| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Report |
1
Department of Diagnostic Radiology, University Hospital Essen, Hufelandstr.
55, D-45122 Essen, Germany.
2
Department of Cardiology, University Hospital Essen, D-45122 Essen,
Germany.
Received August 20, 2001;
accepted after revision August 30, 2001.
Address correspondence to J. Barkhausen.
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CONCLUSION. Using real-time trueFISP permits one to assess left ventricular function in a single breath-hold. The dramatic reduction in data acquisition time does require some compromises. The temporal and spatial resolutions of images obtained with real-time trueFISP were considerably lower than those achieved with segmented trueFISP. Further reduction of the TR or the use of sensitivity encoding could improve temporal resolution and eliminate other limitations of real-time trueFISP.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
To date, cine MR imaging has relied exclusively on single-slice techniques requiring multiple breath-holds. Therefore, coverage of the entire ventricle required scan times of approximately 10 min. Furthermore, variations in inspiratory depth led to discontinuous coverage of the ventricle, introducing measurement errors. A recently developed multislice real-time trueFISP cine sequence capable of covering the entire ventricle in a single breath-hold promises to resolve these problems. This study was performed to assess the performance of a real-time trueFISP sequence as a tool in the evaluation of myocardial function compared with the performance of a conventional single-section segmented trueFISP sequence.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Examinations were performed on a 1.5-T scanner (Magnetom Sonata; Siemens Medical Systems, Erlangen, Germany) equipped with high-performance gradients (maximum amplitude, 40 mT/m; slew rate, 200 mT/m per msec). The gradients in the trueFISP sequence used for cardiac imaging were completely balanced in all three directions. At the end of the TR, the transverse magnetization was refocused, and the next excitation (a radiofrequency pulse) could be started without further preparation.
Short-axis cine MR images were collected with a segmented trueFISP sequence as well as with the real-time trueFISP sequence. The TR was set to the minimum for both sequences, and the flip angles were set to the maximum possible under the limitations posed by the specific absorption rate. The segmented trueFISP sequence (TR/TE, 2.4/1.2; flip angle 60°) collected 15 lines per frame and heartbeat. An inplane data acquisition matrix of 120 x 256 was used, resulting in an acquisition time of 8 heartbeats. A rectangular (6/8) field of view of 350 mm2 rendered an inplane spatial resolution of 2.2 x 1.4 mm. Slice thickness was 8 mm, and the entire left ventricle was covered without interslice gaps. The temporal resolution was 36 msec (15 x TR) for the segmented trueFISP sequence without echo-sharing. The phased array torso coil (2 coil elements) and the spine coil (2 coil elements) were used for signal reception.
The real-time trueFISP sequence (2.2/1.1; flip angle, 45°) collects one line every 2.2 msec. Our inplane data acquisition matrix was 63 x 128. Echo-sharing was used to improve temporal resolution. For this purpose, seven additional lines of central k-space were collected between k- of the first-phase image and k+ of the third-phase image (Fig. 1). The second-phase image was then calculated from k+ of phase 1 to k- of phase 3, with the seven additional central k-lines between the two. With this technique, 70 lines were required for each image set, resulting in an acquisition time of 154 msec (70 x 2.2 msec) and an effective temporal resolution of 77 msec using echo-sharing.
|
For the multislice technique, ECG-triggering is needed to synchronize the slices. All phases of the first slice are collected during the first R-R interval. The number of phases per cardiac cycle depends on the heart rate (number of phases = R-R interval/77 msec). The third R wave triggers the acquisition of the second-slice data. The total scan time can be calculated as 2m heartbeats, with m equaling the number of slices. Because we used a sequence with 7 slices, the acquisition time was 14 heartbeats. Slice thickness was 8 mm with variable interslice gaps (10-50% of the slice thickness) to cover the entire ventricle. The positions of the most basal and the apical slices were identical for the real-time and the segmented trueFISP sequences.
The field of view for the real-time trueFISP sequence was minimized for each patient to improve spatial resolution. Only the two coil elements of the phased array torso coil (B1 and B2) were used, and the spine coil was switched off. The decreased signal intensity of the patient's back resulted in reduced wrap-around artifacts and facilitated further reduction of the rectangular (5/8) field of view (Fig. 2A,2B,2C,2D). Depending on the anatomy of the particular patient, the field of view varied between 270 and 340 mm2, rendering a pixel size ranging from 2.7 x 2.1 mm2 to 3.4 x 2.7 mm2. Image intensity correction provided by the scanner was used to improve the homogeneity of the image (Fig. 2A,2B,2C,2D).
|
|
|
|
Contrast-to-noise ratio (CNR) measurements were based on signal intensity measurements taken in standardized regions of interest in the myocardium, the left ventricular cavity, and an external area (outside the patient's body) with minimal artifacts. The diameter of the round regions of interest were 4 mm for myocardium and 10 mm for the ventricular cavity and the noise measurement. The signal intensity of the myocardium was calculated as the mean value of eight regions of interest. Great care was taken to have regions of interest of identical sizes and shapes at similar positions in each patient for both sequences. Noise was defined as the standard deviation of the signal intensity in the external area. CNRs were calculated from the mean of the measurements throughout the entire cardiac cycle (CNR = [signal intensity of left ventricular cavity — signal intensity of myocardium] / noise).
End-diastolic volumes, end-systolic volumes, ejection fractions, and stroke volumes were calculated from both data sets by software (Argus software; Siemens Medical Systems) that takes different slice thicknesses and interslice gaps into account. All endocardial contours were drawn manually by an experienced observer. The results of the cardiodynamic measurements were compared calculating systematic and random differences as well as the correlation coefficients. For CNR and cardiodynamic measurements, Student's t test was performed to determine the statistical significance of observed differences. A p value of less than 0.05 was considered to indicate statistical significance.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
|
|
|
|
|
|
The signal intensity of the intraventricular blood (324 ± 175) on real-time trueFISP images showed no statistically significant difference to the signal intensity on the segmented trueFISP images (310 ± 126), even though the two elements of the spine coil were switched off. CNR values of real-time images (84 ± 24) exceeded those of segmented trueFISP (32 ± 14) by more than 250% (paired Student's t test, p<0.05), reflecting much lower noise values (2.9 ± 1.0 vs 7.5±2.4) because of the larger pixel size.
Cardiodynamic parameters based on segmented and real-time trueFISP images correlated well, as evidenced by r values greater than or equal to 0.97 for all parameters (Table 1). Manual segmentation of real-time trueFISP images resulted in a slight underestimation of end-diastolic volumes, whereas end-systolic volumes were slightly overestimated. Consequently, stroke volumes and ejection fractions were slightly underestimated.
|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
In 1991, Atkinson and Edelmann [4] introduced a segmented turbo fast low-angle shot sequence for single breath-hold cardiac cine MR imaging—a technique that has become the standard for cardiac cine MR imaging [5]. Scanner generations equipped with higher performance gradients can now be used for trueFISP imaging. TrueFISP sequences are characterized by favorable contrast properties that are nearly independent from blood flow and extremely short data-acquisition times. CNRs between the myocardium and the ventricular cavity are significantly improved compared to those seen with fast low-angle shot imaging [3].
Until now, single-slice techniques have typically been used for clinical cardiac cine MR imaging. Alley et al. [6] showed the feasibility of three-dimensional cine MR imaging, but because of the low intrinsic contrast between the myocardium and the left ventricular blood pool, a T1-shortening contrast agent had to be added. The total acquisition time for this sequence to cover the entire heart was 4.1 min.
Echoplanar cine MR imaging [7] enables full cardiac coverage in just a single breath-hold [8]. Using a data acquisition matrix of 52 x 128, one can achieve an acquisition time of as little as 20-25 sec, but image quality is impaired by artifacts inherent to echoplanar imaging [7]. The real-time trueFISP sequence we present is capable of acquiring multiple short-axis slices through the entire left ventricle in a single breath-hold. This sequence combines the excellent image quality and high contrast inherent in trueFISP with ultrafast multislice data acquisition.
The dramatic reduction in data-acquisition time does require some compromises. Thus, spatial resolution with real-time trueFISP was considerably lower than that achieved with segmented trueFISP. Furthermore, the two dorsal coil elements had to be switched off to permit a reduction in the field of view. We compensated for the ensuing signal drop with an image-intensity-correction algorithm without affecting image quality or the contrast between the intraventricular blood pool and the myocardium. Although image intensity correction might influence the absolute values of the CNR calculations, this effect was not addressed in our study because an identical filter was used for both sequences.
Our data suggest that the reduced spatial resolution with a pixel size of 3.47 x 2.73 mm2 or less had only a minor impact on the accuracy of the volumetric measurements. Endocardial contours were well delineated, and the measured cardiac volumes were within 5% of the standard of reference.
The slight overestimation of end-systolic volumes that resulted in an underestimation of stroke volumes and ejection fractions may be related to limited temporal resolution hampering the exact determination of the end of systole, which lasts for as little as 40-50 msec [9]. As suggested by Foo et al. [9], any discussion of temporal resolution should differentiate between effective and true temporal resolution. True temporal resolution is defined as the time required to collect a single frame, which in turn can be calculated by multiplying the number of k-lines by the TR. Hence, the only way to improve true temporal resolution without reducing the acquisition matrix is by reducing the TR. The TR of the real-time trueFISP sequence we presented is 2.2 msec; therefore, not much room is left for further reductions. The effective temporal resolution, however, can be improved without changing the true temporal resolution or acquisition time via view-sharing [9]. Using this technique, we reduced the effective temporal resolution of our real-time trueFISP sequence to 77 msec, whereas the true temporal resolution was 139 msec (63 k-space lines x TR) (Fig. 1).
Another method of improving spatial or temporal resolution is to combine the real-time sequence with simultaneous acquisition of spatial harmonics [10] or sensitivity encoding [11]. Both techniques use arrays of several parallel receiver coils to reduce scan time. The speed of these techniques results from reducing the number of phase-encoding gradient steps by extracting spatial information contained in a radiofrequency coil array. Because of the excellent CNR, trueFISP sequences are well suited to these approaches.
In conclusion, we found that full cardiac coverage real-time trueFISP cine MR imaging, lasting just 14 heartbeats, was possible in all patients in our study and conclude that real-time trueFISP imaging is well suited for assessing ventricular function. The excellent contrast properties of this robust sequence, coupled with extremely short data-acquisition times, may allow real-time trueFISP to emerge as an essential component of any cardiac MR examination.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  F. Saremi, J. D. Grizzard, and R. J. Kim Optimizing Cardiac MR Imaging: Practical Remedies for Artifacts RadioGraphics, July 1, 2008; 28(4): 1161 - 1187. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C Plathow, M Klopp, C Fink, A Sandner, H Hof, M Puderbach, F Herth, A Schmahl, and H-U Kauczor Quantitative analysis of lung and tumour mobility: comparison of two time-resolved MRI sequences Br. J. Radiol., September 1, 2005; 78(933): 836 - 840. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. Kramer, S. O. Schoenberg, K. Nikolaou, A. Huber, A. Struwe, E. Winnik, B. J. Wintersperger, O. Dietrich, B. Kiefer, and M. F. Reiser Cardiovascular Screening with Parallel Imaging Techniques and a Whole-Body MR Imager Radiology, July 1, 2005; 236(1): 300 - 310. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Schlosser, K. Pagonidis, C. U. Herborn, P. Hunold, K.-U. Waltering, T. C. Lauenstein, and J. Barkhausen Assessment of Left Ventricular Parameters Using 16-MDCT and New Software for Endocardial and Epicardial Border Delineation Am. J. Roentgenol., March 1, 2005; 184(3): 765 - 773. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
