Deep learning interpretation of echocardiograms
Abstract Echocardiography uses ultrasound technology to capture high temporal and spatial resolution images of the heart and surrounding structures, and is the most common imaging modality in cardiovascular medicine. Using convolutional neural networks on a large new dataset, we show that deep learn...
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2020-01-01
|
| Series: | npj Digital Medicine |
| Online Access: | https://doi.org/10.1038/s41746-019-0216-8 |
| _version_ | 1797429026464202752 |
|---|---|
| author | Amirata Ghorbani David Ouyang Abubakar Abid Bryan He Jonathan H. Chen Robert A. Harrington David H. Liang Euan A. Ashley James Y. Zou |
| author_facet | Amirata Ghorbani David Ouyang Abubakar Abid Bryan He Jonathan H. Chen Robert A. Harrington David H. Liang Euan A. Ashley James Y. Zou |
| author_sort | Amirata Ghorbani |
| collection | DOAJ |
| description | Abstract Echocardiography uses ultrasound technology to capture high temporal and spatial resolution images of the heart and surrounding structures, and is the most common imaging modality in cardiovascular medicine. Using convolutional neural networks on a large new dataset, we show that deep learning applied to echocardiography can identify local cardiac structures, estimate cardiac function, and predict systemic phenotypes that modify cardiovascular risk but not readily identifiable to human interpretation. Our deep learning model, EchoNet, accurately identified the presence of pacemaker leads (AUC = 0.89), enlarged left atrium (AUC = 0.86), left ventricular hypertrophy (AUC = 0.75), left ventricular end systolic and diastolic volumes ( $${R}^{2}$$ R 2 = 0.74 and $${R}^{2}$$ R 2 = 0.70), and ejection fraction ( $${R}^{2}$$ R 2 = 0.50), as well as predicted systemic phenotypes of age ( $${R}^{2}$$ R 2 = 0.46), sex (AUC = 0.88), weight ( $${R}^{2}$$ R 2 = 0.56), and height ( $${R}^{2}$$ R 2 = 0.33). Interpretation analysis validates that EchoNet shows appropriate attention to key cardiac structures when performing human-explainable tasks and highlights hypothesis-generating regions of interest when predicting systemic phenotypes difficult for human interpretation. Machine learning on echocardiography images can streamline repetitive tasks in the clinical workflow, provide preliminary interpretation in areas with insufficient qualified cardiologists, and predict phenotypes challenging for human evaluation. |
| first_indexed | 2024-03-09T09:07:15Z |
| format | Article |
| id | doaj.art-914a9b08ef9d49d596f7dc8d8a6c2fa6 |
| institution | Directory Open Access Journal |
| issn | 2398-6352 |
| language | English |
| last_indexed | 2024-03-09T09:07:15Z |
| publishDate | 2020-01-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | npj Digital Medicine |
| spelling | doaj.art-914a9b08ef9d49d596f7dc8d8a6c2fa62023-12-02T10:10:18ZengNature Portfolionpj Digital Medicine2398-63522020-01-013111010.1038/s41746-019-0216-8Deep learning interpretation of echocardiogramsAmirata Ghorbani0David Ouyang1Abubakar Abid2Bryan He3Jonathan H. Chen4Robert A. Harrington5David H. Liang6Euan A. Ashley7James Y. Zou8Department of Electrical Engineering, Stanford UniversityDepartment of Medicine, Stanford UniversityDepartment of Electrical Engineering, Stanford UniversityDepartment of Computer Science, Stanford UniversityDepartment of Medicine, Stanford UniversityDepartment of Medicine, Stanford UniversityDepartment of Medicine, Stanford UniversityDepartment of Medicine, Stanford UniversityDepartment of Electrical Engineering, Stanford UniversityAbstract Echocardiography uses ultrasound technology to capture high temporal and spatial resolution images of the heart and surrounding structures, and is the most common imaging modality in cardiovascular medicine. Using convolutional neural networks on a large new dataset, we show that deep learning applied to echocardiography can identify local cardiac structures, estimate cardiac function, and predict systemic phenotypes that modify cardiovascular risk but not readily identifiable to human interpretation. Our deep learning model, EchoNet, accurately identified the presence of pacemaker leads (AUC = 0.89), enlarged left atrium (AUC = 0.86), left ventricular hypertrophy (AUC = 0.75), left ventricular end systolic and diastolic volumes ( $${R}^{2}$$ R 2 = 0.74 and $${R}^{2}$$ R 2 = 0.70), and ejection fraction ( $${R}^{2}$$ R 2 = 0.50), as well as predicted systemic phenotypes of age ( $${R}^{2}$$ R 2 = 0.46), sex (AUC = 0.88), weight ( $${R}^{2}$$ R 2 = 0.56), and height ( $${R}^{2}$$ R 2 = 0.33). Interpretation analysis validates that EchoNet shows appropriate attention to key cardiac structures when performing human-explainable tasks and highlights hypothesis-generating regions of interest when predicting systemic phenotypes difficult for human interpretation. Machine learning on echocardiography images can streamline repetitive tasks in the clinical workflow, provide preliminary interpretation in areas with insufficient qualified cardiologists, and predict phenotypes challenging for human evaluation.https://doi.org/10.1038/s41746-019-0216-8 |
| spellingShingle | Amirata Ghorbani David Ouyang Abubakar Abid Bryan He Jonathan H. Chen Robert A. Harrington David H. Liang Euan A. Ashley James Y. Zou Deep learning interpretation of echocardiograms npj Digital Medicine |
| title | Deep learning interpretation of echocardiograms |
| title_full | Deep learning interpretation of echocardiograms |
| title_fullStr | Deep learning interpretation of echocardiograms |
| title_full_unstemmed | Deep learning interpretation of echocardiograms |
| title_short | Deep learning interpretation of echocardiograms |
| title_sort | deep learning interpretation of echocardiograms |
| url | https://doi.org/10.1038/s41746-019-0216-8 |
| work_keys_str_mv | AT amirataghorbani deeplearninginterpretationofechocardiograms AT davidouyang deeplearninginterpretationofechocardiograms AT abubakarabid deeplearninginterpretationofechocardiograms AT bryanhe deeplearninginterpretationofechocardiograms AT jonathanhchen deeplearninginterpretationofechocardiograms AT robertaharrington deeplearninginterpretationofechocardiograms AT davidhliang deeplearninginterpretationofechocardiograms AT euanaashley deeplearninginterpretationofechocardiograms AT jamesyzou deeplearninginterpretationofechocardiograms |