CDLR-net: a ECG classification network based on deep residual shrinkage networks and LSTM

Many traditional classification networks directly use the limb two-lead signal (MLII) ECG signals as input for training. However, this method suffers from reduced accuracy when ECG features are not obvious, especially for premature heartbeats. To solve the issue, this paper proposed a novel network, namely CDLR-Net, that combines a Deep Residual Shrinkage Network (DRSN) with a Long Short-Term Memory (LSTM). The model combines MLII lead data with RR interval features. ECG signals are first denoised by wavelet decomposition, after which pre-RR, post-RR, local-10 average RR, and overall average RR intervals are extracted from R-wave localization for each heartbeat. Incorporating RR interval information improves classification accuracy. Finally, the classification is achieved through proposed method. Experiments on the MIT-BIH database under inter-patient and intra-patient schemes achieved 97% and 99% accuracy, respectively, demonstrating the effectiveness of the proposed method.