Reproducibility of Semi-automated Three-dimensional Volumetric Analysis using Cardiac Computed Tomography in Patients With Left Ventricular Assist Device.

Abstract

Multi-detector gated cardiac computed tomography (CCT) allows three-dimensional (3D) quantification of cardiac chambers and is clinically indicated to assess left ventricular assist device (LVAD) malfunction and complications. Automated volumetric analysis is, however, disrupted by inflow cannula artifact in patients with LVAD. With this study, we evaluated intra-observer variability in semi-automated 3D cardiac volumetric analysis using CCT in patients with LVADs. Ten clinically indicated CCTs were studied retrospectively from 9 patients with LVADs. 3D chamber quantification included left and right ventricles end-systolic and end-diastolic volumes (ESV, EDV); and left and right atrial ESV. Derived measurements included cardiac output (CO), ejection fraction (EF), and stroke volume (SV). Automated volumetric analysis was performed, and manual corrections were added when necessary. Absolute and relative differences, Bland-Altman plots, and interclass correlation coefficients (ICCs) were used to assess intra-observer reproducibility for these measurements. Intra-observer reproducibility was excellent for volumetric (ICC >0.99) and derived data (ICC >0.91). Comparing right vs left heart volumetric assessments, the former had a higher relative difference (atria 2.8% vs 1.6%, ESV 3.0% vs 1.9%, EDV 2.7% vs 1.3%), which also translated to a greater relative difference in right-side derived data (CO 11.1% vs. 8.8%, EF 10.5% vs. 9.9%, SV 10.9% vs. 9.0%). The mean difference in left ventricular ejection fraction was 0.4% (limits of agreement [LOA]: -2 and 3.2) and right ventricular ejection fraction was 1.2% (LOA: -4.7 and 7.1). Our results for semi-automated 3D volumetric analysis showed excellent reproducibility for both volumetric and derived data. Electrocardiography-gated cardiac computed tomography with semi-automated volumetric analysis has excellent reproducibility in patients with left ventricular assist device making it imaging modality of choice for functional assessment in this patient population, where cardiac magnetic resonance imaging is contraindicated and transthoracic echocardiography may be limited by poor acoustic windows.