Localizing the Origin of Idiopathic Ventricular Arrhythmias From ECG Using a Recurrent Convolutional Neural Network With Attention

Idiopathic ventricular arrhythmias (IVAs) are extra abnormal heartbeats disturbing the regular heart rhythm that can become fatal if left untreated. Cardiac catheter ablation is the standard approach to treating IVAs; however, a crucial prerequisite for the ablation is the localization of the IVA origin. The current IVA localization techniques are invasive, rely on expert interpretation, or are inaccurate. This study aims to develop a deep learning algorithm that can automatically identify the origin of IVAs from electrocardiogram (ECG) signals without the need for expert manual analysis. Our developed deep learning algorithm comprises a spatial feature extraction to extract the most informative features from multichannel ECG data, temporal modeling to capture the evolving pattern of the ECG time series, and an attention mechanism to weigh the most important temporal features and improve the model’s interpretability. The algorithm was validated on a 12-lead ECG dataset collected from 334 patients (230 females, age
$46~\pm ~13$
) who experienced IVAs and successfully underwent a catheter ablation procedure that determined the IVAs’ exact origins. It achieved an area under the curve (AUC) of 93%, an accuracy of 94%, a sensitivity of 97%, a precision of 95%, and an
${F}1$
-score of 96% in locating the origin of IVAs and outperformed existing automatic and semi-automatic algorithms. Our proposed method shows promise toward automatic and noninvasive evaluation of IVA patients before cardiac catheter ablation.