Zisang Zhang, Ye Zhu, Shujun Chen, Yiheng Dong, Yiwei Zhang, Chun Wu, Ziming Zhang, Shuangshuang Zhu, Manwei Liu, Zhenxing Sun, Peige Zhang, Lili Jiang, Hongliang Yuan, Yuman Li
Abstract
Left ventricular ejection fraction (LVEF) and global longitudinal strain (GLS) are essential for the diagnosis, clinical decision-making, and prognosis of cardiovascular disease. However, accurate assessments of LVEF and GLS by echocardiography are hampered by inter-observer variability, time-consuming, and labor-intensive. This study aimed to develop an automated method to accurately and rapidly assess LVEF and GLS.
Introduction
Accurate assessment of left ventricular systolic function is of great value for screening, diagnosis, and therapeutic decision-making, and prognostic evaluation of cardiac disease [1–5]. Echocardiography is considered an essential first-line imaging modality for cardiac function evaluation, providing non-invasive, real-time quantitative assessments of global and regional left ventricular function [1–5].
Methods
Ethics statement
The study was approved by the Ethics Committee of Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China. It was conducted in accordance with the ethical principles of the Declaration of Helsinki. The requirement for informed consent was waived.
Results
Clinical characteristics of the study cohorts
The clinical characteristics of subjects were obtained from the electronic medical record system of the internal and external centers. The baseline data for the internal, external validation, and CMR validation datasets are detailed in Table 2.
Discussion
This study developed a dual-flow AI model based on segmentation and optical flow algorithms called Echo-DFCNN. The integrated pipeline of the segmentation and optical flow modules enables simultaneous measurements of LVEF and GLS. Indeed, the strategy of collaboratively processing complementary information via dual-flow architectures has been widely validated across various domains, including medical imaging [40] and natural scene analysis [41].
Conclusions
This study proposes a dual-flow AI model based on segmentation and optical flow algorithms called Echo-DFCNN for automatic analysis of echocardiogram videos. Echo-DFCNN achieves simultaneous and precise assessment of LVEF and GLS in the study cohorts, demonstrating its potential for robust performance across different cardiac functions, image qualities and machine types. The application of this model has the potential to improve the efficiency of radiologists and provide support for more effective management of patients with cardiovascular diseases.
Citation: Zhang Z, Zhu Y, Chen S, Dong Y, Zhang Y, Wu C, et al. (2026) Dual-flow convolutional neural network for automatic measurement of left ventricular ejection fraction and global longitudinal strain in echocardiography. PLOS Digit Health 5(8): e0001128. https://doi.org/10.1371/journal.pdig.0001128
Editor: Iqram Hussain, Weill Cornell Medicine, UNITED STATES OF AMERICA
Received: November 3, 2025; Accepted: August 5, 2026; Published: August 28, 2026
Copyright: © 2026 Zhang et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: Datasets: https://kaggle.com/datasets/0d99e1782aee1a92fdacaea03a5256d9169aacd515ea4535eada050175c9b2ea Code: https://github.com/yz-hust/Echo-DFCNN.
Funding: This work was supported by the National Key R&D Program of China (Grant No. 2022YFF0706504 to L. Z.), funded by the Ministry of Science and Technology of the People’s Republic of China; the National Natural Science Foundation of China (Grant Nos. 82371991 to Y. L., 82230066 to M. X., 82302226 to Z. Z., 82302229 to C. W.), funded by the National Natural Science Foundation of China; the Fundamental Research Funds for the Central Universities (Grant Nos. YCJJ20252422 to Z. Z., YCJJ20252427 to Y. Z.), funded by Huazhong University of Science and Technology. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The URLs of the funder websites are: https://en.most.gov.cn/, https://www.nsfc.gov.cn/, and https://www.hust.edu.cn/.
Competing interests: The authors have declared that no competing interests exist.
Abbreviations: LVEF, left ventricular ejection fraction; LVEDV, left ventricular end-diastolic volume; LVESV, left ventricular end-systolic volume; GLS, global longitudinal strain; AI, artificial intelligence; ED, end-diastolic; ES, end-systolic; CMR, cardiac magnetic resonance; DSC, Dice similarity coefficient; ASD, average surface distance; HD, Hausdorff distance; ICC, intraclass correlation coefficients; LOA, limits of agreement; SD, standard deviation; IQR, interquartile range; AUC, area under the curve
