Quantifying epicardial fat volume on cardiac CT: the necessity of ECG-gating versus contrast enhancement

Objective To investigate the feasibility of measuring Epicardial fat volume (EFV) on non-gated, non-contrast, and non-gated non-contrast computed tomography (CT) images. Materials and methods A total of 79 patients (male/female: 46/33, mean age: 57.91 ± 13.30 years old) who underwent triple-rule-out CT (TRO-CT) examinations were retrospectively enrolled, where electrocardiography (ECG)-gated contrast-enhanced coronary CT angiography (CCTA), non-gated non-contrast routine chest CT (RCCT), non-contrast coronary artery calcium scoring CT (CACS-CT), and non-gated CT pulmonary angiography (CTPA) and aortic CT angiography (ACTA) were simultaneously available. Three experienced radiologists manually measured the EFV on each image set. EFVs obtained on CACS-CT, CTPA, ACTA, or RCCT were compared with those obtained on CCTA, which were taken as the reference. Subgroup comparison was made for patients identified with (n = 36) and without coronary plaques (n = 43). Inter-observer agreements were analyzed using intraclass correlation coefficients (ICCs). Differences in EFVs were assessed using paired Wilcoxon signed-rank tests. For EFVs significantly deviating from the reference, correlation coefficients were calculated for the possibility of correction. Results The ICCs for all measurements were > 0.88. EFVs obtained on CACS-CT showed no significant differences from the reference (p = 0.899), while those on CTPA, ACTA or RCCT were significantly underestimated (all p < 0.05). Similar results of comparison were found for patients without plaques, even though EFVs obtained on CTPA were found comparable to the reference for patients identified with plaques (p = 0.187). Underestimated EFVs were fitted using linear functions with respect to those on CCTA, where excellent fitting results were found (all R2 > 0.97). Conclusion EFVs obtained on non-contrast CT images (e.g., CACS-CT) were sufficiently accurate, while those obtained on non-gated CT images (e.g., CTPA and ACTA) or non-gated non-contrast CT images (e.g., RCCT) were significantly underestimated. Fortunately, all underestimations can be eliminated through simple correction functions.