Abstract 12093: Inter-Reader Reliability of Coronary Calcium Estimation From Non-Gated Chest Computed Tomography Scans Using a Semi-Quantitative Scoring Method

Abstract

Introduction: Coronary artery calcium (CAC) is strongly associated with cardiovascular risk. CAC burden can be qualitatively estimated from non-gated chest computed tomography (CT) imaging and may be used to predict future cardiovascular outcomes. We evaluated whether resident physician trainees can reliably estimate CAC burden from existing non-gated chest CTs using a semi-quantitative scoring method. Methods: A group of 4 resident physicians without formal training in CT image interpretation underwent a training session to qualitatively estimate CAC burden from non-gated, contrast and non-contrast CT scans. CAC was estimated for each coronary artery as absent, mild (1-24% vessel length), moderate (25-49% vessel length) and severe (>50% vessel length). All Chest CTs were reviewed by two physicians, each blinded to the interpretations of the other reviewer, to assess (a) the presence or absence of CAC, (b) the presence or absence of moderate-severe CAC, and (c) a semi-quantitative CAC index, at a per-patient level and by vessel. Inter-reader reliability (IRR) was assessed by percent agreement, Cohen's kappa coefficient, and the intraclass correlation coefficient (ICC). Results: Using a semi-quantitative scoring system to estimate an index of overall CAC burden, the ICC was 0.96 (CI 0.94-0.98). The percent agreement for the presence of any CAC was 74% (kappa 0.82) and 89% for moderate-to-severe CAC (kappa 0.85). The IRR stratified for each coronary vessel is shown in Table 1. CAC estimated from both contrast and non-contrast scans was associated with substantial agreement between readers ( Table 1 ). Conclusions: With a single training session, resident physicians can quickly and reliably estimate CAC burden from non-gated chest CT imaging. This may facilitate cardiovascular risk stratification in patients without dedicated cardiac testing. 1b