| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Clinical Radiology, University of Muenster,
Albert-Schweitzer-Straße 33, D-48149 Muenster, Germany.
2 Department of Cardiology and Angiology, University of Muenster, Muenster,
Germany.
Received November 13, 2004;
accepted after revision February 16, 2005.
Address correspondence to K. U. Juergens
(kujuerg{at}uni-muenster.de).
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
SUBJECTS AND METHODS. This study consisted of 20 patients with either coronary artery disease or dilated cardiomyopathy. We evaluated MDCT results using cine MRI as the reference technique.
RESULTS. LV end-diastolic volume (LVEDV) and end-systolic volume (LVESV) were significantly overestimated by the axial MDCT approach in comparison with volume measurements from short-axis CT image reformations. The mean LV ejection fraction (LVEF) was not significantly different (41.2% vs 42.7%). Short-axis and axial MDCT determination of LVEF revealed a systematic underestimation by a mean ± SD of -2.1% ± 3.6% versus -3.6% ± 8.2%, respectively, when compared with LVEF values based on cine MRI. The interobserver variability for volume and function measurements from axial images (LVEDV = 8.5%, LVESV = 10.8%, LVEF = 9.6%) was slightly higher than those measurements from short-axis reformations (LVEDV = 7.2%, LVESV = 9.5%, LVEF = 8.7%). The mean total evaluation time was significantly shorter using axial images (14.1 ± 3.9 min) compared with short-axis reformations (16.9 ± 5.2 min) (p < 0.05).
CONCLUSION. Determination of LV volumes and assessment of global LV function from axial MDCT image reformations is feasible and time efficient. This approach might be a clinically useful alternative to established short-axis-based measurements in patients with normal or near-normal LV function. A progressive underestimation of LVEF with increasing LV volumes may limit the clinical applicability of the axial approach in patients with dilated cardiomyopathy.
Keywords: axial image reformations • cine MRI • left ventricular function • MDCT • semiautomated MDCT data analysis
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Assessment of LV function using MDCT usually is based on planimetric measurements performed on short-axis image reformations. Image postprocessing to generate short-axis reformations for function evaluation prolongs analysis time needed for MDCT compared with cine MRI [5]. Because MDCT of the heart is a high-resolution volume acquisition, slice orientation should not significantly influence the results of volume measurements, and LV volume determination based on axial images should allow faster global LV function assessment. We therefore evaluated the feasibility, postprocessing time, and interobserver variability of LV volumetry and global LV function assessment from axial MDCT images in comparison with short-axis image reformations in patients with normal and impaired LV function and we used cine MRI as the reference technique.
|
|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Scan Protocol and Image Acquisition
MDCT studies were performed on a 16-MDCT system (Somatom Sensation 16,
Siemens Medical Solutions) using standard parameters for coronary artery CT
angiography: detector configuration of 16 x 0.75 mm, 120 kV, 550 mAs,
rotation time of 420 msec, and table speed of 3.4 mm/rotation. Patients
received a ß-adrenoreceptor antagonist 45 min before the examination (80
mg of propanolol orally) if resting heart rate exceeded 60 beats per minute
(bpm). One hundred milliliters of a nonionic contrast material (iomeprol, 300
mg I/mL) was injected via an antecubital vein at 3.5 mL/sec followed by a
50-mL saline chaser bolus using a power injector (Injektron CT2, Medtron).
Image Reconstruction
ECG-gated image reconstruction was performed in 5% steps through the entire
R-R interval yielding 20 phases (axial slice orientation; section thickness, 1
mm; increment, 0.6 mm; medium-soft convolution kernel; field of view, 180 mm;
reconstruction matrix, 512 x 512 pixels). The resulting multiphase image
series were then used to produce multiplanar reformations in the short-axis
orientation to cover the entire LV cavity using the system's standard 3D
software. The section thickness was set to 8 mm without an intersection gap.
The short-axis images were reviewed at the midventricular level to select the
end-diastolic and end-systolic phases, which were identified as images showing
the largest and the smallest LV cavity area, respectively. Axial images for
LVEDV and LVESV measurements were created by fusing the source axial image
sections to thicker-section axial reconstructions (5-mm section thickness, no
intersection gap). Only end-diastolic and end-systolic image series in the
axial and short-axis orientations were used for further analysis (Figs.
1A and
1B).
MDCT Data Analysis
Global LV function assessment based on axial images and short-axis
end-diastolic and end-systolic image reformations was performed using a
commercially available software package
[9] for cardiac function
analysis with CT (CT MASS [version 6.1], Medis). The software supports
automatic endocardial contour detection in short-axis and axial images with a
discernible LV cavity. The outlined borders of the LV cavity were visually
checked and manually corrected if necessary. LV trabeculations and papillary
muscles were treated as LV cavity. For short-axis reformations, the most basal
slice was defined as the image closest to the mitral valve annulus showing LV
myocardium in at least 50% of its perimeter. The most apical image was the
last image with a detectable LV lumen. For the axial images, all sections
showing an LV cavity were included in the evaluation. Because the mitral valve
plane was depicted in all images, the plane connecting the anterior and
posterior mitral valve annulus was used as the basal border of the LV
cavity.
LVEDV and LVESV, LV ejection fraction (LVEF), and LV stroke volume (LVSV) were calculated by the software. Measurements were performed independently by two experienced reviewers trained in the CT analysis software with 5 years (reviewer 1) and 4 years (reviewer 2) of experience in cardiac CT to assess interobserver variability.
The time from loading the thin-slice axial postprocessed images into the scanner's 3D software to create either axial thick-slice images or short-axis reformations until saving the new images was recorded as postprocessing time. The analysis time was the total time needed to load the images into the analysis software, perform contour detection, allow time for data calculation, and save the results. The postprocessing time and analysis time were recorded as the total evaluation time.
Cine MRI Scan Protocol and Data Analysis
MRI was performed on a 1.5-T unit (Gyroscan Intera, release 8.1.3, Philips
Medical Systems). A standard five-element cardiac synergy coil with the
Vectorcardiogram option (Philips Medical Systems) was used for signal
reception. After survey scout images in the axial, coronal, and sagittal
orientations were obtained, a prospectively ECG-gated breath-hold steady-state
free precession cine sequence (balanced fast-field-echo: TR/TE, 3.5/1.7; flip
angle, 50°; temporal resolution, 38 msec; matrix, 256 x 256; field
of view, 380 x 300 mm; section thickness, 8 mm) was acquired in the
short-axis image orientation at end-expiratory suspension.
Data analysis was performed by reviewer 1 using the cine-MRI-compatible version of the analysis software (MASS suite 6.1, Medis), which had been used for MDCT image assessment [10].
Statistical Analysis
The LV volumes and LVEFs are expressed as mean values ± SD and 95%
confidence intervals for reviewer 1. Results from the axial images and
short-axis reformations were compared using the Wilcoxon's test for paired
samples; statistical significance was assumed at a p value of less
than 0.05. The interobserver variability of the MDCT measurements was assessed
by dividing the absolute difference of the measurements from reviewer 1 and
reviewer 2 by the mean of the two measurements. Bland-Altman analysis
[11] was performed to
calculate the systematic error and limits of agreement between results from
axial and short-axis image evaluations. Linear regression between all
variables was tested by computing Pearson's correlation coefficient. The MDCT
data were examined for any pattern of bias related to the magnitude of the
measurements by plotting the data and computing Pearson's correlation
coefficient between the mean value from axial and short-axis MDCT measurements
and the absolute value of the difference between axial and short-axis MDCT
measurements, with a correlation coefficient statistically different from zero
suggesting a significant bias. All computations were done using SPSS software
(version 11.0, Statistical Package for the Social Sciences).
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
|
|
|
|
By correlating the mean values of axial and short-axis MDCT measurements to the absolute values of the differences of MDCT results, we detected a significant bias: Increasing LV volumes were observed and found to show a correlation coefficient of 0.601 for LVEDV and 0.565 for LVESV (all p < 0.05). No bias was found for LVEF (r = -0.019).
The interobserver variability between volume and function measurements from axial images (LVEDV = 8.5%, LVESV = 10.8%, LVEF = 9.6%) was slightly higher than from short-axis reformations (LVEDV = 7.2%, LVESV = 9.5%, LVEF = 8.7%), but only the difference for LVEDV was significant (Table 2).
|
A comparison of the LVEF values determined based on MDCT with those based on cine MRI revealed a systematic underestimation of LVEF of -3.6% ± 8.2% with axial images and -2.1% ± 3.6% with short-axis MDCT image reformations (Figs. 3A, 3B, 3C, 3D, 3E, and 3F).
|
|
|
|
|
|
|
|
|
|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
This short-axis approach has been adapted for determination of LV volume and assessment of global function by MDCT. Recent studies using short-axis image reformations from 4-MDCT indicate that LVEF values determined using that technique correlate well with LVEF values determined using cine MRI in patients with suspected or manifest coronary artery disease [1-3]. Because images are acquired in an axial plane, image postprocessing is necessary to create short-axis images, which prolongs evaluation time compared with cine MRI. We conducted the present study to evaluate the feasibility and interobserver variability of LV volumetric and function analyses from axial MDCT images. To our knowledge, this study is the first to use this approach and to show that global cardiac function can be reliably determined from axial MDCT images.
Because axial images and short-axis reformations were generated from the same primary thin-slice axial MDCT data set, comparable volume results were expected. However, LV volumes assessed using axial MDCT reformations were significantly larger than those obtained using short-axis reformations and overestimated volume by 18.6 mL (LVEDV) and 11.6 mL (LVESV) when compared with values from cine MRI. This effect can most likely be attributed to partial volume averaging, which is more important in axial images than in images adapted to the cardiac axes (Figs. 1A and 1B), and has already been reported in previous validating studies on LV volume measurements by cine MRI [14-16]. The definition of the basal slice in short-axis images may form a systematic error.
Despite significantly larger absolute LV volumes, relative measurements, such as the LVEF, were in good concordance with a systematic error of less than 1%. The limits of agreement between axial and short-axis MDCT results with regard to LVEF were similar to the interobserver variability of 8.7% found for short-axis images. Overestimation of LVEDV results in progressive underestimation of the LVEF with increasing LV volume, which may be clinically significant especially in patients with dilated cardiomyopathy. Considering the observed deviation from the established short-axis measurements, the axial MDCT analysis can be suggested only as an alternative approach in patients with normal or near-normal LV function.
We found a slight underestimation of LVEF by MDCT compared with cine MRI.
This effect was more pronounced for axial images but was still smaller than in
a recent study (
-7.9% to -11.5%) performed on a 4-MDCT system
[3]. We used a 16-MDCT system
with faster rotation speed (420 msec) and better z-axis resolution.
Since underestimation of LVEF is explained by a systematic overestimation of
LVESV due to its inability to capture the maximum systolic contraction because
of the inferior temporal resolution of MDCT, any improvement in temporal
resolution should improve measurement results. In terms of a clinical
application, the small mean differences between the axial and the short-axis
approaches and even the slightly better correlation of short-axis MDCT
measurements with cine MRI (2.1% vs 3.6%) do not seem to be relevant. Taking
these named limitations into consideration, we can therefore suggest that LV
function determination is feasible using only axial images.
The use of axial images resulted in significant reduction of the overall evaluation time. The mean total evaluation time in our study was 14.1 min using axial images and 16.9 min using short-axis image reformations, whereas a recent investigation on a 4-MDCT system reported a postprocessing time of 27 ± 3 min (mean ± SD) using a short-axis approach [17]. This difference may be explained by the different postprocessing software and the faster CT system used in our study. With regard to the total evaluation time, the resulting time savings of the axial MDCT approach occur during the data postprocessing while dispensing additional short-axis image reformations. However, this significant difference might clinically be important only when many cardiac CT examinations per day are performed, with a cumulative effect. Furthermore, one has to speculate that the difference found in this study might become insignificant with improving CT postprocessing technology in the future.
The measurement variability for the short-axis images in our study was comparable to previous studies reporting a 1.5-11.5% variability for LV volumes and 5.1-7.4% variability for LVEF [3, 4, 6]. We did find a slightly higher interobserver variability for LV volumes and LVEF from axial MDCT images compared with short-axis reformations, which can be explained by an inferior definition of LV contours on axial images. A thinner section thickness should improve the contour delineation but, in turn, would prolong analysis time.
A few limitations of our study must be considered: the number of patients is small and there is considerable heterogeneity with regard to the abnormality causing LV dilatation or dysfunction (or both). We did not test the influence of the section thickness of axial images on measurement reliability, but rather used a fixed section thickness.
In conclusion, determination of LV volumes and assessment of global LV function from axial MDCT image reformations is feasible and time efficient. Despite overestimation of LV volumes, LVEF is reliably determined by volumetry of axial images. Our approach might be a clinically useful alternative to the established short-axis-based measurements in patients with normal or near-normal LV volumes.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  K. U. Juergens, H. Seifarth, F. Range, S. Wienbeck, M. Wenker, W. Heindel, and R. Fischbach Automated Threshold-Based 3D Segmentation Versus Short-Axis Planimetry for Assessment of Global Left Ventricular Function with Dual-Source MDCT Am. J. Roentgenol., February 1, 2008; 190(2): 308 - 314. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Nakayama, M. Asano, K. Matsumoto, N. Nomura, T. Saito, A. Mizuno, and A. Mishima Reconstruction of the Right Heart in an Elderly Woman With Ebstein's Anomaly and Severe Heart Failure Ann. Thorac. Surg., November 1, 2007; 84(5): 1745 - 1746. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Brodoefel, U. Kramer, A. Reimann, C. Burgstahler, S. Schroeder, A. Kopp, and M. Heuschmid Dual-Source CT with Improved Temporal Resolution in Assessment of Left Ventricular Function: A Pilot Study Am. J. Roentgenol., November 1, 2007; 189(5): 1064 - 1070. [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]
|
|
|
|
|
|
P. M. Colletti Cardiac Imaging 2006 Am. J. Roentgenol., June 1, 2006; 186(6_Supplement_2): S337 - S340. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
