| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Cardiovascular Center, Rakuwakai Otowa Hospital, 2, Otowachoinji-cho,
Yamashina-ku, Kyoto 607-8062, Japan.
2 Department of Cardiology, Takase Clinic, Takasaki, Japan.
3 Department of Cardiology, Kusatsu Heart Center, Kusatsu, Japan.
4 Department of Radiological Technology, Rakuwakai Otowa Hosptial, Kyoto,
Japan
5 Department of General Internal Medicine, Rakuwakai Otowa Hospital, Kyoto,
Japan.
Received March 5, 2008;
accepted after revision June 16, 2008.
Address correspondence to H. Matsumoto
(matsumoto.hidenari{at}gmail.com).
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
SUBJECTS AND METHODS. Nineteen consecutively registered patients with atrial fibrillation and 19 patients in sinus rhythm taking heart-rate-lowering agents as needed underwent MDCT. Images were reconstructed with a half-scan reconstruction algorithm after ECG editing (deletion of short R-R intervals, insertion of additional temporal windows into the middiastolic phase of long R-R intervals, and shift of R points). We used a 5-point scale (4, no motion artifacts; 0, unevaluable) to evaluate motion artifacts and coronary artery image discontinuities greater than 1 mm on the curved multiplanar reconstruction images. Each coronary artery image with a motion score of 2 or greater for all segments and with 2 or fewer discontinuities was considered acceptable for diagnosis.
RESULTS. Middiastolic images of patients with atrial fibrillation showed fewer motion artifacts and image discontinuities than did end-systolic images of patients with atrial fibrillation. Despite greater heart rate variability under the condition of similar mean heart rates in patients with atrial fibrillation, motion artifacts and image discontinuities on middiastolic images were not significantly different from those on sinus rhythm images. Acceptable quality was achieved on 91% of middiastolic atrial fibrillation images and 93% of sinus rhythm images.
CONCLUSION. ECG-edited middiastolic atrial fibrillation images with aggressive heart rate control were of better quality than end-systolic images in patients with atrial fibrillation. The diagnostic image quality of the middiastolic images was comparable with that of sinus rhythm images.
Keywords: atrial fibrillation • coronary CT angiography • ECG editing • middiastolic phase reconstruction
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Atrial fibrillation is the most common type of arrhythmia, and the incidence increases markedly with aging [5]. Patients with atrial fibrillation may have symptoms mimicking coronary artery disease. Therefore, visualization of the signs of coronary artery disease is an important issue in the treatment of patients with atrial fibrillation. In patients with atrial fibrillation, however, it is difficult to obtain MDCT coronary angiographic images because of high heart rate variability, which results in temporal windows in different cardiac phases. This timing shift causes spatial inconsistency (typical stairstep appearance) on transverse images and motion artifacts (blurring). At end systole during isovolumetric relaxation and at middiastole during the slow filling phase, coronary motion is relatively quiescent [6–10]. The absolute end-systolic phase is reported to be the best time to obtain optimal MDCT coronary angiographic images of patients with atrial fibrillation [11, 12] because the duration of systole is relatively constant [7, 10, 13]. Oncel et al. [14] reported that dual-source CT with a temporal resolution of 83 milliseconds [15] has diagnostic image quality at end-systolic phase reconstruction in patients with atrial fibrillation who have high heart rates. However, end-systolic phase reconstruction results in image artifacts in most standard protocols of coronary angiography, including those with conventional 64-MDCT scanners. The artifacts occur because a scan acquisition time less than 100 milliseconds is needed to obtain coronary artery images with few motion artifacts at end systole [16]. At slower heart rates, middiastolic phase reconstruction images have better image quality than end-systolic phase reconstruction images because of the longer motion-free time in middiastole than in end systole [6–10].
ECG editing, which arbitrarily modifies the position of the temporal windows within the cardiac cycle, enables correction of and compensation for the artifacts produced by heart rhythm irregularities. Cademartiri et al. [17] reported improvement of diagnostic accuracy with 16-MDCT coronary angiography performed with ECG editing in patients with mild irregularities of heart rhythm, including atrial fibrillation. However, only four patients in that study had atrial fibrillation, and image quality was not evaluated.
We hypothesized that for patients with atrial fibrillation, ECG-edited middiastolic phase reconstruction MDCT coronary angiographic images with heart rate control are of better quality than end-systolic phase images. The aims of this study were to evaluate the image quality of 64-MDCT coronary angiography in two reconstruction phases with ECG editing for patients with atrial fibrillation and to compare the images with those of patients in sinus rhythm.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Patients with atrial fibrillation and a heart rate greater than 65 beats/min received oral administration of 20–40 mg of metoprolol (Lopressor, Novartis) and/or 40 mg of verapamil (Vasolan, Eisai) 1 hour before image acquisition. Those who had a persistently high heart rate greater than 65 beats/min after oral administration of heart-rate-lowering agents received additional IV administrations of 2–6 mg of propranolol (Inderal, Sumitomo) and/or 5–10 mg of verapamil while ECG monitoring was performed. Patients in sinus rhythm with a heart rate greater than 65 beats/min received oral administration of 20–100 mg of metoprolol 1 hour before image acquisition. Those who had a persistently high heart rate greater than 65 beats/min after oral administration of metoprolol received additional IV administration of 2–6 mg of propranolol. Two minutes before CT, all patients received a single 0.3-mg sublingual dose of nitroglycerin (Myocor, Yamanouchi).
Exclusion criteria in both groups were unstable clinical status, known allergy to iodinated contrast material, creatinine level greater than 1.5 mg/dL (132.6 µmol/L), pregnancy, inability to hold a breath during scanning, and previous bypass surgery or implantation of a coronary stent. Exclusion criteria among the patients in sinus rhythm were a mean heart rate greater than 70 or less than 40 beats/min, presence of premature heartbeats, second- or third-degree heart block, and mistriggering of the R wave during scanning. Patients with atrial fibrillation and an elevated baseline heart rate were not excluded from this study.
MDCT Protocol
CT was performed with a 64-MDCT scanner (Somatom Sensation Cardiac 64,
Siemens Medical Solutions). Scans were acquired in the craniocaudal direction
with simultaneous recording of the ECG signal to allow image reconstruction
based on retrospective ECG gating. The scanning range covered the entire heart
from the level of the tracheal bifurcation to the diaphragm. Scanning
parameters were 64 x 0.6-mm collimation, z-flying focus spot
technique [18],
330-millisecond rotation time, 120-kV tube voltage, 750- to 850-mAs tube
current, and 61.9-mGy volume CT dose index. The pitch was 0.18–0.2 in
sinus rhythm patients and 0.18 in patients with atrial fibrillation. Digital
ECG was performed during data acquisition, and the data were stored. A bolus
(50–80 mL) of iopamidol (Iopamiron 370, Bayer) followed by saline
solution was injected into the antecubital vein through a 20-gauge IV
catheter. The injection rate for both the contrast agent and saline solution
was body weight in kilograms x 0.06 mL/s. Bolus timing was achieved with
an automated bolus-triggering system (CARE Bolus, Siemens Medical Solutions)
with a threshold of 150 HU detected within a region of interest placed on the
ascending aorta.
CT Data Postprocessing
Image reconstruction, including ECG editing, was performed in consensus by
two operators using retrospective ECG gating. A half-scan reconstruction
algorithm was used wherein each single transverse section contained data from
only one R-R cycle, resulting in a temporal resolution equivalent to one half
of the rotation time (165 milliseconds)
[18]. Image reconstruction
based on absolute timing was performed as previously described
[19]. Absolute timing meant
that image reconstruction was started at a defined time (measured in
milliseconds) after the previous R wave
[10,
19]. The intervals of the
cardiac cycle with minimal cardiac motion were identified with a preview
series that consisted of 0.75-mm transverse sections in the same
z-axis position at the midlevel of the heart. Image reconstruction
parameters were the individually adapted field of view, matrix size of 512
x 512 pixels, medium soft-tissue convolution kernel (B25f), and section
thickness of 0.75 mm with a reconstruction increment of 0.4 mm.
Images of patients with atrial fibrillation were reconstructed at the absolute end-systolic phase and absolute middiastolic phase after ECG editing. If the R-R interval and heart rate variability remained within an acceptable range, no additional editing was needed. With the reconstructed data, we aimed to reduce heart rate and heart rate variability with ECG editing and to reconstruct images from small numbers of relatively long R-R interval data. Taking into account the physiologic properties of the heart, at atrial fibrillation middiastolic reconstruction we obtained 20 equidistant absolute starting points of the reconstruction interval between the end of the T wave (approximately the phase of rapid ventricular filling) and the beginning of the following R wave. The phase depicting the fewest motion artifacts in the coronary arteries was selected. The technique was repeated with a shorter interval of starting points (5–10 milliseconds), and the optimal reconstruction intervals were chosen. The average starting point of the reconstruction interval for middiastole was 710 milliseconds (range, 570–890 milliseconds) after the R wave.
The ECG editing protocol for atrial fibrillation middiastolic reconstruction was as follows. The isolated short R-R interval data (> 65 beats/min) were deleted. If contiguous R-R intervals exhibited a slow heart rate, additional temporal windows were inserted into the middiastolic phase to avoid data gaps. If there were two or more contiguous short R-R intervals (> 65 beats/min), the relative short R-R interval data were deleted as much as possible. When the intervals of the cardiac cycle with minimal cardiac motion at midlevel of the heart caused the occurrence of different cardiac phases in some cardiac cycles, we repositioned the former R point backward to lead the temporal window to be in the slow-filling phase.
At atrial fibrillation end-systolic reconstruction, we chose the optimal starting point of the reconstruction interval between the beginning of the T wave (approximately the end of maximum systolic ejection) and the end of the T wave. The average starting point of the reconstruction interval for end systole was 244 milliseconds (range, 130–320 milliseconds) after the R wave. The ECG editing protocol for atrial fibrillation end-systolic reconstruction included deletion of short R-R intervals and repositioning of R points as needed. The threshold interval between temporal windows to avoid data gaps was approximately 30 beats/min.
For the sinus rhythm patients, image reconstruction was performed at the middiastolic phase because in all of those patients the optimal reconstruction cardiac phase exhibiting minimal cardiac motion occurred during middiastole. The average starting point of the reconstruction interval was 691 milliseconds (range, 520–830 milliseconds) after the R wave. ECG editing was not performed in reconstruction of images of sinus rhythm patients.
Image Analysis
Two independent observers using an external workstation (Aquarius
Workstation, TeraRecon) evaluated the images with a standardized window level
of 100 HU and window width of 700 HU. The observers were blinded to clinical
and image reconstruction data, including ECG editing. The coronary arteries
were classified into 15 segments according to the scheme proposed by the
American Heart Association
[20]. All segments with a
diameter of at least 1.5 mm at the origin were included. Diameter was measured
with an electronic caliper tool. The right coronary artery was defined as
including segments 1–4; the left main artery and left anterior
descending artery, segments 5–10; and the left circumflex artery,
segments 11–15. The left circumflex artery included the intermediate
artery, if present.
The image quality for each coronary segment was evaluated on the curved multiplanar reconstruction images. Image quality, including the presence of motion artifacts, was graded on a previously described [21] 5-point scale (motion score 4, no motion artifacts; 3, minor motion artifacts; 2, moderate motion artifacts; 1, severe motion artifacts; 0, unevaluable or vessel structures not differentiable). Coronary artery image discontinuity was defined as a stairstep artifact larger than 1 mm in one direction in any of the directions. In case of interobserver disagreement about motion score and coronary artery image discontinuity, a consensus interpretation was added. Acceptable image quality for routine clinical diagnostic purposes was considered an image of the coronary artery with a motion score of 2 or greater in all coronary segments and two or fewer discontinuities in each coronary artery.
Statistical Analysis
Statistical analysis was performed with commercially available statistical
software (SPSS version 12.0 for Windows, SPSS). Quantitative variables were
expressed as mean ± SD, and categoric variables as frequencies or
percentages. Comparisons of continuous variables in clinical characteristics
and scan conditions were performed with an unpaired Student's t test
or Welch test when applicable. Clinical variables were compared with the
chi-square test or Fisher's exact test. Agreement between investigators in the
grading of motion scores was calculated with the kappa statistic. Results were
interpreted as poor (
0.20), fair ( = 0.21–0.40),
moderate ( = 0.41–0.60), good ( = 0.61–0.80), very
good ( = 0.81–0.90), or excellent ( 
0.91). Wilcoxon's
signed rank test was used to compare the motion scores and number of
discontinuities between atrial fibrillation end-systolic images and atrial
fibrillation middiastolic images. The Mann-Whitney U test was used to
compare motion scores and numbers of discontinuities between atrial
fibrillation middiastolic images and sinus rhythm images. Frequency of
acceptable image quality was compared by use of the chi-square test or
Fisher's exact test. A value of p < 0.05 was considered a
statistically significant difference for all statistical tests.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Patient Characteristics and MDCT Procedure
Patient characteristics and scan conditions are shown in
Table 1. All atrial
fibrillation patients received heart-rate-lowering agents daily or before CT.
No major adverse reactions to heart-rate-lowering agents were recorded. Mean
heart rate during scanning was controlled to less than 70 beats/min in all
atrial fibrillation patients. Atrial fibrillation patients had higher heart
rate variability than sinus rhythm patients under the condition of similar
mean heart rate (Table 1, Figs.
1A and
1B). ECG editing was done on
atrial fibrillation middiastolic reconstruction images of 18 patients and on
atrial fibrillation end-systolic reconstruction images of five patients.
|
|
|
Image Quality in Patients with Atrial Fibrillation
A total of 250 of the expected 285 segments with a diameter of at least 1.5
mm were evaluated in atrial fibrillation patients. Among the 35 missing
segments, five were absent anatomically, 26 were excluded because the vessel
diameter was less than 1.5 mm, and four were excluded owing to heavy
calcification. Motion scores are shown in
Table 2. Atrial fibrillation
middiastolic images had fewer motion artifacts than did atrial fibrillation
end-systolic images. On atrial fibrillation middiastolic images, no motion
artifacts (score 4) were found in 104 of the 250 coronary segments (42%),
minor blurring (score 3) was found in 126 segments (50%), moderate blurring
(score 2) in 19 segments (8%), and severe blurring (score 1) in one segment
(0.4%). No coronary segment visualized on images was rated as unevaluable
(score 0) on atrial fibrillation middiastolic images. Numbers of coronary
artery image discontinuities are shown in
Table 3. There were fewer
discontinuities per segment on the atrial fibrillation middiastolic images
than on the atrial fibrillation end-systolic images. In 57 coronary arteries,
atrial fibrillation middiastolic images showed no discontinuities in 21
arteries (37%), one discontinuity in 22 arteries (39%), two discontinuities in
10 arteries (18%), and three or more discontinuities in four arteries (7%).
Acceptable image quality was achieved for 91% (52 of 57) of the coronary
arteries on atrial fibrillation middiastolic images and 46% (26 of 57) of the
coronary arteries in atrial fibrillation end-systolic images (p <
0.001) (Figs. 2A,
2B,
2C,
2D,
3A,
3B,
3C, and
3D).
|
|
|
|
|
|
|
|
|
|
Comparison of Image Quality Between Patients with Atrial Fibrillation and Patients in Sinus Rhythm
A total of 251 of the expected 285 segments with a diameter of at least 1.5
mm were evaluated in patients in sinus rhythm. Among the 34 missing segments,
four were absent anatomically, 25 were excluded because the vessel diameter
was less than 1.5 mm, and five were excluded because of heavy calcification.
As shown in Table 2, motion
artifacts were not significantly different between atrial fibrillation
middiastolic images and sinus rhythm images. As shown in
Table 3, the numbers of
discontinuities per segment were not significantly different between the two
groups but tended to be more prevalent on atrial fibrillation middiastolic
images than on sinus rhythm images. Acceptable quality of sinus rhythm images
was achieved for 53 of the 57 coronary arteries (93%), a number not
significantly different from that for atrial fibrillation middiastolic images
(p = 0.50).
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Previous reports [11, 12] have shown that absolute end-systolic phase reconstruction improves the quality of images of patients with atrial fibrillation for the relatively constant duration of the systolic phase. In our study, however, end-systolic phase reconstruction deteriorated image quality. There are several possible explanations for our findings. First, we strove to reduce heart rate to a target of 50–65 beats/min by administration of heart-rate-lowering agents (combination of β-blocker and calcium channel blocker). As a result, the mean heart rate during scanning was controlled to less than 70 beats/min in all patients with atrial fibrillation. As heart rate increases, the duration of diastole decreases. The motion-free time in middiastole shortens more than that in end systole at high heart rates (> 65 beats/min) [7, 22]. In patients with high heart rates, the optimal cardiac phase for acquisition of MDCT coronary angiographic images shifts from middiastole to end systole [7–10]. The mean heart rates of study patients were not given in the previous reports [11, 12], but we believe that data collected at fast heart rates may deteriorate the image quality of middiastolic reconstructions.
A second explanation for the difference between our findings and those of others is that we performed ECG editing, which enables reconstruction of images from a small number of relatively slow heart rate data and leads temporal windows in the slow-filling phase. After these preparations and postprocessing, it was necessary to reconstruct images from the relatively short R-R interval data on some patients with atrial fibrillation who had two or more contiguous short R-R intervals during scanning. However, the longer slow-filling time in atrial fibrillation patients without atrial systole than in sinus rhythm patients with equivalent R-R intervals was conspicuous. Moreover, in end-systolic phase reconstruction, insertion of hypothetical R points results in temporal windows in different cardiac phases. Thus data gaps cannot be avoided in a long R-R interval.
A third explanation for our findings is that we used half-scan reconstruction with a temporal resolution of 165 milliseconds, whereas multisegment reconstruction was used in the previous study [11]. In multisegment reconstruction, temporal resolution is improved with the use of scanning data from two or more consecutive heartbeats for reconstruction of a single transverse image. Although it improves temporal resolution, this approach requires that the heart follow identical motion patterns for the consecutive heartbeats that occur during projection sampling for reconstruction of a single section. This exact reproduction of motion patterns does not seem feasible, particularly in patients with irregular heart rates, such as atrial fibrillation, given the variability of cardiac motion patterns under physiologic conditions. In our study, image discontinuities tended to be more prevalent on atrial fibrillation middiastolic images than on sinus rhythm images. Spatial inconsistencies in the data inevitably occur in multisegment reconstruction, and the benefits of improved temporal resolution outweigh the risk of spatial inconsistency [23, 24]. For this reason, we avoided the use of multisegment reconstruction, although a scan acquisition time less than 100 milliseconds is needed to obtain coronary artery images with few motion artifacts at end systole [16]. Therefore, it is difficult to adapt the temporal window in the quiescent period for coronary artery motion in end systole. In our study, these differences explain the compensation for image artifacts in the middiastolic phase reconstruction in patients with atrial fibrillation.
Coronary artery image discontinuity (stairstep artifact) occurs when the coronary artery does not return to the same position for consecutive heartbeats. The relative delay or absolute reverse method leads temporal windows in different cardiac phases in patients with irregular rhythm and results in coronary artery image discontinuity. In patients with atrial fibrillation, the duration of the systolic phase and the rapid filling phase (early diastolic phase) remains relatively constant, whereas the duration of the slow filling phase is variable [25]. Thus the absolute delay method is advantageous in imaging of patients with irregular heart rhythms [7, 10]. According to this physiologic point of view, after aggressive heart rate control and ECG editing, the absolute delay method enables temporal windows in the slow-filling phase and allows the coronary arteries to be in the same position with every heartbeat on the basis of the reconstructed data. Nevertheless, our results showed that atrial fibrillation middiastolic images tended to have more discontinuities than did sinus rhythm images.
High heart rate variability results in variable end-diastolic cardiac volume and the duration of the subsequent systolic phase. ECG editing enables deletion of short R-R intervals but does not adequately compensate for the effect of variable end-diastolic cardiac volume. In our study, patients with atrial fibrillation had longer scanning times because of lower helical pitch than did patients in sinus rhythm. These factors explain, in part, the image discontinuities in our study. However, the control subjects, that is, the patients in sinus rhythm, had good scan conditions, such as a low heart rate and no premature beat. Image discontinuities of atrial fibrillation middiastolic images were not overly severe in most cases, and 93% of atrial fibrillation middiastolic images had two or fewer discontinuities in each coronary artery, which were acceptable for routine clinical diagnostic purposes.
Our study had several possible limitations. First, image quality scoring might have been influenced by subjective bias. The good interobserver agreement, however, argues against such bias. Second, the number of patients was small, and the diagnostic accuracy of MDCT coronary angiography compared with invasive coronary angiography was not evaluated. Furthermore, only a small proportion of the patients with atrial fibrillation underwent subsequent invasive coronary angiography because of normal findings at MDCT coronary angiography, stress myocardial perfusion scintigraphy, or stress perfusion cardiac MRI. Further investigation is necessary to determine diagnostic accuracy. Third, atrial fibrillation limits the use of dose modulation. The algorithm for dose modulation works with prospective triggering based on the R wave, and the location of the low-dose period is variable. The increased radiation dose, however, seems to be acceptable for diagnostic image quality, which would prevent the use of unnecessary radiation. Last, additional time is required to edit the ECG. Improvements in software may allow patients with atrial fibrillation to undergo routine imaging with high image quality.
For patients with atrial fibrillation, ECG-edited absolute middiastolic phase reconstruction images have better image quality than do end-systolic phase images and have diagnostic image quality comparable with that of images of patients in sinus rhythm, suggesting that atrial fibrillation is not an absolute limitation to MDCT coronary angiography. At 64-MDCT coronary angiography of patients with atrial fibrillation, aggressive heart rate control and ECG-edited absolute middiastolic phase reconstruction are recommended to compensate for coronary artery image artifacts resulting from atrial fibrillation. Stairstep artifacts cannot be fully compensated, but the image quality remains acceptable for diagnostic purposes.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  S. S. Pasricha, D. Nandurkar, S. K. Seneviratne, J. D. Cameron, M. Crossett, M. E. Schneider-Kolsky, and J. M. Troupis Image Quality of Coronary 320-MDCT in Patients With Atrial Fibrillation: Initial Experience Am. J. Roentgenol., December 1, 2009; 193(6): 1514 - 1521. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Yang, Z. Zhang, Z. Fan, C. Xu, L. Zhao, W. Yu, and Z. Yan 64-MDCT Coronary Angiography of Patients With Atrial Fibrillation: Influence of Heart Rate on Image Quality and Efficacy in Evaluation of Coronary Artery Disease Am. J. Roentgenol., September 1, 2009; 193(3): 795 - 801. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
