Deep Learning-Based Identification of Echocardiographic Abnormalities From Electrocardiograms
Goro Fujiki, MD,
Satoshi Kodera, MD, PhD,
Naoto Setoguchi, MD,
Kengo Tanabe, MD, PhD,
Kotaro Miyaji, MD, PhD,
Shunichi Kushida, MD, PhD,
Mike Saji, MD, PhD,
Mamoru Nanasato, MD, PhD,
Hisataka Maki, MD, PhD,
Hideo Fujita, MD, PhD,
Nahoko Kato, MD, PhD,
Hiroyuki Watanabe, MD, PhD,
Minami Suzuki, MD,
Masao Takahashi, MD, PhD,
Naoko Sawada, MD, PhD,
Jiro Ando, MD,
Masataka Sato, MD,
Shinnosuke Sawano, MD, PhD,
Susumu Katsushika, MD, PhD,
Hiroki Shinohara, MD, PhD,
Norifumi Takeda, MD, PhD,
Katsuhito Fujiu, MD, PhD,
Hiroshi Akazawa, MD, PhD,
Hiroyuki Morita, MD, PhD,
Issei Komuro, MD, PhD
Affiliations
Goro Fujiki, MD
Department of Cardiovascular Medicine, The University of Tokyo Hospital, Tokyo, Japan
Satoshi Kodera, MD, PhD
Department of Cardiovascular Medicine, The University of Tokyo Hospital, Tokyo, Japan; Address for correspondence: Dr Satoshi Kodera, Department of Cardiovascular Medicine, The University of Tokyo Hospital, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8655, Japan.
Naoto Setoguchi, MD
Division of Cardiology, Mitsui Memorial Hospital, Tokyo, Japan
Kengo Tanabe, MD, PhD
Division of Cardiology, Mitsui Memorial Hospital, Tokyo, Japan
Kotaro Miyaji, MD, PhD
Department of Cardiovascular Medicine, Asahi General Hospital, Chiba, Japan
Shunichi Kushida, MD, PhD
Department of Cardiovascular Medicine, Asahi General Hospital, Chiba, Japan
Mike Saji, MD, PhD
Department of Cardiology, Sakakibara Heart Institute, Tokyo, Japan
Mamoru Nanasato, MD, PhD
Department of Cardiology, Sakakibara Heart Institute, Tokyo, Japan
Hisataka Maki, MD, PhD
Division of Cardiovascular Medicine, Saitama Medical Center, Jichi Medical University, Omiya, Japan
Hideo Fujita, MD, PhD
Division of Cardiovascular Medicine, Saitama Medical Center, Jichi Medical University, Omiya, Japan
Nahoko Kato, MD, PhD
Department of Cardiology, Tokyo Bay Urayasu Ichikawa Medical Center, Urayasu, Japan
Hiroyuki Watanabe, MD, PhD
Department of Cardiology, Tokyo Bay Urayasu Ichikawa Medical Center, Urayasu, Japan
Minami Suzuki, MD
Department of Cardiology, JR General Hospital, Tokyo, Japan
Masao Takahashi, MD, PhD
Department of Cardiology, JR General Hospital, Tokyo, Japan
Naoko Sawada, MD, PhD
Department of Cardiology, NTT Medical Center Tokyo, Tokyo, Japan
Jiro Ando, MD
Department of Cardiology, NTT Medical Center Tokyo, Tokyo, Japan
Masataka Sato, MD
Department of Cardiovascular Medicine, The University of Tokyo Hospital, Tokyo, Japan
Shinnosuke Sawano, MD, PhD
Department of Cardiovascular Medicine, The University of Tokyo Hospital, Tokyo, Japan
Susumu Katsushika, MD, PhD
Department of Cardiovascular Medicine, The University of Tokyo Hospital, Tokyo, Japan
Hiroki Shinohara, MD, PhD
Department of Cardiovascular Medicine, The University of Tokyo Hospital, Tokyo, Japan
Norifumi Takeda, MD, PhD
Department of Cardiovascular Medicine, The University of Tokyo Hospital, Tokyo, Japan
Katsuhito Fujiu, MD, PhD
Department of Cardiovascular Medicine, The University of Tokyo Hospital, Tokyo, Japan; Department of Advanced Cardiology, The University of Tokyo, Tokyo, Japan
Hiroshi Akazawa, MD, PhD
Department of Cardiovascular Medicine, The University of Tokyo Hospital, Tokyo, Japan
Hiroyuki Morita, MD, PhD
Department of Cardiovascular Medicine, The University of Tokyo Hospital, Tokyo, Japan
Issei Komuro, MD, PhD
Department of Frontier Cardiovascular Science, Graduate School of Medicine, The University of Tokyo, Tokyo, Japan; International University of Health and Welfare, Tochigi, Japan
Background: Heart failure should be diagnosed as early as possible. Although deep learning models can predict one or more echocardiographic findings from electrocardiograms (ECGs), such analyses are not comprehensive. Objectives: This study aimed to develop a deep learning model for comprehensive prediction of echocardiographic findings from ECGs. Methods: We obtained 229,439 paired ECG and echocardiography data sets from 8 centers. Six centers contributed to model development and 2 to external validation. We identified 12 echocardiographic findings related to left-sided cardiac abnormalities, valvular heart diseases, and right-sided cardiac abnormalities. These findings were predicted using convolutional neural networks, and a composite label was analyzed using logistic regression. A positive composite label indicated positivity in any of the 12 findings. Results: For the composite findings label, the area under the receiver-operating characteristic curve was 0.80 (95% CI: 0.80-0.81) on hold-out validation and 0.78 (95% CI: 0.78-0.79) on external validation. The composite findings label applying logistic regression had an area under the receiver-operating characteristic curve of 0.80 (95% CI: 0.80-0.81) with accuracy of 73.8% (95% CI: 73.2-74.4), sensitivity of 81.1% (95% CI: 80.5-81.8), and specificity of 60.7% (95% CI: 59.6-61.8). Conclusions: We have developed convolutional neural network models that predict a wide range of echocardiographic findings, including left-sided cardiac abnormalities, valvular heart diseases, and right-sided cardiac abnormalities from ECGs and created a model to predict a composite findings label by logistic regression analysis. This model has potential to serve as an adjunct for early diagnosis and treatment of previously undetected cardiac disease.
