Assessment of left ventricular ejection fraction with late-systolic and mid-diastolic cardiac phases using multi-slice computed tomography.
- 2018
CONCLUSIONS: Fixed-phase MSCT assessment using late-systole and mid-diastole agreed in defining normal and abnormal LVEF in 76% of patients when compared with echocardiography. Quantitation of LVEF by this method yielded significantly lower values of LVEF and showed no correlation. Thus, accurate quantitation of LVEF by MSCT requires the acquisition of end-systolic and end-diastolic phases. Copyright (c) 2018 The College of Radiographers. Published by Elsevier Ltd. All rights reserved. INTRODUCTION: Multi-slice computed tomography (MSCT) is an accurate tool for the assessment of left ventricular ejection fraction (LVEF). However, in order to reduce radiation dose, prospective acquisition protocols are currently used, in which the end-systole and end-diastole are not scanned. Our aim was to study the accuracy of the assessment of LVEF using fixed late-systolic and mid-diastolic cardiac phases compared with echocardiography. METHODS: MSCT-derived LVEF was measured with off-line commercially available software packages, and compared with echocardiography-derived LVEF using the Simpson's method. LVEF was categorized as normal vs. abnormal (50% cut-off) and was also analyzed as a quantitative parameter. Bland-Altman plots and Pearson correlations were used for inter-technique comparisons. RESULTS: 58 patients were included. The sensitivity and specificity of fixed-phase MSCT when compared with echocardiography for detection of LVEF <=50% was 79% (95% CI = 65-89%) and 43% (10-82%). Misclassification was associated with older age (68 +/- 12 vs. 54 +/- 13 years, p < 0.01), faster heart rate (79 +/- 14 vs. 68 +/- 10 bpm, p = 0.01), and LV hypertrophy (86% vs. 52%, p = 0.03). The quantitative comparison revealed no correlation (r = 0.095, p = 0.478) and a significantly different LVEF (median[IQR], 57.0[50.5-63.1]% vs. 61.0[57.3-64.3]%, p = 0.03). The observed bias between the two methods was -3.7% with broad limits of agreement (+/-25.5%).