Background: There is great interest in measuring and tracking atherosclerosis using electron beam tomography (EBT). We sought to assess the reproducibility of two new software systems, InSight and AccuImage, and the console workstation of an EBT scanner for measuring coronary calcification.
Methods: Two sets of non-contrast EBT scans were obtained in 85 subjects. The calcium volume (CV) score and the Agatston score (AS) were analyzed and the relative differences were compared on three workstations.
Results: The intra- and inter-observer variabilities by InSight and AccuImage were both significantly better than variabilities on the console workstation. Both intra- and inter-observer differences for the AS were significantly smaller than those for the CV on each workstation. However, inter-scan variability was lower for the volume method (13.3%) as compared to the AS (17%). Scores were divided into tertiles (T), and the relative inter-scan differences for the AS in T-I (scores < 66) were higher than those in others (T-I 21.0%, T-II 11.9%, T-III 6.8%, p < 0.01). However, there were no significant differences for the CV method (T-I 19.9%, T-II 9.4%, and T-III 5.3%). Thus, while intra- and inter-observer differences with the AS method was lower than volume scoring, the CV inter-scan variability was significantly better.
Conclusions: Both workstations using volumetric and Agatston methods have higher reliability than the console workstation. Inter- and intra-observer reproducibility was excellent (> 98%). There is minimal inter-scan variability for subjects with higher scores (> 65) for both scoring methods.