The automatic analysis of echocardiography is of great significance in the diagnosis of cardiovascular diseases, and accurate view classification is the basis for achieving the automatic analysis of echocardiography. However, the existing automation methods are confronted with numerous challenges: they rely heavily on expensive human annotations, have insufficient generalization ability among people and devices, and fail to fully utilize the inherent anatomical consistency among different standard views. To address these limitations, this study proposes a novel structure-aware self-supervised learning framework. The core idea is to guide the model to learn the view-invariant representation by constructing positive sample pairs between different echocardiogram views of the same patient, thus eliminating the need for manual annotation. This method enables the model to effectively capture anatomical consistency across views, providing more robust features for downstream tasks. The experimental results show that the viewpoint classification performance of this method on four standard echocardiogram datasets has been significantly improved, effectively verifying its effectiveness and clinical application potential.

Jiawei Han, Xuande Zhang, Long Xu, Kunjing Pang, Xin Huang. Contrastive Cross-View Representation Learning for Echocardiographic View Classification. Academic Journal of Computing & Information Science (2025), Vol. 8, Issue 11: 1-13. https://doi.org/10.25236/AJCIS.2025.081101.

[1] Benjamin EJ, Muntner P, Alonso A, et al. Heart Disease and Stroke Statistics—2019 Update: A Report From the American Heart Association. 2019;139(10):e56-e528.

[2] Rudski LG, Lai WW, Afilalo J, et al. Guidelines for the Echocardiographic Assessment of the Right Heart in Adults: A Report from the American Society of Echocardiography: Endorsed by the European Association of Echocardiography, a registered branch of the European Society of Cardiology, and the Canadian Society of Echocardiography. Journal of the American Society of Echocardiography. 2010;23(7):685-713.

[3] Nagueh SF, Smiseth OA, Appleton CP, et al. Recommendations for the Evaluation of Left Ventricular Diastolic Function by Echocardiography: An Update from the American Society of Echocardiography and the European Association of Cardiovascular Imaging. European Heart Journal - Cardiovascular Imaging. 2016;17(12):1321-1360.

[4] Lang RM, Badano LP, Mor-Avi V, et al. Recommendations for cardiac chamber quantification by echocardiography in adults: an update from the American Society of Echocardiography and the European Association of Cardiovascular Imaging. 2015;16(3):233-271. 

[5] Madani A, Arnaout R, Mofrad M, et al. Fast and accurate view classification of echocardiograms using deep learning. npj Digital Medicine. 2018/03/21, 2018;1(1):6. 

[6] Østvik A, Smistad E, Aase SA, Haugen BO, Lovstakken L. Real-Time Standard View Classification in Transthoracic Echocardiography Using Convolutional Neural Networks. Ultrasound in Medicine & Biology. 2019/02/01/, 2019;45(2):374-384. 

[7] Naser JA, Lee E, Pislaru SV, et al. Artificial intelligence-based classification of echocardiographic views. European Heart Journal - Digital Health. 2024;5(3):260-269. 

[8] Huang M, Lin W, Chen Y, et al. Explainable deep neural network for echocardiography view classification. European Heart Journal - Cardiovascular Imaging. 2022;23(Supplement_1)

[9] Kusunose K, Haga A, Inoue M, Fukuda D, Yamada H, Sata M. Clinically Feasible and Accurate View Classification of Echocardiographic Images Using Deep Learning. Biomolecules. 2020; 10(5):665.

[10] Wu L, Dong B, Liu X, et al. Standard echocardiographic view recognition in diagnosis of congenital heart defects in children using deep learning based on knowledge distillation.Frontiers in Pediatrics. 2022;9:770182.

[11] Gungor DG, Rao B, Wolverton C, et al. View classification and object detection in cardiac ultrasound to localize valves via deep learning.arXiv preprint arXiv:2311.00068 (2023).

[12] Smistad E, Østvik A, Salte IM, et al. Real-time automatic ejection fraction and foreshortening detection using deep learning.IEEE transactions on ultrasonics, ferroelectrics, and frequency control. 2020;67(12):2595-2604. 

[13] Tian Y, Xie L, Zhang X, et al. Semantic-aware generation for self-supervised visual representation learning.arXiv preprint arXiv:2111.13163 (2021).

[14] Le-Khac PH, Healy G, Smeaton AFJIA. Contrastive representation learning: A framework and review. IEEE Access. 2020;8:193907-193934. 

[15] Chen T, Kornblith S, Norouzi M, Hinton G. A simple framework for contrastive learning of visual representations. PmLR; 2020:1597-1607. 

[16] He K, Fan H, Wu Y, et al. Momentum contrast for unsupervised visual representation learning.Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2020:9729-9738.

[17] Grill J-B, Strub F, Altché F, et al. Bootstrap your own latent-a new approach to self-supervised learning.Advances in neural information processing systems. 2020;33:21271-21284. 

[18] Chen X, He K. Exploring simple siamese representation learning.Proceedings of the IEEE/CVF conference on computer vision and pattern recognition.  2021:15750-15758.

[19] Zhou Z, Sodha V, Rahman Siddiquee MM, et al. Models genesis: Generic autodidactic models for 3d medical image analysis. Springer; 2019:384-393.

[20] Chen S, Ma K, Zheng YJapa. Med3d: Transfer learning for 3d medical image analysis. arXiv preprint arXiv:1904.00625 (2019).

[21] Grill J-B, Strub F, Altché F, et al. Bootstrap your own latent-a new approach to self-supervised learning.Advances in neural information processing systems. 2020;33:21271-21284. 

[22] Goyal P, Dollár P, Girshick R, et al. Accurate, large minibatch sgd: Training imagenet in 1 hour. arXiv preprint arXiv:1706.02677 (2017).

[23] Loshchilov I, Hutter FJapa. Sgdr: Stochastic gradient descent with warm restarts. arXiv preprint arXiv:1608.03983 (2016).