Head-to-Head Comparison between MRI and CT in the Evaluation of Volume and Quality of Epicardial Adipose Tissue

Purpose To systematically compare MRI- and CT-based measurements of both the volume and quality of epicardial adipose tissue (EAT). Materials and Methods This prospective study included participants from a subset of the Swedish CArdioPulmonary bioImage Study (SCAPIS) who underwent MRI and CT between November 2017 and July 2018. Dixon fat-water separation MR images were manually segmented, and a threshold-based approach based on a fat signal fraction (FSF) map was used to obtain the EAT volume. Within this EAT volume, the mean FSF was quantified as a measure of fat quality. EAT segmentation from CT images was performed using deep learning techniques, and the EAT volume and its mean attenuation were quantified. Correlation between MRI- and CT-based measurements of EAT volume and quality was assessed using the Pearson correlation coefficient. Results Ninety-two participants (mean age, 59 years ± 5 [SD]; 60 male participants) were included. The intermodality correlation for EAT volume was very strong (<i>r</i> = 0.92, <i>P</i> < .001), with systematically larger values for CT versus MRI (<i>P</i> < .001). There was a strong negative correlation between MRI FSF and CT attenuation (<i>r</i> = -0.72, <i>P</i> < .001). Repeatability analysis for assessment of MRI EAT volume showed good interreader agreement (intraclass correlation coefficient, 0.86) and excellent intrareader agreement (intraclass correlation coefficient, 0.96). Conclusion Correlation between MRI and CT was very strong for EAT volume and strong for EAT quality. <b>Keywords:</b> Cardiac, Adipose Tissue (Obesity Studies), Epicardial Fat, Heart, Tissue Characterization, Comparative Studies, Magnetic Resonance Imaging, Computed Tomography, Fat Signal Fraction, Fat Attenuation Published under a CC BY 4.0 license.