Optimal ECG Lead System for Automatic Myocardial Ischemia Detection

Misha Glazunov1, Alfonso Aranda2, Carlo Galuzzi3
1TU Delft, 2Maastricht University & Medtronic, 3Swinburne University of Technology


We study several features that are not commonly used in clinical practice but may play a role in automatic myocardial ischemia detection using a reduced 3-lead ECG system. In particular, we consider features that were suggested to have diagnostic significance concerning QRS complex such as fragmented QRS in time domain and intra-QRS in time-frequency domain. Moreover, we apply chaos theory to reconstruct attractors from ECG assuming non-linear nature of underlying heart dynamics. Based on that, we devise geometrical features as well as two main dynamical invariants: the correlation dimension and the Lyapunov exponent. For validation, we use the Physionet STAFF III dataset. It contains 12-lead ECG recordings of 104 patients at both baseline and coronary artery occlusion during percutaneous transluminal coronary angiography. We apply statistical analysis to check if there is a significant difference in acquired parameters between these two groups. Furthermore, we use the gradient boosting machine for automatic classification. As a result, we identify the optimal 3-lead ECG system, based on the considered features, and achieve promising results in ischemia diagnostics: the area under the ROC curve (AUC) is 0.91. It improves the results obtained with the baseline features such as ST-segment elevation and T-wave inversion: AUC 0.85. Finally, we combine new parameters with the baseline features and enhance the final model with previously introduced pseudo-vectorcardiography parameters. The results allow accounting for all the regions of the heart ischemia: anterior, inferior, and posterior. The proposed automatic algorithm allows in the easiest possible way to determine the first signs of ischemic heart disease in a patient’s ECG based on the input from the minimal number of leads. Overall, it seems possible to get the diagnosis in the ambulance during transportation or even at home by patients themselves.