Abstract
Artificial intelligence (AI) is developing rapidly in the medical technology field, particularly in image analysis. ECG-diagnosis is an image analysis in the sense that cardiologists assess the waveforms presented in a 2-dimensional image. We hypothesized that an AI using a convolutional neural network (CNN) may also recognize ECG images and patterns accurately. We used the PTB ECG database consisting of 289 ECGs including 148 myocardial infarction (MI) cases to develop a CNN to recognize MI in ECG. Our CNN model, equipped with 6-layer architecture, was trained with training-set ECGs. After that, our CNN and 10 physicians are tested with test-set ECGs and compared their MI recognition capability in metrics F1 (harmonic mean of precision and recall) and accuracy. The F1 and accuracy by our CNN were significantly higher (83 ± 4%, 81 ± 4%) as compared to physicians (70 ± 7%, 67 ± 7%, P < 0.0001, respectively). Furthermore, elimination of Goldberger-leads or ECG image compression up to quarter resolution did not significantly decrease the recognition capability. Deep learning with a simple CNN for image analysis may achieve a comparable capability to physicians in recognizing MI on ECG. Further investigation is warranted for the use of AI in ECG image assessment.
Highlights
Artificial intelligence (AI) is developing rapidly in the medical technology field, in image analysis
We have assessed the usefulness of an AI, which is comprised of a simple neural network architecture, and trained by the ECGs as “images”, not by electric signals, based on a relatively small ECG database, for diagnosing myocardial infarction (MI)
We evaluated the influence of ECG-lead reduction and image compression on the recognition capability of the AI
Summary
Structure of the convolutional neural network (CNN). The structures of our convolutional neural networks (CNN) are shown in Figs. 1b and 1c. To prepare the dataset for each trial, a certain section of image in PNG format (744*368) was clipped out from each ECG (offsets: top 36, left 202, bottom 780, right 570) as our original dataset (Fig. 2a) From this original dataset we have randomly chosen test-sets (n = 25) and validation-sets (n = 25). We built the variants of training/validation/test ECG sets from our original ECG datasets that differed by the number of the leads (Fig. 2c) or in its image-quality (Fig. 2d). The output of this model is Yes (MI) or No (non-MI) to the unknown ECGs, i.e. ECG-images not used during the training (Fig. 2) We repeated this training and test process 10 times for 10 different training/validation/test ECG sets to confirm its reproducibility. For each test-set, the pairs of one board-certified cardiologist and one non-board-certified physician were randomly assigned They judged whether the patient of each ECG had experienced a MI or not. Analyses were performed using python 3.6.0 or JMP (SAS, Version 11)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
