Comparison of dual-source and electron-beam CT for the assessment of coronary artery calcium scoring

Abstract

Objective: Cardiac CT allows the detection and quantification of coronary artery calcification (CAC). Electron-beam CT (EBCT) has been widely replaced by high-end CT generations in the assessment of CAC. The aim of this study was to compare the CAC scores derived from an EBCT with those from a dual-source CT (DSCT).

Methods: We retrospectively selected 92 patients (61 males; mean age, 60.7 ± 12 years) from our database, who underwent both EBCT and DSCT. CAC was assessed using the Agatston score by two independent readers (replicates: 1, 2; 3=mean of reading 1 and 2).

Results: EBCT scores were on average slightly higher than DSCT scores (281 ± 569 vs 241 ± 502; p<0.05). In regression analysis R(2)-values vary from 0.956 (1) to 0.966 (3). We calculated a correction factor as EBCT=(DSCT+1)(1.026)-1. When stratifying into CAC categories (0, 1-99, 100-399, 400-999 and ≥1000), 79 (86%) were correctly classified. From those with positive CAC scores, 7 out of 61 cases (11%, κ=0.81) were classified in different categories. Using the corrected DSCT CAC score, linear regression analysis for the comparison to the EBCT results were r=0.971 (p<0.001), with a mean difference of 6.4 ± 147.8. Five subjects (5.4%) were still classified in different categories (κ=0.84).

Conclusion: CAC obtained from DSCT is highly correlated with the EBCT measures. Using the calculated correction factor, agreement only marginally improved the clinical interpretation of results. Overall, for clinical purposes, face value use of DSCT-derived values appears as useful as EBCT for CAC scoring.

Figure 1

Graphs comparing the coronary artery calcification using the Agatston scores of both electron-beam CT (EBCT) and dual-source CT (DSCT) with respect to mild (≥1–99), moderate (≥100–399), severe (≥400–999) and extensive (≥1000) coronary artery calcification (CAC) categories.

Figure 2

Bland–Altman plot comparing mean values of electron-beam CT (EBCT) and dual-source CT (DSCT) scans with the difference between both scans. The 95% limits of agreement are represented by the dashed lines. The standard deviation (SD) is 0.14±0.56 (95% CI −0.97 to 1.25).

Figure 3

Passing–Bablok regression (PBR) for the mean of two replicates. Identity line and upper 95% confidence interval to the regression line coincide within the line width of the plot. EBCT, electron-beam CT; DSCT, dual-source CT.