| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Radiology, David Geffen School of Medicine, University of
California, Los Angeles, 10945 Le Conte Ave., Ste. 3371, Los Angeles, CA
90095.
2 Siemens Medical Solutions, Los Angeles, CA.
Received August 16, 2005;
accepted after revision October 16, 2005.
Address correspondence to K. Nael
(nkambiz{at}mednet.ucla.edu).
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
SUBJECTS AND METHODS. Thirty-two consecutive patients (13 men, 19 women; age range, 28-88 years) with suspected abdominal arterial disease underwent abdominal 3D CE-MRA on a 3.0-T MR system, using a high-spatial-resolution (0.7 x 0.82 x 0.8 mm3) 3D gradient-refocused echo (GRE) sequence, integrated with a generalized autocalibrating partially parallel acquisitions (GRAPPA) technique with an acceleration factor of 3. Two vascular radiologists evaluated image quality and the presence and degree of arterial stenoses. Interobserver variability was calculated, using the kappa coefficient. The sensitivity and specificity of the technique were calculated and comparative analysis was performed with those of conventional catheter angiography (in eight patients) as the standard of reference.
RESULTS. The abdominal arterial vasculature was visualized with
diagnostic image quality in all subjects. Arterial stenoses were detected in
148 and 142 arterial segments by observer 1 and observer 2, respectively, with
good interobserver agreement (
= 0.75; 95% confidence interval [CI]:
0.69-0.81). The sensitivity and specificity values for CE-MRA for the
detection of significant (> 50%) arterial stenoses were 100% and 96% for
observer 1 and 100% and 92% for observer 2, respectively. There was a
significant correlation between CE-MRA and conventional angiography
(R = 0.96 and 0.93 for observers 1 and 2, respectively) for the
assessment of the degree of stenosis.
CONCLUSION. The outlined MR angiography protocol at 3.0 T combined with parallel acquisition technique renders highly reliable and isotropic high-spatial-resolution imaging of the abdominal vasculature.
Keywords: abdomen • conventional angiography • MR angiography • MR technique • 3.0 T or high magnetic field • parallel acquisition
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The introduction of parallel imaging [10-12] is one of the most exciting advances in MRI. Implementation algorithms such as simultaneous acquisition of spatial harmonics (SMASH) [12] and sensitivity encoding (SENSE) [10] or generalized autocalibrating partially parallel acquisitions (GRAPPA) [11] provide the framework for improved spatial resolution, temporal resolution, or both. In general, with parallel imaging the individual signals from component coil elements in a radiofrequency coil array are used to substitute for some phase-encoding gradient steps, which can then be omitted. This allows either a reduction of the acquisition time or an increase in the spatial resolution by what is termed the "acceleration factor." However, the signal-to-noise ratio (SNR) penalty of parallel imaging, depending on the degree of k-space undersampling, limits the maximum acceleration factor to approximately 2 for most applications at 1.5 T.
The introduction of 3.0-T MR systems into clinical practice, with higher available SNR, has the potential to improve the performance of CE-MRA significantly. The higher SNR gain at 3.0 T can be used to reduce acquisition time and improve spatial resolution, and it is advantageous when parallel imaging with higher acceleration factors is considered.
The purpose of this study was to evaluate the diagnostic performance of a CE-MRA protocol at 3.0 T in a population of patients with suspected abdominal arterial disease. Applying parallel acquisition technique with a high acceleration factor (GRAPPA x3), the protocol allows acquisition of high-spatial-resolution data with isotropic voxel volumes over an extended coverage within a single breath-hold.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The clinical indications for MRA were as follows: uncontrolled high blood pressure (n = 28), clinical suspicion of abdominal aortic aneurysm (n =2), and suspected mesenteric ischemia (n = 2). All studies were performed in accordance with the institutional review board guidelines under an approved protocol. Written informed consent for study participation was obtained from all subjects. Eight patients underwent conventional catheter angiography within a maximum of 2 weeks after the MR examination.
Imaging Technique
All studies were performed on a whole-body 3.0-T imaging system (Magnetom
Trio, Siemens Medical Solutions) with a fast gradient system (peak gradient
amplitude, 40 mT/m; slew rate, 200 mT/m/ms). Subjects were entered headfirst
in the supine position into the magnet. The arms were positioned above the
head and supported with padding and fastener straps to avoid wraparound
artifact. For signal reception, a commercially available wraparound body array
coil was placed around the abdomen. The coil has four linear elements in a
left-to-right direction for both the anterior and posterior components.
Initial survey MR images of the abdomen were acquired in all subjects with non-breath-hold fast imaging with steady-state free-precession (SSFP) in coronal and transverse planes. After IV injection of 2 mL of gadolinium-based contrast agent (gadodiamide [Omniscan, Amersham-GE Healthcare]), a timing bolus was used to measure the transit time between the injection site and the abdominal aorta. The mean delay time for the contrast arrival was 23 seconds (range, 19-30 seconds).
Subsequently, high-spatial-resolution CE-MRA was performed in the coronal plane by using a fast 3D Fourier transform gradient-refocused echo (GRE) sequence with spoiling gradients. An asymmetric k-space sampling scheme (6/8 partial Fourier) and zero interpolation were applied in all three axes to minimize the TE and the acquisition time. Parallel imaging was performed with a GRAPPA algorithm based on autocalibrating SMASH and parallel acquisition [11]. Because of the integrated reference scan, a linear k-space acquisition mode had to be chosen. The GRAPPA parameters were set to a 30 reference k-space lines for calibration in the right-to-left phase-encoding direction and an acceleration factor of 3. Because the integrated parallel acquisition technique (iPAT) requires an additional 20 lines in the center of k-space to measure the coil sensitivity distribution, the effective time saving factor therefore was 2.5. These settings allowed for the acquisition of a full 3D data set with an isotropic spatial resolution of 0.70 x 0.82 x 0.96 (0.8 interpolated) mm3 in a 22-second breath-hold. Table 1 details the imaging parameters.
|
The contrast injection protocol involved 0.2 mmol/kg gadodiamide, injected at a rate of 1.5 mL/s followed by 30 mL of saline at the same rate using an electronic power injector (Spectris, Medrad). Patients were instructed to hyperventilate briefly before MRA and to hold their breath during data acquisition, which was timed to coincide with the arrival of the contrast bolus to the abdominal aorta.
Image Processing
After data acquisition, image processing was performed on a commercially
available workstation (Leonardo, Siemens Medical Solutions), using a
maximum-intensity-projection (MIP) algorithm. The entire 3D volume was
reconstructed in thin MIP (10 mm thick with 9-mm overlap) and full-thickness
rotational MIPs (cover 360° with 10° increments) for each 3D data set.
For image interpretation, the reconstructed data, along with the source data,
were available for both observers on the workstation. Postprocessing time was
approximately 15 minutes per patient.
Catheter Angiography
Catheter angiography was performed by an interventional radiologist with 25
years of experience. After femoral artery catheterization, an abdominal
aortogram was obtained, followed by selective catheterization of the target
abdominal vessels such as the renal, celiac, and mesenteric arteries. The
injected volume of nonionic contrast medium (Omnipaque 240 [iohexol],
Amersham-GE Health) was 7-10 mL for each injection. Images were obtained in
anteroposterior, lateral, and, if necessary, two oblique projections (-45°
and 45°). Eight patients underwent conventional catheter angiography
within a maximum of 2 weeks after the MR examination (1-14 days). As a result,
conventional angiography images were available for 54 arterial segments for
comparative analysis.
Image Analysis
MR angiograms were interpreted independently by two vascular radiologists
with at least 5 years of experience. The study coordinator, who attended all
interpretation sessions, arranged separate image-interpreting sessions for
both observers. The observers were instructed to use the postprocessed data in
a first step and, if necessary, to use the source data for interactive
reformatting in a second step.
Catheter angiography images, which served as the standard of reference, were interpreted by both observers who had reached agreement by consensus. All interpretations (MRA and conventional angiography) were performed blinded to the other examinations and to the patient data.
For analysis purposes, the arterial vascular system was divided into 15 segments (Table 2). When accessory renal arteries were present, these arteries were also included for image analysis.
|
The image quality and visualization of each arterial segment was evaluated by each observer independently, using a 1-4 scoring scale: 1 = poor image quality and blurring of the arterial segment; 2 = fair image quality, not inadequate arterial enhancement for confident diagnosis; 3 = good image quality and arterial enhancement, adequate for confident diagnosis; 4 = excellent image quality and arterial enhancement, for highly confident diagnosis. Image quality of an arterial segment was rated to be diagnostic (3 3) if all clinically relevant diagnostic information could be obtained with good differentiation of arterial vasculature from background tissue. Image quality was considered nondiagnostic if diagnostic information could not be derived because of blurring of the arterial segment or inadequate vessel enhancement. Contaminating venous signal was scored on a scale of 0-2 (none or minimal = 0; mild to moderate, not interfering with diagnosis = 1; significant, interfering with diagnosis = 2).
Arterial disease in each arterial segment was evaluated by each observer separately, applying a 1-4 grading scale (vessel irregularity, < 10% luminal narrowing = 1; mild stenosis [10-50%] = 2; significant stenosis [51-99%] = 3; and occlusion = 4). When two or more stenotic luminal changes were detected in the same vessel segment, the most severe change was used for grading and analysis. Aneurysm was diagnosed in the presence of a focal increase in arterial diameter that exceeded the diameter of the adjacent vessel by more than 50%. Changes consistent with fibromuscular dysplasia (FMD) were noted separately.
Statistical Evaluation
A Wilcoxon's rank sum test was used to evaluate the significance of the
image quality grading differences between two observers (p value of
< 0.05 was an indicator of statistically significant difference).
Interobserver agreement for the definition and image quality of arterial
segments and the assessment of the degree of stenoses between two observers
were determined by calculating the kappa coefficient (poor agreement,
= 0; slight agreement, = 0.01-0.2; fair agreement, = 0.21-0.4;
moderate agreement, = 0.41-0.6; good agreement, = 0.61-0.8;
and excellent agreement, = 0.81-1)
[13]. The relationship between
CE-MRA and catheter angiography in terms of categorized stenosis was analyzed
using Spearman's rank correlation coefficient (R). The sensitivity
and specificity of MRA for stenosis of more than 50% were calculated for each
observer.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The abdominal aorta and its major intraabdominal branches were identified
by both observers with the definition in the diagnostic range
(Table 2). Comparison of the
image quality scores between two observers did not show any significant
difference (p = 0.88), and the overall interobserver agreement was
good ( = 0.74). Eight arterial segments were not rated or evaluated
because of the presence of a stent in three segments (two renal and one
external iliac artery), one missing renal artery related to a previous
nephrectomy, and four arterial segments that were not included in the 3D data
set (two common hepatic and two gastroduodenal arteries). In addition to the
described arterial segments, both observers identified 14 additional accessory
renal arteries, resulting in a total number of 486 arterial segments available
for evaluation. Of these, the image quality of 14 and 12 arterial segments
were graded as not sufficient for diagnosis by observers 1 and 2,
respectively.
Venous contamination was scored as none to minimal in 11 patients (35%) or mild to moderate in 21 (65%) subjects, and it never interfered with the diagnosis.
Evaluation of Arterial Disease
Observer 1 graded 338 arterial segments as nonstenotic and 148 arterial
segments with disease, including 93 arterial irregularities (luminal narrowing
< 10%), 17 segments with mild stenoses (10-50%), 32 segments with
significant stenoses (51-99%), and six segmental occlusions.
Observer 2 graded 344 arterial segments as nonstenotic and 142 arterial segments with stenoses including 80 arterial irregularities (luminal narrowing < 10%), 20 segments with mild stenoses (10-50%), 36 segments with significant stenoses (51-99%), and six segmental occlusions.
The comparison of the arterial stenoses grading scores between the two
observers was not significant (p = 0.67). There was good
interobserver agreement ( = 0.75; 95% confidence interval [CI],
0.69-0.81) for the grading of all the degrees of arterial stenoses (grades
1-4). Table 3 shows a summary
of the interobserver correlation for the detection of arterial stenoses.
|
Of a total number of 54 arterial segments, catheter angiography showed 30 segments without stenoses, 10 segments with mild irregularities (< 10%), five segments with mild stenoses (10-50%), seven significant stenoses (51-99%), and two segmental occlusions. When compared with CE-MRA, both observers correctly identified all significant stenoses and occlusions (grades 3 and 4). However, one overestimation and one underestimation of a mild stenosis for observer 1 and two overestimations of mild stenosis for observer 2 occurred. A significant correlation between catheter angiogram and CE-MRA was found for the degree of stenoses for both observer 1 (R = 0.96, p < 0.0001) and observer 2 (R = 0.93, p < 0.0001) (Table 4). The sensitivity and specificity of CE-MRA for the depiction of stenoses greater than 50% was 100% and 96% for observer 1 and 100% and 92% for observer 2, respectively.
|
Figures 1A, 1B, and 1C shows the CE-MRA and comparison catheter angiogram of a patient referred for the evaluation of renovascular hypertension. Figures 2A, 2B, 2C, and 2D shows the CE-MRA and comparison catheter angiogram in a patient with severe abdominal angina.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Despite the promising results of abdominal 3D CE-MRA with sensitivities and specificities of more than 90% for the detection of significant arterial disease [4, 5, 7, 15-17], most CE-MRA techniques, even with state-of-the-art systems, still operate with an in-plane resolution of above 1 mm2 and a voxel volume of above 5 mm3 [18-20] compared with conventional angiography in which an in-plane resolution of up to 0.3 mm2 is achievable.
Both relatively lower spatial resolution of MRA in comparison to alternative diagnostic techniques such as CTA or conventional catheter angiography and larger voxel volumes may result in lower diagnostic accuracy, particularly in high-grade stenosis or in small-caliber vessels.
Advances in the development of fast MRA sequences with short readout time and the introduction of parallel acquisition techniques [10-12] have the potential to improve the spatial resolution of the MR applications significantly.
Parallel imaging techniques such as GRAPPA [11] were recently introduced. These techniques use the spatial distribution of the MR signal received by multiple radiofrequency array coils, so the number of phase-encoding gradient steps can be reduced without decreasing the spatial resolution or matrix size of the acquired image. This permits either the reduction of acquisition time or an increase of the spatial resolution by a degree defined as the "acceleration factor." The theoretical maximum attainable acceleration factor is equal to the number of coil elements along the phase-encoding direction. In practice, however, SNR represents a fundamental challenge and is increasingly limiting as acceleration advances. The SNR limitation of parallel imaging manifests in a spatially varying noise amplification pattern that intrinsically depends on both the degree of k-space sampling and the coil array geometry [10, 21]. An increase of the acceleration factor is paralleled by an increase in image noise, which can limit the SNR available at 1.5 T [12, 22].
The main advantage of MRI at 3.0 T is that the SNR gain scales approximately linearly with the field strength [23, 24]. The higher SNR gain at 3.0 T can be used flexibly to reduce acquisition time, improve spatial resolution, and support more aggressive parallel acquisition [23-26], and it would likely improve visualization of small-vessel segments [23, 24, 27, 28], as shown in our study.
The experience with CE-MRA at 3.0 T for imaging of the abdominal arteries is limited. A recent report showed the feasibility of CE-MRA with voxel volumes in the range of 1 mm3 for imaging of the renal arteries at 3.0 T [26]. To our knowledge, our study is the first to address the clinical value of abdominal CE-MRA at 3.0 T. The results of this study indicate that high-spatial-resolution 3D CE-MRA at 3.0 T is promising for a comprehensive evaluation of the abdominal arteries. The option to acquire data with higher SNR at 3.0 T has prompted the more aggressive use of a parallel acquisition technique (GRAPPA) with an acceleration factor of 3. Thus generation of high-spatial-resoluion 3D data sets with voxel size on the order of 0.55 mm3 with highly diagnostic image quality over an extended coverage has become possible. This is reflected in the favorable comparative analysis with catheter angiography and the good interobserver agreement.
The relative short acquisition time and test injection technique used for accurate measurement of the contrast timing window proved to be adequate to avoid significant venous contamination in most of the examinations.
In this study, the image quality of the minority of small arterial segments (2%) was graded as not sufficient for diagnosis, mainly as a result of motion artifact. Although very trivial, this is almost inevitable in clinical practice and can be avoided by providing the patient with better breathing instructions, further shortening the scanning time, or both.
The obtained overall sensitivities and specificities in this study exceed 90%, in accordance with the results of previously published studies at 1.5 T [4, 5, 7, 15-17]. This may indicate that by having an isotropic high-spatial-resolution 3D data set, the times of gross overestimation of stenoses with CE-MRA probably have passed.
Our study was limited by the small number of conventional angiograms available that provided comparative image analysis for only 54 arterial segments in eight patients. However, this circumstance reflects the decreasing role of conventional angiography for the diagnosis of arterial disease. In this study, conventional angiography was performed only in patients with severe arterial stenosis as part of a therapeutic intervention. This selection bias may be partially responsible for the high values of the sensitivities and specificities in our study. These numbers may be different in a broader clinical setting or if catheter angiography is available for comparison in all patients who undergo MRA.
In conclusion, 3D CE-MRA of the abdominal arteries at 3.0 T is feasible and promising, reliably displaying the abdominal arteries. The initial results indicate that the implementation of 3D CE-MRA with high-acceleration parallel acquisition at 3.0 T can provide isotropic high-spatial-resolution data over an extended coverage and within a comfortable breath-hold. The technique is highly accurate for the detection of abdominal arterial disease, although further studies are required to confirm the accuracy of the technique in a broader clinical setting.
It seems likely that the introduction of arrays of more sensitive coils and flexible table movement at 3.0 T will allow extended field-of-view high-spatial-resolution CE-MRA throughout the whole body, as with 1.5 T [29].
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  U. Kramer, J. Wiskirchen, M. C. Fenchel, A. Seeger, G. Laub, G. Tepe, J. P. Finn, C. D. Claussen, and S. Miller Isotropic High-Spatial-Resolution Contrast-enhanced 3.0-T MR Angiography in Patients Suspected of Having Renal Artery Stenosis Radiology, April 1, 2008; 247(1): 228 - 240. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
