STAR MR Angiography for Rapid Detection of Vascular Abnormalities in Patients With Acute Cerebrovascular Disease
Background and Purpose We undertook to investigate the usefulness of signal targeting with alternating radiofrequency magnetic resonance angiography (STAR MRA) in the diagnosis of acute cerebrovascular disease. The potential advantage of the technique is that angiographic images can be acquired in less than 1 minute.
Methods We studied 19 patients (11 men and 8 women, ranging in age from 36 to 84 years [mean age, 66 years]) presenting with signs and symptoms of acute stroke. Patients underwent STAR MRA and three-dimensional fast imaging with steady-state precession (3D FISP) MRA. The MRAs were analyzed as to image quality and vascular abnormalities in the vascular territory of stroke as defined by diffusion-weighted imaging abnormalities and compared using a Wilcoxon signed-rank test.
Results STAR MRAs had slightly inferior image quality compared with 3D FISP MRA (P<.05). STAR MRA and 3D FISP MRA agreed in 18 of 19 cases regarding vascular abnormalities in the territory of the infarct (occlusion, n=8; stenosis, n=4; no abnormality, n=6). In one patient, the techniques disagreed, when 3D FISP MRA was normal and STAR MRA demonstrated a vessel occlusion in the vascular territory of a stroke as defined by diffusion-weighted imaging abnormalities (P>.05).
Conclusions Despite slightly inferior image quality compared with 3D FISP MRA, STAR MRA is comparable with 3D FISP MRA in depicting abnormalities in the proximal parts of the cerebral arteries corresponding to ischemic regions on diffusion-weighted imaging, in a strikingly shorter acquisition time. Further studies are necessary to confirm that the smaller branches are better shown by using longer inversion times.
Magnetic resonance angiography has gained wide acceptance in screening for abnormalities of the intracranial vasculature.1 2 3 4 5 The noninvasive nature of the technique and the three-dimensional image display make it an attractive alternative to x-ray angiography. MRA data may be readily combined with conventional MRI findings, allowing correlation of tissue abnormalities with blood flow abnormalities. This combination is particularly attractive in the setting of acute cerebrovascular disease, where evaluation of a vessel may reveal the etiology of a stroke by demonstrating stenosis, occlusion, or embolus. Such findings may be important in acute patient management regarding questions of intravenous or intra-arterial thrombolysis.6
A few studies have used MRA to identify intracranial arterial occlusion or severe stenosis in the initial evaluation of stroke patients.7 8 9 10 11 These studies demonstrated a high degree of correlation of vascular abnormalities on MRA with the distribution of infarcts as depicted on conventional MRI images. Occlusions in larger vessels such as the supraclinoid portion of the ICA, MCA (stem and branches),7 posterior cerebral artery, or basilar artery have been well demonstrated.10 Occlusions of smaller vessels such as the lenticulostriate and thalamic perforating vessels are often missed, mostly due to saturation effects related to their slower flow. Blood flowing in these vessels is exposed to the MR radiofrequency pulse multiple times while crossing the imaging plane and therefore does not give enough signal.12 Overall, MRA may provide information concerning vascular stenosis or occlusion in up to 80% of cases, superior to conventional MRI in up to 50% of patients. In addition, once the diagnosis of an occlusion is established, collateral flow distal to an occlusion can be further investigated using phase-contrast techniques or selective presaturation pulses.8 11 13
However, the acquisition of an MRA in a patient with an acute stroke remains a challenge. Acquisition times for MRAs of the intracranial vessels are still quite long (5 to 10 minutes), and the images are commonly degraded by motion artifact in these patients, who are often clinically unstable and unable to hold still for the acquisition time required. Such motion artifact may render the images uninterpretable or may mimic vascular stenosis.9 A reduction in scanning time is therefore desirable.
While initial 3D phase-contrast MRAs had acquisition times of 37 minutes,14 recent time-of-flight subtraction methods have substantially reduced imaging times, varying from 30 seconds for a 2D STAR MRA15 to 1 minute 30 seconds for a 3D acquisition.16 The STAR technique consists of a preparation phase during which the longitudinal magnetization of the target tissue is inverted on alternate acquisitions and the background tissue is presaturated. In the 2D acquisition, this is followed by a readout phase using a cine-segmented turboFLASH sequence with a shared echo modification to improve temporal resolution; in the 3D acquisition, it is followed by an echo-planar readout. With appropriate alteration of the phases of the radiofrequency excitation pulses, there is cancellation of the background signal intensity, but flow signal is optimized.
The purpose of this study was to compare 2D and 3D STAR MRA techniques to a 3D FISP sequence to define an MRA technique that is feasible for the acutely ill patient.
Subjects and Methods
We evaluated 19 patients (11 men and 8 women ranging in age from 36 to 84 years [mean age, 66 years]). Patients were admitted for MRI because of symptoms suspicious for acute stroke. Eleven patients were imaged with 2D STAR MRA and 8 patients with 3D STAR MRA using segmented EPI readouts. All patients underwent conventional MRA using a 3D FISP sequence for comparison. All patients or authorized representatives gave written informed consent for the MRI procedure, which was approved by the Committee on Clinical Investigations of the Beth Israel Hospital.
Imaging was performed using a prototype whole-body 1.5-T EPI system (Siemens Medical Systems) with a gradient resonant circuit tuned at 1000 Hz. A maximum gradient amplitude of 35 mT/m with 250-μs rise times for all the gradient axes was possible. A circularly polarized head coil was used for excitation and signal reception.
DWI was performed in the usual manner using a multislice, single-shot, spin-echo EPI sequence.17
Conventional MRA was performed using a 3D FISP sequence. Imaging parameters for the 3D FISP sequence were TR/TE/flip angle=35 ms/6 ms/20°, 175×200 mm field of view, 192×256 matrix size, and 64 partitions. Effective slice thickness was 1 mm (voxel size, 0.78×0.78×1.0 mm); acquisition time was 7 minutes 12 seconds.
2D STAR MRA15 was acquired in the transverse plane using a segmented turboFLASH sequence with first-order flow compensation and a 50% asymmetrical echo: TR/TE/flip angle=12 ms/7 ms/12°. To minimize the duration of the readout and TR, a bandwidth of 78 Hz per pixel with zero filling of high spatial frequency samples was used. Field of view was 250×250 mm, with matrix size 256×256 (voxel size, 0.97×0.97×16.0 mm); 7 lines per segment were acquired with a segment duration of 91 ms, a standard sequential interleaved order, and slice thickness of 16 mm. On alternate acquisitions, an inversion pulse is applied over a slab thickness of 6 to 10 cm to tag the arterial inflow with a 23-ms-duration hyperbolic secant pulse below the slab of interest. After the TI has elapsed, the slice of interest is acquired. TIs varying from 350 to 600 ms were used to follow the tagged blood from proximal to distal parts of the arterial circulation, with a recovery time of 300 ms. No cardiac triggering was used; acquisition time varied from 28 seconds (TI=350 ms) to 37 seconds (TI=600 ms).
3D STAR MRA using segmented EPI readout16 images were acquired with the following parameters: TR/TE/TI/flip angle=10 ms/3 ms/400 ms/16°, 160×320 mm field of view, 128×256 matrix size, slab thickness of 64 mm with 32, 2-mm-thin partitions (voxel size, 1.25×1.25×2.0 mm), and acquisition time of 1 minute 8 seconds. Recovery time was 300 ms. The hyperbolic secant inversion pulse was applied similarly as in the 2D STAR sequence. Six echoes per RF excitation were acquired.
Image quality of STAR and 3D FISP MRA was ranked on the following scale: (1) image was degraded by artifact, vessels not clearly visible; (2) vessel trunks were visible, ie, MCA was not depicted further than M1; (3) major vessels are visible, ie, MCA was not depicted further than M2 and/or M3; and (4) smaller branches originating from M2 and M3 were visible.
MRAs were analyzed without knowledge of the result of the DWI so that the reviewer was not influenced in the analysis by the presence or site of an abnormality in the vasculature. The reviewer was instructed to observe any occlusion or stenosis of the ACA, MCA, and PCA or their branches. 2D STAR MRAs were analyzed in a cine mode; 3D STAR MRA was analyzed as maximum intensity projections.
DW images were analyzed regarding the location of a stroke (depicted as an area of high signal intensity on the DW images with the highest b-value). Findings on MRAs in the vascular territory of the stroke were then ranked on a scale from 1 to 3 (1, normal vessel; 2, stenosis; and 3, occlusion). Statistical analysis comparing STAR MRA to 3D FISP MRA was performed using a Wilcoxon signed-rank test.
Seventeen patients demonstrated infarcts on DWI. The vascular territories of the infarcts are listed in the Table⇓. Two patients had no infarct on DWI.
Conventional MRA depicted only the M1 segment in 1 patient, M1 plus M2 and/or M3 segment without branches in 9 patients, and branches of M2 or M3 segment in the 9 other patients. STAR MRA depicted only the M1 segment in 13 patients, M1 plus M2 and/or M3 segment without branches in 5 patients, and branches of M2 or M3 segment in 1 patient. Results were statistically significant (P<.05).
Results are summarized in the Table⇑. 3D FISP MRA and STAR MRA agreed in 18 cases as to the presence or absence of a vascular abnormality in the territory of the infarct as demonstrated on DWI: arterial occlusions corresponding to the ischemic zone were identified with both techniques in 8 cases (Fig 1⇓). Stenoses were demonstrated with both techniques in 4 cases (Fig 2⇓).
In 4 cases, both MRAs were normal, although the DW images demonstrated an infarct in the territory of the left MCA. In 1 case, the stroke was found to be due to an embolus from a thrombosed subclavian vein through a patent foramen ovale. In 3 cases, the etiology of the stroke remained obscure, since both MRA and Doppler evaluations of the extracranial and intracranial arteries were normal, and no cardiac source for emboli was found.
In 2 cases with no abnormalities on DWI, the techniques agreed, demonstrating normal vessels in 1 patient and an occlusion of the left PCA in another patient.
In 1 patient with a stroke in the territory of the left MCA, the techniques disagreed: STAR MRA depicted an occlusion of the left MCA and a stenosis of the right MCA (Fig 3⇓), while 3D FISP MRA was normal. This patient had had myocardial infarction and atrial fibrillation, but no intracardiac thrombi were found. There was no statistically significant difference between the two techniques (P>.05).
In addition, the following findings were made with both techniques: occlusion of the right ICA (n=1), occlusion of the left ICA (n=2), and occlusion of the left MCA and stenosis of the right MCA (n=1). While the occlusion of the ICAs was found in 3 patients with additional occlusions of the MCAs on the corresponding side, the occlusion of the left MCA was due to an older infarct on that side.
The techniques differed in depicting abnormalities in the posterior circulation. In 6 patients, both PCAs were not seen with STAR MRA. In 2 of these patients, the PCAs were also not shown with 3D FISP MRA. In a third patient, 3D FISP MRA demonstrated occlusion of the left PCA. Finally, stenoses of the PCA were found in 2 patients with STAR MRA, while 3D FISP MRA was normal (Fig 2⇑).
In this study, STAR MRA was able to identify the M1 segment in all patients but could not identify secondary and tertiary vessels in most cases. This resulted in a slightly inferior image quality compared with 3D FISP MRA. Even though spatial resolution was highest for 3D FISP MRA, this may not be the only explanation for the finding. Another factor is the short TI used in this study, such that the inverted spins had not yet filled the secondary and tertiary branches of the cerebral arteries at the time the image was acquired. With longer TIs, smaller branches should be more easily visualized.
However, STAR MRA did not differ statistically from 3D FISP MRA in demonstrating a lesion in the vascular territory of the acute infarct as demonstrated on DWI and was able to detect a vascular abnormality in 12 of the 17 patients with acute stroke. This finding likely reflects the observation that the majority of strokes (45% to 50%) result from pathology of large vessels (ICA, MCA, and PCA).18 19 These vessels were well demonstrated on all the STAR MRAs in this study and revealed occlusion (n=8) and stenoses (n=4) in the vascular supply to the abnormal area on DWI.
Four of the 17 patients with a lesion on DWI had normal MRAs. This result is in keeping with reports in the literature and may reflect the following: (1) For up to one third of infarctions, no definite cause can be established at the completion of the workup.20 (2) Furthermore, 25% of strokes were found to be due to small-vessel (lacunar) infarcts,18 19 and the changes in these vessels may not be detectable by MRA. This was perhaps the case in one of our patients who had a 3-mm infarct in the internal capsule. (3) Some of these patients may have suffered emboli from an underlying cardiac source related to myocardial infarction, valvular disease, or cardiomyopathy, which have been reported to account for some 15% of acute strokes.18 19 It is difficult to diagnose embolic stroke by MRA. With x-ray angiography, embolic events may occasionally be visualized directly as intraluminal filling defects or secondarily as vascular occlusions. Embolic occlusion is the most likely diagnosis when multiple peripheral occlusions with meniscal margins are seen.21 However, since angiography is generally performed more than 12 hours after the acute event, 50% or more of emboli may lyse or migrate distally and may not be detected. A slight irregularity at the site of previous occlusion may be the only trace of a recanalized embolic (or thrombotic) occluded artery.
In our study, discrepancies between normal MRAs and the presence of a stroke on DWI were due to emboli from a documented source in one patient. While the diagnosis of embolus may be difficult with MRA, it can be made on conventional proton-density spin-echo images. An intraluminal focus of increased signal intensity, distinct from the signal void from flowing blood, may be detected. However, MRI is limited in its ability to depict intraluminal filling defects in any but the large central arteries.22 In three of our patients, the etiology of the stroke remains unclear. One of these patients, imaged within 24 hours, had suffered myocardial infarction and atrial fibrillation but had no evidence of intracardiac thrombi.
There was a discordant finding in one patient, in whom 3D FISP MRA was normal and STAR MRA demonstrated occlusion of the left MCA, the vessel supplying the infarct, and stenosis of the right MCA (Fig 3⇑). Even though the etiology of the stroke is most likely a cardiac embolus (the patient had suffered a myocardial infarction and had atrial fibrillation at the time of admission), it seems unlikely that the STAR MRA performed immediately before 3D FISP MRA depicted emboli that on subsequent 3D FISP MRA had migrated distally. Therefore, this finding has to be regarded as a false-positive result for STAR MRA. While false-positive results on conventional time-of-flight MRAs are often due to dephasing from turbulent flow or saturation effects from exposure of spins to multiple radiofrequency pulses, STAR MRA has its own potential pitfalls. First, an important consideration for the acquisition of STAR MRAs is the TI. In the patient mentioned above, STAR MRA was acquired for only one TI (400 ms). Therefore, the occlusion depicted might be due to incomplete filling of the vessel at the time of the image acquisition because of slow flow (Fig 1⇑). Second, the positioning of the imaging slab is also important. For example, false-positive results in the territory of the posterior circulation on STAR MRA most likely reflect the position of the thin imaging slice. The portion of the vessel that is not depicted may well be located outside of the imaging slice, and the problem is therefore more of a concern with the 2D technique with thin 16-mm-thick slices. (Five of the six patients with false-positive occlusion/stenosis in the PCA territory were imaged using 2D STAR MRA.) Such error can be avoided by acquiring adjacent slices. The 3D acquisition slab is larger (64 mm); therefore, the problem is not encountered as frequently.
Three-dimensional acquisitions permit the reconstruction of multiple views using maximum intensity projections, whereas the 2D approach yields only a single projection. A pilot scan can be acquired with 3D STAR, permitting tailored imaging of a particular vessel of interest, and using 2D STAR with higher in-plane resolution and thinner slices (4 to 10 mm).
We did not address the potential advantage of STAR MRA in imaging vessels selectively and assessing directionality of flow. A thin tag applied perpendicular to the vessel of interest may be followed along the length of the vessel as a function of inflow time, as has been done with other bolus tracking techniques.23 In addition, sensitivity to slow-flowing blood may be heightened by applying the tag closer to the region of interest and using longer inflow times. The limiting factor for longer inflow times and very slow flow would be the overall signal-to-noise ratio, given that blood relaxes between the tagging pulse and the readout time according to T1.
Although MRA has won increasing acceptance in the clinical community for the evaluation of cerebrovascular disease, conventional x-ray angiography remains the gold standard. Bozzao et al24 correlated angiographic and sequential CT findings in patients with evolving cerebral infarction and found that 6 of their 36 patients (17%) had a normal angiogram, while the remaining 30 showed complete occlusion of the ICA, the MCA, or an MCA branch vessel. Our finding using MRA corresponds to theirs: 4 of our 17 patients (24%) with a lesion on DWI had a normal angiogram.
In conclusion, STAR MRA is as good as conventional time-of-flight MRA in demonstrating abnormalities in the proximal parts of the cerebral arteries but has a significantly shorter acquisition time. Further studies are necessary to confirm that the smaller branches are better shown by using longer TIs.
Selected Abbreviations and Acronyms
|ACA||=||anterior cerebral artery|
|EPI||=||echo planar imaging|
|FISP||=||fast imaging with steady-state precession|
|FLASH||=||fast low-angle shot|
|ICA||=||internal carotid artery|
|MCA||=||middle cerebral artery|
|MRA||=||magnetic resonance angiography|
|PCA||=||posterior cerebral artery|
|STAR MRA||=||signal targeting with alternating radiofrequency MRA|
This work is supported by National Institutes of Health grants NS01634 and NS33660, American Heart Association grant 95012410, the Harcourt General Charitable Foundation, and the Friends of Beth Israel Hospital.
- Received February 14, 1997.
- Revision received February 14, 1997.
- Accepted March 14, 1997.
- Copyright © 1997 by American Heart Association
Ross JS, Masaryk TJ, Modic MT, Ruggieri PM, Haacke EM, Selman WR. Intracranial aneurysms: evaluation by MR angiography. AJNR Am J Neuroradiol. 1990;11:449-455.
Vogl TJ, Bergman C, Villringer A, Einhaupl K, Lissner J, Felix R. Dural sinus thrombosis: value of venous MR angiography for the diagnosis and follow-up. Radiology. 1994;162:1191-1198.
Ross MR, Pelc NJ, Enzmann DR. Qualitative phase contrast MRA in the normal and abnormal circle of Willis. AJNR Am J Neuroradiol. 1993;14:19-25.
Fujita N, Hirabuki N, Fujii K, Hashimoto T, Miura T, Sato T, Kozuka T. MR imaging of middle cerebral artery stenosis and occlusion: value of MR angiography. AJNR Am J Neuroradiol. 1994;15:335-341.
Edelman RR, Mattle HP, O’Reilly GV, Wentz KU, Cheng L, Zhao B. Magnetic resonance imaging of flow dynamics in the circle of Willis. Stroke. 1990;21:56-65.
Tsuruda J, Saloner D, Norman D. Artifacts associated with MR neuroangiography. AJNR Am J Neuroradiol. 1992;13:1411-1422.
Pernicone JR, Siebert JE, Laird TA, Rosenbaum TL, Potchen EJ. Determination of blood flow direction using velocity-phase image display with 3D phase contrast MR angiography. AJNR Am J Neuroradiol. 1992;13:1435-1438.
Pernicone JR, Siebert JE, Potchen EJ, Pera A, Dumoulin CL, Souza SP. Three-dimensional phase-contrast MR angiography in the head and neck. AJNR Am J Neuroradiol. 1990;155:457-466.
Bradac GB. Angiography in cerebral ischemia. Riv Neuroradiol. 1990;3(suppl 2):57-66.
Special report from the National Institute of Neurological Disorders and Stroke: classification of cerebrovascular diseases III. Stroke. 1990;21:637-676.
Ring BA. Diagnosis of embolic occlusions of smaller branches of the intracerebral arteries. AJR Am J Radiol. 1966;97:575-582.
Bozzao L, Bastianello S, Fantozzi LM, Angeloni U, Argentino C, Fieschi C. Correlation of angiographic and sequential CT Findings in patients with evolving cerebral infarction. AJNR Am J Neuroradiol. 1989;10:1215-1222.