Identification of patients with dilated phase of hypertrophic cardiomyopathy using a convolutional neural network applied to multiple, dual, and single lead electrocardiograms

Figures

• 

  Fig. 1. Convolutional neural network (CNN) analysis.

• 

  Fig. 2. The proportion of patients according to the model output in CNN-derived models using eight-lead ECGs The vertical scale indicates the proportion of the patients in dHCM group (orange) and non-dHCM group (blue). The horizontal scale indicates the diagnostic probability for dHCM yielded by CNN model. dHCM1 indicates patients diagnosed as dHCM at baseline in the original database (n = 11). dHCM2 indicates patients registered as having HCM at baseline in the original database and progressed to dHCM later during the time course (n = 4). dHCM3 indicates patients with a diagnosis of dHCM who were registered to our database before 2009 and with available ECG recorded between February 2010 and March 2018 (n = 39). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

• 

  Fig. 3. The proportion of patients according to the model output in CNN-derived models using single-lead or double-lead ECGs The vertical scale indicates the proportion of the patients in dHCM group (orange) and non-dHCM group (blue). The horizontal scale indicates the diagnostic probability for dHCM yielded by each CNN model. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

• 

  Fig. 4. Representative images of GradCAM with eight-lead ECGs The difference in color chart cannot be compared between the models or leads, because the color chart was determined within each model or lead,

• 

  Fig. 5. Representative images of GradCAM with single-lead and double-lead ECGs The difference in color chart cannot be compared between the models or leads, because the color chart was determined within each model or lead,

• 

  Supplementary figure 1.

• 

  Supplementary figure 2.