Key result
Novel deep learning model improves epicardial fat segmentation accuracy by ~5 percentage points on non-contrast CT.
Why the study?
Epicardial fat segmentation is challenging because fat infiltrates grooves between cardiac chambers and is contiguous with myocardium, requiring both location and voxel intensity information.
Absolute Event Rate: 90.06% vs 84.87%
p-value: p=<0.05
Integrating attention gates and deep supervision into a 3D U-Net model significantly enhances the automated segmentation accuracy of epicardial fat on both non-contrast and contrast-enhanced cardiac CTs.
Advances automated epicardial fat segmentation on CT; extends DL tools but leaves clinical validation open.
Epicardial fat (ECF) is localized fat surrounding the heart muscle or myocardium and enclosed by the thin-layer pericardium membrane. Segmenting the ECF is one of the most difficult medical image segmentation tasks. Since the epicardial fat is infiltrated into the groove between cardiac chambers and is contiguous with cardiac muscle, segmentation requires location and voxel intensity. Recently, deep learning methods have been effectively used to solve medical image segmentation problems in several domains with state-of-the-art performance. This paper presents a novel approach to 3D segmentation of ECF by integrating attention gates and deep supervision into the 3D U-Net deep learning architecture. The proposed method shows significant improvement of the segmentation performance, when compared with standard 3D U-Net. The experiments show excellent performance on non-contrast CT datasets with average Dice scores of 90.06%. Transfer learning from a pre-trained model of a non-contrast CT to contrast-enhanced CT dataset was also performed. The segmentation accuracy on the contrast-enhanced CT dataset achieved a Dice score of 88.16%.
No takes yet. Share an insight, caveat, or question.
Siriapisith et al. (2021) studied Epicardial fat segmentation in cardiac CT (n=260). AG-DSV-U-Net deep learning model vs. Standard 3D U-Net was evaluated on Dice similarity coefficient (DSC) on non-contrast CT testing dataset (p=<0.05). The AG-DSV-U-Net deep learning model improved epicardial fat segmentation on non-contrast CT datasets, achieving a mean Dice score of 90.06% versus 84.87% for standard U-Net (p<0.05).
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: