| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Technical Developments |
1 From the Department of Radiology, Cardiovascular Institute and FuWai Hospital, Peking Union Medical College, and Chinese Academy of Medical Sciences, 167 Bei-Li-Shi St, Beijing 100037, China (B.L., N.Z.); and Department of Medicine, Division of Cardiology, Harbor-UCLA Medical Center and Saint John’s Cardiovascular Research Center, Torrance, Calif (J.C., S.S.M., S.C., M.J.B.). Received June 17, 2003; revision requested August 27; final revision received March 4, 2004; accepted May 12. Address correspondence to B.L. (e-mail: blu@vip.sina.com).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMaterials and Methods ResultsDiscussionREFERENCES |
|---|
50%) were 69% and 82% (group 1) and 76% and 92% (group 2) (P > .05 and P < .001, respectively). Baseline heart rate–adjusted ECG triggering improves image quality at coronary artery CT angiography for detection of coronary artery disease. © RSNA, 2004
Index terms: Computed tomography (CT), angiography, 54.12116 • Computed tomography (CT), electron beam • Computed tomography (CT), image quality • Coronary vessels, stenosis or obstruction, 54.76
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and Methods ResultsDiscussionREFERENCES |
|---|
50% stenosis), with sensitivity of 74%–92%, specificity of 79%–100%, and accuracy of 81%–93%. However, on the basis of the results of these studies, 8%–29% of coronary arteries were nonassessable with current electron-beam CT angiographic techniques (3–10). The nonassessability of coronary artery segments was caused by impaired image quality due to multiple types of image artifacts, which included those that resulted from coronary artery motion. Investigators in other studies showed that optimal (baseline heart rate–adjusted) electrocardiographic (ECG) triggering could reduce coronary artery motion artifact (11–14) and yield further improvements in the reproducibility of coronary artery calcium scoring (15–17). We believe that baseline heart rate–adjusted ECG triggering should be set to occur at the end of systole (T wave on an ECG) and the image quality would be improved, as compared with the time with conventional triggering at 80% of the R-R interval (middiastole) during the cardiac cycle (14). Thus, the purpose of our study was to compare conventional ECG triggering with baseline heart rate–adjusted ECG triggering for coronary artery electron-beam CT angiography.
|
Materials and Methods |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and Methods ResultsDiscussionREFERENCES |
|---|
Fifty-three patients (group 1) underwent scanning between January 16, 1996, and April 15, 1999, with conventional ECG triggering (at 80% of the R-R interval on the ECG) and our electron-beam CT scanner. The other 54 patients (group 2) underwent scanning between May 27, 1999, and August 13, 2001, with baseline heart rate–adjusted ECG triggering and the same CT scanner. The overall time of this study was from January 16, 1996, to August 13, 2001. The power of the study that was determined with sample size analysis between groups 1 and 2 was 0.95 (JMP 4.0; SAS Institute, Cary, NC). During this study, CT detectors and image reconstruction systems of the scanner were not changed. The two groups were comparable with respect to the following: mean age (57 vs 58 years, P = .10), mean number of men (33 vs 37 patients, P = .88), mean weight (76 vs 81 kg, P = .11), mean height (173 vs 173 cm, P = .95), and mean baseline heart rate (72 vs 77 beats per minute, P = .06).
Coronary Artery Electron-Beam CT Angiography
CT angiographic studies were performed with an electron-beam CT scanner (C-150XP; Imatron, South San Francisco, Calif). The procedure of coronary artery electron-beam CT angiography has been previously described (1–10). First, a 35-section nonenhanced scan was obtained craniocaudally (3-mm section thickness without gap) to determine the coronary artery calcium scores. Then, a flow study was performed (8-mm section thickness with 4-mm intersection gap) to obtain the circulation time (the time from the contrast material injection into the vein to the peak visualization of the ascending aorta) (3–7). The circulation time (mean, 19 seconds) in each of 107 patients was determined with the time-attenuation curve at the origin of the aorta for enhancement. For the flow study, 15–20 mL of nonionic contrast material (Iopamidol 370; Bracco Diagnostics, Plainsboro, NJ), injected with a power injector (Medrad 100; Medrad, Indianola, Pa) at a rate of 8 mL/sec, was used.
The final contrast-enhanced coronary artery volume scan was obtained with 3-mm collimation, 3-mm section thickness, and 2-mm table feed increments (without intersection gap) to provide overlap for optimal resolution of the coronary artery tree. All the CT images were obtained with a 100-msec acquisition time per image and were reconstructed with a 512 x 512 matrix and an 18-cm field of view. The same nonionic contrast material was administered through an antecubital vein with an injection rate of 4 mL/sec and a total volume of 120–160 mL.
Image acquisition was performed with triggering at 80% of the R-R interval on the patient’s ECG (group 1) or triggering at an individually selected time that was based on the baseline heart rate (group 2). For group 2, the protocol for ECG triggering was derived from results of our study of coronary artery motion (14). We used the linear regression equation of y = 0.3755 · x to obtain the point of least motion of the coronary artery during the cardiac cycle in patients with different heart rates, where y represents the percentage of the R-R interval and x represents the baseline heart rate prior to scanning (14,16). On the basis of this equation, the optimal ECG triggering time in terms of the percentage of the R-R interval and resting heart rates of patients were as follows: approximately 30%, 50 beats per minute or less; 32%, 51–60 beats per minute; 35%, 61–70 beats per minute; 38%, 71–80 beats per minute; 41%, 81–90 beats per minute; 44%, 91–100 beats per minute; and 45%, more than 100 beats per minute.
Evaluation of Electron-Beam CT Angiographic Images
Electron-beam CT angiographic images were transferred to a personal computer–based workstation (IBM, Triangle Park, NC). Calcium scoring and three-dimensional reconstruction (volume-rendering techniques) of coronary arteries were performed with a commercially available software package (Insight; NeoImagery Technologies, City of Industry, Calif). The Agatston method was used to measure the coronary artery calcium score (18).
Two experienced readers (B.L. and M.J.B., with 6 and 10 years of experience with electron-beam CT, respectively), who were blinded to the conventional coronary artery angiographic results, independently evaluated three-dimensional electron-beam CT angiographic image quality and determined the final diagnosis on the basis of segmental classification in consensus. In our study, segmental classification of the coronary artery tree included the left main coronary artery and the proximal, middle, and distal segments of the left anterior descending artery (LAD), the left circumflex artery (LCx), and the right coronary artery (RCA). To minimize overlapping of adjacent segments and improve correlation between the location of stenosis identified with electron-beam CT angiography and that identified with conventional coronary artery angiography, the segments of each coronary artery on CT images were defined and analyzed on three-dimensional volume-rendered images. For the distal RCA, the posterior descending artery was not analyzed. For a small LCx, the first obtuse marginal branch was evaluated. In patients with left dominant coronary arteries (two patients), the first acute marginal branch of the RCA was evaluated. Significant CAD was defined as 50% or greater luminal diameter stenosis determined visually in that coronary artery segment for both techniques.
A nonassessable coronary artery segment on three-dimensional electron-beam CT angiographic images was defined as no coronary vessel visualized in that segment or an inability to determine a diagnosis because of poor image quality. We defined coronary artery motion as in-plane coronary artery movement of a distance greater than its diameter on sectional electron-beam CT angiographic images.
Comparison of Electron-Beam CT Angiography and Reference Standard
Conventional coronary artery angiography served as the reference standard for determination of CAD. The coronary artery angiograms were analyzed with visual assessment by two experienced readers who were unaware of electron-beam CT angiographic findings. The two readers, who had 11 and 15 years of experience, independently interpreted coronary artery angiograms. To improve the correlation between electron-beam CT angiography and the reference standard in regard to the location of the stenosis, the location of each lesion was recorded segment by segment on a standard diagram (19). Significant CAD was defined as 50% or greater stenosis of the luminal diameter in the coronary artery segment for both techniques.
Data and Statistical Analysis
Calcium scores were compared in patients with obstructive CAD and nonobstructive CAD. Odds ratios, which were determined with the same software package as mentioned previously, were used for prediction of possible CAD. The 
2 analysis of the contingency table was used to compare the differences between patients with CAD and those with nonobstructive CAD (Excel; Microsoft, Seattle, Wash).
Diagnostic values of electron-beam CT angiography were blindly compared with those of the reference standard and expressed as sensitivity, specificity, positive and negative predictive values, and accuracy. These values were compared between group 1 and 2 patients with 2 analysis. Receiver operating characteristic curve analysis, conducted with software mentioned previously, was employed to compare the reliability of electron-beam CT angiography performed with conventional ECG triggering and with baseline heart rate–adjusted ECG triggering. The Cohen 
statistic, determined with software previously mentioned, was used to evaluate interobserver agreement. A difference with a P value of less than .05 was considered significant.
|
Results |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and Methods ResultsDiscussionREFERENCES |
|---|
50% stenosis in at least one coronary artery segment) was confirmed at conventional coronary artery angiography in 65 (61%) of 107 subjects. In group 1, 33 (62%) of 53 subjects had obstructive CAD, whereas in group 2, 32 (59%) of 54 subjects had obstructive CAD (P > .05). At conventional angiography, patients in groups 1 and 2, respectively, had one-, two-, and three-vessel disease as follows: six versus 13, 15 versus 11, and 12 versus eight. A total of 150 (14%) of 1070 (89 in group 1 and 61 in group 2 patients) coronary artery segments with 50% or greater luminal narrowing were identified on angiograms. Coronary artery calcium with a score greater than 0 was highest (92% [60 of 65]) in patients with CAD. Patients with CAD had a statistically significant higher calcium score (533.5 ± 606.5) than did those with nonobstructive CAD (142.2 ± 293.6) (P < .01). Three hundred five of 316 coronary artery segments without calcium did not have obstructive disease, yielding a negative predictive value of 97%.
For all subjects (n = 107), sensitivity, specificity, and accuracy of electron-beam CT angiography for detection of CAD were 73%, 88%, and 86%, respectively; positive and negative predictive values were 49% and 95%, respectively. Sixty of 87 (sensitivity, 69%) versus 39 of 51 (sensitivity, 76%) lesions were detected in patients in groups 1 and 2, respectively (difference with a P value was not significant); 340 of 415 (specificity, 82%) versus 417 of 451 (specificity, 92%) normal coronary artery segments were correctly identified in patients in groups 1 and 2, respectively (P < .001). Overall accuracy of electron-beam CT angiography in group 2 (91%) was higher than that in group 1 (80%) (P < .001). Diagnostic values were all improved in three major coronary arteries by using baseline heart rate–adjusted ECG triggering (Table 1). Receiver operating characteristic curve analysis showed that baseline heart rate–adjusted ECG triggering improved diagnostic reliability of electron-beam CT angiography for detection of obstructive CAD, yielding an area under the curve of 0.88 (0.78 for conventional ECG triggering) (P < .01) (Fig 1). Electron-beam CT angiography for detection of obstructive CAD could be improved by using baseline heart rate–adjusted ECG triggering (odds ratio, 21.3; 95% confidence interval: 5.4, 112.7) rather than conventional ECG triggering (odds ratio, 9.3; 95% confidence interval: 1.9, 68.9). The Cohen statistic for interobserver agreement was 0.50 in group 1 and 0.67 in group 2 for segmental classification (grouping normal arteries with those with 50% luminal stenosis).
|
|
|
|
|
Discussion |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and Methods ResultsDiscussionREFERENCES |
|---|
ECG triggering plays a very important role in the reduction of image misregistration by means of synchronous acquisition of CT images during the cardiac cycle. Conventional ECG triggering as a default method on an electron-beam CT scanner was widely used in previous studies (1–10). More recently, ECG triggering has been useful in reduction of coronary artery motion artifact with optimal setting of the scanning time in the areas with least coronary artery motion during the cardiac cycle (12–14). The time in the areas with least motion was at T wave (end systole) rather than at 80% of the R-R interval (middiastole) on an ECG (12–14), and it changed with baseline heart rate in percentages after the R wave of the cardiac cycle. Optimal ECG triggering times were thus determined according to the baseline heart rate of the patients (14,16,17). For patients with a baseline heart rate of less than 70 beats per minute, the points of least motion during the cardiac cycle were at 60%–70% of the R-R interval for both the RCA and the LCx (11–14). In a study with multi–detector row CT (21), results confirmed again that selection of appropriate triggering delays and decreasing heart rate were effective in the reduction of cardiac motion artifacts. With an acquisition time of 125–250 msec at multi–detector row CT, approximately 29% of coronary artery segments could not be evaluated because of frequent motion artifacts (22). Hence, for retrospectively ECG-gated multi–detector row CT angiography, it was critical to control patient baseline heart rate and to select the optimal image reconstruction window (23).
In two studies (4,10), researchers used conventional ECG triggering and demonstrated that findings in 24% and 25% of all coronary arteries were noninterpretable. In the study of Schmermund et al (7), results indicated that nonassessability of coronary arteries was 28%. With regard to coronary artery segments, findings in 19% of the proximal and middle coronary artery segments were noninterpretable (5). By using baseline heart rate–adjusted ECG triggering, we demonstrated that only 13% (compared with 35% with conventional ECG triggering) of coronary artery segments were nonassessable. Individual vessel analysis demonstrated that the number of nonassessable coronary artery segments was significantly decreased in the LAD, LCx, and RCA, except for the left main coronary artery segment.
In using conventional ECG triggering, nonassessability rates were higher for segments in the LCx and RCA than they were for those in the left main coronary artery and in the LAD (7). This result probably explains the phenomena that more left main coronary arteries and the LAD are adequately visualized and that higher sensitivity and specificity are achieved for identification of significant luminal narrowing than are achieved with the LCx and the RCA (2,6,8). In using baseline heart rate–adjusted ECG triggering, our data showed that nonassessability was more reduced in the RCA (36%) than in the LCx (17%) and the LAD (16%), and this finding probably caused the equivalent diagnostic accuracy in the RCA, compared with that in other coronary arteries depicted on our scans obtained with baseline heart rate–adjusted ECG triggering.
We demonstrated that compared with ECG triggering at 80% of the R-R interval (group 1), baseline heart rate–adjusted ECG triggering (group 2) resulted in a 22% reduction in motion artifacts. In group 2 patients, images of the RCA still had more motion artifacts than did those of the LCx and LAD; however, more of the middle (96%) and distal segments of the RCA (81%) were analyzable than were those of the LAD (93% and 70%, respectively) and of the LCx (73% and 67%, respectively). This was most likely due to the significantly larger luminal diameter in the distal segments of the RCA, compared with the luminal diameter of the LAD and the LCx (P < .01) (24).
The most important development in the use of baseline heart rate–adjusted ECG triggering was that diagnostic values such as sensitivity and specificity for electron-beam CT angiography could be improved for the detection of obstructive coronary artery stenosis. False-positive (75 vs 34 in groups 1 and 2, respectively) and false-negative (27 vs 12 in groups 1 and 2, respectively) diagnoses could be reduced.
Some issues limited our study. First, two groups in this study had different numbers and distributions of coronary artery obstructive lesions. Fortunately, most of these lesions were distributed in proximal (46%) and middle (38%) portions of coronary arteries in both groups. The same readers of electron-beam CT angiograms evaluated all of the patients in both groups with the same criteria to avoid bias and intraobserver and/or interobserver variability.
Second, the prospective approach with ECG triggering used in this study may be challenged in patients with cardiac arrhythmia. However, no patient with cardiac arrhythmia was enrolled in this study. A study with patients who have cardiac arrhythmia needs to be performed in future research. Another critical limitation of prospective ECG triggering is the inability to adapt image reconstruction to the particular motion patterns of different coronary arteries. The use of one fixed time point in the cardiac cycle for image reconstruction does not provide optimal image quality (23). Fortunately, a prospective fixed point of ECG triggering for the RCA provided optimal image quality for that artery and also for that of the LCx (14,16). With prospective ECG triggering, the motion artifact on the images of the LAD was minimal and disregardable (12–14,16,20).
Third, the detection and evaluation of coronary artery luminal stenosis were restricted by coronary artery calcium. As shown with results in this study and with those in other studies, patients with obstructive CAD tend to have a higher amount of calcium. This will affect image quality and cause a false-positive or false-negative diagnosis.
Finally, our data confirmed that the reliability of current electron-beam CT angiographic techniques was still limited. The inability to assess coronary artery anatomy in a large portion of distal coronary artery segments and small branches requires further improvement of the scanner techniques, including spatial resolution and scanning speed (20,25–27). The in-plane spatial resolution of more than 7 line pairs per centimeter, section thickness less than 2 mm, and acquisition time less than 50 msec are needed to accurately visualize coronary arteries of less than 2 mm in diameter.
We conclude that a successful coronary artery electron-beam CT angiographic study is dependent on good image quality and motion-free images. Baseline heart rate–adjusted ECG triggering at an individually selected percentage of the R-R interval should be employed to acquire CT data during the time of least motion of the coronary artery. Furthermore, baseline heart rate–adjusted ECG triggering improves the image quality and accuracy of electron-beam CT angiography as a tool to evaluate the presence and severity of CAD.
|
FOOTNOTES |
|---|
Authors stated no financial relationship to disclose.
Author contributions: Guarantors of integrity of entire study, B.L., M.J.B.; study concepts and design, B.L., N.Z., S.S.M., M.J.B.; literature research, B.L., N.Z., S.S.M.; clinical studies, B.L., M.J.B.; data acquisition, B.L., N.Z., J.C., S.C., M.J.B.; data analysis/interpretation, B.L., N.Z., S.S.M., M.J.B.; statistical analysis, B.L., N.Z., S.S.M.; manuscript preparation and revision/review, all authors; manuscript definition of intellectual content, B.L., N.Z., J.C., S.C., M.J.B.; manuscript editing, B.L., N.Z., S.S.M., M.J.B; manuscript final version approval, B.L., N.Z., M.J.B.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and Methods ResultsDiscussionREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
