European Radiology

, Volume 17, Issue 2, pp 321–329

The accuracy of 1- and 3-mm slices in coronary calcium scoring using multi-slice CT in vitro and in vivo

Authors

    • Department of Diagnostic RadiologyUniversity Hospital (RWTH) Aachen
  • Ernst Klotz
    • Siemens Medical Solutions, Computed Tomography
  • Joachim E. Wildberger
    • Department of Diagnostic RadiologyUniversity Hospital (RWTH) Aachen
  • Ralf Koos
    • Department of CardiologyUniversity Hospital (RWTH)
  • Marco Das
    • Department of Diagnostic RadiologyUniversity Hospital (RWTH) Aachen
  • Matthias Niethammer
    • Siemens Medical Solutions, Computed Tomography
  • Christian Hohl
    • Department of Diagnostic RadiologyUniversity Hospital (RWTH) Aachen
  • Dagmar Honnef
    • Department of Diagnostic RadiologyUniversity Hospital (RWTH) Aachen
  • Christoph Thomas
    • Department of Diagnostic RadiologyUniversity Hospital (RWTH) Aachen
  • Rolf W. Günther
    • Department of Diagnostic RadiologyUniversity Hospital (RWTH) Aachen
  • Andreas H. Mahnken
    • Applied Medical Engineering, Helmholtz InstituteRWTH-Aachen University
Cardiac

DOI: 10.1007/s00330-006-0332-0

Cite this article as:
Mühlenbruch, G., Klotz, E., Wildberger, J.E. et al. Eur Radiol (2007) 17: 321. doi:10.1007/s00330-006-0332-0

Abstract

The accuracy of coronary calcium scoring using 16-row MSCT comparing 1- and 3-mm slices was assessed. A thorax phantom with calcium cylinder inserts was scanned applying a non-enhanced retrospectively ECG-gated examination protocol: collimation 12×0.75 mm; 120 kV; 133 mAseff. Thirty-eight patients were examined using the same scan protocol. Image reconstruction was performed with an effective slice thickness of 3 and 1 mm. The volume score, calcium mass and Agatston score were determined. Image noise was measured in both studies. The volume score and calcium mass varied less than the Agatston score. The overall measured calcium mass compared to the actual calcium mass revealed a relative difference of +2.0% for 1-mm slices and −1.2% for 3-mm slices. Due to increased image noise in thinner slices in the patient study (26.1 HU), overall calcium scoring with a scoring threshold of 130 HU was not feasible. Interlesion comparison showed significantly higher scoring results for thinner slices (all P<0.001). A similar accuracy comparing calcium scoring results of 1- and 3-mm slices was shown in the phantom study; therefore, the potentially necessary increase of the patient's dose in order to achieve assessable 1-mm slices with an acceptable image-to-noise-ratio appears not to be justified.

Keywords

Computed tomographyCoronary calcificationCalcium scoringSlice thickness

Introduction

Coronary artery calcifications are known to be a constituent of atherosclerosis [14]. As several studies proved the number of coronary calcification to be correlated with the risk of severe cardiac events, detection and quantification of coronary calcifications can be a useful diagnostic tool in the workup of patients with suspected coronary artery disease [5, 6]. Based on a meta-analysis of different electron beam CT (EBCT) studies, a risk stratification and management scheme for patients undergoing coronary calcium screening based on the Agatston score has been introduced [7]. Besides EBCT, considered the reference standard for the detection and quantification of coronary calcifications [8, 9], multislice spiral CT (MSCT) is now widely accepted for coronary calcium scoring and correlates well with the results of EBCT [1013]. Moreover, retrospectively ECG-gated MSCT showed higher reproducibility than EBCT [14]. Diastolic image reconstruction at 60% of the RR interval has been recommended for retrospectively ECG-gated MSCT [15, 16]. Comparing three different scoring algorithms (volume score, calcium mass and Agatston score), the calcium mass showed the highest reproducibility [17].

Recently, Janssen et al. reported a significant correlation between the coronary calcium score and dobutamine cardiac magnetic resonance imaging [18]. The impact of coronary calcium scoring is presently being evaluated in follow-up trials and as a tool of therapy control. For these purposes, a high level of reproducibility for the measurement of coronary calcification is crucial. Effects of different scanning and reconstruction protocols on calcium scoring results have been reported [19, 20]. The commonly used imaging protocols for the detection of coronary calcifications typically include a slice thickness of 3 mm; this might lower the sensitivity for detecting small low-attenuating calcifications. Some authors suggest that thinner-slice protocols may substantially improve the accuracy of calcium scoring as a result of decreased partial volume effects [2123]. For EBCT an improved accuracy in determining the volume score of 1.5-mm slices compared to 3-mm slices was shown [24]. Moreover, a relevant effect on the coronary calcium scoring results was shown with significantly higher values for thinner slices [24], whereas no significant change in calcium scoring results was found in a different EBCT study comparing 3- and 1.5-mm slices [25]. For MSCT, Hong et al. observed no relevant changes in the calcium mass comparing a 3-mm with a 1.25-mm scan protocol; however, a non-enhanced 3-mm scan protocol was compared to a contrast-enhanced 1.25-mm scan protocol [26]. Comparing slice thicknesses of 3 and 1 mm in a non-enhanced setting in MSCT higher calcium scoring results have been reported for thinner slices [27]. However, a scoring threshold of 350 HU has been used in the latter study, and no comment could be made on the accuracy of calcium scoring in 1-mm slices.

The purpose of this study was to compare the accuracy of coronary calcium scoring in MSCT comparing a slice thickness of 1 and 3 mm in a phantom and a patient study.

Materials and methods

Phantom study

A stationary anthropomorphic cardiac CT phantom (Cardio CT Phantom; QRM, Möhrendorf, Germany) (Fig. 1), which has been used and described in previous studies, was scanned [24, 28]. A cylindrical insert with nine cylinders containing calcium hydroxyapatite with densities of 200 mg/cm3, 400 mg/cm3 and 800 mg/cm3 was placed at the position of the heart. Diameters of the cylinders were 1, 3 and 5 mm with the cylinder height being equal to the diameter. The phantom was positioned in the gantry of a 16-slice CT scanner (SOMATOM Sensation 16; Siemens Medical Solutions, Forchheim, Germany) with the long axis of the cylinders perpendicular to the scanning plane. A standardized examination protocol with a tube voltage of 120 kV, an effective tube current-time product of 133 mAseff, a collimation of 12×0.75 mm, a table feed of 2.8 mm per rotation and a rotation time of 420 ms (pitch: 0.311) was used for all helical scans. A sinus rhythm of 80 bpm was simulated using the Laerdal Cardiac Rhythm Simulator (Stavanger, Norway). The scan was repeated ten times with the phantom being repositioned after each scan.
https://static-content.springer.com/image/art%3A10.1007%2Fs00330-006-0332-0/MediaObjects/330_2006_332_Fig1_HTML.gif
Fig. 1

Cardio CT phantom (a) and detailed view of insert with calcium cylinders (b)

Patient study

Thirty-eight consecutive patients (21 male and 17 female; mean age 62.1±6.4 years) with suspected coronary artery disease who were referred to our department for routine calcium scoring were included. Informed consent of all patients was obtained. Throughout the scan the ECG signal was digitally recorded. The mean heart rate during the scan was 76.2±14.2 bpm; no medication to slow down the heart rate was administered. Image acquisition was performed using the same CT scanner as in the phantom study during a single breath-hold of 19.3-28.4 s (median 21.5 s). The same examination protocol as in the phantom study was used for all patient scans: tube voltage of 120 kV, effective tube current-time product of 133 mAseff, collimation of 12×0.75 mm, a table feed of 2.8 mm per rotation and a rotation time of 420 ms (pitch: 0.311). The dose-length-product (DLP), CTDIvol and effective doses were calculated for all protocols using a commercially available dose calculation software (CT Expo 1.4, G. Stamm, Medizinische Hochschule Hannover, Germany) [29].

Image reconstruction

In both the phantom and patient study, axial images were reconstructed at 60% of the RR-interval using a field of view of 180×180 mm2 and a 512×512 matrix. A dedicated convolution kernel (B35f), specifically designed to enhance the depiction of calcifications, was applied. Two stacks of images were reconstructed from the raw data of each scan: first, a slice thickness of 3 mm with an increment of 2 mm, and second a slice thickness of 1 mm with an increment of 0.6 mm. To investigate the effect of different reconstruction increments on the scoring results exploratively, images of a randomly chosen subgroup of phantom scans (n=5) and patient scans (n=5; 26 lesions) were additionally reconstructed as follows: slice thickness of 3 mm with an increment of 1 mm and second a slice thickness of 1 mm with an increment of 1 mm.

Calcium scoring and image noise

All image sets were transferred to an external workstation (Leonardo, Siemens, Germany) for coronary calcium scoring (Syngo Calcium Scoring CT, Siemens, Germany). A consensus reading of two radiologists, both experienced in cardiac CT imaging, was performed. Data acquired included the volume score, calcium mass and Agatston score. Image noise in the phantom and the patient study was determined. The mean CT values in a large homogeneous insert in the phantom study and a water syringe fixed to the patient’s chest in the patient study as described by Wildberger et al. were measured [27]. The standard deviation of these measurements was used to estimate image noise. Due to increased image noise in the thinner slices [noise level phantom (patient) study: 25.5 HU (26.1 HU) in 1-mm slices and 15.9 HU (16.0 HU) in 3-mm slices] a reliable discrimination between small calcified lesions and image noise in the patient study was very difficult (Fig. 2), and classical calcium scoring from the 1-mm slices was practically not feasible. Therefore, direct interlesion comparison and not comparison of overall calcium scoring results was performed. After a specific lesion was identified and scored on the 3-mm slices, the same lesion was identified and scored on the 1-mm slices. To exclude depiction errors, the screens of two identical workstations were placed side by side, evaluating the 1- and 3-mm slices of the same patient simultaneously.
https://static-content.springer.com/image/art%3A10.1007%2Fs00330-006-0332-0/MediaObjects/330_2006_332_Fig2_HTML.jpg
Fig. 2

A 52-year-old man with coronary artery disease. Applying 1-mm slices for coronary calcium scoring, larger calcified lesions can reliably be depicted (arrows), but due to relatively high image noise discrimination between rather small lesions and image noise is indistinct (arrowheads). Note: All pixels with CT values >130 HU are brightened for better image display

Statistical analysis

Statistical analysis was performed using statistical software (Med Calc Version 7.4.2.0, Medcalc Software, Mariakerke, Belgium). In the phantom study values for the volume score, calcium mass and Agatston score are given as the mean±standard deviation (SD). To assess measurement variation in the phantom study, variation coefficients, defined as the SD divided by the mean, were calculated. The effects of slice thickness were analyzed applying a paired Student’s t-test.

As scoring results in the patient study were not normally distributed, the medians and interquartile ranges of all scores were computed. A paired Wilcoxon test was performed to compare the volume score, calcium mass and Agatston score for 1- and 3-mm slices. Lesions were classified into quartiles based on the calcium mass from the 3-mm slices. The quartile limits for calcium mass quartiles for the 3-mm slices were then used to categorize the calcium masses of the 1-mm slices. The systematic error and the limit of agreement between the measurements were determined according to the method described by Bland and Altman [30]. The level of significance was chosen as 5%.

Results

Phantom study

The 3- and 5-mm cylinders were reliably detected in all scans. The three smaller calcium cylinders with a diameter of 1 mm were not reliably detected. In 3-mm slices the densest 1-mm cylinder (800 mg/cm3) was detected in nine of ten scans; the other two cylinders (400 and 200 mg/cm3) were always missed. In 1-mm slices the least dense cylinder was detected in six of ten scans with the two more dense cylinders being detected in all scans. As the comparable data of these small cylinders was only fragmentary, it was not included in the statistical analysis further down. Explorative image reconstruction (1- and 3-mm slices both reconstructed with an increment of 1 mm) did not show different scoring values compared to the standard protocol. Therefore, all further results refer to a reconstruction increment of 0.6 mm for 1-mm slices and a reconstruction increment of 2 mm for 3-mm slices. Tables 1, 2 and 3 display the mean values±SD for the volume score, calcium mass and Agatston score, the relative accuracy (in percent change comparing the measured value and the true value) and the variation coefficients for all six cylinders.
Table 1

Volume score results of the phantom study

Volume score

 

1-mm slice thickness

3-mm slice thickness

 

Calcium cylinder diameter and density (mg/cm3)

True cylinder volume

Volume score (mean±SD)

Accuracy (%)

Variation coefficient (%)

Volume score (mean±SD)

Accuracy (%)

Variation coefficient (%)

P value*

3 mm

        

 200

21.2

21.9±1.2

3.3

5.6

17.9±1.5

−15.6

8.6

<0.001

 400

21.2

37.5±1.5

76.7

3.9

42.7±1.7

101.1

3.9

<0.001

 800

21.2

52.7±0.9

148.5

1.8

69.0±1.4

225.2

2.1

<0.001

5 mm

        

 200

98.2

109.4±2.3

11.4

2.1

109.5±2.6

11.5

2.4

0.896

 400

98.2

147.8±3.2

50.5

2.1

166.7±2.7

69.8

1.6

<0.001

 800

98.2

188.6±2.5

92.1

1.3

225.4±4.4

129.6

2.0

<0.001

Mean

   

2.8

  

3.4

 

*Paired t-test to compare volume scores of the two protocols

Table 2

Calcium mass results of the phantom study

Calcium mass

 

1-mm slice thickness

3-mm slice thickness

 

Calcium cylinder diameter and density (mg/cm3)

True calcium mass

Calcium mass (mean±SD)

Accuracy (%)

Variation coefficient (%)

Calcium mass (mean±SD)

Accuracy (%)

Variation coefficient (%)

P value*

3 mm

        

 200

4.24

3.21±0.19

−24.2

5.9

2.52±0.29

−40.5

11.5

<0.001

 400

8.48

7.99±0.29

−5.8

3.6

7.75±0.39

−8.7

5.0

<0.001

 800

19.96

17.98±0.39

−9.9

2.2

18.28±0.46

−8.4

2.5

0.011

5 mm

        

 200

19.63

17.99±0.48

−8.4

2.7

16.90±0.76

−13.9

4.5

<0.001

 400

39.27

37.85±0.69

−3.6

1.8

36.09±0.64

−8.1

1.8

<0.001

 800

78.54

86.71±0.70

10.4

0.8

84.74±1.21

7.9

1.4

0.045

Mean

   

2.8

  

4.5

 

*Paired t-test to compare calcium masses of the two protocols

Table 3

Agatston score results of the phantom study

Agatston score

1-mm slice thickness

3-mm slice thickness

 

Calcium cylinder diameter and density (mg/cm3)

Agatston score (mean±SD)

Variation coefficient (%)

Agatston score (mean±SD)

Variation coefficient (%)

P value*

3 mm

     

 200

16.2±2.2

13.4

11.0±0.9

7.8

<0.001

 400

45.1±2.2

5.0

41.9±2.2

5.2

0.275

 800

65.4±1.5

2.2

83.1±3.8

4.5

<0.001

5 mm

     

 200

100.0±4.2

4.2

81.5±13.1

16.1

<0.001

 400

185.0±3.7

2.0

183.7±5.78

3.1

0.275

 800

240.9±2.6

1.1

256.4±10.8

4.2

0.001

Mean variation

 

4.7

 

6.8

 

*Paired t-test to compare Agatston scores of the two protocols

Mean calcium scoring results (A: volume score; B: calcium mass; C: Agatston score) for the calcium cylinders of the phantom study; n=10 scans. SD: standard deviation

Significant differences of the absolute calcium scoring results of the different hydroxyapatite inserts in the phantom comparing 1- and 3-mm slices were seen (Tables 1, 2 and 3). Addressing the volume score, more cylinders showed significantly higher values in 3-mm slices, particularly the more dense cylinders, compared to 1-mm slices. Calcium mass, however, revealed significantly higher scoring results in thinner slices for almost all cylinders. The Agatston score from 1-mm slices showed significantly higher values in less dense cylinders, but on the other hand significantly lower values in more dense cylinders.

Comparing the volume score and calcium mass measured in the phantom study to the true volumes and masses of the cylinders the volume score showed a strong tendency to overestimate the true value (mean accuracy of the absolute values in 1-mm slices: 63.8%, and in 3-mm slices: 92.1%). However, the calcium mass showed a less severe tendency to underestimate the true value (mean accuracy of the absolute values in 1-mm slices: 10.4%, and in 3-mm slices: 14.6%). Comparing the overall calcium mass of all six cylinders measured with the actual calcium mass of the six inserts (168.28 mg), the relative difference was +2.0% for 1-mm slices and −1.2% for 3-mm slices. For both, the volume score and the calcium mass overall accuracy was slightly higher in thinner slices.

The measurement variation ranged from 0.8 to 13.4% for the results of the 1-mm slices and from 1.4 to 16.1% for the results of the 3-mm slices. For the majority of measurements (16 of 18, 88.9%), the measurement variation was equal or lower in 1-mm slices compared to 3-mm slices. Mean variation coefficients of the volume score and calcium mass were lower than those of the Agatston score (Tables 1, 2 and 3). Maximum measurement variance was observed for the Agatston score of the least dense calcium cylinders. Minimum measurement variance was observed for the calcium mass of the largest and most dense calcium cylinder.

Patient study

With a mean scan range of 142.3±16.5 mm (143.5±17.4 mm) in male (female) patients, the calculated CTDIvol was 14.9 mGy (14.9 mGy), the dose length product was 209 mGy*cm (209 mGy*cm) and the effective radiation dose was calculated as 2.9 mSv (4.3 mSv) for male (female) patients.

Twenty-eight patients (73.7%) presented with coronary calcifications on the standard protocol with a slice thickness of 3 mm. In total 97 lesions were scored using both 1- and 3-mm slices. Like in the phantom study, explorative image reconstruction (1- and 3-mm slices both reconstructed with an increment of 1 mm) in five patients did not show different scoring values compared to the standard protocol. Therefore, all further results refer to a reconstruction increment of 0.6 mm for 1-mm slices and a reconstruction increment of 2 mm for 3-mm slices. The mean number of lesions per patient was 3.5 with a range of 1 to 10. The distribution of calcium scoring results in the patient study was highly skewed. Median values and interquartile ranges of the volume score, calcium mass and Agatston score for 1- and 3-mm slice thickness are given in Table 4. The median differences (interquartile ranges) were 7.4 mm3 (2.9–16.0 mm3) for the volume score, 2.33 mg (1.09–6.43 mg) for the calcium mass and 9.1 (3.2–20.0) for the Agatston score with higher values for thinner slices. A paired Wilcoxon test showed significantly higher scoring results for all three scoring values from the 1-mm slices compared to a slice thickness of 3 mm (all P values<0.001).
Table 4

Scoring results of the patient study

 

1-mm slice thickness median (interquartile range)

3-mm slice thickness median (interquartile range)

Median difference of 1- and 3-mm slice thickness (interquartile range)

P value*

Volume score (mm3)

43.3 (15.2 to 125.0)

32.2 (8.8 to 95.1)

7.4 (16.0 to 2.9)

<0.001

Calcium mass (mg)

8.55 (2.79 to 28.64)

6.52 (1.58 to 20.58)

2.33 (6.43 to 1.09)

<0.001

Agatston score

49.1 (16.9 to 157.9)

36.7 (8.1 to 125.1)

9.1 (20.0 to 3.2)

<0.001

*Paired Wilcoxon test to compare scoring results of 1- and 3-mm slice thickness

Median and interquartile range of scoring results for all lesions in the patient study (n=97)

Table 5 shows the classification of the calcium mass of all lesions for the two protocols. Looking at 1-mm mass values 22.7% (n=22) of all lesions were classified into a different category compared to classification based on 3-mm mass values; all of these lesions were classified in a higher category.
Table 5

Cross-tabulation of classification according to calcium mass categories comparing 1- and 3-mm slices

 

Calcium mass quartile (3-mm slices)

 

Calcium mass category (1-mm slices)

1

2

3

4

Total

1

14

0

0

0

14

2

10

19

0

0

29

3

0

5

17

0

22

4

0

0

7

25

32

Total

24

24

24

25

97

Cut-offs for categories were based on calcium mass quartiles for 3-mm slices

Discussion

The purpose of this study was to compare the accuracy and scoring results of 1- and 3-mm slice thicknesses for coronary calcium scoring using MSCT in vivo and in vitro. An improved accuracy of 1.5-mm slices compared to 3-mm slices was shown for the volume score in EBCT [24]. Our phantom study also showed improved overall accuracy of the volume score and calcium mass and reduced measurement variation for all three scoring algorithms (volume score, calcium mass and Agatston score) in thinner slices. In accordance with the literature, variation coefficients of the volume score and calcium mass were lower than those of the Agatston score on retrospectively ECG-gated MSCT [14]. Drawbacks of quantification of calcification according to Agatston et al. include the dependence on the peak voxel attenuation of a calcified lesion, which varies between scans, and an arbitrary scaling factor based on the peak attenuation [9]. Increased image noise in thinner slices as seen in both the phantom and the patient study influences the peak voxel attenuation and therefore also affects the variability of the Agatston score.

Using the same cardiac phantom Vliegenthart et al. described higher volume scores in thicker slices (3 mm compared to 1.5 mm) in more dense calcium cylinder inserts [24]. In our study the same was observed for the volume score (Tables 1, 2 and 3). The main cause for higher scoring results in 3-mm slices focusing on more dense lesions is the partial volume effect. Applying a 3-mm protocol compared to a 1-mm protocol, the size of a voxel is three times as big. Voxels that contain part of a high density calcification are considered as part of a calcified lesion if the average attenuation of the entire voxel is above 130 HU. On the other hand, the partial volume effect, particularly for less attenuating lesions, may also lead to underestimation of a calcified lesion when the average attenuation of a voxel that contains part of a calcification decreases below 130 HU.

Calcium mass revealed slightly higher scoring results for thinner slices (1 mm) in five out of six cylinders. However, absolute differences of the mean values were the smallest of all scoring algorithms; when comparing the overall calcium mass of all six cylinders measured with the actual calcium mass of the six inserts (168.28 mg), the relative difference was +2.0% for 1-mm slices and −1.2% for 3-mm slices, both very accurate compared to, e.g., an interscan variability of 21.6% for the Agatston score and 17.8% for the calcium mass [31]. In accordance to this, Mao et al. reported no significant difference between the true and measured values from 1.5- and 3-mm images for the volume score and calcium mass in EBCT [25].

The very small calcium cylinder inserts in the phantom study with a diameter of 1 mm were not reliably detected even with a slice thickness of 1 mm. This again shows how partial volume alters calcium scoring results, especially when the phantom is being repositioned after each scan. Future studies with the newest generation of CT scanners and a further increase in spatial resolution will have to show whether lesions of this size and density will be reliably detectable.

In the patient study, the interlesion comparison showed significantly higher calcium scoring results for 1-mm slices compared to the standard 3-mm protocol for all three scoring algorithms. For EBCT the same finding was described by Vliegenthart et al. for the volume score [24]. In MSCT also higher scoring results for 1-mm slices compared to 3-mm slices have been reported; however, an arbitrary scoring threshold of 350 HU had been used to overcome the problem of increased image noise in thinner slices; moreover, in this study no comment on the accuracy of calcification measurements could be made [27]. As many calcium deposits in atherosclerotic plaques have a density comparable to or lower than that of the calcium cylinder with the lowest density in the phantom, partial volume effects may lead to underestimation or even complete disregard of small or relatively low-density calcifications using a 3-mm slice thickness. Previous studies showed that the CT density of a vascular plaque or lesion is highly influenced by the surrounding or adjacent medium, e.g., a contrast-enhanced vessel lumen [32, 33].

As expected, increased image noise was seen in thinner slices; to exclude false-positive depiction of image noise as calcification in our study, a direct interlesion comparison and not a comparison of overall calcium scoring results was performed. Therefore, the observed increase of scoring results is caused by an underestimation of the size of lesions in the 3-mm protocol only (Fig. 3), and not by false-positive detection of image noise as calcifications. The effect of increased image noise on the scoring results of a true calcified lesion can be neglected, as increased and decreased attenuation of voxels inside and adjacent to this lesion statistically cancel each other out. However, discrimination between small calcified lesions and virtual small lesions caused by image noise is not feasible. When scanning obese patients, applying low-dose scan protocols, or when using thinner slices, an increase of image noise is immanent. At present, one has to state that applying a standard coronary calcium scoring radiation exposure, calcium scoring from 1-mm slices is practically not feasible. The value of individually, e.g., weight-adapted scan protocols or different calcium scoring thresholds for coronary calcium scoring needs to be specified in future studies [34, 35]. Increased tube current in retrospectively ECG-gated cardiac CT and the use of a prospectively ECG-trigger scan protocol both can provide thin slices with less image noise. Evaluation of the necessary increase of the patient dose in order to achieve a tolerable image-to-noise ratio may be the subject of future studies. The superiority of retrospectively ECG-gated helical scanning compared to ECG-triggered sequential scanning in terms of reproducibility has already been shown [14].
https://static-content.springer.com/image/art%3A10.1007%2Fs00330-006-0332-0/MediaObjects/330_2006_332_Fig3_HTML.jpg
Fig. 3

A 47-year-old woman with coronary artery disease. Images with 1-mm slice thickness (a) and 3-mm slice thickness (b) at the corresponding scan level are shown. Note the depicted calcified lesion in the right coronary artery, which is considerably larger in the 1-mm slices (a) than in the 3-mm slices (b) (calcium mass: 11.79 mg compared to 6.99 mg). Note: All pixels with CT value >130 HU are brightened for better image display

In contrast to EBCT, the necessity for longer breath-holds, even at thinner collimations, is no longer a concern with MSCT. With a median breath-hold of 21.5 s in this study, the occurrence of breathing artifacts is negligible; with 64-slice CT scanners, even shorter acquisition times can be reached. However, thinner slices increase the amount of data transfer and storage. MSCT of the heart is a reliable noninvasive method to detect calcified lesions of the coronary arteries. Coronary calcium scoring may not only provide a tool for clinical risk stratification [5], but also a means to evaluate the progression or regression of atherosclerosis [36, 37]. Accurate measurement of the coronary calcium burden, especially for the latter purpose, is compulsory. Again, this study proves that in MSCT the least measurement variability is observed for the volume score and calcium mass and that measurements of the Agatston score vary significantly more.

In relation to the quartile classification of all lesions in the patient study based on the calcium mass of the 3-mm slices, the categorization of the calcium mass of 1-mm slices led to an aberrant classification of 22.7% of all lesions. When comparing calcium scoring results of different section thicknesses, one needs to be aware of significant changes of the scoring results caused by different slice thicknesses. This effect is in accordance with both EBCT and MSCT studies and again emphasizes the need for a standardized scan protocol [24, 27].

In summary, the results of this study show that MSCT coronary calcium scoring with a slice thickness of 1 mm shows less measurement variation than the standard 3-mm protocol applying a standard scoring threshold of 130 HU. The volume score and calcium mass vary less than the Agatston score, independent of the slice thickness. Interlesion comparison in the patient study revealed significantly higher calcium scoring results for thinner slices due to the decreased partial volume effect. However, applying 1-mm slices in clinical practice, an increase of the patient dose in order to achieve assessable slices with an acceptable image-to-noise ratio would be necessary. As 1-mm slices show a similar accuracy of volume score and calcium mass compared to 3-mm slices, a scan protocol with 3-mm slices appears reasonable for clinical practice.

Copyright information

© Springer-Verlag 2006