| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1
Department of Radiology, Hamamatsu University School of Medicine, 3600 Handa,
Hamamatsu 431-3192, Japan.
2
General Electric Yokogawa Medical Systems, Ltd., 4-7-127 Asahigaoka, Hino,
Tokyo 191-8503, Japan.
Received August 28, 2000;
accepted after revision October 17, 2000.
Address correspondence to H. Masunaga.
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
SUBJECTS AND METHODS. Thirty-nine patients underwent MR angiography using an ultrafast 3D Fourier transform spoiled gradient-recalled acquisition in the steady state (TR/TE range, 2.6/0.7-0.8). Five seconds after administration of 15-20 mL gadodiamide hydrate, four or five consecutive data sets with imaging times of 7.0-7.6 sec were acquired during a single breath-hold. A timing examination was not performed. Image quality was assessed using quantitative analysis (signal-to-noise, contrast-to-noise, and venous-to-arterial enhancement ratios) and qualitative analysis (presence of venous overlap, presence of artifacts, and degree of renal arterial enhancement). MR angiography depiction of the renal artery stenosis was evaluated using conventional angiography as the standard of reference.
RESULTS. On the best arterial phase, average aortic signal-to-noise ratio (±SD) was 74.5 ± 24.4, aorta-to—inferior vena cava contrast-to-noise ratio was 70.8 ± 23.4, and inferior vena cava—to-aorta venous-to-arterial enhancement ratio was 0.03 ± 0.04. No venous overlap was seen in 38 of 39 patients. Substantial enhancement of renal arteries was seen in all patients without any noticeable artifacts. MR angiography correctly depicted the degree of stenosis in 44 of 47 normal arteries, 13 of 16 mildly stenotic arteries, five of five moderately stenotic arteries, three of four severely stenotic arteries, and one of one occluded artery. Sensitivity and specificity for revealing greater than 50% stenosis was 100%.
CONCLUSION. Time-resolved 3D MR angiography can provide high-quality arteriograms. Its performance in revealing renal artery stenosis is comparable with that of conventional angiography.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
MR angiography has a number of advantages over conventional angiography. It is noninvasive and uses gadolinium-based contrast material, which has a lower risk of nephrotoxicity [2], and provides sufficient vascular enhancement without saturation effects [3]. In addition, the introduction of high-performance gradients and improved pulse sequences has allowed breath-hold three-dimensional (3D) MR angiography to produce high-quality images with adequate spatial resolution for diagnostic imaging. During the last several years, many groups of researchers have reported its accuracy in revealing renal artery stenosis [4,5,6,7,8,9,10,11,12,13]. Recently, gadolinium-enhanced time-resolved 3D MR angiography with an acquisition time of less than 10 sec has been implemented, resulting in reports of its usefulness in assessment of vessel abnormalities [14,15,16]. In theory, the rapid acquisition method has some benefits compared with non—time-resolved 3D MR angiography that requires longer imaging time (20-30 sec). With the use of consecutive rapid acquisitions, at least one data set will be obtained during arterial peak enhancement, producing a sufficiently enhanced arteriogram without venous overlap. In addition, by consecutively acquiring several data sets, no timing examination is needed.
The purpose of this study was to perform a clinical feasibility assessment of image quality of time-resolved 3D MR angiography and to evaluate its accuracy in revealing renal artery stenosis compared with using the findings of conventional angiography as the standard of reference.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
MR Imaging
All patients were examined using a 1.5-T MR scanner (Signa Horizon
EchoSpeed; General Electric Medical Systems, Milwaukee, WI) with an enhanced
gradient system with a gradient capability of 2.3 G/cm and a rise time of 190
µsec. A phased array body surface coil was used. The parameters for
time-resolved 3D MR angiography using an ultrafast 3D Fourier transform
spoiled gradient-recalled acquisition in the steady-state sequence were TR/TE
range of 2.6/0.7-0.8 msec, 20° flip angle, and 62.5 kHz. The matrix was
256 x 128 over a 34- to 38-cm three-quarter rectangular field of view.
The average pixel size was 1.37 x 2.74 mm. The acquisition volume, which
was adapted to ensure coverage of the entire length of the abdominal aorta and
its major branches, ranged from 86.4 to 144 mm (mean, 105.6 mm). After
zero-filled interpolation (factor of 2), the 48 contiguous sections had an
effective section thickness of 4-6 mm. Sequential data acquisition was used
for full k-space ordering. The acquisition time for each 3D data set was
7.0-7.6 sec (mean, 7.5 sec). Oxygen was administrated at a rate of 2-4 L/min
by nasal cannula for patients who had difficulty in suspending respiration. A
dose of 0.1-0.2 mmol/kg (mean, 0.14 mmol/kg) of gadodiamide hydrate (Omniscan;
Daiichi-Pharmaceutical, Tokyo, Japan) was administered IV through a 20- or
22-gauge peripheral catheter. The contrast material was injected using a power
injector (Nemoto Kyorindo, Tokyo, Japan) followed by a 20-mL saline flush at a
rate of 2 mL/sec. Data acquisition started 5 sec after the commencement of
contrast medium injection. Depending on each patient's respiratory condition,
four or five acquisitions were repeated for 28-38 sec during a single
breath-hold. In addition, two to four acquisitions were performed to obtain
venous phase images after the patients caught their breath
(Fig. 1).
|
All MR angiograms were postprocessed on a computer workstation (Advantage Windows; General Electric Medical Systems). Additional 3D data sets (subtracted 3D data sets) were calculated by subtracting the first precontrast data set (the mask) from the subsequent postcontrast data sets.
Conventional Angiography
Conventional angiography was performed within 0 to 28 days (mean, 13 days)
of the MR examination. Anteroposterior abdominal aortic intraarterial digital
subtraction angiography was performed with a 5-French pigtail catheter
positioned at or above the level of the renal arteries. A dose of 20-40 mL of
Iopamidol (Iopamiron 370; Nihon-Schering, Osaka, Japan) was injected at a rate
of 10-20 mL/sec. If necessary, selective renal arterial catheterization was
performed using a 5-French shepherd-hook catheter.
Image Analysis
Image quality.—For quantitative analysis of image quality,
mean signal intensities were measured from both original MR angiographic
images and the subtracted images in the following locations: the aorta at the
level of the renal arteries, the right and left renal arteries, the suprarenal
inferior vena cava, the right and left renal veins, retroperitoneal fat, and
paraspinal muscle. Images acquired during the venous phase were used to define
the locations of the venous structures on the precontrast and arterial phase
images. The signal-to-noise ratios of these measurements were calculated using
the standard deviation of the aortic noise as previously reported
[3]. Aorta-to—inferior
vena cava, right renal artery—to—right renal vein, left renal
artery—to—left renal vein, aorta-to—retroperitoneal fat, and
aorta-to—paraspinal muscle contrast-to-noise ratios were also
calculated. In addition, the venous-to-arterial enhancement ratios were
calculated for inferior vena cava—to-aorta, right renal
vein—to—right renal artery, and left renal
vein—to—left renal artery by the following equation:
venous-to-arterial enhancement ratio = (arterial phase V — precontrast
V) / (arterial phase A — precontrast A), where V represents venous
signal intensity and A represents arterial signal intensity. Ideally, this
ratio should be zero [17,
18]. Differences between
contrast-to-noise ratios of the original and those of the subtracted images
were determined using the Student's t test for paired data.
For qualitative analysis of images, each patient's subtracted 3D data set of arterial phase, which was defined as the phase with maximal renal artery—to—renal vein contrast-to-noise ratio, was evaluated by two experienced radiologists in consensus. The evaluation was performed using maximum-intensity-projection algorithms and multiplanar reformatting algorithms on the computer workstation. Image quality was assessed for presence of venous overlap, presence of artifacts, and degree of renal arterial enhancement. For the assessment of presence of venous overlap, grading was applied as follows: 0, renal arteries were clearly visible without venous overlap; 1, renal veins were barely visible; and 2, renal veins were depicted equal to or stronger than renal arteries. For the assessment of presence of artifacts, grading was applied as follows: 0, artifacts were absent; 1, mild artifacts were present; and 2, severe artifacts were present. For assessment of degree of renal arterial enhancement, grading was applied as follows: 0, no enhancement; 1, mild enhancement; and 2, substantial enhancement.
Detection of renal artery stenosis.—For detection of the number of renal arteries and the degree of renal artery stenosis, each patient's subtracted 3D data set of the arterial phase was evaluated by two experienced radiologists in consensus. They were unaware of the results of conventional angiography and other clinical information. The evaluation was performed with maximum-intensity-projection algorithms, multiplanar reformating algorithms, and cine-loop display of source images on the computer workstation. The renal arteries were assessed from the aorta to the renal hilum. The sensitivity of stenosis was categorized with the following criteria: grade 0, no stenosis; grade 1, mild stenosis (1-49%); grade 2, moderate stenosis (50-69%); grade 3, severe stenosis (70-99%); and grade 4, occlusion.
The severity was assessed by measuring the width of the narrowest section of the stenotic segment and comparing it with the width of the uninvolved segment of the renal artery proximal or distal to the stenosis. A vessel with totally interrupted intraarterial high signal intensity was defined as an occlusion. The sensitivity and specificity in revealing renal artery stenosis were calculated using the findings of conventional angiography as the standard of reference.
Conventional angiograms were reviewed by two vascular interventional radiologists in consensus. They were unaware of the MR angiographic results and other clinical information. The number of renal arteries and the degree of renal artery stenosis were recorded with the same grading protocol used for MR angiography.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Image Quality
The signal-intensity measurements are summarized in
Table 1. Before contrast medium
administration, all but fat signal intensities were similar. In the arterial
phase, significant arterial enhancement was seen, but the enhancement of
veins, fat, and muscle was negligible. This resulted in high artery-to-vein
contrast-to-noise, artery-to-muscle contrast-to-noise, and nearly zero
venous-to-arterial enhancement ratios. Using the subtraction method,
aorta-to-fat contrast-to-noise ratio was improved over that found before
subtraction with a statistically significant difference (p <
0.0001). Figure 2 illustrates
the changes of vascular and background tissue signal-to-noise at the different
phases on the subtracted time-resolved 3D MR angiograms in 28 patients in whom
arterial phase images were obtained in the third data set.
|
|
In 38 of 39 patients, no venous overlap (grade 0) was visible on MR angiograms. In one patient, both renal veins were barely visible (grade 1). In all 39 patients, no artifacts affected image quality (grade 0), and substantial enhancement of renal arteries was seen (grade 2).
Detection of Renal Artery Stenosis
Inclusive of six accessory renal arteries, 84 renal arteries were
visualized on time-resolved 3D MR angiography as well as conventional
angiography (Figs.
3A,3B,3C,4A,4B,4C,4D,4E,5A,5B).
For the degree of renal artery stenosis, 73 of 84 renal arteries were
evaluated. Eleven renal arteries were excluded from the analysis, because
their conventional angiograms were nondiagnostic due to super-imposition on
other structures (n = 7) or poor contrast medium filling caused by
branching from false lumina in aortic dissections (n = 4) (Fig.
4A,4B,4C,4D,4E).
Results comparing time-resolved 3D MR angiography with conventional
angiography are shown in Table
2. Time-resolved 3D MR angiography correctly depicted 44 of 47
normal arteries (grade 0) (Fig.
3A,3B,3C),
13 of 16 mildly stenotic arteries (grade 1), five of five moderately stenotic
arteries (grade 2), three of four severely stenotic arteries (grade 3) (Fig.
5A,5B),
and one of one occluded artery (grade 4). Three normal arteries (grade 0) were
overestimated as mildly stenotic (grade 1), three mildly stenotic arteries
(grade 1) were underestimated as normal (grade 0), and one severely stenotic
artery (grade 3) was underestimated as moderately stenotic (grade 2). For the
diagnosis of mildly stenotic (grade 1), moderately stenotic (grade 2),
severely stenotic (grade 3), and occluded (grade 4) lesions, the sensitivity
was 81%, 100%, 75%, and 100%, respectively, and the specificity was 95%,
98.5%, 100%, and 100%, respectively. Sensitivity and specificity for revealing
greater than 50% stenosis were 100%.
|
|
|
|
|
|
|
|
|
|
|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The time-resolved 3D MR angiographic technique used in this study produced superior images for the following reasons. Our contrast medium injection time is 5-10 sec. According to Van Hoe et al. [16], using similar technique, the minimal bolus prolongation is estimated as 5-7 sec. Therefore, it is assumed that contrast medium concentration is maintained for at least 10-17 sec. The acquisition time of 7.0-7.6 sec for each data set using time-resolved 3D MR angiography is much shorter than the 10- to 17-sec duration. Consequently, with this method, k-space can be filled while a high concentration of contrast medium bolus remains in the arteries and little has reached the veins, and clear visualization of arterial vessels without venous overlap is provided. In non—time-resolved 3D MR angiography, the acquisition time of 20-30 sec is too lengthy to fill the entire k-space during peak arterial enhancement.
Short acquisition time can minimize motion artifacts caused by respiration and intestinal peristalsis. Furthermore, intravoxel dephasing is avoided using a short TE (0.7-0.8 msec), and it may help to decrease the overestimation of stenoses at the stenotic portion caused by turbulent flow [5, 18]. A short TE is also effective in minimizing susceptibility artifacts produced by bowel gas and metallic implants.
The image subtraction method allows better depiction of vessels by eliminating the background signals that are mainly produced by surrounding adipose tissues with short T1 values. Subtraction also eliminates wraparound artifacts when a limited field of view is used to shorten the data acquisition time [14]. However, subtraction may lead to misregistration artifacts caused by motion between mask image acquisition and postcontrast image acquisition. Time-resolved 3D MR angiography can keep these misregistration artifacts negligibly small by acquiring precontrast data together with postcontrast data in a single breath-hold. In non—time-resolved 3D MR angiography, subtraction is not as useful because each data set of the mask or postcontrast image acquisition requires separate breath-holds, which may result in inconsistent breath-holding positions.
In contrast-enhanced MR angiography, the signal intensity of the vessels is determined by the concentration of contrast medium within the vessels while data are being acquired at the central portion of the k-space (low-frequency component) [21]. In non—time-resolved 3D MR angiography with a long acquisition time compared with arterial peak enhancement, it is necessary to strictly synchronize the data acquisition of the k-space center with the arrival of the contrast medium bolus. For this reason, various timing examinations have been developed, including test bolus injection [8,9,10, 13, 18, 19], automatic triggering [20], and fluoroscopic triggering [17]. In time-resolved 3D MR angiography, however, a data set of optimal arterial phase can be selected from four or five consecutive data sets without using a timing examination.
In addition, non—time-resolved 3D MR angiography with a timing examination will not reveal entire vessels that have different flow velocities because of differences in transit time of the contrast medium bolus. In a patient with aortic dissection, if data acquisitions are synchronized with the true lumen, the depiction of the false lumen and its branch vessels may not be adequate. In a patient who has undergone bypass grafting from the subclavian artery because of abdominal aortic occlusion, it will be difficult to show both the bypass grafting and the abdominal aorta proximal to the occlusive portion. On the other hand, time-resolved 3D MR angiography can show each vessel with different flow velocities in different time-frame images, and it also provides additional information about the hemodynamics of affected vessels (Fig. 4A,4B,4C,4D,4E).
This study has some limitations. First, a small number of patients were evaluated. For confirmation of sufficient detectability of renal artery stenosis, a larger group of patients must be studied. Second, 18 of the 39 patients underwent graft replacement of the abdominal aorta with infrarenal aortic clamping between MR examinations and conventional angiography. Previous studies reported that infrarenal aortic clamping might produce sustained alterations in renal hemodynamics, but it was completely reversible in most patients [22, 23]. In our patients, the serum creatinine levels did not change significantly before or after the surgery. Therefore, we believe that no significant hemodynamic change occurred in the patients after the surgical intervention. Third, although we believe time-resolved 3D MR angiography will be superior to non—time-resolved 3D MR angiography at many points as stated in the "Discussion" section, this study is lacking in direct correlation between these two MR angiographic techniques. Fourth, time-resolved 3D MR angiography sacrificed the spatial resolution to shorten the acquisition time, especially in slice-select directions (antero-posterior directions in this study). Therefore, renal artery stenoses having their origin in the anteroposterior directions could theoretically be overlooked. In spite of this limitation, time-resolved 3D MR angiography was quite consistent with digital subtraction angiography findings in this study. We believe our findings were a result of the following: anteroposterior digital subtraction angiography, which was used as the standard of reference in this study, also had limitations caused by two-dimensional projection images. Finally, although all patients were successfully examined using the current protocol, patients who have a long circulation time will not have sufficiently clear MR angiograms. For patients with suspected heart failure, more than five consecutive data sets or a longer imaging delay may be required. On the other hand, for patients with short circulation time, the imaging delay of 5 sec will be too long to obtain precontrast images as the mask. In previous reports, four to five data sets with an imaging delay of 5 sec [15], and six data sets with an imaging delay of 10 sec [16] were acquired. Further investigation will be required to optimize the number of data sets and imaging acquisition delay.
In conclusion, time-resolved 3D MR angiography can provide high-quality arteriograms free of venous overlap without using any timing examination. The performance in revealing renal artery stenosis is comparable with that of conventional angiography. Time-resolved 3D MR angiography is considered to be a useful substitute for conventional angiography in evaluating renovascular disease.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  C. Y. Kim, R. A. Mirza, J. A. Bryant, E. D. Whiting, D. M. Delong, C. E. Spritzer, and E. M. Merkle Central Veins of the Chest: Evaluation with Time-resolved MR Angiography Radiology, May 1, 2008; 247(2): 558 - 566. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. Soulez, M. Pasowicz, G. Benea, L. Grazioli, J. P. Niedmann, M. Konopka, P. C. Douek, G. Morana, F. K. W. Schaefer, A. Vanzulli, et al. Renal Artery Stenosis Evaluation: Diagnostic Performance of Gadobenate Dimeglumine-enhanced MR Angiography--Comparison with DSA Radiology, April 1, 2008; 247(1): 273 - 285. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Prokop, G. Schneider, A. Vanzulli, M. Goyen, S. G. Ruehm, P. Douek, M. Dapra, G. Pirovano, M. A. Kirchin, and A. Spinazzi Contrast-enhanced MR Angiography of the Renal Arteries: Blinded Multicenter Crossover Comparison of Gadobenate Dimeglumine and Gadopentetate Dimeglumine Radiology, February 1, 2005; 234(2): 399 - 408. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. L. Coenegrachts, R. M. Hoogeveen, J. A. Vaninbroukx, H. T. Bosmans, D. J. Bielen, G. Maleux, F. Maes, P. Hamaekers, R. H. Oyen, and G. J. Marchal High-Spatial-Resolution 3D Balanced Turbo Field-Echo Technique for MR Angiography of the Renal Arteries: Initial Experience Radiology, April 1, 2004; 231(1): 237 - 242. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. Golman, J. H. Ardenkjaer-Larsen, J. S. Petersson, S. Mansson, and I. Leunbach Molecular imaging with endogenous substances PNAS, September 2, 2003; 100(18): 10435 - 10439. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
= aorta,
= right renal artery, 
= left renal artery, x = inferior
vena cava, asterisks = right renal vein, [UNK] = left renal vein, + = fat,
horizontal lines = muscle.
