| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Cardiac Imaging |
1 From the Department of Internal Medicine/Cardiology, German Heart Institute Berlin, Augustenburger Platz 1, 13353 Berlin, Germany. Received September 12, 2005; revision requested November 10; revision received December 13; accepted January 11, 2006; final version accepted February 9. Address correspondence to C.J. (e-mail: jahnke{at}dhzb.de).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
Materials and Methods: The study was conducted according to standards of the Charité and Virchow-Klinikum Ethics Committee. Patients gave written consent. Dobutamine stress magnetic resonance (MR) imaging was conducted in 65 patients (mean age, 63 years ± 9 [standard deviation]; 49 men) with conventional cine steady-state free precession (SSFP). Accelerated four-dimensional (4D) k-t BLAST single-breath-hold imaging with complete left ventricular (LV) coverage was also performed at rest and during stress. For the cine SSFP and accelerated cine techniques, duration of imaging at rest and LV end-diastolic volume and ejection fraction were assessed. Segmental agreement for resting and inducible wall motion abnormalities was determined. In a subgroup (n = 40), direct comparison between SSFP and accelerated cine was performed for coronary stenosis detection. A paired Student t test was used to assess significance of continuous variables. Pearson correlation was used to test correlation between the techniques. Sensitivity, specificity, and diagnostic accuracy were calculated (standard definitions). For quantitative measurement of agreement, Cohen 
was applied.
Results: For accelerated cine, imaging duration at rest was shortened by 40%. Correlations between cine SSFP and accelerated cine for LV parameters were 135 mL ± 37 versus 129 mL ± 31 (r = 0.89) for end-diastolic volume and 59% ± 8 versus 58% ± 7 (r = 0.95) for ejection fraction. Values for segmental wall motion at rest and stress ranged from 0.77 to 0.91. Sensitivity, specificity, and diagnostic accuracy for coronary stenosis (
50%) detection based on arterial territory were 82%, 87%, and 86%, respectively, for cine SSFP and 82%, 86%, and 85%, respectively, for accelerated cine imaging.
Conclusion: Accelerated 4D k-t BLAST wall motion imaging at rest and at dobutamine stress is rapid and feasible; LV measurements were nearly identical between the imaging approaches. Segmental wall motion analysis at rest and at stress show excellent agreement and reliable depiction of myocardial territories supplied by coronary arteries with 50% or more luminal narrowing.
Supplemental material: http://radiology.rsnajnls.org/cgi/content/full/2413051522/DC1
© RSNA, 2006
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
The dobutamine stress MR imaging procedure follows a highly standardized protocol—that is, there is an initial planning phase at rest to define the standard short-axis and long-axis views being acquired during multiple breath holds (duration of 6–10 seconds).
Especially in patients with preexisting wall motion abnormalities, a prerequisite for accurate determination of LV volumes and dimensions is to acquire multiple short-axis views with complete coverage of the heart (Simpson rule) (9). Next, the standard views are repetitively obtained during increasing doses of dobutamine infusion (10, 20, 30, and 40 [µg · kg–1]/min for 3 minutes each).
A new strategy to achieve faster dynamic imaging by means of reduced data acquisition based on exploiting correlations in k-space and time (k-t) is available, due to the recent development and implementation of k-t broad-use linear acquisition speed-up technique (k-t BLAST) imaging (10,11). This technique uses the quasi-periodic motion of the heart, allowing for the acquisition of a reduced data set and recovering the omitted data afterward by using a small set of training data to learn the signal correlations (10,12). In phantom and initial volunteer studies, k-t BLAST two-dimensional cine imaging did not significantly compromise image quality (11). In addition, the feasibility of three-dimensional cine imaging of the heart during a single breath hold by using k-t BLAST imaging has been reported (13).
Thus, k-t BLAST appears to be a promising technique for accelerated cardiac three-dimensional cine MR imaging in patients. The use of a three-dimensional cine data set acquired in a single breath hold may have several advantages in comparison with the current approach that uses multiple two-dimensional standardized views: First, full coverage of the heart facilitates accurate determination of LV volumes by means of the Simpson method and visual evaluation of segmental wall motion at once. Second, reconstruction of standard views from a three-dimensional volume reduces planning effort and shortens imaging time.
The purpose of our study was to prospectively determine the feasibility and diagnostic performance (with angiography as a reference standard) of k-t BLAST accelerated cine imaging during dobutamine stress for the identification of inducible cardiac wall motion abnormalities.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
MR Imaging
MR imaging was performed with the patient in the supine position by using a 1.5-T MR imager (Intera CV; Philips, Best, the Netherlands) equipped with a Nova gradient system (33 mT/m, 160 [mT · m–1]/msec) and custom-made software (Gyrotools, Zurich, Switzerland) that was based on the imager's software (release 11). A five-element cardiac synergy coil was used for signal reception. Cardiac synchronization was performed with four electrodes placed on the left anterior hemithorax (vector electrocardiography), and imaging was triggered during the R wave of the electrocardiography examination (14).
The following stepwise protocol for standardized image planning was applied: (a) a rapid multistack multisection survey, allowing for localization of the heart in the three standard planes (transversal, sagittal, and coronal), (b) single-angulated, single-section cine acquisition of the LV on a transverse view, (c) double-angulated, single-section cine acquisition of the LV planned on the previous view, (d) cine acquisition of 10–12 short-axis views covering the whole LV (SSFP sequence, no gap between sections, acquisition of two sections per breath hold), (e) cine acquisition of three standard short-axis views (apical, equatorial, and basal short-axis views), (f) cine acquisition of three standard long-axis views (four-, two-, and three-chamber views), and (g) three-dimensional cine acquisition of the whole LV in a single breath hold by using k-t BLAST with the geometry planned on the previously acquired double-angulated view.
The time period for planning and acquisition of the standard views was recorded for the conventional cine SSFP technique (steps a–f) and for the accelerated cine technique (steps a–c and step g).
Conventional Cine SSFP MR Imaging
For conventional cine imaging, each standard view was acquired during an expiratory breath hold of 4–6 seconds by using an SSFP sequence in combination with parallel image acquisition (sensitivity encoding) and retrospective gating (25 phases per cardiac cycle, 2.7/1.4 [repetition time msec/echo time msec], and a 60° flip angle). Typical in-plane spatial resolution was 1.8 x 1.7 mm (reconstructed to 1.5 x 1.5 mm), with a section thickness of 8 mm.
Three-dimensional Accelerated Cine Imaging
For k-t BLAST accelerated cine imaging, a three-dimensional sequence adapted from Kozerke et al (13) and recently optimized by Tsao et al (11) was used. Measured in-plane spatial resolution was 2.1 x 2.4 mm (reconstructed to 1.3 x 1.3 mm), with a section thickness of 8 mm. The acquisition time was 34 msec per phase, and a total of 24 phases were acquired by using prospective gating (3.3/1.7; flip angle, 45°). The average breath-hold duration for acquisition of 12 sections covering the whole LV in a short-axis orientation was 18 seconds for acquisition of the image data and 8 seconds for the training data. The k-t sampling acceleration factor was set to 6.0, which resulted in an effective net acceleration factor of 4.1.
After acquisition of the four-dimensional data set, the commercially available multiplanar reformatting software of the imager was used to reconstruct images in any desired geometry. The three standard long-axis views (ie, a four-, a two-, and a three-chamber view) were reconstructed (Fig 1; Movie 1, http://radiology.rsnajnls.org/cgi/content/full/2413051522/DC1). These user-defined geometries were then applied to all 24 phases, which resulted in movies (Movie 2, http://radiology.rsnajnls.org/cgi/content/full/2413051522/DC1) of the complete cardiac cycle in each of the standard views.
|
Image Analysis
Visual score.—With both MR imaging techniques, image quality of each standard view was graded on a four-point scale from excellent (score of 4) to nondiagnostic (score of 1) at rest and at each stress level as determined in consensus by two observers (C.J. and I.P., with 4 and 8 years of experience in cardiac MR imaging, respectively). The visual score refers to the visibility of the endocardial border (score of 1, barely or not visible; 2, moderately or partly visible; 3, well visible; and 4, excellently visible).
LV dimensions.—The following parameters were determined for conventional cine SSFP and accelerated cine imaging at rest: LV end-diastolic volume and end-systolic volume determined by using the Simpson rule, LV ejection fraction, LV end-diastolic diameter, and end-diastolic wall thickness of the septal and lateral wall. LV dimensions were evaluated by one observer (C.J.).
Wall motion analysis.—Segmental wall motion analysis of the cine images was performed by two observers (C.J., I.P.) in consensus, who were blinded to the results of invasive coronary angiography, by using a randomized blinded order of patients. A synchronized quadscreen image display was used for analysis by applying the standard scoring system (score of 1, normokinesis; 2, hypokinesis; 3, akinesis; and 4, dyskinesis) (16) and classifying all 17 standardized segments of the LV (16) at rest and at each stress level. For dobutamine stress MR images, ischemia was defined as one or more segments showing an inducible wall motion abnormality (ie, an increase in the segmental wall motion score of 1) or a biphasic response (ie, an increase of regional contractile function in a dysfunctional segment during low-dose dobutamine stress imaging, with subsequent deterioration at infusion of higher doses of dobutamine). Segmental agreement between the two imaging techniques was assessed at rest, at low-dose dobutamine stress, and at maximum stress.
Coronary arterial territories.—The 17 myocardial segments were assigned to one of the three major coronary arteries (left anterior descending, left circumflex, or right coronary artery) according to the guidelines of the American Heart Association (16). Segments that showed an inducible wall motion abnormality during dobutamine stress were considered indicative of significant coronary luminal narrowing (ie, 50%) in the respective coronary arterial territory.
Patient Subgroups
Feasibility subgroup.—To initially evaluate the feasibility of the three-dimensional accelerated cine approach, 25 consecutive patients undergoing a conventional dobutamine stress MR examination were evaluated. The MR procedure is described above in detail. Image analysis comprised a visual score for overall image quality, assessment of LV dimensions, and assessment of segmental wall motion at rest and during dobutamine stress.
Diagnostic performance subgroup.—For evaluation of the diagnostic performance of accelerated cine imaging, a subgroup of 40 patients who were referred to our institution for cardiac catheterization underwent the identical dobutamine stress MR protocol and image analysis as was previously described for the feasibility group. In addition, by using invasive conventional (x-ray) angiography as the standard of reference, the diagnostic values were calculated on a territorial basis.
Conventional Coronary Angiography
Invasive coronary angiography was performed with the observer unaware of the results of MR imaging. Significant coronary artery stenosis was defined visually as a reduction of 50% or more in diameter in vessels with a diameter of 2 mm or larger. In patients with significant CAD, the most severe stenosis was defined by an experienced interventionalist (E.F.) and was assigned to a coronary arterial territory (left anterior descending, left circumflex, or right coronary artery) that was deemed to be responsible for the occurrence of an inducible wall motion abnormality. Similarly, in patients with bypass grafts, significant arterial or vein graft stenoses were considered to be responsible for ischemia in the myocardial territory supplied by the recipient native coronary vessel.
Statistical Analysis
Statistical analysis was performed by using a statistical software package (SPSS, release 12.0.1; SPSS, Chicago, Ill). For all continuous parameters, data are given as the mean ± standard deviation. All tests were two-tailed, and P < .05 was considered to indicate a statistically significant difference. The paired Student t test was used to assess statistical significance of continuous variables. Pearson correlation was used to test statistical correlation between the two imaging techniques. Sensitivity, specificity, and diagnostic accuracy were calculated according to standard definitions; the respective 95% confidence intervals (CIs) were calculated on the basis of a binomial model (17).
Bland-Altman analysis was performed to compare conventional cine SSFP and accelerated cine imaging with regard to LV function and dimensions; the degrees of agreement between these two methods were determined as mean absolute difference, 95% CI of the mean difference, and mean relative difference (difference of the two techniques divided by their mean value) (18).
To allow quantitative measure of agreement, the Cohen was applied (19) by using the following grading: grades of 0–0.20, poor agreement; 0.21–0.40, fair agreement; 0.41–0.60, moderate agreement; 0.61–0.80, substantial agreement; and 0.81–1.0, nearly perfect agreement. For all values, the respective 95% CIs are given.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
|
Visual score.—For both imaging approaches, all standard views at rest and at each stress level were considered diagnostic (ie, visual score of 2; Table 2). For conventional cine SSFP and accelerated cine imaging, the percentages of standard views graded as good to excellent were 100% and 99%, respectively, at rest and 99% and 98%, respectively, at maximum stress.
|
|
|
of 0.86 (95% CI: 0.74, 0.98).
|
|
= 0.91; 95% CI: 0.73, 1.09) and at maximum stress ( = 0.84; 95% CI: 0.74, 0.94).
Diagnostic Performance Group
Patient characteristics and dobutamine stress testing.—Thirty-six (90%) of 40 patients reached target heart rate (average doses of dobutamine and atropine were 36 [µg · kg–1]/min ± 7 and 0.4 mg ± 0.4, respectively) (Table 1). The reasons for premature termination included chest pain and dyspnea in three (8%) of 40 patients and a newly detected wall motion abnormality in one (2%) of 40 patients.
Time period at rest.—The time period for planning and acquisition of the standard views at rest was significantly shorter with accelerated cine imaging than with conventional cine SSFP imaging (558 seconds ± 97 vs 959 seconds ± 181; P < .01, paired Student t test).
Visual score.—For both imaging approaches, all standard views at rest and at each stress level were diagnostic (ie, a visual score of 2; Table 2). For conventional cine SSFP and accelerated cine imaging, the percentages of standard views graded as good to excellent were 100% for both techniques at rest and 98% and 94%, respectively, at maximum stress.
LV dimensions.—LV end-diastolic volume, diameter, and septal and lateral wall thickness, as well as the calculated ejection fraction, were excellently correlated between conventional cine SSFP and accelerated cine imaging (Fig 2, Table 3).
Wall motion analysis at rest.—With conventional cine SSFP imaging, 18 (45%) of 40 patients had wall motion abnormalities at rest (eight patients with hypokinesis, nine with akinesis, and one with dyskinesis); with accelerated cine imaging, 17 patients (43%) had wall motion abnormalities at rest (seven patients with hypokinesis, nine with akinesis, and one with dyskinesis). Segmental agreement between the conventional cine SSFP and accelerated cine approaches for regional wall motion at rest was nearly perfect ( = 0.84; 95% CI: 0.76, 0.92).
Wall motion analysis during dobutamine stress.—With the use of conventional cine SSFP imaging, 27 (68%) of 40 patients demonstrated an inducible wall motion abnormality, and with accelerated cine imaging, 28 patients (70%) had an inducible wall motion abnormality (Figs 4, 5). Segmental agreement between conventional cine SSFP and accelerated cine imaging was substantial at low-dose dobutamine stress ( = 0.77; 95% CI: 0.65, 0.89) and was nearly perfect at maximum stress ( = 0.81; 95% CI: 0.75, 0.87).
|
|
|
|
|
|
Diagnostic value.—With conventional cine SSFP imaging, the presence of significant CAD was correctly diagnosed in 25 of 28 patients, and the absence of significant CAD was correctly diagnosed in 10 of 12 patients. With accelerated cine imaging, the presence of significant CAD was correctly diagnosed in 25 of 28 patients, and the absence of significant CAD was correctly diagnosed in nine of 12 patients.
Coronary arterial territories.—For the assignment of inducible wall motion abnormalities that were considered indicative of a significant coronary stenosis of the left anterior descending, left circumflex, or right coronary artery territory, conventional cine SSFP imaging helped to correctly identify 23 of 28 arterial territories (left anterior descending, 15 patients; left circumflex, four patients; and right coronary, four patients). The absence of an inducible wall motion abnormality was correctly identified in 80 of 92 territories. Sensitivity, specificity, and diagnostic accuracy for the detection of significant stenosis of a coronary arterial territory were 82% (95% CI: 66%, 98%), 87% (95% CI: 80%, 94%), and 86% (95% CI: 79%, 93%), respectively. With accelerated cine imaging, inducible wall motion abnormalities were correctly assigned to the supplying coronary arterial territory in 23 of 28 arterial territories (left anterior descending in 15 patients, left circumflex in three patients, and right coronary in five patients). The absence of an inducible wall motion abnormality was correctly identified in 79 of 92 territories. Sensitivity, specificity, and diagnostic accuracy per coronary arterial territory were 82% (95% CI: 66%, 98%), 86% (95% CI: 78%, 94%), and 85% (95% CI: 78%, 92%), respectively. In a comparison of the conventional cine SSFP and accelerated cine approaches, the average value of agreement for the assignment of inducible wall motion abnormalities to the coronary arterial territories was found to be substantial ( = 0.78; 95% CI: 0.66, 0.90).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
The new k-t BLAST approach has been suggested as a promising technique for accelerated four-dimensional imaging of the LV (13) but has not yet been evaluated with regard to the assessment of LV dimensions and regional wall motion abnormalities at rest or during dobutamine stress. With the use of k-t BLAST, an effective shortening of the examination duration at rest was expected. In addition, taking advantage of its four-dimensional nature of data acquisition, k-t BLAST may prove useful for regional wall motion analysis, because it facilitates reconstruction of any desired geometry during the complete cardiac cycle. However, the combined acquisition of a reduced set of image data and a small set of training data may limit the diagnostic applicability of the accelerated cine technique because spatial misregistration might occur during stress testing (12). Thus, in the present study, we performed a direct comparison between accelerated cine imaging and the conventional, widely recognized reference standard of cine SSFP imaging. Our results indicate that a similarly high image quality can be achieved with accelerated cine imaging as with conventional cine SSFP imaging: With both approaches, more than 90% of standard views were considered good or excellent, although with accelerated cine imaging there was a tendency toward a higher proportion of standard views (about 20%) being judged as good.
At present, cardiac MR imaging with the cine SSFP approach represents the reference standard for determination of LV volumes and, thus, has recently been designated a class I indication because of its accuracy, reproducibility, and profound validation (1). Routinely, to be independent of geometrical assumptions (Simpson rule), multiple cine short-axis views obtained during multiple breath holds with complete LV coverage are used. This proved to be the most accurate method for assessment of LV volumes and ejection fraction (9) and is clearly preferable in all patients with preexisting wall motion abnormalities.
With accelerated cine imaging, the whole LV was imaged with a four-dimensional acquisition during a single breath hold; thus, the measurement of LV volumes can be accomplished according to the Simpson rule, and all standard views can be reconstructed from the data set at once. LV parameters as assessed from conventional cine SSFP and accelerated cine images were nearly identical. However, accelerated cine imaging yielded a significantly shorter examination time, thereby increasing patient comfort.
Visual assessment of LV regional contractile function at rest and during dobutamine stress is widely recognized as highly accurate for the detection of wall motion abnormalities. For both—that is, for the detection of preexisting wall motion abnormalities at rest and for the detection of inducible wall motion abnormalities during dobutamine stress—accelerated cine imaging showed a high level of agreement with the well-established cine SSFP approach. With conventional coronary angiography used as the standard of reference, the diagnostic accuracies of accelerated cine and cine SSFP imaging for the detection of significant coronary luminal narrowing were similarly high (85% and 86%, respectively).
Our study had some limitations. In previous large-scale patient studies, a similar diagnostic performance of dobutamine stress MR has been reported (2,5,6). However, it is well known that a high prevalence of CAD might have an effect on sensitivity, specificity, and diagnostic accuracy. Thus, the diagnostic performance as reported in our study might not fully reflect the diagnostic performance that can be achieved with either imaging approach. We wanted to show the equivalence of conventional and k-t–accelerated cine imaging. In regard to this, the accurate assessment of segmental wall motion (ie, the agreement of the two methods on a per-segment basis) was deemed to be the most clinically relevant parameter. Future studies are needed to evaluate the applicability of three-dimensional accelerated cine dobutamine stress MR imaging in more unselected patient populations with a lower prevalence of CAD.
Accelerated three-dimensional cardiac cine imaging with use of k-t BLAST proved to be rapid and feasible at rest and during dobutamine stress. When compared with conventional cine SSFP imaging, the assessment of LV volumes and dimensions resulted in nearly identical values. In addition, segmental wall motion analysis at rest and during dobutamine stress showed an exceptionally high level of agreement and enabled reliable identification of patients with angiographically significant coronary artery stenoses.
|
ADVANCES IN KNOWLEDGE |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
|
FOOTNOTES |
|---|
Abbreviations: CAD = coronary artery disease • CI = confidence interval • k-t = k-space and time • k-t BLAST = k-t broad-use linear acquisition speed-up technique • LV = left ventricle • SSFP = steady-state free precession
Authors stated no financial relationship to disclose.
Author contributions: Guarantors of integrity of entire study, C.J., I.P., E.F., E.N.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; manuscript final version approval, all authors; literature research, C.J., I.P., A.B.; clinical studies, C.J., I.P., R.G.; statistical analysis, C.J., I.P.; and manuscript editing, all authors
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  T. Lockie, E. Nagel, S. Redwood, and S. Plein Use of Cardiovascular Magnetic Resonance Imaging in Acute Coronary Syndromes Circulation, March 31, 2009; 119(12): 1671 - 1681. [Full Text] [PDF]
|
|
|
|
 |
|
 W P Bandettini and A E Arai Advances in clinical applications of cardiovascular magnetic resonance imaging Heart, November 1, 2008; 94(11): 1485 - 1495. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Gebker, C. Jahnke, I. Paetsch, B. Schnackenburg, S. Kozerke, A. Bornstedt, E. Fleck, and E. Nagel MR Myocardial Perfusion Imaging with k-Space and Time Broad-Use Linear Acquisition Speed-up Technique: Feasibility Study Radiology, December 1, 2007; 245(3): 863 - 871. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. R. Nandalur, B. A. Dwamena, A. F. Choudhri, M. R. Nandalur, and R. C. Carlos Diagnostic Performance of Stress Cardiac Magnetic Resonance Imaging in the Detection of Coronary Artery Disease: A Meta-Analysis J. Am. Coll. Cardiol., October 2, 2007; 50(14): 1343 - 1353. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
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 |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
