EchoEFNet: Multi-task deep learning network for automatic calculation of left ventricular ejection fraction in 2D echocardiography. (April 2023)

• Record Type: Journal Article
• Authors: Li, Honghe; Wang, Yonghuai; Qu, Mingjun; Cao, Peng; Feng, Chaolu; Yang, Jinzhu
• Is Part Of: Computers in biology and medicine. Volume 156(2023)
• Publication Date: 2023-04
• Subjects: Ejection fraction -- Echocardiogram -- Multitasking -- Deep learning; Medicine -- Data processing -- Periodicals; Biology -- Data processing -- Periodicals; 610.285
• DOI: 10.1016/j.compbiomed.2023.106705
• Languages: English
• ISSNs: 0010-4825

Abstract

Abstract: 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. Highlights: A multi-task learning model is designed for both segmentation and landmark detection tasks. The proposed landmark detection mechanism is able to reduce the burden of human annotate. The indices are calculated in a manner consistent with the clinical workflow. The model is validated on two datasets. The results are highly correlated with the true value.