The reproducibility of coronary calcium scoring on multiple software platforms

Abstract

Introduction Coronary Artery Calcium (CAC) Scoring has been validated as an accurate tool to risk stratify patients without known coronary artery disease. The aim of this analysis is to evaluate the reproducibility of CAC calculated on different commercial softwares. Methods We included 159 patients who underwent CAC scoring with use of 64- slice multidetector computed tomography (CT) with prospective electrocardiographic gating for clinical reasons. The data sets were evaluated on two different commercially available softwares (4DM from INVIA, Ann Arbor, MI (software A) and Smart score from General Electric, Milwaukee, WI (software B)) by two blinded independent readers using the method of Agatston with a threshold of 130 Hounsfield units. Comparative analysis of CAC scores between the different software was performed by using Spearman rank correlation and Bland Altman analysis. Results Each software produced different absolute numeric results for Agatston score. CAC was detected on 107 scan on both softwares. A total of 59 scans (37%) had the same reading of which 50 patients are without detected calcium. In contrast, CAC reading were within 10 units in 86 scans (52%) There was excellent statistical correlation between the two softwares (r = 0.948, p 400), 132 (87%) of the scans were in the same group by both softwares. Conclusion Our analysis shows that there is a close correlation between the different software calculation of CAC, although the different CAC software different absolute CAC scores. The two software concordantly classified 87% of the study population prognostically.