Radiology
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online before print October 2, 2003, 10.1148/radiol.2292021016
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
2292021016v1
229/2/375    most recent
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Hoffmann, U.
Right arrow Articles by Brady, T. J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Hoffmann, U.
Right arrow Articles by Brady, T. J.
(Radiology 2003;229:375-381.)
© RSNA, 2003

Head and Neck Imaging

Vascular Calcification in ex Vivo Carotid Specimens: Precision and Accuracy of Measurements with Multi–Detector Row CT1

Udo Hoffmann, MD, Dylan C. Kwait, BS, Jason Handwerker, MD, Raymond Chan, PhD, Glenn Lamuraglia, MD and Thomas J. Brady, MD

1 From the Departments of Radiology (U.H., D.C.K., J.H., R.C., T.J.B.) and Vascular Surgery (G.L.), Massachusetts General Hospital and Harvard Medical School, 100 Charles River Plaza, Suite 400, Boston, MA 02114. Received August 21, 2002; revision requested October 24; final revision received March 26, 2003; accepted April 14. Funded in part by the Center for the Integration of Medicine and Innovative Technology (CIMIT), Boston, Mass, and the New York Cardiac Center. Address correspondence to U.H. (e-mail: uhoffman@partners.org).


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
 REFERENCES
 
PURPOSE: To test the accuracy and precision of multi–detector row computed tomography (CT)–derived measurements of vascular calcification in ex vivo human carotid endarterectomy (CEA) specimens.

MATERIALS AND METHODS: Sixteen ex vivo CEA specimens were imaged with multi–detector row CT. Multi–detector row CT–derived calcium scoring algorithms (ie, mineral mass and volume score) were compared with the mass and volume of ashed remnants of the CEA specimens. Bland-Altman analysis was performed to assess the mean (ie, bias) and SD (ie, precision) of differences between multi–detector row CT– and ashing-derived measurements. In addition, conventional Agatston score, volume score, mineral mass, and modified Agatston score were calculated for each specimen by using a number of scanning protocols. Images were obtained at a section thickness of 1.25 mm by using different tube energy settings and tube currents. Specimens were also imaged at different section thicknesses with fixed combinations of tube energy and tube current. To compare measurement variability of scoring methods, coefficients of variation for all protocols were calculated.

RESULTS: Both mean multi–detector row CT–derived mineral mass and mean ashing-derived mineral mass were 0.129 g ± 0.173 (r = 0.99, P < .001). Mean multi–detector row CT– and ashing-derived volumes were 339.94 mm3 ± 395.4 and 39.48 mm3 ± 55.76, respectively (r = 0.95, P < .001). Measurement bias relative to the reference ashing values was high (2,800.0%) for volume score and low (2.58%) for mineral mass. Measurement precision was 0.6% for volume score and greater than 0.0005% for mineral mass. Mean coefficients of variation for all CT protocols were 5.0% ± 4.2 and 4.9% ± 4.2 for mineral mass and modified Agatston score, respectively, and 16.0% ± 9.2 and 14.5% ± 3.9 for conventional Agatston and volume scores, respectively (P < .001).

CONCLUSION: Compared with the conventional volume score, multi–detector row CT–derived mineral mass is a less biased and more precise measurement of the mineral content of nonmoving ex vivo CEA specimens. Mineral mass and modified Agatston score are more reproducible than conventional volume and Agatston scores.

© RSNA, 2003

Index terms: Arteries, calcification, 904.721 • Carotid arteries, CT, 904.12911 • Carotid arteries, stenosis or obstruction, 904.721 • Computed tomography (CT), multi– detector row, 904.12911


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
 REFERENCES
 
Vascular calcification has been suggested as a surrogate marker of atherosclerosis (1–3). Clinically, vascular calcification represents a potential predictor of risk for acute coronary events in selected patient populations, as well as a tool to track the natural history and progression of coronary artery disease (4–8). Although initial research involving the use of electron-beam computed tomography (CT) has been promising, there is continued debate regarding whether calcium scoring should be used in clinical decision making (9–14). A key component of this debate is the substantial measurement variability (up to 30%) of the conventional Agatston score, a semiquantitative measure, and of the volume score of coronary vascular calcium (15,16).

Recent study results suggest that the use of calibration phantoms for calculation of mineral mass can potentially lower measurement variability (17–19). The calibration is based on the direct relation of CT-derived attenuation to the density of the tissue. Phantoms containing known amounts of hydroxyapatite have been shown to be useful for calibrating the attenuation of calcified lesions (20). However, it remains to be demonstrated in a direct comparison whether the mineral mass is more accurate than the volume score for calcium measurement. Another approach to assessing the measurement reproducibility of multi–detector row CT–derived calcium scoring algorithms is the variation of section thickness to simulate partial volume effects and the variation of tube energy and tube current to simulate various levels of CT attenuation and image noise. Although helpful, these variation approaches may not be feasible in human subjects.

The mass of remnants of ashed tissue samples accurately reflects the mineral content of such specimens. This measurement closely correlates with atomic absorption spectrophotometry and is routinely used to measure the mineral content of bones (21–25). Hydroxyapatite has been shown to be the major component (99%) of ashing remnants of atherosclerotic plaque (26).

The purpose of this study was to test the accuracy and precision of multi–detector row CT–derived measurements of vascular calcification in ex vivo human carotid endarterectomy (CEA) specimens.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
 REFERENCES
 
CT Imaging
Carotid artery imaging.—Sixteen CEA specimens were retrieved consecutively from 16 patients (10 men, six women; mean age, 64.3 years ± 7.6 [SD]; age range, 55–76 years) who underwent CEA because of recently symptomatic severe carotid artery stenosis (<90% luminal patency). There was no statistically significant difference between the mean ages of the male and female patients (P = .34, unpaired t test). The samples contained the carotid bifurcation with a proximal segment of the internal carotid artery (n = 3) or a proximal segment of the internal carotid artery only (n = 13). The mean length of the specimens was 2.1 cm ± 0.7. The study was approved by the institutional review board of Massachusetts General Hospital, with patient consent waived.

Imaging was performed with a four-row multi–detector row CT scanner (Lightspeed Plus, GE Medical Systems, Milwaukee, Wis). For CT imaging, the CEA specimens were placed in small (length, 9 cm; diameter, 2 cm) plastic tubes filled with saline solution. The plastic tubes were fixed to the table surface with tape to reduce possible motion artifacts resulting from table movement (Fig 1). The revolution time of the x-ray gantry was 500 msec. A simulated electrocardiographic signal at 70 beats per minute was used to engage the scanner in the clinical cardiac scanning mode. In this mode, only 240° of gantry rotation was used for data acquisition, resulting in a temporal resolution of 330 msec. CT scanning was prospectively initiated at 50% of the RR interval to simulate end-diastolic triggering.



View larger version (46K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1. Transverse CT scan (sequential scanning mode, 1.25-mm section thickness, 140 kVp, 140 mA) demonstrates the experimental setup. CEA specimens are embedded in plastic tubes filled with saline. The bone window shows the calcified plaques (bright structures) in excised vessels.

 
CT imaging of each specimen included the use of various scanning protocols with different section thicknesses (in millimeters), tube energies (in peak kilovolts), and tube currents (in milliamperes). All acquisitions were performed by using a sequential scanning mode. A protocol involving a section thickness of 1.25 mm, a tube energy of 140 kVp, and a tube current of 140 mA was used to compare multi–detector row CT–derived calcium scoring algorithms with the mass and volume of ashed remnants of the CEA specimens. In addition, each specimen was imaged at a section thickness of 1.25 mm with different tube energies—80, 100, 120, and 140 kVp—and different tube currents—10, 50, 100, 140, and 180 mA. All specimens were also imaged at different section thicknesses—0.60, 1.25, 2.50, 3.75, and 5.00 mm—with fixed combinations of tube energy and tube current: 80 kVp and 50 mA, 120 kVp and 50 mA, and 120 kVp and 100 mA. Use of a small field of view, 9.6 cm, and a pixel matrix of 512 x 512 resulted in a pixel area of approximately 0.035 mm2.

Phantom imaging and score correction.—To calibrate CEA specimen measurements, we repeated every CT protocol by using a standard phantom (QCT Phantom; Image Analysis, Lexington, Ky). The phantom consisted of calcified tubes with various concentrations (50, 100, and 200 mg/cm3) of hydroxyapatite (Ca10[OH]2[PO4]6, tissue density = 3.153 g/cm3). The hydroxyapatite concentration and the attenuation (in Hounsfield units) of the calcified tubes, as determined by using standardized region-of-interest measurements (region-of-interest area, 10 mm2) (by D.C.K.), were used to determine a linear regression equation describing the relationship between attenuation (in Hounsfield units) and the corresponding true density (described in following text). These equations were used for each CT protocol to normalize mineral mass calculations derived with ex vivo CEA specimen imaging, as previously described (20).

CT Image Analysis
Two observers (including U.H.) analyzed the CT images independently at a workstation (Advanced Windows; GE Medical Systems) with a semiautomatic software program (Smartscore 3.2.2.; GE Medical Systems). Calcified lesions were identified on the basis of the presence of at least two adjacent pixels with an attenuation threshold higher than 90 HU so the low image noise (SD of the image background and the saline tube, <10) would not interfere with the depiction of any calcification. The use of this threshold in patients has been supported previously (27). Only the depicted areas that showed at least two connected pixels (>0.07 mm2) with an attenuation higher than 90 HU were recorded as lesions. The following scoring methods were used:

1. Agatston score: The Agatston score (28) is calculated as the product of the area of a lesion and the weighted signal intensity score (SIS) (ie, or weighted attenuation score), which is dependent on the maximal Hounsfield unit value (HUmax) within the lesion. The total Agatston score (ASt) represents the sum of all single calcifications in a sample:

where A is the area of the calcification and

2. Volume score: The volume score (VS) is calculated as the product of the number of voxels with an attenuation greater than 90 HU (VXn) and the voxel volume (VXv): VS = VXn · VXv.

3. Modified Agatston score: The modified Agatston score (ASm) is calculated as the product of the mean attenuation for all pixels in the lesion with an attenuation higher than 90 HU (CTmean) and the volume of the calcified lesion (V): ASm = CTmean · V.

The relationship between x-ray energy (ie, tube energy) and attenuation is well known and routinely applied in dual x-ray absorptiometry to measure body composition (29). Therefore, we introduced a conversion of the mean attenuation obtained at 80, 100, and 120 kVp to a 140-kVp setting to allow calculation of modified Agatston scores independent of tube energy. This conversion uses the linear relationship between the mean attenuation of the mineral bone phantom inserts at (a) 80, (b) 100, and (c) 120 kVp: (a) y = 0.6588x - 0.642, (b) y = 0.8008x - 0.6274, and (c) y = 0.9147x - 1.4185. The x value corresponds to the attenuation at 80, 100, or 120 kVp, and y is the converted attenuation, which is then used for further calculation of the modified Agatston score. A sample conversion plot is shown in Figure 2.



View larger version (12K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2. Graph illustrates tube energy setting conversion factors used to calculate modified Agatston scores. The attenuation of phantom structures—that is, hydroxyapatite cylinders—was measured with different tube energies and converted to a 40-kVp standard setting. Graph shows an example of attenuation values (in Hounsfield units) measured at 120-140 kVp. Similar plots were generated for 80- and 100-kVp settings.

 
4. Mineral mass: The mineral mass (MM) was calculated directly from the CTmean and the volume of the calcification: MM {propto} CTmean · V.

The pixel attenuation and the assigned CTmean are linearly related to tissue density. A mathematic model that has been previously described in detail (20) was used to calculate the mineral mass depicted on the CT images. This model involves the use of a standard bone mineral phantom for calibration to establish a direct relationship between CTmean and the corresponding mineral density. The following equation was used to calculate the actual mineral density ({delta}L) from the mean Hounsfield unit value within a calcified lesion (HUmean L) for each scanning protocol:

where {delta}HA is the density of hydroxyapatite and HUmean HA is the mean Hounsfield unit value for hydroxyapatite. The mineral mass of the lesion is then calculated as the product of the calcified lesion density and the volume: MM = {delta}L · V.

Ashing
Each CEA specimen was placed in a flask and heated in a furnace (Moldatherm 55000 Series Hinged Tube Furnace [Koyo Thermo Systems, Nara, Japan] with Eurotherm 818 PID Controller [Eurotherm, Worthing, United Kingdom]) at 700° for 72 hours. The mass of the dry empty flask and the mass of the flask containing the remaining ashes were measured by using a precise balance (Mettler H35AR; Mettler Instruments, Columbus, Ohio). The mineral mass of each specimen was calculated as the difference between these measurements. The mineral volume of the ashing remnants was calculated by using the density of hydroxyapatite (3.153 g/cm3).

Validation of Ashing
We validated our experimental setup by ashing four samples with known amounts of ß-tricalcium phosphate (Ca3O5P2 >= 96%, molecular weight, 310.18; Fluka Chemika, Buchs, Switzerland). The samples contained quantities of mineral similar to those in the CEA specimens to ensure that measurement accuracy was maintained throughout the range of studied material (mean quantity of mineral in the mass, 0.66 g ± 0.72; range, 0.006–1.542 g). There was high agreement between the given and measured masses of ß-tricalcium phosphate (r = 0.99, P < .001), as calculated with the paired t test. The measurement difference was 0.001 g ± 0.0004.

Statistical Analyses
Conventional (mean vs difference) and normalized (mean vs difference/reference value) Bland-Altman analyses were carried out to evaluate the bias and precision of multi–detector row CT–derived mineral mass and volume as compared with those of the ashing-derived mass and volume. We used the mean difference in measurement between multi–detector row CT and ashing to determine bias and the SD of the difference in measurement between these two methods as an estimate of measurement precision. Multi–detector row CT–derived mineral mass, modified Agatston score, conventional Agatston score, and volume score were correlated with the mineral mass and mineral volume determined with ashing by using paired two-tailed t test and linear regression analyses. The coefficient of variation (ie, SD/mean) was calculated for variations in tube energy and section thickness and among all CT protocols. P < .05 was considered statistically significant.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
 REFERENCES
 
Measurement Accuracy and Precision
The multi–detector row CT–derived mineral mass and the mineral mass of ashing remnants were in high agreement: The mean multi–detector row CT–derived mineral mass was 0.129 g ± 0.173 (range, 0.001–0.573 g) versus the mean ashing-derived mineral mass of 0.129 g ± 0.173 (range, 0.005–0.636 g) (r = 0.99, P < .001) (Fig 3a). The calcified plaque volume measured by using multi–detector row CT was significantly larger than the mineral plaque volume calculated from the ashing remnants: The mean multi–detector row CT–derived volume was 339.94 mm3 ± 395.4 (range, 11–1,250 mm3) versus the mean ashing-derived volume of 39.48 mm3 ± 55.8 (range, 1.2–201.0 mm3) (r = 0.95, P < .001) (Fig 3b).



View larger version (22K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3a. Graphs illustrate comparison of multi-detector row CT- and ashing-derived mineral mass and volume values for 16 ex vivo CEA specimens performed with a standard multi-detector row CT protocol (140 kVp, 140 mA, 1.25-mm section thickness). (a) Correlation between multi-detector row CT-derived mineral mass and ashing-derived mineral mass (r = 0.99, P < .001). (b) Correlation between multi-detector row CT-derived calcified plaque volume and ashing-derived mineral volume (r = 0.95, P < .001). The volume of ashing remnants was calculated as the density of the major component of calcified plaque minus the density of hydroxyapatite (3.153 g/cm3).

 


View larger version (26K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3b. Graphs illustrate comparison of multi-detector row CT- and ashing-derived mineral mass and volume values for 16 ex vivo CEA specimens performed with a standard multi-detector row CT protocol (140 kVp, 140 mA, 1.25-mm section thickness). (a) Correlation between multi-detector row CT-derived mineral mass and ashing-derived mineral mass (r = 0.99, P < .001). (b) Correlation between multi-detector row CT-derived calcified plaque volume and ashing-derived mineral volume (r = 0.95, P < .001). The volume of ashing remnants was calculated as the density of the major component of calcified plaque minus the density of hydroxyapatite (3.153 g/cm3).

 
Conventional Bland-Altman plots demonstrate that differences between multi–detector row CT– and ashing-derived mass values were small (y = -0.0043x + 0.0002; mean difference, -0.00036 g) (Fig 4a). In contrast, differences between multi–detector row CT– and ashing-derived measurements of calcified volume were large and had a systematic error (y = 1.5195x + 12.186; mean difference, 300.4 mm3), as shown in Figure 4c. Bland-Altman plots normalized for systematic measurement errors (Figs 4b, 4d) show that the calibrated mineral mass algorithm was less biased (2.58% vs 2,800%) and yielded higher measurement precision (>0.0005% vs 0.6%) than the conventional volume score.



View larger version (21K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 4a. (a) Conventional and (b) normalized Bland-Altman scatterplots for multi-detector row CT-derived mineral mass (MMCT) as compared with the mineral mass of ashing remnants (MMA), the independent reference standard. The difference between the two mass values is calculated as follows: MMCT - MMA. The mean mineral mass is calculated as follows: (MMCT + MMA)/2. The normalized difference between the two mass values is calculated as follows: (MMCT - MMA)/MMA. (c) Conventional and (d) normalized Bland-Altman scatterplots for multi-detector row CT-derived calcified volume (VCT) as compared with the volume of ashing remnants (VA). The difference between the two volume values is calculated as follows: VCT - VA. The mean calcified volume is calculated as follows: (VCT + VA)/2. The normalized difference between the two volume values is calculated as follows: (VCT - VA)/VA. The multi-detector row CT-derived mineral mass is a less biased and more precise measurement of the mineral content of nonmoving ex vivo CEA specimens as compared with the conventional volume score. Volume measurements systematically are overestimations of the mineral content.

 


View larger version (21K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 4b. (a) Conventional and (b) normalized Bland-Altman scatterplots for multi-detector row CT-derived mineral mass (MMCT) as compared with the mineral mass of ashing remnants (MMA), the independent reference standard. The difference between the two mass values is calculated as follows: MMCT - MMA. The mean mineral mass is calculated as follows: (MMCT + MMA)/2. The normalized difference between the two mass values is calculated as follows: (MMCT - MMA)/MMA. (c) Conventional and (d) normalized Bland-Altman scatterplots for multi-detector row CT-derived calcified volume (VCT) as compared with the volume of ashing remnants (VA). The difference between the two volume values is calculated as follows: VCT - VA. The mean calcified volume is calculated as follows: (VCT + VA)/2. The normalized difference between the two volume values is calculated as follows: (VCT - VA)/VA. The multi-detector row CT-derived mineral mass is a less biased and more precise measurement of the mineral content of nonmoving ex vivo CEA specimens as compared with the conventional volume score. Volume measurements systematically are overestimations of the mineral content.

 


View larger version (21K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 4c. (a) Conventional and (b) normalized Bland-Altman scatterplots for multi-detector row CT-derived mineral mass (MMCT) as compared with the mineral mass of ashing remnants (MMA), the independent reference standard. The difference between the two mass values is calculated as follows: MMCT - MMA. The mean mineral mass is calculated as follows: (MMCT + MMA)/2. The normalized difference between the two mass values is calculated as follows: (MMCT - MMA)/MMA. (c) Conventional and (d) normalized Bland-Altman scatterplots for multi-detector row CT-derived calcified volume (VCT) as compared with the volume of ashing remnants (VA). The difference between the two volume values is calculated as follows: VCT - VA. The mean calcified volume is calculated as follows: (VCT + VA)/2. The normalized difference between the two volume values is calculated as follows: (VCT - VA)/VA. The multi-detector row CT-derived mineral mass is a less biased and more precise measurement of the mineral content of nonmoving ex vivo CEA specimens as compared with the conventional volume score. Volume measurements systematically are overestimations of the mineral content.

 


View larger version (23K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 4d. (a) Conventional and (b) normalized Bland-Altman scatterplots for multi-detector row CT-derived mineral mass (MMCT) as compared with the mineral mass of ashing remnants (MMA), the independent reference standard. The difference between the two mass values is calculated as follows: MMCT - MMA. The mean mineral mass is calculated as follows: (MMCT + MMA)/2. The normalized difference between the two mass values is calculated as follows: (MMCT - MMA)/MMA. (c) Conventional and (d) normalized Bland-Altman scatterplots for multi-detector row CT-derived calcified volume (VCT) as compared with the volume of ashing remnants (VA). The difference between the two volume values is calculated as follows: VCT - VA. The mean calcified volume is calculated as follows: (VCT + VA)/2. The normalized difference between the two volume values is calculated as follows: (VCT - VA)/VA. The multi-detector row CT-derived mineral mass is a less biased and more precise measurement of the mineral content of nonmoving ex vivo CEA specimens as compared with the conventional volume score. Volume measurements systematically are overestimations of the mineral content.

 
There was also a good correlation of the conventional Agatston score and the modified Agatston score to ashing-derived mass (r = 0.93 and r = 0.97, respectively; P < .001). The interobserver agreement for all multi–detector row CT–derived scoring algorithms was excellent (r = 0.99, P < .001).

Imaging Variables
Mineral mass– and modified Agatston score–based measurements of the mineral content of CEA specimens, as compared with conventional Agatston score– and volume score–based measurements, were insensitive to variations in section thickness (mean coefficients of variation: 6.8% ± 5.9 and 6.5% ± 5.6 for mineral mass– and modified Agatston score–based measurements, respectively, vs 14.6% ± 14.5 and 12.2% ± 3.7 for conventional Agatston score– and volume score–based measurements, respectively; P < .001) (Fig 5). The sensitivity to variations in section thickness is demonstrated in greater detail in Figure 6. The most dramatic measurement difference observed was the difference in modified Agatston score, conventional Agatston score, and volume score between the 1.25- and 2.50-mm section thickness protocols. There were larger measurement differences in conventional Agatston and volume scores (21.5% and 26.3%, respectively) than in mineral mass and modified Agatston score (6.5% and 7.6%, respectively) between the 1.25- and 5.00-mm-thick sections. Measurement differences in conventional Agatston score and volume score (r = 0.87 and r = 0.89, respectively; P < .005) increased linearly with section thickness.



View larger version (16K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 5. Bar graph illustrates measurement variability of scoring algorithms for quantification of ex vivo vascular calcifications. Mineral mass (MM) and modified Agatston score (ASM) measurements are more reproducible than conventional Agatston score (AS) and volume score (VS) measurements. Mean coefficients of variation (±SD) for 16 CEA specimens are depicted. Black bars: overall measurement variability among all multi-detector row CT protocols, including those involving variations in tube current, tube voltage, and section thickness (described in Materials and Methods). Gray bars: measurement variation with 0.60-, 1.25-, 2.50-, 3.75-, and 5.00-mm section thicknesses; measurements were obtained at 120 kVp and 50 mA. White bars: measurement variation with 80, 100, 120, and 140 kVp; measurements were obtained with a 1.25-mm section thickness and 150 mA.

 


View larger version (15K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 6. Graph illustrates variations in section thickness and resulting changes in calcium scoring measurements. Data calculated at 1.25-mm (black), 2.50-mm (dark gray), 3.75-mm (light gray), and 5.00-mm (white) section thicknesses are compared with data obtained at a 0.6-mm section thickness with 120 kVp and 50 mA. AS = conventional Agatston score, ASM = modified Agatston score, MM = calibrated mineral mass, VS = volume score.

 
A similar observation was made with regard to variations in tube energy setting (Fig 5). Again, conventional measurements were significantly more sensitive to variations in tube energy than were mineral mass– and modified Agatston score–based measurements (mean coefficients of variation: 12.2% ± 10.6 and 9.4% ± 4.3 for conventional Agatston score and volume score, respectively, vs 2.8% ± 2.1 and 2.2% ± 1.7 for modified Agatston score– and mineral mass–based measurements, respectively; P < .001). Detailed analysis of volume scores revealed that calcified volume inversely decreased in a linear fashion with tube energy (r = -0.95, P < .001). Volume decreased by 14.0% (P < .03), 17.3% (P < .01), and 20.2% (P < .006) with 100-, 120-, and 140-kVp settings, respectively, as compared with the volume measured with the 80-kVp acquisition (Fig 7).



View larger version (16K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 7. Graph illustrates variations in tube voltage (80, 100, 120, and 140 kVp) and tube current (50 mA [dashed line] and 140 mA [solid line]) and resulting changes in multi-detector row CT volume score measurements in 16 ex vivo CEA specimens. Measurements were performed at a 1.25-mm section thickness. Calcified plaque volume was calculated as the sum of measurements in the 16 CEA specimens. The calcified plaque volume is inversely related to tube voltage (y = -395.4x = 6,703, r = -0.95, P < .001) but unaffected by changes in tube current (50-140 mA, P = 0.45) owing to the experimental setup (ie, with small embedding volume and low image noise).

 
Overall mineral mass and modified Agatston score were not substantially sensitive to variations in CT scanning parameters, as compared with the conventional Agatston and volume scores (mean coefficients of variation: 5.0% ± 4.2 and 4.9% ± 4.2 for mineral mass and modified Agatston score, respectively, vs 16.0% ± 9.2 and 14.5% ± 3.9 for conventional Agatston and volume scores, respectively; P < .001) (Fig 5).


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
REFERENCES
 
The results of this study demonstrate that multi–detector row CT–derived mineral mass is a precise and nonbiased measurement of the actual mineral content of the vessel wall of immobile explanted CEA specimens. In addition, mineral mass– and modified Agatston score–based measurements are significantly less sensitive to changes in section thickness and tube energy than are conventional Agatston and volume scores.

The concept of phantom-based calibration with use of known amounts of hydroxyapatite for CT-derived measurements of vascular calcification has previously been introduced for electron-beam CT and multi–detector row CT (19,20,30). The critical importance of reduced measurement variability as a key component in evaluating the potential of coronary calcification as a risk marker and predictor of coronary events has been recognized by many observers.

Current study results show that under carefully controlled conditions, the calibrated mineral mass in ex vivo carotid specimens is a more precise and less biased measurement of vascular calcification than the conventional volume score. In fact, the multi–detector row CT–derived calcified plaque volume is systematically an overestimation of the volume of ashing remnants by an order of magnitude. One might expect an overestimation since the wet mineral plaque contains about 15% of nonmineral material—predominantly proteins—which is removed by means of ashing (26). However, the large difference observed in our study can only be explained by the heterogeneity of atherosclerotic plaque (31) and the subsequent partial volume effects. Because of the spatial resolution of 1.25 mm in the z-axis direction, a single pixel most often contains other plaque components, such as fibrous tissue, in addition to calcifications.

The phantom data in this study also confirmed that intrinsic blurring of high-attenuating materials such as calcifications may also contribute to this overestimation; the SD of attenuation in a homogeneous compound increases with the density of the material, as reflected by attenuation coefficients and subsequently Hounsfield unit values. One could argue that the threshold of 90 HU might have been a source of the overestimation of mineral volume; however, the image noise was nonsignificant—the SD of the attenuation within the surrounding saline solution and the image background was less than 10—so it is unlikely that the low threshold was responsible for the volume overestimation.

Although a larger saline volume has the advantage in that it better simulates in vivo conditions, the experimental setup in this study yielded low image noise and thus allowed rigorous analysis of the effects of section thickness and tube voltage. Consequently, interobserver agreement was excellent (r = 0.99); however, this was not the focus of this study and could be better addressed in an in vivo study.

Because nonmoving specimens were studied, partial volume effects and variations in CT attenuation were simulated by varying section thickness, tube energy, and tube current. These factors have a pivotal role in the reproducibility of calcification measurements. We demonstrated, in a direct comparison, that mineral mass and modified Agatston score are less sensitive to variations in section thickness or partial volume effects, tube energy or attenuation, and tube current than are conventional Agatston and volume scores and thus have superior reproducibility.

We included the Agatston score in our analysis for two reasons: First, this score is commonly used for coronary calcium measurement and has been shown to predict the risk of coronary events in selected patients. Second, the Agatston score served as a measurement of comparison with the newly introduced modified Agatston score. We believe this is particularly interesting since the Agatston score is calculated in a semiquantitative way by using a density-weighting factor based on a large scale: 200–299 HU equals a score of 2, and greater than 400 HU equals a score of 4. Intuitively, one might expect such a semiquantitative method to be less sensitive to changes in CT attenuation than, for example, the modified Agatston score method.

The inverse correlation between calcified volume measurements and tube current (r = -0.95) indicates that the introduction of a conversion factor for different tube energies is one reason for the improved reproducibility of measurements based on the modified Agatston score.

Another important source of the decreased accuracy and reproducibility that are closely related to partial volume effects is the size of single calcifications. In the current study, for example, we observed the largest measurement difference in volume and Agatston scores between the 1.25- and 2.50-mm section thickness protocols. Lesions that are smaller than the given section thickness may be missed at CT. The introduction of variable Hounsfield unit thresholds independent of section thickness and image noise, as recently suggested, could potentially address this problem (32).

Although it was not proved in the present study, mineral mass and modified Agatston score may prove to be useful to compare calcium scores between different CT scanners, such as electron-beam CT and multi–detector row CT units. Although the clinical value of the modified Agatston score remains unknown, the lower measurement variability and more quantitative nature of this score, with use of the mean density, possibly could enhance the ability to track calcified plaque development and regression in response to lipid-lowering therapy.

Although an in vivo human study involving the use of CT in patients scheduled to undergo CEA would enable the evaluation of tissue in vivo with histologic correlation and even ashing, it is easier to obtain quantitative measurements if explanted tissue is used. The experimental setup that we used for ex vivo CEA specimens provided ideal imaging conditions (ie, large vessels, no motion, nonsubstantial image noise) that cannot be replicated in vivo. The fact that all specimens were stationary and only 16 were studied is a limitation of the study. Although this study may be a good first step, patient studies are warranted to prove the benefits of mineral mass and modified Agatston score, as compared with those of conventional Agatston and volume scores.

In conclusion, the results of this study demonstrate that multi–detector row CT–derived mineral mass represents a precise, robust, and highly reproducible equivalent of the actual mineral content of vascular calcifications in nonmoving ex vivo CEA specimens. In addition, mineral mass and modified Agatston score are significantly less sensitive to changes in section thickness and tube energy than are conventional scoring methods such as those based on Agatston and volume scores.


   ACKNOWLEDGMENTS
 
The authors thank Henry Tsai, MD, Matthew Philp, BS, Ryan Millea, BS, and Michael Rosol, PhD for their assistance with data acquisition.


   FOOTNOTES
 
Abbreviation: CEA = carotid endarterectomy

Author contributions: Guarantors of integrity of entire study, U.H., T.J.B.; study concepts, U.H., J.H.; study design, G.L., T.J.B., U.H.; literature research, U.H., D.C.K.; clinical studies, U.H., G.L.; experimental studies, U.H., J.H.; data acquisition, D.C.K., U.H.; data analysis/interpretation, D.C.K., J.H.; statistical analysis, U.H., D.C.K.; manuscript preparation, U.H., D.C.K., T.J.B.; manuscript definition of intellectual content, U.H., T.J.B.; manuscript editing, R.C.; manuscript revision/review, R.C., J.H., G.L.; manuscript final version approval, U.H., T.J.B.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
 REFERENCES
 

  1. Sangiorgi G, Rumberger JA, Severson A, et al. Arterial calcification and not lumen stenosis is highly correlated with atherosclerotic plaque burden in humans: a histologic study of 723 coronary artery segments using nondecalcifying methodology. J Am Coll Cardiol 1998; 31:126-133.[Abstract/Free Full Text]
  2. Rumberger JA, Simons DB, Fitzpatrick LA, Sheedy PF, Schwartz RS. Coronary artery calcium area by electron-beam computed tomography and coronary atherosclerotic plaque area: a histopathologic correlative study. Circulation 1995; 92:2157-2162.[Abstract/Free Full Text]
  3. Arad Y, Spadaro LA, Roth M, et al. Correlations between vascular calcification and atherosclerosis: a comparative electron beam CT study of the coronary and carotid arteries. J Comput Assist Tomogr 1998; 22:207-211.[CrossRef][Medline]
  4. Callister TQ, Raggi P, Cooil B, Lippolis NJ, Russo DJ. Effect of HMG-CoA reductase inhibitors on coronary artery disease as assessed by electron-beam computed tomography. N Engl J Med 1998; 339:1972-1978.[Abstract/Free Full Text]
  5. Budoff MJ, Raggi P. Coronary artery disease progression assessed by electron-beam computed tomography. Am J Cardiol 2001; 88:46E-50E.[CrossRef][Medline]
  6. Budoff MJ, Lane KL, Bakhsheshi H, et al. Rates of progression of coronary calcium by electron beam tomography. Am J Cardiol 2000; 86:8-11.[Medline]
  7. Mitchell TL, Pippin JJ, Wei M, et al. Progression of volume of coronary artery calcification In: Advances of electron beam computed tomography, proceedings of an international symposium. Iowa City, Iowa: University of Iowa Press, 1998; 29.
  8. Goodman WG, Goldin J, Kuizon BD, et al. Coronary-artery calcification in young adults with end-stage renal disease who are undergoing dialysis. N Engl J Med 2000; 342:1478-1483.[Abstract/Free Full Text]
  9. Budoff MJ, Diamond GA, Raggi P, et al. Continuous probabilistic prediction of angiographically significant coronary artery disease using electron beam tomography. Circulation 2002; 105:1791-1796.[Abstract/Free Full Text]
  10. American College of Cardiology/American Heart Association expert consensus document on electron-beam computed tomography for the diagnosis and prognosis of coronary artery disease. Circulation 2000; 102:126-140.[Free Full Text]
  11. Georgiou D, Budoff MJ, Kaufer E, Kennedy JM, Lu B, Brundage BH. Screening patients with chest pain in the emergency department using electron beam tomography: a follow-up study. J Am Coll Cardiol 2001; 38:105-110.[Abstract/Free Full Text]
  12. Wayhs R, Zelinger A, Raggi P. High coronary artery calcium scores pose an extremely elevated risk for hard events. J Am Coll Cardiol 2002; 39:225-230.[Abstract/Free Full Text]
  13. Kaufmann RB, Sheedy PF, II, Maher JE, et al. Quantity of coronary artery calcium detected by electron beam computed tomography in asymptomatic subjects and angiographically studied patients. Mayo Clin Proc 1995; 70:223-232.[Medline]
  14. Keelan PC, Bielak LF, Ashai K, et al. Long-term prognostic value of coronary calcification detected by electron-beam computed tomography in patients undergoing coronary angiography. Circulation 2001; 104:412-417.[Abstract/Free Full Text]
  15. Agatston AS, Janowitz WR, Hildner FJ, Zusmer NR, Viamonte M, Detrano R. Quantification of coronary artery calcium using ultrafast computed tomography. J Am Coll Cardiol 1990; 15:827-832.[Abstract]
  16. Achenbach S, Ropers D, Mohlenkamp S, et al. Variability of repeated coronary artery calcium measurements by electron beam tomography. Am J Cardiol 2001; 87:210-213.[CrossRef][Medline]
  17. Hopper KD, Strollo DC, Mauger DT. Comparison of electron-beam and ungated helical CT in detecting coronary arterial calcification by using a working heart phantom and artificial coronary arteries. Radiology 2002; 222:474-482.[Abstract/Free Full Text]
  18. Callister TQ, Cooil B, Raya SP, Lippolis NJ, Russo DJ, Raggi P. Coronary artery disease: improved reproducibility of calcium scoring with an electron-beam CT volumetric method. Radiology 1998; 208:807-814.[Abstract/Free Full Text]
  19. McCollough CH, Kaufmann RB, Cameron BM, Katz DJ, Sheedy PF, II, Peyser PA. Electron-beam CT: use of a calibration phantom to reduce variability in calcium quantitation. Radiology 1995; 196:159-165.[Abstract/Free Full Text]
  20. Gutfinger DE, Leung CY, Hiro T, et al. In vitro atherosclerotic plaque and calcium quantitation by intravascular ultrasound and electron-beam computed tomography. Am Heart J 1996; 131:899-906.[CrossRef][Medline]
  21. Klein RF, Mitchell SR, Phillips TJ, Belknap JK, Orwoll ES. Quantitative trait loci affecting peak bone mineral density in mice. J Bone Miner Res 1998; 13:1648-1656.[CrossRef][Medline]
  22. Robin JC, Sirianni SR, Pragay DA, Ambrus JL. Studies on osteoporosis. V. Comparison of methods of evaluation of osteoporosis and study of chromosome changes induced by neutron activation. J Med 1981; 12:251-256.
  23. Sabin MA, Blake GM, MacLaughlin-Black SM, Fogelman I. The accuracy of volumetric bone density measurements in dual x-ray absorptiometry. Calcif Tissue Int 1995; 56:210-214.[CrossRef][Medline]
  24. Pouilles JM, Collard P, Tremollieres F, et al. Accuracy and precision of in vivo bone mineral measurements in sheep using dual-energy X-ray absorptiometry. Calcif Tissue Int 2000; 66:70-73.[CrossRef][Medline]
  25. Gala Paniagua J, Diaz-Curiel M, de la Piedra Gordo C, Castilla Reparaz C, Torralbo Garcia M. Bone mass assessment in rats by dual energy X-ray absorptiometry. Br J Radiol 1998; 71:754-758.[Abstract]
  26. Schmid K, McSharry WO, Pameijer CH, Binette JP. Chemical and physicochemical studies on the mineral deposits of the human atherosclerotic aorta. Atherosclerosis 1980; 37:199-210.[CrossRef][Medline]
  27. Carr JJ, Crouse JR, III, Goff DC, Jr, D’Agostino RB, Jr, Peterson NP, Burke GL. Evaluation of subsecond gated helical CT for quantification of coronary artery calcium and comparison with electron beam CT. AJR Am J Roentgenol 2000; 174:915-921.[Abstract/Free Full Text]
  28. Agatston AS, Janowitz WR, Hildner FJ, Zusmer NR, Viamonte M, Detrano R. Quantification of coronar artery calcium using ultrafast computed tomography. J Am Coll Cardiol 1990; 15:827-832.
  29. Brunton JA, Bayley HS, Atkinson SA. Validation and application of dual-energy x-ray absorptiometry to measure bone mass and body composition in small infants. Am J Clin Nutr 1993; 58:839-845.[Abstract/Free Full Text]
  30. Hong C, Becker CR, Schoepf UJ, Ohnesorge B, Bruening R, Reiser MF. Coronary artery calcium: absolute quantification in nonenhanced and contrast-enhanced multi-detector row CT studies. Radiology 2002; 223:474-480.[Abstract/Free Full Text]
  31. Scott DS, Arora UK, Farb A, Virmani R, Weissman NJ. Pathologic validation of a new method to quantify coronary calcific deposits in vivo using intravascular ultrasound. Am J Cardiol 2000; 85:37-40.[CrossRef][Medline]
  32. Raggi P, Callister TQ, Cooil B. Calcium scoring of the coronary artery by electron beam CT: how to apply an individual attenuation threshold. AJR Am J Roentgenol 2002; 178:497-502.[Abstract/Free Full Text]



This article has been cited by other articles:


Home page
 RadiologyHome page
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]

Home page
 J. Am. Soc. Nephrol.Home page
D. A. Towler
Calciotropic Hormones and Arterial Physiology: "D"-lightful Insights
J. Am. Soc. Nephrol., February 1, 2007; 18(2): 369 - 373.
[Full Text] [PDF]

Home page
 Am. J. Neuroradiol.Home page
N.F. Fanning, T.D. Walters, A.J. Fox, and S.P. Symons
Association between Calcification of the Cervical Carotid Artery Bifurcation and White Matter Ischemia.
AJNR Am. J. Neuroradiol., February 1, 2006; 27(2): 378 - 383.
[Abstract] [Full Text] [PDF]

Home page
 Arterioscler. Thromb. Vasc. Bio.Home page
R. L. Wolf, S. L. Wehrli, A. M. Popescu, J. H. Woo, H. K. Song, A. C. Wright, E. R. Mohler III, J. D. Harding, E. L. Zager, R. M. Fairman, et al.
Mineral Volume and Morphology in Carotid Plaque Specimens Using High-Resolution MRI and CT
Arterioscler. Thromb. Vasc. Biol., August 1, 2005; 25(8): 1729 - 1735.
[Abstract] [Full Text] [PDF]

Home page
 CirculationHome page
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]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
2292021016v1
229/2/375    most recent
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Hoffmann, U.
Right arrow Articles by Brady, T. J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Hoffmann, U.
Right arrow Articles by Brady, T. J.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
RADIOLOGY RADIOGRAPHICS RSNA JOURNALS ONLINE