Background: Coronary artery calcium scoring is used as a method for cardiovascular risk stratification and monitoring of coronary heart disease. Automatic software-based calcium mass calculation has been proposed to improve the performance of the procedure.
Purpose: To compare two algorithms of calcium mass measurement, automatic and phantom calibrated, with respect to correlation, measurement error, and accuracy in vitro and in vivo.
Material and Methods: A cardiac phantom with calcium cylinder inserts was scanned with sequential non-overlapping collimation 4×2.5 mm, at 120 kV and 165 mAs. Fifty adults (37 men; mean age 46.2 years) were examined with the same settings using prospective electrocardiographic triggering to detect and quantify coronary artery calcifications. Calculations were performed with two methods: software-based automatic calcium mass measurement (ACM) and phantom-calibrated calcium mass measurement (CCM).
Results: The total phantom calcium masses measured with ACM and CCM were 175.0±13.8 mg and 163.0±4.4 mg, respectively (P<0.0001), and ACM produced a higher mean error (4.5 vs. 3.2; P<0.05). Results of ACM and CCM were strongly correlated to each other (R=0.73–0.96; P<0.0001). Mean image noise in the patient study was 8.72±1.68 HU. Results of patient calcium scoring with ACM and CCM were significantly different (median 70.3 mg and 59.7 mg, respectively; P<0.0001), with a mean systematic error of 17.5% (limit of agreement between 14.6% and 20.4%). The use of ACM resulted in an altered quartile classification for 14% of patients, as compared to CCM; all of these patients were classified into a higher category.
Conclusion: Our data indicate that multidetector-row computed tomography coronary calcium mass determination based on dedicated phantom calibration shows lower measurement error than an automatic software-based calculation method. The tested automatic software does not yet seem to be a reliable option for calcium mass measurement.
