Effect of electrocardiogram triggering on reproducibility of coronary artery calcium scoring

PURPOSE

To test the hypothesis that computed tomographic (CT) scanning during early rather than middle diastole can significantly reduce the interscan variability of coronary artery calcium (CAC) scores.

MATERIALS AND METHODS

Five hundred thirty-eight patients were initially enrolled; 282 of them were found to have CAC at electron-beam CT and underwent repeat scanning to measure interscan variability with different electrocardiogram (ECG) triggers. Eight patients were excluded owing to respiratory motion; thus, 274 asymptomatic patients were examined. Patients were randomly assigned to different ECG trigger interval groups: 40% (group 1), 50% (group 2), 60% (group 3), and 80% (group 4). Patients in whom more than one-third of sections had greater than 10% ECG trigger variability were classified in the untriggered group (group 5). Interscan variation was compared among all five groups.

RESULTS

Interscan variabilities in CAC groups 1-5 were 11.5%, 15.3%, 20.3%, 17.4%, and 33.1%, respectively, for total calcium area, and 15.0%, 23.3%, 25.6%, 24.0%, and 42.4%, respectively, for total calcium score. CAC score variability was reduced by 34%; and calcium area variability, by 38% in group 1, as compared with the reduced variabilities in group 4 (P <.01 for both measures). Breath holding was adequate in 812 cases, and ECG triggering was correct in 790 of cases.

CONCLUSION

Study results strongly support the use of an ECG trigger of 40% rather than 80% of the R-R interval in electron-beam CT calcium studies.

Improved reproducibility of coronary artery calcium scoring by electron beam tomography with a new electrocardiographic trigger method

RATIONALE AND OBJECTIVES

To improve the interscan reproducibility with electron beam tomography (EBT) by choosing an optimal electrocardiographic (ECG) trigger time.

METHODS

Two hundred fourteen asymptomatic subjects found to have coronary artery calcium (CAC) on EBT were rescanned immediately to measure the interscan variability. Subjects were randomized to one of two different ECG trigger interval groups: the new trigger method (group 1) and the 80% R-R interval trigger method (group 2). The new trigger method was derived from a previous study of motion in the coronary arteries. In group 1 (new trigger method), the ECG trigger was programmed for a certain time (in ms) after the R wave, based on the resting heart rate. The triggers for group 1 were 360 (heart rate <50 beats per minute [bpm]), 340 (51--60 bpm), 314 (61--70 bpm), 300 (71--80 bpm), 290 (81--90 bpm), 280 (91--100 bpm), and 270 ms (>100 bpm). The interscan variation (CAC area and Agatston score) was compared between the two groups.

RESULTS

The interscan variability was significantly reduced using the new trigger method for both CAC area and score compared with the 80% trigger method. The individual lesion variation was also significantly reduced by the new trigger method compared with the 80% trigger method. Area measure had a significantly lower variability compared with the Agatston score.

CONCLUSIONS

These results strongly support the use of this new ECG trigger that relies on a rate-adjusted millisecond delay after the R wave instead of the more commonly used 80% R-R interval in EBT calcium studies.

Coronary artery motion during the cardiac cycle and optimal ECG triggering for coronary artery imaging

RATIONALE AND OBJECTIVES

Our purpose was to investigate the motion characteristics of the coronary arteries and determine optimal electrocardiographic (ECG) trigger time during the cardiac cycle to minimize motion artifacts.

METHODS

Contrast-enhanced multislice movie studies of electron beam tomography (EBT) images were performed on 70 subjects. The EBT datasets, which covered an entire cardiac cycle at 58-ms intervals, were acquired for a short-axis view of the heart with ECG triggering. The pixel values along x and y axes were measured at multiple intervals during the cardiac cycle to establish the motion distance and velocity of three major coronary arteries.

RESULTS

Coronary artery motion varied greatly throughout the cardiac cycle in three major coronary arteries and increased with the patient's baseline heart rate. The greatest and lowest velocities of coronary arterial movement during the cardiac cycle were determined. Based on the lowest velocity of right coronary artery movement during the cardiac cycle, the optimal ECG trigger times were located at approximately 35% (31.4%-37.6%) or 70% (68.7%-71.4%) of the R-R interval in patients whose resting heart rate was < or =70 beats per minute (bpm); at 50% (47.2%-61.1%) of the R-R interval in the 71- to 100-bpm group; and at 55% (52.8%-59.1%) of the R-R interval in the >100-bpm group. Our data demonstrated that the motion characteristics of the left circumflex artery were quite similar to those of the right coronary artery and that the left anterior descending coronary artery had no significant differences in motion throughout the cardiac cycle. A minimum scan speed of 35.4 to 75.5 ms per slice is needed to completely diminish cardiac motion artifacts (in-plane coronary artery motion with <1-mm displacement).

CONCLUSIONS

For coronary artery screening, the optimal ECG trigger time should be determined according to the patient's heart rate, thus greatly reducing motion and motion artifacts during 100-ms acquisitions.

Comparison of spiral and electron beam tomography in the evaluation of coronary calcification in asymptomatic persons

Recently, investigators have begun evaluating the ability of spiral computed tomography (sequence scan mode-SEQ) to measure coronary calcium. Electron Beam Tomography (EBT) and SEQ studies were performed in 10 women and 23 men, with a mean age of 54+/-9 years. The EBT study was performed within 4 weeks (mean 11+/-4 days) of the SEQ with no clinical interval event (MI, revascularization). The mean EBT calcium score (Agatston method) was 52.1+/-58.6, with a range of 0 to 175. The SEQ mean score was 60.1+/-71.1 (range 0 to 253). There were 7 persons with scores of 0 on both scans, and 9 persons with scores of zero on either EBT or spiral CT, but not both. Three persons had negative EBT studies where SEQ detected calcium, and 6 persons had EBT detected calcium and negative SEQ studies. The six patients with negative SEQ and positive EBT studies had a mean score of 47+/-25.7 (range 9 to 99). The remaining sixteen persons had coronary calcium detected on both studies. As compared to EBT, spiral CT had a sensitivity of 74% and a specificity of 70%, for an overall diagnostic accuracy of 73%. The positive and negative predictive values were 85 and 54%, respectively for SEQ in this study. The absolute difference in scores between the two tests was 29.1+/-28.5 (mean+/-S.D.). The inter-test variability, defined as the mean values of the differences between the calcium scores in the two scans on the same subjects divided by the mean of the two scores (Absolute Difference between tests/mean), was 84.5% in this study. In asymptomatic persons, spiral CT (using SEQ) provides a limited sensitivity (74%) and specificity (70%) for coronary calcium when compared to EBT. Caution should be used when evaluating the results of spiral CT coronary calcium especially in patients with relatively low calcium scores (<200).

A simple single slice method for measurement of left and right ventricular enlargement by electron beam tomography

RATIONALE AND OBJECTIVE

We devised to test the feasibility of measuring the left and right ventricular sizes by non-contrast electron beam tomographic images.

METHODS

Ventricular sizes consist of the sum of the intracavitary cavity and myocardial mass for each ventricle. A total of 50-image studies from subjects undergoing contrast-enhanced studies were used to develop the measurement methodology. About 20 contrast studies were used to test the measure. The methodology was then prospectively tested on 75 patients with non-contrast studies to estimate the intra-observer, inter-observer and inter-study reproducibility.

RESULTS

Multiple linear regression analysis was completed and the correct regression formulas to calculate ventricular volumes were acquired by using the area and span from the contrast studies. There was excellent correlation between the estimate of LV (r > 0.97, p < 0.001) and RV (r > 0.93, p < 0.001) sizes between measured and calculated (contrast, single slice) left and right ventricular volumes. The intra-observer, inter-observer and inter-study reproducibility demonstrated excellent results with < 7% difference in absolute values and a high correlation (r > 0.89, p < 0.001).

CONCLUSION

We conclude that the left and right ventricular sizes can be accurately estimated from a single mid-ventricular slice on non-contrast electron beam tomographic images.

Accurate measures of left ventricular ejection fraction using electron beam tomography: a comparison with radionuclide angiography, and cine angiography

BACKGROUND

Quantitative determination of ejection fraction is predicated on precise measurement of end-diastolic and end-systolic volumes of the left ventricle. Contrast enhanced electron beam tomography (EBT), with excellent temporal and spatial resolution, has the potential for highly accurate measures of ejection fraction.

METHODS

EBT protocol used a short axis scan of the left ventricle (8-12 levels, apex to base) during infusion of iodinated contrast. To assess the accuracy of the measured left ventricular ejection fraction (LVEF), we compared EBT with first-pass radionuclide angiography (RNA) and cine angiography (CINE).

RESULTS

A total of 41 patients (26 men and 15 women) underwent all three tests within 1 week. Resting ejection fraction using each modality was assessed in a linear regression model to assess inter-test correlation with the other two modalities. Correlation between CINE and EBT was high (r = 0.90, intercept 4.67, p < 0.001). Similarly, correlation of CINE and RNA (r = 0.87, intercept -5.48, p < 0.001) and between EBT and RNA (r = 0.87, intercept -4.6, p < 0.001) were high. In a subset of those patients with LVEF < or = 40%, correlation was consistently high between EBT and CINE. However, correlations were poor for the comparisons between RNA and CINE (r = 0.40), and between the RNA and EBT (r = 0.47). The mean differences of measured ejection fractions between each of the imaging modality were small. However, there was only modest agreement between each of the comparisons as measured using 95% confidence interval (CI) on Bland-Altman plots.

CONCLUSION

These data indicate that the LVEF results are comparable among EBT, RNA, and CINE and can be used interchangeably to assess ventricular function for LVEF > 40%. For LVEF < or = 40%, we demonstrated some disparate results between cine angiography and RNA and between EBT and RNA, indicating that CINE or EBT may provide more accurate assessment.

Preview method for electron-beam CT scanning of the coronary arteries

RATIONALE AND OBJECTIVES

The purpose of this study was to evaluate a method that uses electron-beam computed tomography to obtain the most appropriate starting level for complete imaging of the coronary tree and to compare it with the existing method. A second aim was to evaluate the spatial location of the coronary arteries relative to different anatomic cardiac and chest landmarks.

MATERIALS AND METHODS

Two hundred forty consecutive patients were randomly assigned to imaging with either a six-level preview (new) method or the traditional preview method. The accuracy of each preview method to encompass the coronary anatomy was compared.

RESULTS

All coronary arteries were included in 3-mm scans obtained starting three levels above the origin of the left main coronary artery. The left anterior descending coronary artery extended 3-9 mm above the left main artery in 33 patients (14%). The coronary arteries were encompassed by scans depicting 72-105 mm; thus, with 3-mm sections, up to 35 scans are necessary to cover the entire coronary tree. No stable relationship between the coronary arteries and the pulmonary artery or carina could be found.

CONCLUSION

The six-level preview method that identifies the left main coronary artery and begins 9 mm above this level is the most accurate method for depicting the coronary anatomy. No anatomic landmarks in the heart or chest can be used reliably to identify the position of the coronary arteries in individual patients.

Rates of progression of coronary calcium by electron beam tomography

In this study, we sought to determine the rate of progression of atherosclerosis using coronary calcium scores derived from electron beam tomography (EBT). We studied a variety of disease states (hypertension, high cholesterol, tobacco use, diabetes mellitus) followed for 1 to 6.5 years. We evaluated 299 asymptomatic persons (227 men and 72 women) who underwent 2 consecutive EBT scans at least 12 months apart. The average change in the calcium score (Agatston method) for the entire group was 33.2 +/- 9.2%/year. The treated group (receiving statins) demonstrated an average increase in calcium scores of 15 +/- 8%/year compared with 39 +/- 12%/year for untreated patients (p <0.001). Among the 60 patients on statin monotherapy, 37% had a decrease in the calcium score from baseline to follow-up scan. The relative increase in calcium scores did not vary significantly by gender or risk factors, with the exception of statin-treated hypercholesterolemic subjects. Scores of zero on the initial scan portend a low likelihood of significant calcific deposits on repeat scanning. Only 2 of 81 participants (2%) with scores of zero at baseline had scores >10 on repeat study. In this study, statin therapy induced a 61% reduction in the rate of coronary calcium progression. This study demonstrates that EBT may be a useful tool in assessing efficacy of different interventions to retard progression of atherosclerosis, noninvasively, over relatively short time periods.

Evaluation of fluid collection in the pericardial sinuses and recesses: noncontrast-enhanced electron beam tomography

RATIONALE AND OBJECTIVES

To evaluate the attenuation, size, and volume of the pericardial sinuses and recesses by using electrocardiographically triggered, noncontrast-enhanced electron beam tomography (EBT) and to consider its relation with sex, age, and heart volume.

METHODS

Findings in 213 consecutive patients without known pericardial disease were studied. The patients underwent EBT scanning of the heart to evaluate coronary artery calcification. Incremental electrocardiographically triggered noncontrast images were obtained with a 100-ms exposure time and a 3-mm slice thickness. The appearance, density, and volume of the pericardial sinuses and recesses were calculated.

RESULTS

Among the 213 patients, 97.2% had at least one of the sinuses or recesses visible on EBT. The sinuses or recesses were seen with the following frequency: transverse sinus (93.9%), oblique sinus (71.8%), and superior aortic recess (51.2%). The mean attenuation and volume were 9.9 +/- 7.3 Hounsfield units (HU), 12.6 +/- 8.1 HU, and 12.6 +/- 8.7 HU, and 1.9 +/- 1.3 mL, 1.3 +/- 1.0 mL, and 0.8 +/- 0.8 mL, respectively. The total volume of the pericardial sinuses (3.3 +/- 2.2 mL) had no significant relation with the total heart volume.

CONCLUSIONS

Pericardial sinuses and recesses were frequently and well depicted on noncontrast EBT images. In patients without obvious pericardial effusion, physiological fluid collections were observed in the transverse and oblique sinuses or other recesses. Location, attenuation, and volume were helpful in the differentiation of normal pericardial sinuses from pericardial effusions and mediastinal lymph nodes.

Coronary artery motion in electron beam tomography

PURPOSE

The purpose of this work was to evaluate coronary artery motion characteristics and determine optimal electron beam tomography (EBT) scan time during the cardiac cycle to image the coronary arteries.

METHOD

This study evaluated the movement of coronary arteries in 20 EBT cine studies, at rest and during stress, obtained for evaluating coronary artery disease. The proximal, middle, and distal segments of each coronary artery were measured at multiple times during the cardiac cycle. The motion distance (mm) and velocity (mm/s) of each segment of the coronary arteries were then measured to establish the motion that occurs in the x and y axes during different times in the cardiac cycle.

RESULTS

Coronary artery velocity ranged from 22.4 to 108.6 mm/s. The least motion (and slowest speed) occurred between 30-50 and 40-60% of the R-R interval at rest and stress, respectively. The right coronary artery moved the greatest in the x and y planes (highest speed and spatial change), followed in decreasing order by the circumflex, left main, and left anterior descending arteries. The phase of the cardiac cycle with the greatest coronary artery motion was between 0 and 20% of the R-R interval.

CONCLUSION

Coronary artery motion varies greatly throughout the cardiac cycle. To minimize cardiac motion during tomographic imaging of the coronary arteries, we recommend 40-50% R-R interval as an electrocardiographic trigger time and avoiding the use of image acquisition times of >100 ms.

Effect of exercise on left and right ventricular ejection fraction and wall motion

OBJECTIVE

We evaluated the diagnostic value of response of left and right ventricular ejection fraction and wall motion to exercise using electron beam computed tomography.

METHODS AND RESULTS

We attempted to determine the value of exercise electron beam computed tomography for detecting coronary artery disease, including evaluation of the right ventricular ejection fraction and wall motion abnormalities. A study of 35 patients undergoing electron beam tomography exercise cine studies and coronary artery angiography for the evaluation of chest pain was performed. Of the 18 patients with significant coronary disease (> or = 50% luminal diameter stenosis in at least one coronary artery), 17 (94%) had failure to increase global left ventricular ejection fraction with exercise. Fourteen of 18 (78%) developed a wall motion abnormality during peak exercise, and eight (44%) developed a regional right ventricular wall motion abnormality during peak exercise. Of the 17 patients without obstructive disease, 14 (82%) had a increase in ejection fraction > or = 5% and none had an abnormal response in left ventricular wall motion during peak exercise (specificity = 100%). The change in right ventricular ejection fraction with exercise was not a significant predictor of obstructive coronary disease in this study (P=NS). Using different criteria during stress to predict coronary disease, the accuracy was 89% (31/35) using an increase of <5% in ejection fraction, 89% (31/35) using the development of a new or worsened wall motion abnormality, and 91% (32/35) using both left ventricular criteria.

CONCLUSION

Our study suggests that exercise electron beam computed tomography appears to be a useful tool for the detection of coronary disease. A increase of <5% in ejection fraction and abnormal left ventricular response to exercise were important predictors, while the exercise induced changes of right ventricular ejection fraction was not a significant predictor of obstructive disease. Both left and right ventricular wall motion abnormalities are useful and important parameters in identifying patients with obstructive disease from those with normal coronary arteries.

Simple single-section method for measurement of left and right atrial volumes with electron-beam CT

RATIONALE AND OBJECTIVES

The authors estimated left and right atrial volumes by using a simple method of measurement with nonenhanced electron-beam computed tomography (CT).

MATERIALS AND METHODS

One hundred sixty-four contrast material-enhanced electron-beam CT studies were divided into two groups. Group 1, which included 104 studies, was used to develop the measurement method (i.e., the formulas) for measuring left and right atrial volumes from a nonenhanced study. Group 2 consisted of 60 studies on which the validity of the method was tested. Measurement of left and right atrial volumes was performed on all section levels by tracing the respective atrial borders for each section, then multiplying the area by section thickness and summing the resultant volumes.

RESULTS

Calculated left and right atrial volumes were derived by using the biggest atrial area and cephalic-caudal span. The span was equal to section thickness times the number of sections in which the atria were present. Linear regression analysis formulas were acquired with the biggest atrial area and cephalic-caudal span. Left and right atrial calculated volumes were obtained with these formulas and demonstrated a significant good relation (r > .95, P < .001) and a difference of less than 11% (P < .05) in absolute values between measured and calculated volumes. Intraobserver, interobserver, and interstudy reproducibility were excellent, with less than 10% difference in absolute values.

CONCLUSION

Left and right atrial volumes can be accurately estimated from a single midventricular section by using nonenhanced electron-beam CT.

Intravenous three-dimensional coronary angiography using contrast enhanced electron beam computed tomography

Coronary angiography remains the diagnostic standard for establishing the presence, site, and severity of coronary artery disease (CAD). Electron beam computed tomography (EBCT), with its 3-dimensional capabilities, is an emerging technology with the potential for obtaining essentially noninvasive coronary arteriograms. The purpose of this study was to (1) test the accuracy of intravenous coronary arteriography using the EBCT to conventional coronary arteriographic images; (2) establish the inter-reader variability of this procedure; (3) determine the limitations due to location within the coronary tree; and (4) identify factors that contributed to improved image quality of the 3-dimensional EBCT angiograms. Fifty-two patients underwent both EBCT angiography and coronary angiography within 2 weeks. The coronary angiogram and the EBCT 3-dimensional images were analyzed by 2 observers blinded to the results of the other techniques. EBCT correctly identified 43 of 55 significantly stenosed arteries (sensitivity 78%), and correctly identified 118 of 130 of the nonobstructed arteries, yielding a specificity of 91% (p <0.001, chi-square analysis). The overall accuracy for EBCT angiography was 87%. Significantly more left main and anterior descending coronary arteries were adequately visualized than the circumflex and right coronary vessels (p = 0.003). Overall, 23 of 208 (11%) major epicardial vessels were noninterpretable by the blinded EBCT readers, primarily due to motion artifacts caused by cardiac and respiratory motion and poor electrocardiographic gating. The inter-reader variability was similar to that of angiography, and its high accuracy makes this a clinically useful test. This study demonstrates, by using intravenous contrast enhancement, that EBCT can clearly depict the coronary artery anatomy and can permit identification of coronary artery stenosis.

Variation of heart rate and electrocardiograph trigger interval during ultrafast computed tomography

OBJECTIVES

Electrocardiographic (ECG) trigger records obtained during cardiac ultrafast computed tomography (UFCT) scanning were analyzed to estimate the variability in heart rate and ECG trigger interval to develop a protocol that would allow the development of better ECG triggering software.

METHODS

One-hundred-eighteen patients underwent cardiac UFCT imaging for diagnostic purposes. All subjects were divided into three groups according to the heart rate and ECG trigger condition. Thirty slices were obtained in the high-resolution volume mode for each patient.

RESULTS

A decrease in heart rate and ECG trigger interval was found during image acquisition of the first four slices in all three groups. The nadir of the heart rate occurred during acquisition of the 4th slice, 5.3, 3.5, and 5.6 beats per minute less than the initial heart rate in groups 1, 2, and 3 respectively, with a 6.9%, 2.8%, and 5.0% shorter ECG trigger interval (p < .001, p = .08, p < .05, respectively). From the 4th to the 30th slices, heart rate and ECG trigger interval progressively increased, but less variability was found in the last 20 slices in all three groups.

CONCLUSIONS

Significant variation in heart rate and ECG trigger interval was seen during 30-level cardiac UFCT imaging, especially during image acquisition of the first four slices (approximately 1-6 seconds after breatholding). This can result in scanning during the suboptimal phase of the cardiac cycle by the current UFCT triggering software. A delay in the initiation of scanning to approximately 6 to 10 seconds after breatholding would result in imaging during a time when the heart rate is relatively stable, and a smaller variability in ECG trigger interval occurs. Recalculation of the required delay before each heart beat may improve the precision of ECG triggering.