| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Radiological Sciences, David Geffen School of Medicine,
University of California at Los Angeles, 10945 Le Conte Ave., Ste. 3371, Los
Angeles, CA 90095-7206.
2 Department of Radiology, University Hospitals Grosshadem,
Ludwig-Maximilians-University Munich, Munich, Germany.
3 Siemens Medical Solutions, Malvern, PA.
Received December 8, 2005;
accepted after revision March 15, 2006.
Address correspondence to K. Nael.
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
SUBJECTS AND METHODS. Fifty consecutive patients with suspected PVD (31 men, 19 women; age range, 46-91 years) underwent multistation whole-body contrast-enhanced MR angiography (CE-MRA) on a 32-channel 1.5-T MR system equipped with multicoil technology. A two-step contrast injection protocol was used: After the first injection, images of the most proximal station (station I, head and neck) were acquired, followed by the most distal station (station IV, calves). Images of the intermediate two stations (station II, chest and abdomen; station III, pelvis and thighs) were acquired during the second injection. Conventional catheter angiography was performed for symptomatic vascular regions in 30 patients. The image quality of the arterial segments and the presence and degree of the arterial stenosis were evaluated by two radiologists. The interobserver variability was calculated by kappa statistics, and comparative analysis between CE-MRA and catheter angiography was performed by means of the Spearman's rank correlation coefficient.
RESULTS. Most of the vascular segments (1,912/1,976 [97%]) were
visualized on wholebody CE-MRA with diagnostic image quality. Significant
arterial disease (3 50%) was detected in 167 (observer 1) and 177
(observer 2) segments with excellent interobserver agreement (
= 0.84).
There was a significant correlation between CE-MRA and conventional
angiography for the degree of stenosis (R = 0.92 and 0.89 for
observers 1 and 2, respectively). The sensitivity and specificity of CE-MRA
for the detection of arterial stenoses 50% or greater were 92% and 96% for
observer 1 and 93% and 97% for observer 2, respectively, compared with those
of conventional angiography.
CONCLUSION. Using a multichannel radiofrequency system with multicoil technology, the whole-body CE-MRA approach outlined in this article is able to provide high-spatial-resolution data sets with high diagnostic image quality for evaluation of arterial occlusive disease in most vascular territories.
Keywords: arterial atherosclerotic disease • MR angiography • multicoil technology • parallel acquisition techniques • peripheral vascular disease (PVD) • whole-body MRA
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CTA has been applied successfully for the evaluation of PVD in large part because of its high spatial resolution and recent advances in multidetector technology [3, 4]. Nevertheless, the radiation exposure inherent to CTA, the need for a nephrotoxic contrast agent [5] in patients with PVD who have a high prevalence of renal impairment, and the sensitivity of CTA to artifacts from vascular calcification [6] make it less than ideal.
Despite a relatively lower spatial resolution than its alternative diagnostic techniques, such as CTA or conventional catheter angiography, MR angiography (MRA) has been widely implemented in evaluation of PVD with high diagnostic accuracy [7-9]. The lack of ionizing radiation and the use of a relatively safe contrast agent [10, 11] are the appealing features for the broad acceptance of MRA as the technique of choice in the workup of patients with suspected PVD.
|
Earlier implementations of whole-body MRA have used either a single-element body coil or a device for sliding the patient through a stationary body array coil (AngioSURF [angiography system for unlimited rolling field of views]) [18] for signal reception. The introduction of parallel acquisition strategies [19, 20], together with recent developments in multichannel MR systems with multiarray coil technology, has further improved spatial resolution and the speed of data acquisition for whole-body MRA protocols [21, 22].
Recently, multireceiver channel wholebody MR scanners with multicoil technology have become commercially available. Thus, the challenge for whole-body CE-MRA is now reduced to defining an optimal algorithm for multistation acquisitions that incorporate an appropriate contrast injection scheme, choice of sequence parameters, and table movement protocol.
The purpose of this study was to investigate and describe our initial clinical experience with a multistation whole-body CE-MRA protocol using a 32-channel MR system to display clinically relevant atherosclerotic lesions inside and beyond the peripheral vasculature in patients with suspected PVD.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The indication for imaging was clinical presentation of PVD including lower extremity claudication (n = 33), resting leg or foot pain (n = 12), and resting foot ulcer (n = 5). Associated relevant history was history of hypertension (n = 14), history of diabetes (n = 9), history of chronic renal impairment (n = 7 [four of whom had already undergone renal transplantation]), history of transient ischemic attack (n = 6), and history of coronary arterial disease (n = 8). Exclusion criteria were contraindications for MRI (pacemaker, claustrophobia, contrast reaction, and so on). All studies were performed in accordance with institutional review board guidelines under an approved protocol. Prospective written informed consent for study participation was obtained from all subjects after the nature of the procedure had been fully explained.
Thirty patients underwent selective catheter angiography for symptomatic vascular regions within 1-28 days after CE-MRA.
Imaging Technique
All studies were performed on a 32-channel 1.5-T MR system (Magnetom
Avanto, Siemens Medical Solutions) with high-gradient performance (amplitude,
45 mT/m; slew rate, 200 T/m/s) and a radiofrequency coil design that allows
full flexibility in whole-body parallel imaging. The system is equipped with
32 independent receiver channels and a total imaging matrix system that allows
connecting up to 76 coil elements simultaneously to the system. All
radiofrequency coils are designed for parallel imaging in all three directions
including transverse (superior to inferior), sagittal (right to left), and
coronal (anterior to posterior).
Patients were positioned supine and head-first in the magnet bore. A combination of surface phasedarray coils, together with the spine coil elements that are integrated in the scanner table, was activated in sequential stations and assigned to 32 independent receiver channels (Fig. 1). The numbers of activated coil elements for each station were as follows: station I (head and neck, n = 22), station II (chest and abdomen, n = 24), station III (pelvis and thighs, n = 24), and station IV (calves, n = 21). All flexible adjacent coils were set up with a 2-cm overlap and supported by hooks-and-fastener straps. The total imaging range for the system is 2,050 mm, which obviated repositioning of most patients. A z-axis coverage of up to 1,850 mm was accomplished in the four overlapping stations (stations I-IV) using a 500-mm field of view with a 50-mm overlap between each station (Fig. 2).
|
|
After an initial measurement of contrast transit time to the left heart and calves, the entire body was imaged using a two-phase contrast injection scheme [23]. The timing measurement involved a single injection of 2 mL of contrast material (gadodiamide [Omniscan, Amersham-GE Health]) at 1.2 mL/s, followed by a 30-mL saline flush, using an automated power injector (Spectris, Medrad). Arrival of the timing bolus in the left heart was monitored in real-time, at which point the technologist initiated table movement to the calf station. The times of bolus arrival in the left heart and subsequently in the calf were noted.
For MRA, a total dose of gadodiamide (0.2 mmol/kg) was infused in two separate injections at a rate of 1.2 mL/s, followed by 30 mL of saline administered at the same rate. After the first contrast injection, images of the most proximal station (station I) were acquired; then, the table moved automatically down to station IV to image the calves. Because bolus transit times to the calves can be asymmetric in patients with PVD, the CE-MRA data sets of the calves was acquired twice in sequential measurements. Images of the intermediate two adjacent stations (II and III) were acquired after the injection of the second half of contrast material. There was a mean delay of approximately 5 minutes between the first and second contrast injections. The phase-encoding order was linear for stations I and II and centric for stations III and IV. A preinjection "mask" measurement was acquired at each station immediately before the respective contrast infusion and MRA. Subjects were instructed to hold their breath during acquisitions of stations I and II. Figure 3 summarizes the imaging protocol in our study.
|
Catheter Angiography
Thirty patients underwent follow-up on conventional catheter angiography
within 1 month (1-28 days) after the MR examination. After femoral artery
catheterization, selective angiography of the symptomatic vascular region was
performed. Images were obtained in anteroposterior, lateral, and two oblique
projections (-45° and 45°) for each catheterization, and
interventional angioplasty was performed as needed. The injected volume of
nonionic contrast medium (iohexol [Omnipaque 240, Amersham-GE Health]) was 9
mL for each injection. The catheter angiography images were available on a
computer workstation for image analysis and served as the standard of
reference.
Image Analysis
MR angiograms were interpreted independently by two vascular radiologists
with 10 and 5 years of experience, respectively.
For analysis purposes, the arterial tree was divided into 40 segments including segments 1 and 2, popliteal arteries; 3 and 4, tibioperoneal trunks; 5 and 6, anterior tibial arteries; 7 and 8, posterior tibial arteries; 9 and 10, peroneal arteries in the calves; 11 and 12, common iliac arteries; 13 and 14, external iliac arteries; 15 and 16, internal iliac arteries; 17 and 18, femoral arteries; 19 and 20, superficial femoral arteries in the pelvis and thighs; 21 and 22, supra- and infrarenal aorta; 23, celiac artery; 24, superior mesenteric artery (SMA); 25, inferior mesenteric artery (IMA); 26 and 27, renal arteries in the abdomen; 28 and 29, subclavian arteries; 30 and 31, common carotid arteries; 32 and 33, internal carotid arteries; 34 and 35, external carotid arteries; 36 and 37, vertebral arteries; 38, brachiocephalic trunk; and 39 and 40, thoracic ascending and descending aorta in the head-neck station.
The image quality and visualization of each arterial segment were evaluated
by both observers, who reached agreement by consensus, using a 1- to 4-point
scoring scale. The scores were based on the sharpness of specific vessel
segments and overall image quality, with a score of 1 indicating poor image
quality and blurring of the arterial segment; 2, fair image quality,
inadequate arterial enhancement for confident diagnosis; 3, good image quality
and arterial enhancement, adequate for confident diagnosis; and 4, excellent
image quality and arterial enhancement, for highly confident diagnosis. The
image quality of an arterial segment was rated to be diagnostic (score of 
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 in each station was graded on a scale
of 0-2: none or minimal, score of 0; mild to moderate, not interfering with
diagnosis, 1; and significant, interfering with diagnosis, 2.
Arterial disease in each arterial segment was recorded and scored by each observer independently applying the following 1-4 grading scale: vessel irregularity, < 10% luminal narrowing, grade 1; mild stenosis (10-49%), 2; significant stenosis (50-99%), 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. Both observers were blinded to each patient's information and clinical data. Separate image review sessions were organized for both observers by the study coordinator, who attended all review sessions. 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.
Conventional angiography images in 30 patients (total number of arterial segments, n = 359) were interpreted by both observers as the standard of reference. The observers reached agreement by means of consensus. Both observers were blinded to each patient's name and clinical history and findings on the MRA examination.
Statistical Evaluation
A Wilcoxon's rank sum test was used to evaluate the significance of the
grading differences between the two observers. Interobserver agreement for the
definition of the degree of stenoses between two observers was determined by
calculating kappa statistics (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)
[24]. The relationship between
CE-MRA and catheter angiography in terms of categorized stenosis was analyzed
using the Spearman's rank correlation coefficient (R). The
sensitivity and specificity of MRA for the detection of stenoses that were 50%
or greater were calculated for each observer using conventional catheter
angiography as the standard of reference.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
For image analysis, 24 arterial segments were absent for a variety of
reasons, including four renal arteries because of previous nephrectomies and
two patients without some segments because of unilateral below-the-knee
amputation. Not counting those 24 segments, results in a total number of 1,976
arterial segments were available for evaluation. The image quality of 1,912
(97%) of the 1,976 segments was rated with the definition in the diagnostic
range (score 3) (Fig. 4A),
and 64 arterial segments (3%) were rated as showing definition that was less
than adequate for diagnosis (score < 3) as a result of the presence of
stent artifact (n =9), venous contamination (n = 13), and
not-adequate contrast enhancement for the confident diagnosis (n =
42). However, the image quality scores were consistently lower in the presence
of arterial disease (narrowing > 10%)
(Fig. 4B). In nine subjects,
the observers noted artifactual signal loss in the subclavian artery on the
side of the contrast injection (T2*) and, in four subjects, this
was severe enough to impair evaluation of the ipsilateral subclavian artery.
Venous contamination was scored as none to minimal in 126 stations (63%), mild
to moderate in 67 stations (33.5%), and severe in the following seven stations
(3.5%): abdomen (n = 2), thigh (n = 3)
(Fig. 5), and calf (n
=2). This resulted in impairing the diagnosis of 13 arterial segments (<
1%): renal arteries (n = 4), superficial femoral artery (n =
3), peroneal artery (n =3), posterior tibial artery (n = 2),
and anterior tibial artery (n =1).
|
|
|
Evaluation of Arterial Disease
Observer 1 identified 1,338 arterial segments with no stenoses and 574
arterial segments with disease, including 313 arterial irregularities (luminal
narrowing < 10%), 94 segments with mild stenosis (10-49%), 101 segments
with significant stenoses (50-99%), and 66 segmental occlusions. Image
analysis was based on the postprocessed images (MIP) alone except in 30
segments, for which interactive reformatting was necessary.
Observer 2 identified 1,360 arterial segments with no stenoses and 552 arterial segments with disease, including 291 arterial irregularities (luminal narrowing < 10%), 84 segments with mild stenosis (10-49%), 109 segments with significant stenoses (50-99%), and 68 segmental occlusions. Image analysis was based on the postprocessed images (MIP) alone except in 18 segments, for which interactive reformatting was necessary. The need for interactive reformatting was due to the angulation and tortuosity of the arterial segments and was mainly necessary for evaluation of the internal iliac arteries, inferior mesenteric arteries, and celiac trunk origins.
The breakdown of arterial stenoses segments in the evaluation of PVD is summarized in Table 2.
|
Concomitant Arterial Disease
In the present study, 10 (20%) of the patients with PVD also showed
evidence of significant ( 50% luminal narrowing) carotid artery disease,
and significant renal artery stenosis (3 50% luminal narrowing) was
diagnosed in six (12%) patients. Significant (3 50% luminal
narrowing) mesenteric arterial disease (including celiac trunk, SMA, and IMA)
was diagnosed in four (8%) patients. These patients with significant
concomitant disease required midterm follow-up, and 12 of them were referred
for conventional angiography and therapeutic intervention. In two patients, a
change of plan and immediate therapeutic intervention were necessary
(occlusion of both celiac trunk and SMA in one patient, and severe stenosis of
the brachiocephalic trunk and left common carotid artery in the other
patient).
Table 3 shows the assessment of the arterial tree outside the peripheral vascular territory.
|
The comparison of the arterial stenoses grading scores between two
observers did not show any significant difference (Wilcoxon's test, p
= 0.63). There was excellent interobserver agreement ( = 0.84; 95% CI,
0.82-0.87) for classifying the degree of arterial stenosis.
Table 4 shows the details of
the interobserver correlation for the detection of arterial stenoses using
CE-MRA.
|
The result of conventional angiography for 359 arterial segments showed 121 segments with no stenoses, 68 segments with mild irregularity (< 10%), 48 segments with mild stenoses (10-49%), 77 significant stenoses (3 50-99%), and 45 segmental occlusions.
Table 5 shows the breakdown and comparison of the degree of arterial stenoses with conventional angiography and CEMRA. There was a significant correlation between conventional angiography and CEMRA for the degree of stenosis (observer 1, R = 0.92, p < 0.0001; observer 2, R = 0.89, p < 0.0001).
|
When their CE-MRA interpretations were compared with the findings on conventional catheter angiography, observer 1 had overestimated significant stenosis in three segments and mild stenosis in four segments, and observer 2 had overestimated significant stenosis in two segments and mild stenosis in three segments. More arterial irregularities (17 additional segments) were noted using CEMRA than using conventional angiography. The sensitivity and specificity of CE-MRA in depicting stenoses 50% or greater were 92% and 96% for observer 1 and 93% and 97% for observer 2, respectively.
Figures 6A, 6B, 6C, 7A, 7B, 7C, 7D, 7E, 8A, 8B, 8C, 8D show clinical examples of whole-body CE-MRA images and conventional angiograms in three different patients.
|
|
|
|
|
|
|
|
|
|
|
|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The two-step contrast injection protocol was effective and had additional value in improving the image quality and avoiding venous contamination. The results showed excellent overall diagnostic accuracy, with high sensitivity (92% and 93%) and specificity (96% and 97%), for the detection of significant arterial stenoses, which is in accordance with the previously published data [7-9, 25].
Because of the systemic nature of atherosclerosis, a whole-body MRA protocol that allows comprehensive evaluation of the arterial system from the supraaortic arteries to distal runoff vessels is desirable. Multistation whole-body MRA has a broad acceptance in this regard and has been integrated into routine clinical practice in many centers throughout the world [9, 15-17].
Despite the encouraging results of the initial evaluations [18, 26], whole-body MRA faces some limitations regarding spatial resolution for the depiction of tight stenoses and small infrapopliteal vessels, so delineation of small arteries still remains challenging with the current whole-body approaches. Potential advantages associated with the acquisition of higher-spatial-resolution data sets are offset by the development of venous overlap. Various strategies have been used to improve spatial resolution in whole-body MRA without having significant venous overlap, including the use of parallel acquisition [27, 28], the use of external venous compression [29], or the modification of contrast injection protocols [23].
Recent developments in MR scanner design, including multiple receiver channel MR systems and multicoil technology, are relevant for whole-body MRA applications. Combining large arrays of surface coils with multiple independent detectors allows collection of signals from several object regions in parallel and extends the area from which high sensitivity measurements can be obtained. Promising results indicating the feasibility of these techniques for whole-body MRA have been recently described [21, 22].
Our study is the initial clinical experience using a 32-channel MR system in a population of patients with PVD. In this study, we found that the improved geometry of the multiarray coil system supports the effective implementation of parallel imaging in all four stations and allows acquisition of 3D data sets with a voxel size on the order of 1.5 mm3 throughout the body without impairing image quality. This is reflected in the favorable comparative analysis with catheter angiography and good interobserver agreement. Despite excellent overall image quality in our study (Fig. 4A), it is important to note that most of the arterial segments (82%) scored for image quality were relatively healthy (no stenosis or < 10% irregularities). The mean image quality scores are expected to be lower in the presence of significant arterial occlusive disease, as has been shown in our study (Fig. 4B).
The minimal venous contamination in our study—that is, impaired diagnosis of only 13 of 1,976 arterial segments (< 1%)—is due to a combination of several factors including, first, fast acquisition time due to implementation of parallel imaging; and second, the use of a test bolus to accurately time the contrast arrival and image acquisition. The third factor, applying a centric phase-encoding acquisition in distal stations (thighs and calves), allowed the lowspatial-frequency k-space data (center of k-space) to be acquired at the beginning of the imaging period and enhanced the operator's ability to synchronize imaging with the contrast bolus arrival. Finally, the fourth factor, the use of a two-step contrast injection protocol, also minimized venous contamination.
Typically, researchers who conducted most previous studies of whole-body MRA used a single biphasic protocol for contrast injection and then successive image acquisition from head to foot. However, by using that method, evaluation of the arterial vasculature—especially in distal station (calves vessels)—is partially hampered by venous contamination, as reported in several studies [7, 30]. Also, because of the rapid transit time and bidirectional blood flow between the carotid arteries and descending aorta, venous contamination of the renal arteries cannot be avoided when the head-neck station, is successively followed by acquisition of the thoracoabdominal station in singlestep contrast injection protocols.
Using a two-step contrast injection such as hybrid dual acquisition showed extremely promising results for lower extremity MRA [23]. In our study, we have effectively applied a two-step contrast injection protocol with no additional contrast dose needed. This was achieved primarily because of the coil arrangement and flexible range of table movement (185 cm) that allowed imaging of different vascular territories in a deliberate order.
Several studies have shown that a systematic noninvasive vascular screening strategy can be useful in the detection of concomitant atherosclerotic disease in patients undergoing surgery for vascular disease [12, 31]. Newly diagnosed concomitant diseases in these patients contribute to alterations in surgical treatment and also have important implications for patient prognosis [12]. In our study, 20% of the patients with PVD also showed evidence of significant carotid artery disease, and significant renal artery stenosis was diagnosed in 12% of the patients. Although fewer patients were included in this study, the prevalence of atherosclerotic manifestations in the carotid and renal arteries is comparable to that in the previously mentioned studies [12-14]. In our study, these findings required therapeutic intervention in 12 patients, whereas a change of plan and immediate treatment was necessary in two patients. This emphasizes the value of a noninvasive approach for the assessment of the whole-body vasculature in patients who manifest regional atherosclerotic disease.
We acknowledge several limitations of our study. The intracranial (i.e., beyond the circle of Willis) and coronary arteries are not seen with enough diagnostic visibility on the whole-body CE-MRA approach, and they still require a dedicated approach for diagnostic assessment. The distal arterial branches of the foot were not imaged in all subjects. However, if complete coverage of foot is clinically required, imaging of the whole foot is feasible by increasing the slab thickness, more accurate positioning of the imaging slab, or both.
We cannot validate the accuracy of findings in all detected segmental disease because conventional angiography, the standard of reference, was performed in only a subset of patients (30 of 50 patients). However, at our institution, as at many others, conventional angiography is no longer routinely performed in all patients with suspected arterial disease. Therefore, an estimate of the accuracy of whole-body CE-MRA in a large patient cohort remains undetermined. However, with high diagnostic accuracy and excellent correlation for the detection of arterial disease between MRA findings and those of conventional angiography, we are confident that most additional cases of arterial disease found on whole-body CE-MRA represent true vascular disease.
Because of the lack of a more accurate technique, we used 2D conventional catheter angiography as the standard of reference. Limitations of this projectional technique for the detection and characterization of nonconcentric stenoses are well recognized [8, 32]. Therefore, the sensitivity and specificity should be noted in this context. Conventional angiography was performed only in patients with severe arterial stenosis or for symptomatic arterial regions as a part of therapeutic intervention. This selection bias may be partially responsible for the high values for the sensitivity and specificity in our study.
Finally, there are some practical issues associated with whole-body imaging that need to be explored in more comprehensive clinical trails. These include the liability of physicians for the interpretation of incidental findings, the cost-effectiveness of the examination, and the probable need for additional diagnostic or therapeutic interventions.
In conclusion, using a multichannel radiofrequency system with multicoil technology, we found that the outlined whole-body CE-MRA approach provides high-spatial-resolution data sets with high diagnostic image quality for the evaluation of arterial occlusive disease in most vascular territories. The procedure is feasible and convenient for both the patient and technician, eliminating the need for coil or patient repositioning. Although more extensive clinical testing is warranted to explore the boundaries and limitations, it seems likely that this paradigm will replace those currently in use for lower extremity MRA.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  R. Habibi, M. S. Krishnam, D. G. Lohan, F. Barkhordarian, M. Jalili, R. S. Saleh, S. G. Ruehm, and J. P. Finn High-Spatial-Resolution Lower Extremity MR Angiography at 3.0 T: Contrast Agent Dose Comparison Study Radiology, August 1, 2008; 248(2): 680 - 692. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||
