Spatiotemporal 3-D Variations Modeling With Self-Attention for Multilabel ECG Classification

Multilabel electrocardiogram (ECG) classif- ication is of great significance for the diagnosis of cardiovascular diseases, and it is also a challenging task because of the various of pathological patterns across different cardiac abnormalities. This article analyzes the complex spatial-temporal variations information contained in the 12-lead ECG from a new perspective and proposes an advanced deep neural network (DNN) framework to accurately identify various multilabel ECG abnormalities. First, the 12-lead ECG was modeled as 3-D spatial-temporal variations involving the intraperiod and interperiod temporal dimensions and cross-lead spatial dimension. Then, a 3-D spatial-temporal learning network (3D-STLNet) was carefully designed to extract the complex pathological features from the 3-D spatial-temporal variations. Second, the multihead attention (MHA) mechanism was introduced into the proposed network to extract global temporal dependencies in ECG segments, which further improved the performance of the proposed network. By integrating a residual backbone network, a 3D-STLNet, and the MHA module, our proposed network allows end-to-end learning and prediction in a unified framework. The influence of several classic threshold selection strategies on the performance of multilabel ECG classification algorithm was also analyzed, which provides a reference for subsequent research on optimal threshold selection for multilabel ECG classification. Finally, our proposed network was evaluated on multiple large multilabel ECG public datasets, and the performance of our network outperforms that of current state-of-the-art networks, which demonstrates that our network has great competitiveness in identifying a wider range of multilabel ECG abnormalities.