EchoEFNet automatically calculated left ventricular ejection fraction, achieving correlations between predicted and true values of 0.854 on the CAMUS dataset and 0.916 on the CMUEcho dataset.
Does the EchoEFNet deep learning network accurately and automatically calculate LVEF in 2D echocardiography compared to manual assessment?
EchoEFNet provides an automated, deep learning-based method for calculating LVEF from 2D echocardiograms that correlates highly with manual measurements.
Effect estimate: Correlation 0.854 (CAMUS), 0.916 (CMUEcho)
Left ventricular ejection fraction (LVEF) is essential for evaluating left ventricular systolic function. However, its clinical calculation requires the physician to interactively segment the left ventricle and obtain the mitral annulus and apical landmarks. This process is poorly reproducible and error prone. In this study, we propose a multi-task deep learning network EchoEFNet. The network use ResNet50 with dilated convolution as the backbone to extract high-dimensional features while maintaining spatial features. The branching network used our designed multi-scale feature fusion decoder to segment the left ventricle and detect landmarks simultaneously. The LVEF was then calculated automatically and accurately using the biplane Simpson's method. The model was tested for performance on the public dataset CAMUS and private dataset CMUEcho. The experimental results showed that the geometrical metrics and percentage of correct keypoints of EchoEFNet outperformed other deep learning methods. The correlation between the predicted LVEF and true values on the CAMUS and CMUEcho datasets was 0.854 and 0.916, respectively.
Li et al. (Fri,) conducted a other in Left ventricular systolic function evaluation. EchoEFNet vs. Other deep learning methods was evaluated on Correlation between predicted LVEF and true values (Correlation 0.854 (CAMUS), 0.916 (CMUEcho)). EchoEFNet automatically calculated left ventricular ejection fraction, achieving correlations between predicted and true values of 0.854 on the CAMUS dataset and 0.916 on the CMUEcho dataset.