| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Cardiac Imaging |
1 From the Mallinckrodt Institute of Radiology, Washington University School of Medicine, Campus Box 8131, 510 S Kingshighway Blvd, St Louis, MO 63110. Received April 7, 2002; revision requested June 17; final revision received October 3; accepted October 23. Address correspondence to K.T.B. (e-mail: baet@mir.wustl.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: A cardiac CT phantom containing predetermined calcified cylinders was scanned. Calcium volume and mass were measured at various threshold values ranging from 80 to 230 HU. In 32 patients, two consecutive CT scans were obtained, and the coronary artery calcium score, volume, and mass were measured by one observer at 130- and 90-HU thresholds. Correlation analysis and analysis of variance were performed to evaluate the measurement errors in the phantom study and the interscan variability in the clinical study.
RESULTS: In the phantom, mass measurement error varied with threshold and calcium density (P < .01). Mass error was strongly correlated with volume error (r = 0.91, P < .01) but with a much smaller range. In the clinical study, interscan variability of mass measurements was significantly lower than that with other measurement methods for both patients and individual vessels. For the patients, the mean interscan variability of calcium score, volume, and mass at the 130-HU threshold was 20.4%, 13.9%, and 9.3%, respectively. For all methods, interscan variability was not significantly different between the 130- and 90-HU thresholds (P > .05).
CONCLUSION: The mass measurement is more accurate, less variable, and more reproducible in coronary calcium quantification than are measurements with other algorithms. Accurate quantification of calcium in each calcified plaque may require that the threshold be set individually, depending on the calcium density.
© RSNA, 2003
Index terms: Computed tomography (CT), experimental studies, 54.12119 • Coronary vessels, calcification, 54.12119 • Coronary vessels, CT, 54.12119 • Phantoms
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
The Agatston scoring system (5), which was developed for electron-beam computed tomography (CT), has been widely used during the past decade as the standard of referencein the quantitative evaluation of calcified coronary atherosclerotic plaques (6–8). According to its definition, the Agatston score is given by the area of a calcified plaque multiplied by a scaling cofactor that is estimated on the basis of the peak attenuation of a calcified lesion. A fixed minimum attenuation threshold of 130 HU is used to differentiate noncalcified and calcified arterial lesions, with the assumption that most noncalcified areas would be excluded if the attenuation threshold remained at 3 SDs above the mean soft-tissue attenuation of the heart at electron-beam CT. Excellent interobserver and intraobserver agreement for calcium quantification for each scan could be achieved by using this scoring system in electron-beam CT (9), but the interscan variability of the scores is rather high, in part, as a result of the arbitrariness of the scaling cofactor used to calculate the score (10–12). To improve interscan reproducibility, Callister et al (13) propose an alternative calcium scoring system that is based on calcium volume measurement. Most recently, calcium mass measurement is suggested to offer a more accurate quantification of the coronary calcium burden (14).
The introduction of multi–detector row CT in the past few years has evoked increased interest in the use of conventional CT to measure coronary calcium. A high correlation is reported between multi–detector row CT and electron-beam CT in the detection and quantification of coronary calcium (15). However, whether the 130-HU threshold in multi–detector row CT is suitable for accurate quantification of coronary calcium volume and mass and whether the reproducibility of coronary calcium measurement can be improved with the volume- and mass-based algorithms have not been addressed, to our knowledge. The purpose of the present study was to evaluate the effects of different thresholds and quantification methods on the accuracy and reproducibility of coronary calcium measurements with multi–detector row CT.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
CT scans were obtained with a multi–detector row CT scanner (Somatom Plus 4 VolumeZoom; Siemens Medical Systems, Forchheim, Germany) in a separate session before the clinical study. The phantom was positioned on the CT scanner table with the longitudinal axis perpendicular to the imaging plane. Sequential CT scans were obtained through the two large calibration inserts and the nine calcified cylinders in the phantom by using the following parameters: 4 x 2.5-mm collimation, 120 kVp, 100 mA, and 500-msec gantry rotation time.
Clinical Study
The study protocol was approved by the institutional clinical study review board, and all patients gave informed consent. Thirty-two consecutive patients (10 women and 22 men; age range, 36–78 years; mean age, 56 years ± 14) participated in this study. The mean heart rate of the patients during scanning was 72 beats per minute ± 17. The patients enrolled in the study after they saw our advertisement for coronary calcium measurement research or after they were recruited at the radiology department during a clinical chest CT visit. None of the patients had acute chest pain or other cardiac-related symptoms. Patients with coronary stents, a cardiac pacemaker, or prior cardiac surgery were excluded from the study.
All CT scans were obtained with the same multi–detector row CT scanner. Two consecutive CT scans of the heart were obtained in each patient to provide the CT data with which to evaluate the reproducibility of coronary calcium measurements. Patients remained stationary on the CT scanner table between the two acquisitions. Scans were obtained in one breath hold from a level approximately 1 cm below the carina to the inferior margin of the heart. Sequential CT images were acquired with 4 x 2.5-mm collimation, 120 kVp, 100 mA, 500-msec gantry rotation time, and prospective electrocardiographic triggering at 50%–60% of the RR interval.
Calcium Measurements
All images acquired in the phantom and clinical studies were transferred to a workstation (NetraMD; ScImage, Los Altos, Calif). One radiologist (C.H.), who has 4 years of experience in cardiac CT imaging, quantitatively analyzed the simulated calcium plaques in the phantom study and the coronary calcium plaques in the clinical study by means of dedicated software installed in the workstation.
The calcified cylinders in the phantom were identified by setting a wide range of CT thresholds, which varied from 80 to 230 HU with a 10-HU interval. The 230-HU threshold was chosen as the upper limit because it resulted in the smallest error in the total CaHA volume of all calcified cylinders, regardless of individual measurement. The 80-HU threshold was chosen as the lower limit to exclude noncalcified soft tissue or background noise. The CaHA volume of calcified cylinders was determined by means of the isotropic interpolation algorithm (13).
To determine CaHA mass, a calibration factor (c) was calculated according to the equation c = 
CaHA/(CTCaHA - CTwater) by using calibration CaHA inserts with known CaHA density (CaHA) (14). The denominator of this equation represents the difference in mean CT numbers between the calibration CaHA insert (CTCaHA) and the calibration water insert (CTwater). The mean CT number and SD in these two large calibration inserts were measured in a circular region of interest (102 mm2 ± 2). This calibration factor was then used to compute the CaHA mass from the measured volume and mean CT number of CaHA.
In each patient scan, the SD of attenuation in the aortic root lumen was taken as a measure of image noise. Coronary calcium was initially detected at the 130-HU threshold in the left main, left anterior descending, circumflex, and right coronary arteries. The calcium score, volume, and mass were measured in each of these arteries. The calcium score was determined according to the algorithm suggested by Agatston et al (5) (cofactor of 1, 130–199 HU; cofactor of 2, 200–299 HU; cofactor of 3, 300–399 HU; cofactor of 4, 
400 HU), where the area represents two or more contiguous pixels with an attenuation of 130 HU or greater. Calcium volume and mass were calculated by means of the algorithms described in the phantom study. Detection of coronary calcium was repeated at the 90-HU threshold, and the calcium score, volume, and mass were remeasured in each coronary artery. This additional threshold value was selected after results of a preliminary analysis of the phantom data suggested that the calcium mass could be quantified more accurately at lower thresholds. Although the choice of 90 HU over others was rather arbitrary, this particular value was tested in previous studies of conventional CT (17,18) as an alternative threshold value to improve the detection of coronary calcifications.
Statistical Analysis
The differences (errors) between the measured and true values of calcium volume and mass in the phantom were calculated as a percentage of the true value for each CT threshold used. The percentage difference between the initial CT scan (scan 1) and the repeat CT scan (scan 2) in each patient was determined to quantify the variability of coronary artery calcium score, volume, and mass. The percentage difference was calculated as the mean of |value 1 - value 2|/[0.5 x (value 1 + value 2)] x 100, where values 1 and 2 are the measurements from scans 1 and 2, respectively.
The Pearson correlation coefficient was calculated to assess the association of calcium measurements obtained with the algorithms (volume and mass) to the CT threshold. Analysis of variance was performed to evaluate the distribution of errors in the measurement of CaHA volume and mass on the basis of CaHA density. A paired t test was used to evaluate the interscan variability. A P value of less than .05 was considered to indicate a statistically significant difference.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
The relationship between the error in CaHA volume measurement and CT threshold is shown in Figure 1. There was a clear tendency for volume error to decrease as CT threshold increased (r = -0.50, P < .01). CaHA volume was overestimated in the cylinders with the CaHA density of 800 mg/cm3 at all CT thresholds and in those with the CaHA density of 400 mg/cm3 at most CT thresholds. CaHA volume was underestimated in the cylinders with the CaHA density of 200 mg/cm3 at most CT thresholds. Minimum errors in CaHA volume at the corresponding CT thresholds are summarized in Table 1, in which errors at the 130-HU threshold are listed as the standard of reference.
|
|
|
|
|
.002).
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Results in our study demonstrate that measurement variation is substantially smaller with the calcium mass method than that with the calcium volume method. Two parameters that affect the variation in calcium measurements are CT threshold and CaHA density, as shown in the phantom study. These findings suggest that accurate prediction of calcium volume and mass may require different CT thresholds for calcified plaques with different density. In a recent investigation (21), Raggi et al report that a large variability in coronary calcium measurements is associated with different compositions of soft tissue that surrounds the heart. They suggest that this variability would be reduced if an adjustable threshold value was used instead of a fixed value.
Certain sources of error that could not be eliminated from the mass measurement are believed to be small. Mass measurement revealed a discrepancy between the reference and measured values. This discrepancy was more prominent at higher CT thresholds and was more apparent in plaques with lower reference density. This finding indicates that the measurement precision was affected by partial volume averaging, especially in small lesions or those with lower density. With a calibration phantom, the variation in calcium mass measurement can be reduced after the pixels that compose each calcified lesion are corrected by an appropriate calibration factor to compensate for the decreased mean CT numbers that result from the partial volume averaging effect. Furthermore, the accuracy of mass measurement may be improved by adjusting the CT threshold for lesions with different density.
In this study, we also demonstrated that the mass measurement method provides the best reproducibility among three methods in the clinical calcium quantification. The mass method resulted in interscan variability that was significantly lower than that with score or volume at the level of both the total patients and the individual vessel. Interscan variability for the circumflex artery and right coronary artery was higher than that for the left anterior descending coronary artery. This difference may be explained by the difference in the motion artifacts related to the diastolic contraction of the right atrium (22). In addition, small lesions are found more frequently in the circumflex and right coronary arteries, and uncertainty in their measurements may contribute to increased variability.
As found in the phantom study, lower CT thresholds are more suitable than high thresholds for accurate quantification of calcium mass. The 90-HU threshold that we chose for the patient study resulted in a significant increase in calcium score, volume, and mass compared with those obtained with the conventional 130-HU threshold. Adjustment of the threshold on the basis of the calcium density may result in calcium mass or volume values that are closer to the actual value, particularly in the case of a newly formed low-attenuation plaque. Adjustment of the CT threshold in calcium mass measurement is likely more efficient than that in volume measurement, as the variation in mass measurement is much smaller than that in volume measurement, as shown in the phantom study. Mass or volume measurements with initial and repeat scans are affected equally by a change in threshold; therefore, interscan variability is independent of the CT threshold as long as the same threshold is used for all scans.
Image noise had a significant influence on the variability of all quantification algorithms in our study. This observation concurs with that in a previous study (23), in which the authors concluded that the volume method was less affected by image noise than was the Agatston score. In our study, 48% (14 of 29) of patients had Agatston scores below 100, which represented a large portion of patients with a small amount of coronary calcification, and the effect of image noise might have been magnified and may have caused the statistical significance of differences for all quantification algorithms. This finding supports the importance of increasing the x-ray tube current to reduce image noise in the detection of coronary calcifications in obese patients and in those with very small calcified coronary plaques.
With increased use of conventional CT (including multi–detector row CT), volume scanning for the detection of coronary calcium becomes more readily available, and three-dimensional quantification algorithms, such as determination of plaque volume, density, and mass, may replace the conventional two-dimensional Agatston score. On the basis of histopathologic findings, Detrano et al (24) report that the relative calcium mass estimated in an explanted human heart at electron-beam CT reliably reflects the actual mass of precipitated calcium phosphate in diseased coronary arteries. Their findings and ours indicate that calcium mass measurement can improve the accuracy and reproducibility of coronary calcium quantification.
Several limitations of the current study have to be addressed. First, this study was performed with a motion-free phantom. Thus, motion artifacts and electrocardiographic misregistration could not be taken into account. However, it is also important to know the actual accuracy of the quantification methods in ideal circumstances. Second, because of the construction of the phantom, there were no data to serially illustrate the most appropriate CT thresholds for various CaHA densities and dimensions. Third, application of the calibration factor for each patient for both scans was a labor-intensive process. Fourth, the sample size of the patient group was relatively small. Our findings should be validated in large cohorts in a future study.
In conclusion, results of this study demonstrate that in the quantification of coronary calcium, the mass measurement is more accurate, less variable, and more reproducible than measurements with the other algorithms. Accurate quantification of calcium in each calcified plaque may require that the threshold is set on an individual basis, depending on the calcium density of the plaque. Since structural information about the calcified plaques can be revealed by the mass index, there is great potential that calcium mass quantification will be increasingly accepted as a tool with which to assess coronary artery disease.
|
FOOTNOTES |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  S. M. Kim, M. J. Chung, K. S. Lee, Y. H. Choe, C. A. Yi, and B.-K. Choe Coronary Calcium Screening Using Low-Dose Lung Cancer Screening: Effectiveness of MDCT with Retrospective Reconstruction Am. J. Roentgenol., April 1, 2008; 190(4): 917 - 922. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Raman, B. Raman, S. Napel, and G. D. Rubin Semiautomated Quantification of the Mass and Distribution of Vascular Calcification with Multidetector CT: Method and Evaluation Radiology, April 1, 2008; 247(1): 241 - 250. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Rutten, I. Isgum, and M. Prokop Coronary Calcification: Effect of Small Variation of Scan Starting Position on Agatston, Volume, and Mass Scores Radiology, December 1, 2007; 246(1): 90 - 98. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. J. Budoff, S. Achenbach, R. S. Blumenthal, J. J. Carr, J. G. Goldin, P. Greenland, A. D. Guerci, J. A.C. Lima, D. J. Rader, G. D. Rubin, et al. Assessment of Coronary Artery Disease by Cardiac Computed Tomography: A Scientific Statement From the American Heart Association Committee on Cardiovascular Imaging and Intervention, Council on Cardiovascular Radiology and Intervention, and Committee on Cardiac Imaging, Council on Clinical Cardiology Circulation, October 17, 2006; 114(16): 1761 - 1791. [Full Text] [PDF]
|
|
|
|
|
|
D. T. Boll, A. S. Bossert, A. J. Aschoff, M. H. Hoffmann, and R. C. Gilkeson Synergy of MDCT and Cine MRI for the Evaluation of Cardiac Motility Am. J. Roentgenol., June 1, 2006; 186(6_Supplement_2): S379 - S386. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. E. Clouse, J. Chen, H. M. Krumholz, M. E. Clouse, J. Chen, and H. M. Krumholz Noninvasive Screening for Coronary Artery Disease With Computed Tomography Is Useful Circulation, January 3, 2006; 113(1): 125 - 146. [Full Text] [PDF]
|
|
|
|
|
|
F. Moselewski, C. J. O'Donnell, S. Achenbach, M. Ferencik, J. Massaro, A. Nguyen, R. C. Cury, S. Abbara, I.-K. Jang, T. J. Brady, et al. Calcium Concentration of Individual Coronary Calcified Plaques as Measured by Multidetector Row Computed Tomography Circulation, June 21, 2005; 111(24): 3236 - 3241. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. S. Halliburton, A. E. Stillman, M. Lieber, J. M. Kasper, S. A. Kuzmiak, and R. D. White Potential Clinical Impact of Variability in the Measurement of Coronary Artery Calcification with Sequential MDCT Am. J. Roentgenol., February 1, 2005; 184(2): 643 - 648. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Koos, A. H. Mahnken, A. M. Sinha, J. E. Wildberger, R. Hoffmann, and H. P. Kuhl Aortic Valve Calcification as a Marker for Aortic Stenosis Severity: Assessment on 16-MDCT Am. J. Roentgenol., December 1, 2004; 183(6): 1813 - 1818. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. Hong, T. K. Pilgram, F. Zhu, and K. T. Bae Coronary Artery Calcification: Effect of Size of Field of View on Multi-Detector Row CT Measurements Radiology, October 1, 2004; 233(1): 281 - 285. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. E. Hokanson, T. MacKenzie, G. Kinney, J. K. Snell-Bergeon, D. Dabelea, J. Ehrlich, R. H. Eckel, and M. Rewers Evaluating Changes in Coronary Artery Calcium: An Analytic Method That Accounts for Interscan Variability Am. J. Roentgenol., May 1, 2004; 182(5): 1327 - 1332. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. A. Rumberger and L. Kaufman A Rosetta Stone for Coronary Calcium Risk Stratification: Agatston, Volume, and Mass Scores in 11,490 Individuals Am. J. Roentgenol., September 1, 2003; 181(3): 743 - 748. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
), volume (
), and mass (
) measurements. Percentage difference tends to increase with image noise (r > 0.48 and P < .01 for all measurements).
= individual value,