer P, Martí-Bonmatí L, Mollá E, Casillas C: [Value of MR myelography in the diagnosis of the spine disorders.] Med Clin (Barc) 115:366–369, 2000 (Span) 8. Freund M, Hutzelmann A, Steffens JC, Buhl R, Heller M: [MR myelography in spinal canal stenosis.] Rofo 167:474–478, 1997 (Ger) 9. Hackenbroch MH, Waldecker B, Prömper KH: [Prolapse of the lumbar vertebral disk—correlation of CT and myelographic findings with surgical findings.] Rontgenblatter 36:50–55, 1983 (Ger) 10. Janssen ME, Bertrand SL, Joe C, Levine MI: Lumbar herniated disk disease: comparison of MRI, myelography, and postmyelographic CT scan with surgical findings. Orthopedics 17: 121–127, 1994 11. Kampmann H, Schroedl P, Spranger M: [Diagnosis of lumbar intervertebral disk prolapse by computed tomography. A comparative clinical study between myelographic, computed tomographic and the surgical research results with 158 patients.] Rontgenblatter 38:387–391, 1985 (Ger) 12. Kanno H, Endo T, Ozawa H, Koizumi Y, Morozumi N, Itoi E, et al: Axial loading during magnetic resonance imaging in patients with lumbar spinal canal stenosis: does it reproduce the positional change of the dural sac detected by upright myelography? Spine (Phila Pa 1976) 37:E985–E992, 2012 13. Kanno H, Ozawa H, Koizumi Y, Morozumi N, Aizawa T, Kusakabe T, et al: Dynamic change of dural sac cross-sectional area in axial loaded magnetic resonance imaging correlates
225
K. Eberhardt, O. Ganslandt, and A. Stadlbauer with the severity of clinical symptoms in patients with lumbar spinal canal stenosis. Spine (Phila Pa 1976) 37:207–213, 2012 14. Katayama H, Yamaguchi K, Kozuka T, Takashima T, Seez P, Matsuura K: Adverse reactions to ionic and nonionic contrast media. A report from the Japanese Committee on the Safety of Contrast Media. Radiology 175:621–628, 1990 15. Kuroki H, Tajima N, Hirakawa S, Kubo S, Tabe R, Kakitsubata Y: Comparative study of MR myelography and conventional myelography in the diagnosis of lumbar spinal diseases. J Spinal Disord 11:487–492, 1998 16. Madsen R, Jensen TS, Pope M, Sorensen JS, Bendix T: The effect of body position and axial load on spinal canal morphology: an MRI study of central spinal stenosis. Spine (Phila Pa 1976) 33:61–67, 2008 17. Maldjian C, Adam RJ, Akhtar N, Bonakdarpour A, Boyko OB: Volume fast spin-echo imaging of the cervical spine. Acad Radiol 6:84–88, 1999 18. Maldjian C, Adam RJ, Akhtar N, Maldjian JA, Bonakdarpour A, Boyko O: Volume (three-dimensional) fast spin-echo imaging of the lumbar spine. Acad Radiol 6:339–342, 1999 19. Mayhew PD, Kapatkin AS, Wortman JA, Vite CH: Association of cauda equina compression on magnetic resonance images and clinical signs in dogs with degenerative lumbosacral stenosis. J Am Anim Hosp Assoc 38:555–562, 2002 20. Melhem ER: Technical challenges in MR imaging of the cervical spine and cord. Magn Reson Imaging Clin N Am 8: 435–452, 2000 21. Morishita Y, Hida S, Naito M, Arimizu J, Takamori Y: Neurogenic intermittent claudication in lumbar spinal canal stenosis: the clinical relationship between the local pressure of the intervertebral foramen and the clinical findings in lumbar spinal canal stenosis. J Spinal Disord Tech 22:130–134, 2009
226
22. Raininko R: The value of CT after total block on myelography. Experience with 25 patients. Rofo 138:61–65, 1983 23. Schäfer M, Riebeling H, Hacker H: [False negative myelography in surgically confirmed lumbar intervertebral disk prolapse.] Neurochirurgia (Stuttg) 19:201–207, 1976 (Ger) 24. Singh K, Samartzis D, Vaccaro AR, Nassr A, Andersson GB, Yoon ST, et al: Congenital lumbar spinal stenosis: a prospective, control-matched, cohort radiographic analysis. Spine J 5: 615–622, 2005 25. Sugahara T, Korogi Y, Hirai T, Shigematu Y, Ushio Y, Takahashi M: Contrast-enhanced T1-weighted three-dimensional gradient-echo MR imaging of the whole spine for intradural tumor dissemination. AJNR Am J Neuroradiol 19:1773–1779, 1998 26. Wang YF, Lirng JF, Fuh JL, Hseu SS, Wang SJ: Heavily T2weighted MR myelography vs CT myelography in spontaneous intracranial hypotension. Neurology 73:1892–1898, 2009 27. Zou KH, Tuncali K, Silverman SG: Correlation and simple linear regression. Radiology 227:617–622, 2003
Manuscript submitted February 2, 2013. Accepted October 29, 2013. Please include this information when citing this paper: published online December 13, 2013; DOI: 10.3171/2013.10.SPINE13119. Address correspondence to: Knut Eberhardt, M.D., District Hospital Castle of Werneck, MRI Center of Excellence, Baltha sar-Neumann-Platz 1, Werneck D-97440, Germany. email: info@ mrt-kompetenzzentrum.de.
J Neurosurg: Spine / Volume 20 / February 2014
Improved magnetic resonance myelography of the lumbar spine using image fusion and volumetry Clinical article Knut Eberhardt, M.D.,1 Oliver Ganslandt, M.D., 2 and Andreas Stadlbauer, Ph.D. 2,3 1 MRI Center of Excellence, District Hospital Castle of Werneck, Germany; 2Department of Neurosurgery, University of Erlangen-Nuremberg, Erlangen, Germany; and 3MR Physics Group, Department of Radiology, Landesklinikum St. Poelten, Austria
Object. The goal of this study was to investigate the usefulness and diagnostic potential of an improved MR myelography approach using a dedicated, heavily T2-weighted, 3D MRI sequence in combination with image fusion. A further aim was to compare dural areas and volumes of normal and pathological altered lumbar spine levels, and accuracy in the assessment of the stenoses, with those obtained by postmyelographic CT. Methods. Fifty patients underwent myelography, postmyelographic CT, and the improved MR myelography approach using a dedicated, heavily T2-weighted, 3D MRI sequence and image fusion. Dural cross-sectional areas and volumes for normal lumbar levels and levels with stenosis were calculated for postmyelographic CT and MR myelography data. The significant differences and Pearson correlations between dural cross-sectional areas and volumes from L1–2 to L5–S1 of postmyelographic CT and MR myelography were analyzed. The 99% CIs for normal and stenotic levels in patients with claudication distances less than 100 meters were calculated. Results. For both dural areas and volumes in normal lumbar levels, the authors found no significant differences and strong correlations between postmyelographic CT and MR myelography. For the lower lumbar levels (L4–5 and L5–S1) they found significantly larger dural areas on MR myelography compared with postmyelographic CT, but not for the upper levels (L2–3 and L3–4). Dural volume analysis revealed significantly larger volumes for MR myelography at all 4 lumbar levels with stenoses in the cohort (L2–3 to L5–S1). Complete separation with no overlap was found between the 99% CIs for normal levels and stenotic levels. Conclusions. Differences in dural areas and volumes in this study may have been caused by the fact that in the case of a severely compressed thecal sac, the viscosity of the intrathecally applied contrast agent is too high in the framework of myelography. The gravitationally dependent component is thus too low to achieve sufficient fluid contrast. An optimized MR myelography approach—a dedicated 3D MR myelography sequence with high spatial resolution in combination with image fusion—is required to achieve a more reliable diagnosis of lumbar spine stenoses, especially with severe compression, compared with postmyelographic CT. This MR myelography approach may be helpful in preventing overestimation of lumbar spine stenoses. The upper limits of 99% CIs for stenotic levels can be interpreted as an indication for surgical treatment. However, further studies that include postoperative outcomes are required. (http://thejns.org/doi/abs/10.3171/2013.10.SPINE13119)
Key Words • lumbar • stenosis • computed tomography • image fusion • magnetic resonance myelography • dural volume
R
compression syndromes play a central role in the pathophysiology of spinal diseases. Symptoms of pain syndromes are often atypical and the assignment to dermatomes may be difficult or deficient. Despite the substantial progress in medical imaging, discrepancies between symptoms and imaging findings may be significant. Patients with large space-occupying lesions frequently have minor symptoms, whereas others show major symptoms without morphological findings on imaging;21,24 this was confirmed by results from a study using an animal model.19 Improvements in sectional imaging techniques have to produce a reduction of the disadicular
Abbreviation used in this paper: HASTE = half-Fourier acquisition single-shot turbo spin-echo.
220
crepancies between symptoms and morphological imaging findings. Lumbar myelography was accepted as the gold standard of detecting radicular compression syndromes for a long time. However, the diagnostic performance of this invasive method is limited,9,23 and undesirable side effects are not uncommon, such as an allergic reaction to the contrast agent or infections.14 The invention of CT allowed for combining conventional myelography with a digital sectional imaging technique (postmyelographic CT). Comparative studies by Raininko22 and Kampmann et al.11 demonstrated the advantages of this technique. Magnetic resonance imaging, a method with excellent soft-tissue contrast, was initially used for diagnosis of intraspinal masses, especially for neuronal compression J Neurosurg: Spine / Volume 20 / February 2014
Improved magnetic resonance myelography symptoms and tumors.10,25 In addition, MRI has the advantage that no ionizing radiation is used for generation of the images, and the imaging orientation can be chosen in any direction. The development of heavily T2-weighted MRI sequences allowed for depicting the dural sac threedimensionally with high spatial resolution and without artifacts due to pulsatile blood flow. Magnetic resonance myelography has the ability to visualize dural structures in a manner comparable to conventional myelography and postmyelographic CT.8 Furthermore, an isotropic 3D MR myelography data set can be reconstructed in any direction of space later by using multiplanar reconstruction. A crucial and well-known disadvantage of heavily T2weighted MR myelography sequences is the associated loss of information about the anatomical vicinity of the dural sac, which is compounded by the fact that the preferred phase encoding direction is in the coronal orientation to achieve optimal image quality. The purpose of this study was to determine whether an improved MR myelography approach using a dedicated, heavily T2-weighted MRI sequence and image fusion could demonstrate similar cross-sectional images of the dural sac area and volumes comparable in size to those detected by postmyelographic CT. Study Population
Methods
Fifty patients with suspected lumbar spinal stenosis and referred for lumbar myelography and postmyelographic CT were also evaluated with our improved MR myelography protocol. There were 15 women and 35 men, ranging in age from 36 to 82 years old (mean ± SD = 61.4 ± 12.0 years). All patients suffered from lower back pain and had neurological symptoms (radicular deficits and/or neurogenic claudication [claudicatio spinalis]). The procedures were explained to the patients and written informed consent was obtained from all patients before the procedures as required by the local institutional review board.
Myelography and Postmyelographic CT
Conventional myelography was performed while patients were prone. The nonionic contrast agent (10 ml Isovist 300 and 5 ml Isovist 240; Schering AG) was applied intrathecally at the levels where the stenoses were assumed. To achieve optimal distribution of the contrast agent in patients with an incomplete allocation of the agent within the range of the stenosis, an abdominal press, the flexion (bent forward) sitting position, and waiting periods of up to 1 hour were applied. To display the lumbar region, 3 standard projections were obtained with anteroposterior, lateral, and left anterior oblique or right anterior oblique orientation. In patients with suspicion of spinal stenosis and an incomplete or complete contrast block on conventional myelography, a postmyelographic CT examination was additionally performed within the following 2 hours using a multislice spiral CT scan (Volume Zoom; Siemens). Contiguous with the stenotic level of the myelography in the postmyelographic CT examination, the subsequent upper and lower levels were examined. The examination was performed while the patient was prone, with breath-holding.
J Neurosurg: Spine / Volume 20 / February 2014
Magnetic Resonance Myelography
Magnetic resonance imaging examinations were performed within 2 days after myelography on a 1.5-T wholebody MRI machine (Sonata; Siemens) equipped with a 6-channel spine array coil, with the patient supine. For MR myelography we used a dedicated MR sequence, a 3D heavily T2-weighted half-Fourier acquisition single-shot turbo spin-echo (HASTE) sequence with the following parameters: TR 10,000 msec, TE 165 msec, slice thickness 0.5 mm, flip angle 180°, voxel size 0.5 × 0.5 × 0.5 mm3, matrix size 256 × 256, and FOV 120 × 120 mm2. Orientation of the volume of interest, as well as phase encoding direction, was coronal. The acquisition time of this dedicated MR myelography sequence was 10 minutes. The sequence scheme included frequency-selective fat suppression and coronal saturation pulses to suppress motion artifacts due to breathing. Additionally, “partial Fourier” was used (that is, 4/8 of the k-space was sampled) to suppress anatomical structures in the vicinity of the CSF and the dural sac. In so doing, the central structures remained unaltered and fully visible, whereas the contrast resolution decreased with increasing distance to the central structures. However, the advantage of an enhanced CSF signal comes with the drawback of loss of anatomical information in the HASTE data. A sagittal T2-weighted turbo spin echo sequence was used as an anatomical-basis data set, an MR data set with anatomical information that was used as the basis data set for image fusion of the MR myelography. The parameters of the sequence were TR 4000 msec, TE 120 msec, slice thickness 2 mm, FOV 250 × 250 mm2, matrix size 256 × 256, and acquisition time 2.5 minutes. Image Data Postprocessing and Evaluation
Coronal sections were calculated from the postmyelographic CT data using multiplanar reconstruction. For each lumbar vertebra the midline of the intervertebral disc was defined as a sectional plane on these images. From this axial section plane along the midline of the intervertebral disc, 4 sectional planes above and 4 sectional planes below were defined in parallel to the central plane, with a section thickness of 2 mm each. Multiplanar reconstruction resulted in 9 axial slices around the intervertebral disc. To determine exact anatomical information within the MR myelography data, the 3D HASTE data sets were fused to the corresponding sagittal T2-weighted turbo spin echo data set. This procedure was performed on a commercially available workstation (Leonardo; Siemens) using the “3D fusion” tool. Section planes along the midline of the intervertebral disc, in the same orientation as described for the postmyelographic CT and multiplanar reconstruction, were used for reformatting to obtain 9 axial slices around the intervertebral disc. The slice thickness of these axial MR myelography images was also 2 mm. The dural area was manually defined on the 9 axial images of both the postmyelographic CT and MR myelography data in an interdisciplinary manner, using an experienced neuroradiologist, an experienced neurosurgeon, and an experienced medical physicist in consensus. The values of the dural area for the central axial image plane were used for comparisons of dural cross-sectional areas 221
K. Eberhardt, O. Ganslandt, and A. Stadlbauer between postmyelographic CT and MR myelography. The area values of all 9 image planes were used for calculation of the dural sac volume. Figure 1 shows the postprocessing steps of both the postmyelographic CT (Fig. 1 upper) and MR myelography (Fig. 1 lower) data for a normal lumbar level: multiplanar reconstruction of the 9 axial slices (Fig. 1 left column), and segmentation of the dural areas (Fig. 1 right column). Statistical Methods
Data were analyzed using SPSS statistical software (version 14.0; SPSS Inc.). To compare differences in dural area and volume between postmyelographic CT and MR myelography, we used a 2-sided paired Student t-test for all levels combined, and a Wilcoxon signed-rank test for individual lumbar levels. Comparisons of dural area and volume in normal levels and in levels with stenosis were performed using a 2-sided Mann-Whitney U-test. The 99% CI was calculated for levels with no pathological changes and stenotic levels in patients with claudication distances less than 100 meters. Pearson correlation coefficients were calculated for measuring relationships between dural area and volume assessed with postmyelographic CT and MR myelography. Correlation coefficients were interpreted under consideration of the specifications given by Zou et al.27 For all tests, the level of significance was set at p < 0.05.
Results
A total of 180 lumbar levels (L1–2 to L5–S1) were evaluated in the 50 patients (mean 3.6 levels per patients, range 2–5 levels) who were suspected to suffer from a stenosis of the lumbar spinal canal. No pathological changes were found on 88 levels in 41 patients. Eleven normal areas were found at L1–2, 19 at L2–3 and L3–4, 15 at L4–5, and 24 at L5–S1. Spinal canal stenoses were detected on 92 levels in 41 patients (31 patients with a claudication distance < 100 meters). Nine patients showed no lumbar spinal canal stenoses. No stenoses were detected at L1– 2. Twelve areas with stenoses were found at L2–3, 25 at L3–4, 34 at L4–5, and 21 at L5–S1. Figure 2 shows the postprocessing results of both the postmyelographic CT (Fig. 2 upper) and MR myelography (Fig. 2 lower) data for a patient with stenosis at L3–4.
Fig. 1. The preparation for multiplanar reconstruction (left column) and segmentation of dural areas (right column) on postmyelographic CT (upper) and improved MR myelography using a fused HASTE and T2-weighted turbo spin echo data set (lower) on a normal lumbar level (L3–4) in a 43-year-old man. The volume of interest for the multiplanar reconstruction is indicated by the black frames. The 9 axial slices are represented by black lines. The red outlines indicate the segmented dural areas.
Comparison and Correlation of Dural Areas
The mean value of the 88 normal dural areas on postmyelographic CT was 169.8 ± 47.6 mm2, and on MR myelography was 170.8 ± 46.8 mm2. These mean values were not significantly different according to a 2-sided paired Student t-test. Analyzed according to lumbar level, we found no significant differences in normal dural areas between postmyelographic CT and MR myelography for all levels (L1–2 to L5–S1). The areas on MR myelography were slightly larger, ranging between 0.0 (0%, at L1–2) and 1.9 mm2 (1.2%, at L4–5) larger (Fig. 3 upper). The mean dural area of the 92 levels with stenosis on postmyelographic CT was 83.2 ± 34.0 mm2 and on MR myelography was 94.0 ± 33.1 mm2. These mean values were significantly different (p = 0.001). Analyzed accord222
Fig. 2. Multiplanar reconstruction (left column) and segmentation of dural areas (right column) on postmyelographic CT (upper) and improved MR myelography using a fused HASTE and T2-weighted turbo spin echo data set (lower) on a lumbar level (L3–4) with stenosis in an 82-year-old man. The volume of interest for the multiplanar reconstruction is indicated by the black frames. The 9 axial slices are represented by black lines. The red outlines indicate the segmented dural areas.
J Neurosurg: Spine / Volume 20 / February 2014
Improved magnetic resonance myelography values were not significantly different. Regarding different lumbar levels we found no significant difference in normal dural volumes between postmyelographic CT and MR myelography for all levels. The volumes on MR myelography were slightly larger, ranging between 3.5 mm3 (0.1%, at L1–2) and 46.4 mm3 (1.5%, at L5–S1) larger (Fig. 4 upper). The mean dural volume of the 92 levels with stenosis on postmyelographic CT was 1794.1 ± 600.5 mm3, and on MR myelography was 1968.0 ± 552.4 mm3. These mean values were also significantly different (p = 0.001). Analyzed according to lumbar level, we found significant differences between postmyelographic CT and MR myelography for all levels (Fig. 4 lower): L2–3 (difference of 115.7 mm3, 5.8%; p = 0.041), L3–4 (difference of 172.6 mm3, 8.9%; p = 0.03), L4–5 (difference of 189.9 mm3, 9.3%; p = 0.001), and L5–S1 (difference of 183.1 mm3, 9.9%; p = 0.021). As with normal areas, Pearson correlation revealed a strong correlation between normal dural volumes on postmyelographic CT and MR myelography for the entirety of levels as well as for all individual levels. For levels with stenosis we found a moderate correlation for the entirety of levels and for all individual levels (Table 1). Comparison of Normal Levels and Levels With Stenosis
As expected, both dural areas and volumes for postmyelographic CT and MR myelography were significantly different (p < 0.001 each) between normal levels and levels with stenosis using a Mann-Whitney U-test. The 99% CIs for dural areas and volumes of normal and stenotic levels (in patients with claudication distances < 100 meters) are shown in Table 2.
Discussion
Fig. 3. Comparison between postmyelographic CT and MR myelography of the dural sac cross-sectional area for normal lumbar levels (upper) and lumbar levels with stenosis (lower). Asterisk = significant differences (p < 0.05) between mean values.
ing to lumbar level, we found significant differences between postmyelographic CT and MR myelography for the levels L4–5 (difference of 14.9 mm2, 15.7%; p = 0.002) and L5–S1 (difference of 13.4 mm2, 13.4%; p = 0.033), but not for L2–3 (difference of 3.5 mm2, 4.0%) and L3–4 (difference of 6.4 mm2, 7.0%; Fig. 3 lower). Pearson correlation revealed a strong correlation between normal dural areas on postmyelographic CT and MR myelography for all levels combined, as well as for all individual levels. However, for levels with stenosis we found a moderate correlation for the entirety of levels, L4– 5, and L5–S1, a strong correlation for L3–4, and a weak correlation that did not reach significance at L2–3 (Table 1). Comparison and Correlation of Dural Volumes
The mean volume of the 88 normal dural levels for the postmyelographic CT was 3141.1 ± 866.0 mm3 and for the MR myelography was 3187.5 ± 828.2 mm3. These mean
J Neurosurg: Spine / Volume 20 / February 2014
Several studies in the literature have compared conventional myelography, postmyelographic CT, and MR myelography. Kuroki et al.15 investigated 40 patients with symptoms of lumbar nerve root compression using conventional and MR myelography. They found that both depiction of intradural structures and image quality were comparable. Studies by Eberhardt et al.5,6 even demonstrated advantages of MR myelography compared with conventional myelography in combination with postmyelographic CT, particularly in the detection of extreme lumbar spinal canal stenoses. In a cohort of 275 patients, Ferrer et al.7 found that MR myelography improved diagnostic accuracy in 16.7% of cases. Studies from Bartlett et al.,1 Birchall et al.,2 Melhem,20 and Wang et al.26 supported these findings. Selection of the appropriate MR sequence technique is essential. Gradient echo sequences were primarily used for heavily T2-weighted MR sequences thus far.25 However, these sequences are prone to susceptibility artifacts, especially in structures with varying proton density.17 This drawback can be reduced by using spin echo sequences,3,18 which additionally have the advantage of a higher signal-to-noise ratio.17 In this study, we went one step further and used an improved 3D turbo spin echo sequence (3D HASTE) in combination with “partial Fourier” (collecting only the 223
K. Eberhardt, O. Ganslandt, and A. Stadlbauer TABLE 1: Correlation of dural areas and volumes from postmyelographic CT and MR myelography* Lumbar Level Dural Area/Volume
All
L1–2
L2–3
L3–4
L4–5
L5–S1
normal areas areas w/ stenosis normal volumes volumes w/ stenosis
0.975‡ 0.582‡ 0.966‡ 0.733‡
0.992‡ NA 0.968‡ NA
0.942‡ 0.333 0.991‡ 0.676†
0.984‡ 0.772‡ 0.985‡ 0.654‡
0.987‡ 0.562‡ 0.992‡ 0.819‡
0.980‡ 0.594‡ 0.957‡ 0.796‡
* All values are correlation coefficients. NA = not applicable. † Significant correlation at p < 0.05. ‡ Significant correlation at p < 0.001.
central parts of the k-space), and frequency selective fat suppression, to improve background contrast and avoid signal overlapping from adjacent structures, respectively. Additionally, this strategy is associated with a reduction of breathing artifacts. We found no significant differences in normal lumbar levels and strong correlations for both dural areas and volumes between postmyelographic CT and MR myelography. The lower lumbar levels (L4–5 and L5–S1) showed significantly larger areas on MR myelography when compared with postmyelographic CT, but not for the upper levels (L2–3 and L3–4). Dural volume revealed significantly larger values for MR myelography at all 4 lumbar levels with stenoses in our cohort (L2–3 to L5–S1). Both dural area and volume of levels with stenoses showed no to minor correlation between postmyelographic CT and MR myelography. Furthermore, the upper limits of the 99% CIs for dural areas and volumes for stenotic levels in patients with claudication distance of less than 100 meters (Table 2) can be interpreted as limit values and an indication for surgical treatment. Our values for the cross-sectional areas of the dural sac of normal lumbar levels are in agreement with previously published findings,16 although their values are tendentially lower. The authors investigated the effect of body position and axial load on dural sac cross-sectional area using conventional MRI, with the result that horizontal MRI with the patient supine and the legs straightened was comparable to vertical MRI, whether axial compression was added or not. This finding stands in marked contrast to the findings of Kanno et al.12,13 These discrepancies might be explained by the differences in the conventional MR protocols (TR, TE, slice thickness, and others) used in the studies, and that only dural areas but not dural volumes were evaluated. Although it is more laborious, the determination of dural volumes has the advantage of allowing for compensation of even minor differences and inaccuracies in slice positioning and orientation. The usefulness and advantages of dedicated MR myelography sequences have been demonstrated previously. Dorenbeck et al.4 compared a T2*-weighted, 2D, spoiled gradient echo multiecho sequence with magnetization transfer saturation pulse, with lumbar myelography and postmyelographic CT. They found that this gradient echo sequence was as accurate as postmyelographic CT in evaluating osteophytes and narrowing of the neural foramina. Eberhardt et al.6 compared conventional myelography and postmyelographic CT with MR myelography using a fat-suppressed 224
3D fast imaging with steady precession sequence for diagnosis of the lumbar root compression syndrome. They found that compared with the conventional method, MR myelography showed comparable sensitivity in the visualization of the spinal nerve roots in the lumbar spine, and MR myelography had significant advantages, especially in
Fig. 4. Comparison between postmyelographic CT and MR myelography of the dural sac volume for normal lumbar levels (upper) and lumbar levels with stenosis (lower). Asterisk = significant differences (p < 0.05) between mean values.
J Neurosurg: Spine / Volume 20 / February 2014
Improved magnetic resonance myelography TABLE 2: Confidence intervals (99%) for dural areas and volumes from postmyelographic CT and MR myelography Area/Volume MR myelography areas (mm2) normal stenotic* postmyelographic CT areas (mm2) normal stenotic* MR myelography volumes (mm3) normal stenotic* postmyelographic CT volumes (mm3) normal stenotic*
All
L1–2
L2–3
L3–4
L4–5
L5–S1
158–184 81–100
143–201 NA
143–181 62–123
146–184 73–106
136–190 76–103
151–222 64–124
157–183 67–87
142–203 NA
143–181 55–107
144–183 64–102
133–189 57–88
149–222 40–101
2960–3415 1738–2037
2928–3891 NA
2762–3421 1630–2600
2804–3539 1676–2155
2552–3557 1668–2105
2617–3899 1190–2002
2903–3379 1484–1835
2880–3876 NA
2765–3399 1388–2299
2772–3491 1409–2077
2564–3538 1401–1913
2441–3845 915–1651
* Calculated for the 31 patients with claudication distances < 100 meters.
cases of extreme spinal canal stenosis. The differences in dural areas and volumes in this study might be explained by the fact that in a case of a severely compressed thecal sac, the viscosity of the intrathecally applied contrast agent is too high in the framework of myelography and that the gravitationally dependent component is too low to achieve sufficient fluid contrast.
Conclusions
An optimized MR myelography approach—a dedicated 3D MR myelography sequence in combination with image fusion—is required to achieve more a reliable diagnosis of lumbar spine stenoses compared with postmyelographic CT, especially in cases of severe compression. This approach may be helpful in preventing overestimation of lumbar spine stenoses. Upper limits of 99% CIs for stenotic levels can be interpreted as an indication for surgical treatment. Further studies that include postoperative outcomes are required, however. Disclosure The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper. Author contributions to the study and manuscript preparation include the following. Conception and design: Eberhardt, Stadlbauer. Acquisition of data: Eberhardt, Ganslandt. Analysis and interpretation of data: all authors. Drafting the article: all authors. Critically revising the article: all authors. Reviewed submitted version of manuscript: all authors. Approved the final version of the manuscript on behalf of all authors: Stadlbauer. Statistical analysis: Stadlbauer. Administrative/technical/material support: Eberhardt. Study supervision: Eberhardt. References 1. Bartlett RJ, Hill CA, Devlin R, Gardiner ED: Two-dimensional MRI at 1.5 and 0.5 T versus CT myelography in the diagnosis of cervical radiculopathy. Neuroradiology 38:142–147, 1996 2. Birchall D, Connelly D, Walker L, Hall K: Evaluation of mag-
J Neurosurg: Spine / Volume 20 / February 2014
netic resonance myelography in the investigation of cervical spondylotic radiculopathy. Br J Radiol 76:525–531, 2003 3. Chiavassa H, Sans N, Galy-Fourcade D, Jarlaud T, Giobbini K, Manelfe C, et al: [HASTE sequence and cine-MRI evaluation of the cervical spinal canal: evaluation in 11 healthy subjects.] J Radiol 81:611–617, 2000 (Fr) 4. Dorenbeck U, Schreyer AG, Grunwald IQ, Held P, Feuerbach S, Seitz J: Degenerative diseases of the lumbar spine. Comparison of the multiecho data image combination sequence with magnetization transfer saturation pulse versus lumbar myelography/postmyelographic computed tomography. Acta Radiol 45:866–873, 2004 5. Eberhardt KE, Hollenbach HP, Huk WJ: [3D-MR myelography in diagnosis of lumbar spinal nerve root compression syndromes. Comparative study with conventional myelography.] Aktuelle Radiol 4:313–317, 1994 (Ger) 6. Eberhardt KE, Hollenbach HP, Tomandl B, Huk WJ: Three-dimensional MR myelography of the lumbar spine: comparative case study to X-ray myelography. Eur Radiol 7:737–742, 1997 7. Ferrer P, Martí-Bonmatí L, Mollá E, Casillas C: [Value of MR myelography in the diagnosis of the spine disorders.] Med Clin (Barc) 115:366–369, 2000 (Span) 8. Freund M, Hutzelmann A, Steffens JC, Buhl R, Heller M: [MR myelography in spinal canal stenosis.] Rofo 167:474–478, 1997 (Ger) 9. Hackenbroch MH, Waldecker B, Prömper KH: [Prolapse of the lumbar vertebral disk—correlation of CT and myelographic findings with surgical findings.] Rontgenblatter 36:50–55, 1983 (Ger) 10. Janssen ME, Bertrand SL, Joe C, Levine MI: Lumbar herniated disk disease: comparison of MRI, myelography, and postmyelographic CT scan with surgical findings. Orthopedics 17: 121–127, 1994 11. Kampmann H, Schroedl P, Spranger M: [Diagnosis of lumbar intervertebral disk prolapse by computed tomography. A comparative clinical study between myelographic, computed tomographic and the surgical research results with 158 patients.] Rontgenblatter 38:387–391, 1985 (Ger) 12. Kanno H, Endo T, Ozawa H, Koizumi Y, Morozumi N, Itoi E, et al: Axial loading during magnetic resonance imaging in patients with lumbar spinal canal stenosis: does it reproduce the positional change of the dural sac detected by upright myelography? Spine (Phila Pa 1976) 37:E985–E992, 2012 13. Kanno H, Ozawa H, Koizumi Y, Morozumi N, Aizawa T, Kusakabe T, et al: Dynamic change of dural sac cross-sectional area in axial loaded magnetic resonance imaging correlates
225
K. Eberhardt, O. Ganslandt, and A. Stadlbauer with the severity of clinical symptoms in patients with lumbar spinal canal stenosis. Spine (Phila Pa 1976) 37:207–213, 2012 14. Katayama H, Yamaguchi K, Kozuka T, Takashima T, Seez P, Matsuura K: Adverse reactions to ionic and nonionic contrast media. A report from the Japanese Committee on the Safety of Contrast Media. Radiology 175:621–628, 1990 15. Kuroki H, Tajima N, Hirakawa S, Kubo S, Tabe R, Kakitsubata Y: Comparative study of MR myelography and conventional myelography in the diagnosis of lumbar spinal diseases. J Spinal Disord 11:487–492, 1998 16. Madsen R, Jensen TS, Pope M, Sorensen JS, Bendix T: The effect of body position and axial load on spinal canal morphology: an MRI study of central spinal stenosis. Spine (Phila Pa 1976) 33:61–67, 2008 17. Maldjian C, Adam RJ, Akhtar N, Bonakdarpour A, Boyko OB: Volume fast spin-echo imaging of the cervical spine. Acad Radiol 6:84–88, 1999 18. Maldjian C, Adam RJ, Akhtar N, Maldjian JA, Bonakdarpour A, Boyko O: Volume (three-dimensional) fast spin-echo imaging of the lumbar spine. Acad Radiol 6:339–342, 1999 19. Mayhew PD, Kapatkin AS, Wortman JA, Vite CH: Association of cauda equina compression on magnetic resonance images and clinical signs in dogs with degenerative lumbosacral stenosis. J Am Anim Hosp Assoc 38:555–562, 2002 20. Melhem ER: Technical challenges in MR imaging of the cervical spine and cord. Magn Reson Imaging Clin N Am 8: 435–452, 2000 21. Morishita Y, Hida S, Naito M, Arimizu J, Takamori Y: Neurogenic intermittent claudication in lumbar spinal canal stenosis: the clinical relationship between the local pressure of the intervertebral foramen and the clinical findings in lumbar spinal canal stenosis. J Spinal Disord Tech 22:130–134, 2009
226
22. Raininko R: The value of CT after total block on myelography. Experience with 25 patients. Rofo 138:61–65, 1983 23. Schäfer M, Riebeling H, Hacker H: [False negative myelography in surgically confirmed lumbar intervertebral disk prolapse.] Neurochirurgia (Stuttg) 19:201–207, 1976 (Ger) 24. Singh K, Samartzis D, Vaccaro AR, Nassr A, Andersson GB, Yoon ST, et al: Congenital lumbar spinal stenosis: a prospective, control-matched, cohort radiographic analysis. Spine J 5: 615–622, 2005 25. Sugahara T, Korogi Y, Hirai T, Shigematu Y, Ushio Y, Takahashi M: Contrast-enhanced T1-weighted three-dimensional gradient-echo MR imaging of the whole spine for intradural tumor dissemination. AJNR Am J Neuroradiol 19:1773–1779, 1998 26. Wang YF, Lirng JF, Fuh JL, Hseu SS, Wang SJ: Heavily T2weighted MR myelography vs CT myelography in spontaneous intracranial hypotension. Neurology 73:1892–1898, 2009 27. Zou KH, Tuncali K, Silverman SG: Correlation and simple linear regression. Radiology 227:617–622, 2003
Manuscript submitted February 2, 2013. Accepted October 29, 2013. Please include this information when citing this paper: published online December 13, 2013; DOI: 10.3171/2013.10.SPINE13119. Address correspondence to: Knut Eberhardt, M.D., District Hospital Castle of Werneck, MRI Center of Excellence, Baltha sar-Neumann-Platz 1, Werneck D-97440, Germany. email: info@ mrt-kompetenzzentrum.de.
J Neurosurg: Spine / Volume 20 / February 2014
