Quantitative prediction of right ventricular and size and function from the electrocardiogram

Figures

• 

  Figure 1.. Classification results for Abnormal Right Ventricular Ejection Fraction and End-Diastolic Volume Receiver operating characteristic curves (Blue) and Precision-Recall curves (Orange) to classify RV systolic dysfunction (A,left) and RV dilation (B, right) in UK Biobank dataset (top), MSH dataset (middle), and MSH prospective validation dataset(bottom). Red-dashed lines on ROC curves represents no model skill, and on PR curves represents baseline incidence of disease. AUROC: Area under the receiver operating curve, AUPRC: Area under the precision recall curve.

• 

  Figure 2.. Regression Models, MSH dataset MSHoriginal regression model with ECG-predicted versus cMRI-expected scatter plots to predict numerical RVEF (Box A) and numerical RVEDV (Box C). Scatter plots plotted with line of perfect concordance (dashed black line) and linear best-fit line (solid blue). Bland-Altman analysis of RVEF (Box B) and RVEDV (Box D) with mean difference (red dashed line) and upper/lower limits of agreement (blue solid line).

• 

  Figure 3.. Saliency Mapping to Identify Key Areas of the ECG relevant to the prediction. Saliency mapping ECG examples from the MSHoriginal test set. Leads are arranged from bottom to top: lead I, II, V1-V6. Increasing shades of red indicate increasingly influential pixel in the model. Figure A) shows a true positive example of RV dysfunction. The P and QRS portions of V1, V5, and V6 are particularly influential. Figure B) shows a true negative example of RV dilation. QRS complex of lead I and P-wave in V1 are particularly influential