| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Neuroradiology |
1 From the Department of Diagnostic Imaging, St James's Hospital, James's Street, Dublin 8, Ireland. Received May 4, 2004; revision requested July 20; revision received November 4; accepted December 29. Address correspondence to N.P.S. (e-mail: niallsheehy{at}moyneroad.com).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
MATERIALS AND METHODS: The institutional research ethics committee approved this study and provided a waiver for informed consent. Data sets were evaluated in 50 consecutive patients referred for contrast-enhanced three-dimensional MR angiography of the carotid arteries (32 male and 18 female patients; age range, 15–80 years; mean age, 59 years). The ASA was defined as a linear area of high signal intensity that is seen anterior to the spinal cord in an arterial phase of enhancement and connects directly to a known arterial structure. If the linear area of high signal intensity was seen in the arterial phase but did not connect to a known arterial structure, it was considered a probable ASA. Venous enhancement was graded on a five-point scale (0–4) with grade 0 (no venous enhancement) or grade 1 (trace venous enhancement) considered to be in the arterial phase.
RESULTS: The ASA was identified with certainty in 37 of 50 patients. A vessel visualized anterior to the spinal cord, which probably represented the ASA, was seen in another 11 of 50 patients. In 29 of 50 patients the vessel was visualized only on the full-volume maximum intensity projection (MIP) image. In the remainder of cases the artery was identified on operator-defined subvolume MIP images. Continuity between the vessel and the vertebrobasilar arterial structures was identified in 35 of 50 patients. The vessel was seen as a continuous structure throughout its length in 34 patients and appeared discontinuous in 14. Radiculomedullary feeders were identified in 24 of 50 patients.
CONCLUSION: The normal cervical ASA was visualized in 48 of 50 of subjects with contrast-enhanced high-spatial-resolution three-dimensional MR angiography.
© RSNA, 2005
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Several reports suggest that the spinal vessels in patients with spinal arteriovenous malformations can be detected by using the high-spatial-resolution approach tailored to carotid artery stenosis evaluation (3–5). The combination of MR imaging and MR angiography has been shown to provide improved screening for dural arteriovenous malformations and to benefit the subsequent angiographic study by helping target the level of the fistula (6). Arteries in patients with vascular malformations, however, are typically larger than those in healthy subjects and are therefore more likely to be demonstrated with MR angiography.
Although there have been reports about the use of contrast-enhanced MR angiography to image the artery of Adamkiewicz and the lumbar anterior spinal artery (ASA) (7,8), the normal appearance of the spinal vessels in the thoracic and cervical cord seen with this technique, to our knowledge, has not been defined. If contrast-enhanced three-dimensional MR angiography is to be used for diagnosis of spinal arteriovenous malformations, knowledge of the normal appearance of the normal spinal vessels with this technique is essential (9). Thus, the purpose of our study was to determine retrospectively whether the ASA is visualized in the cervical region with contrast-enhanced high-spatial-resolution three-dimensional MR angiography of the extracranial carotid and vertebral arteries.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Study Participants
The institutional research ethics committee approved this study and provided a waiver for informed consent. The examinations of 50 sequential patients (32 male and 18 female patients; mean age, 59 years; range, 15–80 years) who were referred for contrast-enhanced three-dimensional MR angiography of the carotid arteries between May 2002 and March 2003 were included in our study. The examinations were performed for assessment of atherosclerotic carotid artery stenosis (38 patients), confirmation of carotid artery occlusion (six patients), diagnosis of carotid artery dissection (four patients), or postoperative assessment (two patients).
Image Acquisition
All images were acquired at 1.5 T (Magnetom Symphony; Siemens, Erlangen, Germany) with quantum gradients. MR angiography was implemented with a coronal three-dimensional radiofrequency–spoiled fast gradient-echo volume acquisition (three-dimensional fast low-angle shot) at the following settings: 4.7/1.6 (repetition time msec/echo time msec); flip angle, 25°; imaging matrix, 256 x 200; and field of view (FOV), 260 x 200 mm. The carotid and vertebral vessels were imaged from the aortic arch to the circle of Willis. A head and neck coil was used. Voxel size was approximately 1 mm3 before zero interpolation, and the imaging time was approximately 30 seconds.
Images were acquired after injection of 30 mL of 0.5 mmol/L gadobenate dimeglumine (Multihance; Bracco, Milan, Italy), delivered with a power injector (Medrad, Pittsburgh, Pa) at an injection rate of 1.2 mL/sec, followed by a 20-mL saline flush. A two-dimensional single thick-section fluoroscopic acquisition (CareBolus; Siemens), which encompassed the entire volume of the aortic arch, carotid artery, and vertebral vessels and was reconstructed and displayed in real time, was used to synchronize acquisition of central k-space data with the arterial peak of contrast enhancement. An elliptic centric sequence was used for k-space acquisition. Because of the long imaging time (30 seconds), breath holding was not employed, but patients were instructed to take regular shallow breaths during imaging.
Image Review
Two radiologists blinded to the clinical details (J.F.M.M. and N.P.S.) independently reviewed the images on one workstation (Syngo; Siemens). Both were experienced in reviewing contrast-enhanced three-dimensional MR angiograms of the carotid and vertebral vessels (J.F.M.M. and N.P.S., with 8 and 3 years of experience, respectively). Each radiologist was presented with the three-dimensional data set and allowed to manipulate the images by using three-dimensional reconstructions, including multiplanar reformatting and full- and partial-volume maximum intensity projection (MIP) images as appropriate. All reconstructions were created with the same workstation as mentioned before. Disagreements were resolved with consensus.
The data sets were analyzed in two stages by each radiologist to assess the suitability of the images for spinal artery evaluation. The images were evaluated for the presence of the cervical cord within the FOV (yes or no) and for venous enhancement, which was graded on a five-point scale. Grade 0 indicated no venous enhancement; 1, trace enhancement; 2, minor enhancement (no interference with diagnostic potential of image); 3, moderate enhancement (some interference with diagnostic potential of image); and 4, major enhancement (image nondiagnostic, imaging needs to be repeated). If venous contamination was graded as minor or better, we considered the image to be in the arterial phase.
Five aspects of the ASA were evaluated, according to the following questions: (a) Is a vessel (a linear high-signal-intensity structure) visible anterior to the spinal cord? (b) Is this vessel visible on the full-volume MIP image? (c) Is it continuous with the vertebrobasilar vessels? (d) Is it continuous within the cervical region or interrupted? (e) Does the vessel have any visible feeding vessels? These criteria were based on anatomic descriptions of the vessel to ensure its proper identification (10).
Data Analysis
A linear area of high signal intensity on the ventral portion of the cervical cord could represent the ASA or the anterior median spinal vein, which are alongside one another. A vessel seen on an appropriate arterial phase image was considered a probable ASA. If the linear area of high signal intensity could be seen to join a definite arterial structure via either a feeder vessel or its origin, we considered it an ASA.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
|
The ASA appeared continuous in 34 of 50 patients and discontinuous in 14 of them. In all studies, the vessel disappeared from view within either the lower cervical or the upper thoracic canal, even though the upper thoracic canal was included within the FOV in all instances. No examination included a paired vessel. The vessel was seen to be continuous with (ie, arising from) the vertebrobasilar arterial structures in 35 of 50 studies (Figs 2 and 3). Therefore, the vessel in these cases was an ASA (Figs 2–4).
|
|
|
|
We therefore identified 37 of 50 ASAs and a further 11 probable ASAs. As 48 of 50 patients in which a vessel was seen showed, at most, trace venous enhancement, we believe that all vessels seen were likely to be arterial (ie, the ASA rather than the anterior median vein).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
The ASA supplies the spinal cord throughout its length. It runs in the ventral median sulcus of the spinal cord within the subpial space and is situated dorsal to the accompanying vein. Usually single, it may occasionally be "duplicated" due to failure of fusion of the two metameric precursor vessels in the developing fetus. It may be focally discontinuous, especially within the thoracic region, and it frequently demonstrates an irregular caliber within the cervical canal. It receives a major contribution to flow from a varying number of radiculomedullary arteries (typically two to four), which arise from the somatic arteries from the skull base to the sacral level.
Spinal angiography and postmortem examinations reveal ASA diameters of 200–500 µm within the cervical region and of 500–800 µm in the lumbosacral region. Diameters of the radiculomedullary vessels are 400–600 µm in the cervical cord but larger in the lumbar region, where the largest of these, the artery of Adamkiewicz, may be as large as 1.2 mm in diameter. The ventromedial spinal vein is 200–600 µm in diameter.
Spinal angiography, the reference standard for evaluation of the intraspinal vasculature, has the high spatial resolution necessary to image these small vessels. Distinction between arteries and veins is achieved by observing the temporal pattern of arterial filling prior to venous filling. Despite this seemingly ideal profile, spinal angiography is not widely used outside specialized centers, as it is invasive and requires trained personnel. Therefore, spinal angiography is rarely used for clarification of the nature of possibly enlarged intraspinal arteries on cross-sectional MR images of the spinal canal. In this group of patients, and in all other patients with known spinal vascular malformations or in those suspected of having them, noninvasive angiography with MR angiography would offer the clinician an attractive alternative.
Challenges in applying contrast-enhanced three-dimensional MR angiography to the imaging of spinal vessels include resolution constraints and difficulties in distinguishing arteries from veins (9). We addressed the resolution issue with a high-spatial-resolution technique optimized for the imaging of the carotid and vertebral arteries (1-mm isotropic voxels). Although this technique is adequate for evaluating the larger carotid and vertebral arteries, it might be assumed that the ASA in the cervical region, with a known diameter of only 200–500 µm, would still be beyond resolution capacity. Nonetheless, we were successful in identifying the ASA in most subjects. If we assume that the diameter of the ASA in our subjects corresponds to previously reported dimensions (200–500-µm diameter), we postulate that the arteries might have been demonstrated because of partial-volume effects. As the signal from each voxel is a combination of high signal from the contrast material in the vessel and low signal from the remaining volume of the voxel, partial averaging within the voxel might have rendered arteries smaller than the voxel size visible.
In all of our examinations, the visualized ASA disappeared once it entered the thoracic region; it is uncertain whether this represented true vessel discontinuity, a decrease in signal at the edge of the coil, or the smaller size of the thoracic ASA. Similarly, we were unable to determine whether discontinuities within the cervical region were real or artifactual. These observations underline one of the major limitations of our study, namely, that spinal angiography was not available as the reference standard. The superior spatial resolution of spinal angiography would allow these issues to be resolved, but its use would be ethically unacceptable in a healthy patient population.
The requirement for high-spatial-resolution imaging of the spinal arteries with contrast-enhanced three-dimensional MR angiography has the inherent drawback of long acquisition time. The angiographic sequence used in this study took approximately 30 seconds to acquire. This time frame would clearly result in visualization of both arterial and venous enhancement in most organ systems. The circulation time between arterial and venous enhancement within the cord is unknown but is likely to be no more than the observed 8–12-second circulation time within the carotid artery circulation imposed by the blood-brain barrier. In our study, we used an elliptic centric pattern of k-space acquisition in combination with fluoroscopic triggering to ensure minimal venous contamination (11,12).
Differentiation of the ASA from the anterior median vein represents another difficult problem. In our data sets, the only way to be certain that a vessel represented the ASA and not the anterior median vein was to establish direct continuity with a known arterial structure. In the carotid-jugular circulation, the vessels run at a fairly constant course and have predictable morphologic characteristics that facilitate differentiation of the artery from vein. This is not possible in the spinal circulation, however, given that the artery and vein are of similar size and shape and run similar courses alongside each other. This limitation does not apply to conventional digital subtraction angiography, as direct intraarterial injection always outlines the arteries first. Arteriovenous differentiation is enabled by side-by-side inspection of pairs of images.
Although dedicated arterial phase imaging with contrast-enhanced three-dimensional MR angiography is possible for most anatomic regions, dedicated examinations of the spinal venous system pose a much greater challenge with this modality. Although we did not address the issue specifically in this article, in the few instances in which we performed delayed imaging and/or subtracted an arterial data set from one acquired in the venous phase, the anterior spinal vein could not be demonstrated due to extensive enhancement within the vertebral venous plexus. This made it impossible to distinguish the ASA from enhancing venous structures. The currently most widely employed techniques of "direct" contrast-enhanced MR venography, in which the target vein can be injected directly with acquisition of central k-space data during infusion of dilute contrast material (to overcome T2* effects) are not applicable to the spinal veins because the veins cannot be accessed directly. To delineate the spinal veins, time-resolved imaging (eg, time-resolved imaging of contrast kinetics, or TRICKS) (13) may be applicable, possibly along with a method to subtract arteries from veins (eg, the venous-enhanced subtracted peak arterial technique) (14). Time-resolved imaging usually involves a trade-off with spatial resolution, which is unfavorable considering the small size of the vessels.
An FOV limitation is another factor that has prevented complete information about the spinal vessels from being available. In many patients, current imagers do not permit a single FOV in the craniocaudal dimension that is sufficient to include the entire spinal axis. We analyzed the data sets that were acquired for an indication unrelated to the spinal vasculature and used an FOV limited to about 10 vertebral levels (to optimize resolution). An examination of the entire spinal cord with this approach would require two to three individual MR angiographic data sets. In any event, high-spatial-resolution data sets that might allow visualization of the ASA within the thoracic or lumbar regions are not acquired for any indication we are aware of, thus depriving the medical community of "free" information in regard to the normal ASA outside the cervical region.
Despite our success in visualizing the ASA in most cases, recent advances in imaging may offer the opportunity to obtain images at much higher spatial resolution. For example, gradient speed is continually improving, and the fastest imagers currently available offer a threefold increase in imaging time compared with the time with our approach (repetition time of 5.0 vs 1.7 msec). Parallel imaging also offers the potential for higher spatial resolution, faster image acquisition (two to four times faster), or a combination of both, which will improve the ability to delineate smaller vascular structures (15).
Although these approaches will undoubtedly improve spatial resolution, addressing contrast agent indexes can offset the unwanted decrease in signal-to-noise ratio. A simple increase in infusion rate will provide a decrease in blood T1, which will at least partially redress the balance. Contrast agents with protein binding (as used in this study) may offer advantages over standard extracellular agents (16). Newer contrast agents formulated at 1.0 mol/L (compared with standard agents with 0.5 mol/L) also offer potential advantages (17). Although the requirement for increased signal-to-noise ratio and higher spatial resolution may be simultaneously addressed by using blood pool agents (18), the issue of differentiation of arteries from veins will probably nullify the benefit of increased resolution. Finally, the imaging efficiency of multichannel coils is constantly improving, an effect that improves imaging with essentially no penalty.
We determined that the cervical ASA is demonstrated in almost all patients who undergo isotropic contrast-enhanced high-spatial-resolution three-dimensional MR angiography of the carotid and vertebral arteries. Progressive improvements in resolution will facilitate more detailed morphologic assessment of the characteristics of this small but important artery that was, until recently, thought to lie beyond the resolution capability of MR angiography.
|
FOOTNOTES |
|---|
Abbreviations: ASA = anterior spinal artery • FOV = field of view
See Materials and Methods for pertinent disclosures.
Author contributions: Guarantor of integrity of entire study, J.F.M.M.; study concepts, N.P.S., J.F.M.M., G.E.B.; study design, N.P.S., G.E.B.; literature research, N.P.S.; clinical studies, N.P.S., J.F.M.M., G.E.B.; data acquisition, N.P.S.; data analysis/interpretation, N.P.S., J.F.M.M.; statistical analysis, N.P.S.; manuscript preparation and definition of intellectual content, N.P.S.; manuscript editing and final version approval, N.P.S., J.F.M.M.; manuscript revision/review, J.F.M.M.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  W.H. Backes and R.J. Nijenhuis Advances in Spinal Cord MR Angiography AJNR Am. J. Neuroradiol., April 1, 2008; 29(4): 619 - 631. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
