| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Innovation |
1 Department of Diagnostic and Interventional Radiology, Center of Diagnostic
Imaging and Intervention, University Medical Center Hamburg-Eppendorf,
Martinistrasse 52, Hamburg 20246, Germany.
2 Department of Medical Physics, Cambridge University Hospitals, NHS Foundation
Trust, Cambridge, United Kingdom.
3 Department of Cardiology/Angiology, Center of Diagnostic Imaging and
Intervention, University Medical Center Hamburg-Eppendorf, Hamburg,
Germany.
4 Roentgeninstitut, 40476 Duesseldorf, Germany.
Received February 21, 2006;
accepted after revision July 17, 2006.
Address correspondence to P. M. Bansmann
(pbansmann{at}uke.uni-hamburg.de).
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionConclusionReferences |
|---|
CONCLUSION. This study shows that the quality of coronary vessel wall imaging of the right coronary artery with radial k-space sampling in 3D turbo field-echo sequences is superior to cartesian k-space sampling at 3 T. Radial k-space sampling at 3 T makes it possible to combine low motion artifact susceptibility with high signal-to-noise ratio.
Keywords: cardiac imaging • cardiovascular imaging • coronary vessels • heart • MR angiography • MRI
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionConclusionReferences |
|---|
The feasibility of black blood coronary MRA has been shown at 1.5 and 3 T. Different pulse sequences have been used to acquire coronary vessel wall images [6-12]; however, the image quality still needs to be improved for clinical use and each sequence has its advantages and disadvantages [12]. Considering the small thickness of the moving coronary vessel wall, a sequence with a high spatial resolution and a low degree of motion artifact susceptibility is needed for coronary vessel wall imaging.
Radial k-space sampling has been shown to be less susceptible to motion artifacts caused by respiratory and cardiac motion than the corresponding cartesian k-space sampling method at 1.5 T [12, 13]. One drawback of the radial k-space sampling method is the reduced—by a factor of 0.81—signal-to-noise ratio (SNR) compared with the corresponding cartesian images [14].
The purpose of our study was to compare the quality of 3D gradient-echo images obtained using radial versus cartesian k-space sampling at 3 T. We addressed the known drawback of a reduced SNR achieved with the radial compared with the cartesian k-space sampling methods at 1.5 T by scanning at a higher magnetic field strength, 3 T. Because the SNR is approximately proportional to the main magnetic field strength, a way to compensate for reduced SNR is to increase the magnetic field strength. Another way to achieve higher SNR is to use 3D imaging methods instead of the previously described 2D methods [15-17]. An in-plane orientation of the imaging volume allows coverage of the main coronary arteries to the point where inverted blood flows.
When using 3D methods combined with double inversion recovery, it is necessary to avoid reinversion of blood in the left ventricle, which is upstream of the coronary arteries. Slab-selective reinversion has been successfully implemented at 3 T using adiabatic pulses, which have a broad bandwidth and are robust to radiofrequency inhomogeneities. Motion artifacts are also reduced by improved magnetization restoration before navigator measurements, which are essential for free-breathing coronary artery imaging [17].
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionConclusionReferences |
|---|
|
To compensate in real time for respiratory motion and the corresponding artifacts during free breathing, imaging data were acquired using previously described prospective respiratory navigator gating and tracking [20]. A 5-mm gating window with a correction factor of 0.6 was used for slice tracking in the caudocephalad direction. The respiratory navigator pencil beam had a flip angle of 25° and was positioned manually over the dome of the right hemidiaphragm. To reduce the off-resonance sensitivity of the pencil beam excitation at 3 T, the number of turns in the k-space was reduced to 9 with a beam width of 40 mm. This gave a short pulse duration of 4 milliseconds.
Before the navigator and imaging sequences, a fat-suppression pulse was applied [21]. Blood signal suppression in the RCA was achieved by a double inversion recovery preparation directly after the R wave trigger. Therefore, the inversion delay was the maximum value possible and varied from patient to patient because of the variable trigger delay. The first nonselective inversion pulse was followed by a second slab-selective pulse to reinvert magnetization in the region of the RCA. Figure 1 shows an example of the slab positioning to reinvert magnetization. Both pulses were hyperbolic secant pulses and had a duration of 12 milliseconds and a bandwidth of 1,040 Hz.
The inversion time was not precisely the theoretic value to suppress the blood signal but still resulted in good blood signal suppression and maximized the inflow of inverted blood. The slab position had to be defined manually and was oriented obliquely to the actual imaging volume to avoid reinversion of blood in the left ventricle, which would flow into the coronary artery and reduce contrast [17]. After the double inversion recovery prepulses, a second reinversion pulse was applied to restore magnetization in the right hemidiaphragm region [22].
To locate the course of the coronary arteries, a high-resolution scout image was obtained using a k-space segmented gradient-echo planar imaging sequence with fat suppression, T2 preparation, and respiratory navigator gating. The imaging parameters were as follows: TR/TE, 13/4.4; flip angle, 30°; turbo field-echo factor, 9; echo-planar imaging factor, 5; acquisition matrix, 256 x 164 x 24; field of view, 350 x 280 x 96 mm; and sensitivity-encoding (SENSE) acceleration factor, 1.9. The course of the RCA was defined using a standard three-point-plan scan tool [23].
Coronary vessel wall images were acquired with a 3D turbo field-echo sequence (flip angle, 30°; field of view, 270 mm; acquired in-plane resolution, 0.7 x 0.7 mm) during free breathing and using prospective navigator gating. The oversampling factor in the partition encode direction was 1.3 for both sequences. A schematic picture of the sequence used is shown in Figure 2.
|
Radial coronary vessel wall imaging of the RCA was performed using 8.9/2.3. Assuming 50% navigator efficiency, the scanning duration of this sequence was 768 heartbeats or 25 minutes 36 seconds. Three hundred eighty-four radial trajectories were acquired in the x-y plane. Both cartesian and radial images were acquired with 6 partitions of 2-mm thickness and were reconstructed to 12 slices at 1-mm intervals. The intershot interval was two heartbeats with an acquisition window of 71 milliseconds per shot. An anterior saturation band and saturation bands to suppress foldover were applied. The specific absorption rate (SAR) was 1.5 W/kg.
Image Analysis
With an interactive visualization and analysis tool (SoapBubble Tool
[Intera, Philips Medical Systems]), the length of visible vessels and the SNRs
for the vessel wall, epicardial fat, and luminal blood were measured. Using
the signal from all six receiver coils, the SNR was measured on a 512 x
512 image matrix without homogeneity correction
[24]. In the described
regions, the signal was measured in a region of interest. The SD in a region
of interest in air outside the subject was measured. The contrast-to-noise
ratio (CNR) between vessel wall and blood and between vessel wall and fat was
calculated. The SNR was defined as the mean signal divided by the SD of noise.
The CNR was defined as the difference between the two appropriate SNR
values.
The identifying information on the acquired images was removed, and the images were evaluated by one experienced radiologist and one experienced cardiologist in consensus, both of whom were blinded to sequence parameters. According to a segmentation scheme recommended by the American College of Cardiology and the American Heart Association (ACC/AHA), the RCA was divided into three segments (proximal, mid, and distal) and evaluated [25]. Due to the inflow time for inverted blood, all proximal segments were assessed. The qualitative analysis of the vessel wall was graded for each image on a scale from 0 to 4 for the depiction of the RCA wall: 0, vessel wall not visualized; 1, insufficient visualization of vessel wall; 2, sufficient visualization of vessel wall with significant artifact level; 3, good visualization of vessel wall with low artifact level; and 4, excellent image quality. In a comparative analysis, both investigators in consensus evaluated vessel diameter and vessel wall thickness in all proximal segments.
|
|
|
|
|
|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionConclusionReferences |
|---|
|
|
Radial k-space sampling resulted in fewer motion artifacts and improved vessel wall visualization compared with cartesian k-space sampling. Streak artifacts could be seen in all radial images. Because these artifacts have a conspicuous appearance, they can be easily identified by the observer. In this study, streak artifacts did not affect the quality of the images obtained using radial k-space sampling. Because streak artifacts are around the center of the acquired field of view, it is best to avoid having the coronary artery in the exact center of the image. No statistically significant difference was found comparing the depicted vessel wall lengths, as shown in Table 2. The SNR and CNR at this higher field strength showed no significant difference between radial and cartesian k-space sampling. There was a significant difference in the average quality score for the depicted RCA vessel wall. The radial images were graded an average score of 2.9 and the cartesian images, an average score of 2.1. In Figures 3A, 3B, 3C, 3D, 3E, and 3F, well-depicted vessel walls of the RCA to the point where inverted blood flowed from the left ventricle are shown. The qualitative and quantitative results averaged for all volunteers are shown in Tables 1 and 2.
In the qualitative comparison of vessel diameters, no recognizable difference was found comparing cartesian and radial k-space sampling.
Visual analysis of vessel wall thickness resulted in assessing a larger vessel wall thickness in six of nine volunteers when using cartesian k-space sampling. In these cases, radial k-space sampling resulted in a sharper delineation of the vessel wall. In the remaining three volunteers, no visual difference could be found comparing vessel wall thickness.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionConclusionReferences |
|---|
In this study, in-plane coronary MRA scans of the vessel wall with cartesian and radial k-space sampling were compared. Besides allowing a higher SNR, our selected approach of 3D in-plane scans provided volumetric coverage of the course of the RCA vessel wall to the point at which inverted blood flowed. Good blood suppression was achieved using a previously described double inversion recovery prepulse [17]. The slab could manually be defined to avoid upstream blood from the left ventricle in all volunteers. Using a vector ECG made reliable ECG-gating at 3 T possible [18]. At a calculated SAR of 1.5 W/kg, it was easy to comply with the energy deposition limits.
Both imaging sequences provided an adequate depiction of the RCA vessel wall to the point where inverted blood flowed. As was shown by previous studies at 1.5 T, radial k-space sampling proved to be less susceptible to motion artifacts caused by respiratory and cardiac motion [12, 13, 26]. Both respiratory and cardiac motion may cause reduced vessel border definition, which is, considering the extremely small thickness of the coronary vessel wall and the relatively strong motion during the respiratory and cardiac cycles, a major impediment to acquiring high-resolution images of the coronary vessel wall. A well-known drawback of the radial k-space sampling method is that only poor SNR can be achieved at 1.5 T [14].
In this study, the known drawback of reduced SNR at 1.5 T could be compensated for at the higher field strength of 3 T. No statistically significant difference was found between radial and cartesian images with regard to SNR measurements. The subjective analysis of image quality and motion artifacts showed that radial k-space sampling is less susceptible to motion artifacts caused by respiratory and cardiac motion, as was shown before at 1.5 T. Our results show the superior potential of radial 3D turbo field-echo vessel wall coronary MRA compared with cartesian 3D turbo field-echo vessel wall coronary MRA.
A limitation of our investigation is the fact that imaging was performed on healthy adult volunteers without ECG changes, with low heart rates (mean, 62.2 bpm), and with regular breathing patterns. It is known that patients with a history of cardiac disease often present with higher or irregular heart rates. The breathing pattern in patients may also be more variable. Also, it is known that the cardiac rest period varies strongly among subjects. At higher heart rates, the duration of the systolic rest period proved to be more consistent than the duration of the diastolic rest period [27]. Data acquisition during the systolic instead of the diastolic rest period might thus be advisable in patients with high heart rates.
The average weight of the volunteers was 67.2 ± 7.6 kg. Compared with the average body weight of patients with coronary heart disease, there might be a significant difference. Well-suppressed fat signal surrounding the RCA might result in better depiction of the vessel wall. Further work is needed to analyze the effects of a larger body weight on the image quality of vessel wall scans. Another limitation is the fact that only RCA wall scans were acquired. Because the data acquisition planned during the diastolic rest period focused on the origin of the RCA, we do not know whether imaging of the vessel walls of the left anterior descending or left circumflex coronary arteries would present similar results. Additional scout images with a focus on the optimal rest period of the left coronary arteries leading to even longer scanning times might be necessary.
Previously described whole-heart methods have been shown to be a promising way to cover all three major coronary arteries in one procedure [4]. For vessel wall imaging, however, special precaution to avoid reinversion of blood within the left ventricle is necessary. The long scanning time of 20-30 minutes per scan presents another problem. The overall limit of 90 minutes of scanning time for healthy volunteers might not be applicable to patients. Even if the absolute scanning time could be tolerated by patients, heart rates and breathing patterns would vary between healthy volunteers and patients. Increasing the acquisition window, reducing the intershot TR, using radial undersampling, and performing parallel imaging could lead to reduced scanning times [28], which would be especially beneficial in imaging patients. Because radial imaging showed its potential with regard to lower motion artifact susceptibility, further work is needed to find out how much scanning duration can be reduced with acceptable losses in SNR and image quality.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionConclusionReferences |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionConclusionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  R. J. Gibbons, P. A. Araoz, and E. E. Williamson The Year in Cardiac Imaging J. Am. Coll. Cardiol., September 4, 2007; 50(10): 988 - 1003. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
