| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Developments |
1 From the Department of Diagnostic Radiology (E.S., A.H.M., R.W.G., A.B.) and Medical Clinic I (J.S., H.P.K., P.H.), University Hospital, Technical University of Aachen, Pauwelsstrasse 30, 52057 Aachen, Germany; Philips Research Laboratories, Hamburg, Germany (T.S.); Department of Medicine (Cardiovascular Division), Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Mass (R.M.B.); and Philips Medical Systems, Best, the Netherlands (R.M.B.). Received March 28, 2002; revision requested May 31; final revision received September 27; accepted October 14. Address correspondence to E.S. (e-mail: spuenti@rad.rwth-aachen.de).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMaterials and Methods ResultsDiscussionREFERENCES |
|---|
Supplemental material: radiology.rsnajnls.org/cgi/content/full/227/3/870/DC1
© RSNA, 2003
Index terms: Heart, ventricles • Magnetic resonance (MR), cine study, 51.121419, 51.12144 • Magnetic resonance (MR), functional imaging, 51.121412, 51.121419 • Magnetic resonance (MR), volume measurement, 51.12144 • Myocardium, MR, 51.121412, 51.121419, 51.12144
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and Methods ResultsDiscussionREFERENCES |
|---|
Recently, first real-time approaches for the assessment of cardiac function have been introduced by using cartesian or spiral k-space readouts, and these approaches allowed real-time data sampling without k-space segmentation (images of all heart phases are acquired during a single R-R interval) (6–9). Thus, such techniques offer the potential for data acquisition during free-breathing and without cardiac triggering. One limitation of these approaches was the lack of real-time reconstruction and interactive section positioning, which is essential for an interactive examination of the heart with real-time image display and interactive continuous section positioning, such as the desired "fly through," as known from echocardiography. Furthermore, radial k-space sampling has not yet been investigated for quantitative assessment of left ventricular function, and in most of the earlier studies, breath holding or cardiac triggering were used.
Consequently, the purpose of our study was to investigate the potential of the quantitative assessment of left ventricular function during an interactive examination of the heart (without cardiac triggering or breath holding). In addition to spiral imaging as used in earlier studies (10,11), we also investigated a newly developed radial steady-state free precession (also referred to as the balanced fast field-echo, the true fast imaging with steady-state precession, or the fast imaging employing steady-state acquisition [or FIESTA]) (12–14) sequence.
|
Materials and Methods |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and Methods ResultsDiscussionREFERENCES |
|---|
Subjects
Standard breath-hold segmented k-space cine steady-state free precession, interactive real-time spiral gradient-echo, and interactive real-time radial steady-state free precession MR imaging of the left ventricle were performed in 18 consecutive subjects (12 men [age range, 21–83 years], six women [age range, 41–81 years]) with a history of subacute (n = 2) or chronic (n = 4) myocardial infarction, hypertrophic obstructive cardiomyopathy (n = 2), aortic valve replacement (n = 2), coronary artery disease (n = 1), aortic regurgitation (n = 1), aortic bulb aneurysm (n = 1), ischemic cardiomyopathy (n = 2), or acute stroke (n = 3). Informed consent was obtained from all subjects who participated in this study, as required by our local review board that approved this study.
Standard Breath-hold Steady-State Free Precession Cine MR Imaging
A cardiac gated segmented k-space cine breath-hold steady-state free precession (balanced fast field-echo) sequence (16–19) performed to acquire images of 20 heart phases was used as the standard. Imaging parameters included repetition time msec/echo time msec, 3.2/1.6; flip angle, 55°; field of view, 370 x 259 mm2; matrix, 256 x 128; section thickness, 10 mm; section gap, none. Each section was acquired in a single end-expiratory breath hold. Seven to 12 short-axis views were acquired to cover the entire left ventricle. For data reconstruction, only the two anterior coil elements (6) were used for image reconstruction.
Interactive Real-Time MR Imaging
Spiral gradient-echo (20) and radial steady-state free precession (balanced fast field-echo) (12,13) sequences were used for interactive real-time MR assessment of left ventricular function. Data were acquired during free breathing and without cardiac triggering. The spiral sequence consisted of four spiral interleaves with 30/4 and a flip angle of 30°. Imaging parameters of the radial steady-state free precession sequence were 2.5/1.2 and a flip angle of 45°. The field of view was 320 mm2 and the matrix was 128 x 128, which was reconstructed to a 256 x 256 matrix by using zero filling. For improved temporal resolution in radial imaging, undersampling with 80 continuous radial k-space lines was performed without compromising spatial resolution (12,13). For both sequences, image reconstruction was performed by using the sliding-window technique (21), which enabled further improvements in temporal resolution. Thus, real-time images obtained with a frame rate of 15 images per second (67 milliseconds per frame) were displayed online at the operator console by using a dedicated real-time reconstructor (15). Image geometry and contrast could be changed interactively (22).
Description of Experiments and Analysis of Data
All images were successfully obtained without complications. In all subjects, short-axis views that included the entire left ventricle could be obtained with all three sequences. Imaging time was less than 10 minutes for both interactive real-time sequences. Frames obtained with both real-time sequences were stored during an interactive examination of the heart for subsequent quantitative analysis. To allow comparison of all three sequences, section positions as derived from the breath-hold segmented k-space sequence were repeated during interactive examination of the heart, and movie clips were stored. Because of real-time reconstruction and real-time online display, the user could identify an end-expiratory position prior to movie storage.
Data were transferred to a commercially available workstation (Easy Vision 4.0; Philips Medical Systems). For each section and each sequence, end-diastolic and end-systolic frames were visually assessed by one author (E.S.) from the movie displayed at the workstation console. Only end-expiratory positions were used for analysis. End systole was defined as the frame with maximum contraction and end diastole was defined as the frame with maximum dilatation of the left ventricle over the entire movie. End-systolic and end-diastolic volume measurements for the quantitative assessment of left ventricular function were performed by using a user-specified image-window display and a user-specified number of points that were positioned manually by one author (E.S.) who was blinded to patient data at the endocardial and epicardial border (23). The endocardial border was defined as the border between the left ventricular muscle and the left ventricular blood pool excluding the papillary muscles. For epicardial volume measurements, the points were placed at the edge of the right ventricular blood pool, epicardial fat, or lung tissue (if no epicardial fat was visible). After initial point placement, each point was captured again and replaced to ensure correct positioning at the endocardial and epicardial border. Sections from the apex to the base were used. The most basal section for analysis was defined as the last section that showed more than 50% circumference of the left ventricular myocardium in all heart phases (6). From these regions of interest (ROIs), the end-diastolic and end-systolic endocardial volume, as well as epicardial volume, of the left ventricle were calculated by using the Simpson method. Furthermore, stroke volume, ejection fraction, and diastolic myocardial mass were derived (6,9,23).
For signal quantification, contrast-to-noise ratio (CNR) (7,13,17) was calculated from a user-specified ROI drawn in the septum (ROI in the myocardium, 117–223 mm2) and in the adjacent left ventricular cavity (ROI in blood, 106–495 mm2) as follows: CNR = (SIblood - SImyo)/SDair, where SI is signal intensity in blood and myocardial ROIs, and SDair was defined as the SD in a region of air that was 530–877 mm2 and in front of the chest. Identical (size, position) ROIs were used for all three investigated sequences.
In addition to quantitative analysis, image quality of the diastolic and systolic frames was assessed by two investigators (A.H.M., E.S.) in consensus by using the 16-segment model (24) and a three-point grading scale for the endocardial and the epicardial borders. The scale was as follows: grade 3, good image quality with a well-defined border and no artifacts; grade 2, moderate image quality with minor artifacts and minor compromised border detection, but with quantitative analysis possible; and grade 1, severe susceptibility artifacts or blurring with insufficient image quality for border detection.
Statistical Evaluation
All values were expressed as the mean ± 1 SD. Quantitative data obtained with the new techniques (interactive spiral gradient-echo and interactive radial steady-state free precession sequences) were compared with those obtained with the standard technique (segmented k-space breath-hold steady-state free precession) by using a bidirectional paired Student t test and by calculating the relative difference (ie, the difference between two techniques divided by their mean value) as previously described (6,25). A P value of less than .05 was considered to indicate a significant difference.
|
Results |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and Methods ResultsDiscussionREFERENCES |
|---|
|
|
|
|
|
|
|
|
In all (18 of 18) subjects, the endocardial and epicardial borders could be successfully defined with a score of 2 or greater (all segments) with all three investigated sequences (all scores 2 or 3 in all segments in all patients and in all three sequences). However, epicardial border definition in segments 9–15 was graded with a mean score of less than 2.50 (score of 3.00 in less than 50% of all cases) if the interactive real-time spiral gradient-echo sequence was used (Table 2; Figs 2, 3). The segments to which the numbers refer are the segments of the 16-segment model described in Methods. Delineation of the segments is included in Table 2. This was mainly due to signal void at the epicardial border (Fig 2). Furthermore, especially in theses segments, blurring at the endocardial border was seen (Fig 3) with the interactive real-time spiral gradient-echo sequence.
|
|
Discussion |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and Methods ResultsDiscussionREFERENCES |
|---|
In contrast to this finding, interactive radial steady-state free precession imaging is less sensitive to susceptibility and T2* artifacts (14,28) due to the short repetition time (repetition time, 2.5 msec) and short echo time (echo time, 1.2 msec). Radial artifacts (radial streaks) (12,13) are perpendicularly oriented to the endocardial and epicardial borders and therefore are more favorable with respect to border delineation (12,13) when compared with the blurring that occurs along the borders with spiral techniques (26). Additionally, radial imaging is less sensitive to motion and allows enhanced temporal resolution by using undersampling without compromising spatial resolution (12–14,29,30). Radial imaging can also be favorably combined with the sliding-window technique, because reduced radial k-space lines allow sufficient border delineation, and a short repetition time is used (12,13). This may explain the excellent border definition in all segments and all heart phases, as well as the excellent agreement of the quantitative assessment of left ventricular function by using interactive real-time radial steady-state free precession with segmented k-space breath-hold steady-state free precession as the standard.
For quantitative assessment of left ventricular function by using cine MR imaging, a high temporal and spatial resolution is needed (7). High temporal resolution (15 frames per second) was accomplished by using the sliding-window technique (12–14,21), which has been shown to allow enhanced temporal resolution, thereby enabling accurate quantification of left ventricular volumes (11). This was combined with undersampling in radial scanning (12,13,30). Spatial resolution was 2.5 x 2.5 mm2 (reconstructed to 1.25 x 1.25 mm2) for both interactive real-time sequences, which was higher than it was in published articles about real-time sequences (2.2 x 4.4 mm2 [6]) but similar to that of our standard breath-hold segmented k-space sequence (1.4 x 2.8 mm2). To avoid foldover artifacts by using cartesian sampling, a larger field of view was necessary with the standard breath-hold sequence when compared with that with the interactive real-time sequences (12,13).
The different contrast properties of steady-state free precession imaging when compared with a gradient-echo approach may have a further influence on the endocardial and epicardial volume measurements. Image contrast with a gradient-echo sequence (including spiral gradient-echo sequences) is typically based on inflow and may be reduced by relatively thick sections (10) or impaired cardiac function. On the other hand, image contrast with a steady-state free precession technique is primarily based on a T2-like contrast (31) rather than on inflow (intrinsic flow compensation). Because the steady-state free precession technique has been shown to be superior for functional cardiac MR imaging in terms of signal intensity, image contrast, and border detection (17–19), we used a segmented k-space breath-hold steady-state free precession technique as the standard sequence. Thus, the contrast in our interactive real-time radial steady-state free precession and that in the standard breath-hold segmented k-space steady-state free precession sequences were similar (12,13). This also may explain the excellent agreement of the calculated functional parameters in our interactive real-time radial steady-state free precession sequence when compared with those of the standard breath-hold technique. Although findings in earlier studies (18) in which segmented gradient-echo techniques were compared with segmented steady-state free precession techniques indicated a slight underestimation of endocardial volumes with gradient-echo imaging, which was similar to the underestimation with our results by using spiral gradient-echo imaging, overestimation of myocardial mass as seen by using those segmented cartesian gradient-echo techniques was not observed in our study. This may be due to the signal void at the epicardial border caused by off-resonance and susceptibility artifacts with the spiral imaging approach (10,26,27).
Quantitative signal analysis demonstrated a higher CNR for the interactive real-time spiral gradient-echo technique when it was compared with the radial steady-state free precession technique, and this finding can be explained by the intrinsic high signal-to-noise ratio in spiral k-space sampling. However, reduced CNR with the interactive radial steady-state free precession sequence did not compromise accurate assessment of left ventricular volumes, which indicated that CNR in our interactive real-time sequences was not a limiting factor for the assessment of left ventricular function.
Recently, findings with several fast MR imaging techniques for real-time data acquisition have been reported (6,11,18), and these findings allowed highly reproducible quantification of left ventricular volumes. However, data reconstruction was often performed off line, and therefore no interactive examination (as known from echocardiography) was possible. Furthermore, electrocardiographic triggering or breath holds were used (6,7,32). The main advantage of real-time imaging, however, is the potential of a completely untriggered data acquisition (no electrocardiographic triggering and no breath holding), which increases patient comfort, reduces operator involvement, and may be especially needed in patients with arrhythmia or reduced breath-holding capabilities (20,33). In the present study, we demonstrated that quantitative assessment of left ventricular function can be accurately performed during an interactive examination of the heart by using interactive real-time radial steady-state free precession imaging.
In our study, we focused on the quantitative assessment of left ventricular function. Further important clinical parameters in functional cardiac MR imaging include wall motion, which remains to be investigated by using the present interactive real-time approaches. For wall motion analysis, however, larger patient groups, including multiple focal wall motion abnormalities, are needed.
In conclusion, interactive real-time radial steady-state free precession imaging with a fly through the heart allows accurate quantitative assessment of left ventricular function when it is compared with standard segmented k-space steady-state free precession cine MR imaging, whereas interactive spiral gradient-echo imaging was less accurate.
|
FOOTNOTES |
|---|
Author contributions: Guarantors of integrity of entire study, E.S., R.W.G., A.B.; study concepts, E.S., T.S., A.B.; study design, E.S., J.S.; literature research, E.S., A.B.; clinical studies, E.S., J.S., A.H.M.; data acquisition, E.S., J.S., A.H.M.; data analysis/interpretation, E.S., A.B., J.S., A.H.M., H.P.K.; statistical analysis, E.S., R.M.B., H.P.K.; manuscript preparation, E.S., R.M.B., A.B.; manuscript definition of intellectual content, E.S., A.B., P.H., T.S.; manuscript editing, E.S., A.B.; manuscript revision/review, R.W.G., R.M.B., H.P.K.; manuscript final version approval, E.S., R.M.B., H.P.K.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and Methods ResultsDiscussionREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  V. Muthurangu, P. Lurz, J. D. Critchely, J. E. Deanfield, A. M. Taylor, and M. S. Hansen Real-time Assessment of Right and Left Ventricular Volumes and Function in Patients with Congenital Heart Disease by Using High Spatiotemporal Resolution Radial k-t SENSE Radiology, September 1, 2008; 248(3): 782 - 791. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 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. Corsi, R. M. Lang, F. Veronesi, L. Weinert, E. G. Caiani, P. MacEneaney, C. Lamberti, and V. Mor-Avi Volumetric Quantification of Global and Regional Left Ventricular Function From Real-Time Three-Dimensional Echocardiographic Images Circulation, August 23, 2005; 112(8): 1161 - 1170. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
