Improving automated labeling with deep learning and signal segmentation for accurate ECG signal analysis

Accurate automated analysis of ECG signals is essential for efficient heart health assessment. This study aims to develop a robust deep learning framework for precise ECG signal labeling. The approach combines signal segmentation and classification to enhance QRS complex and R-peak detection. A two-stage deep learning architecture was employed. The initial stage involves waveform segmentation using an LSTM network, optimized by the Fourier Synchrosqueezed Transform. The second stage identifies QRS complexes and R-peaks within segmented regions using a pre-trained deep learning model and a custom peak finder. The proposed method was evaluated on the QT Database, comprising approximately 15 min of ECG recordings from 105 patients. The proposed framework demonstrated superior performance in QRS complex detection compared to existing methods. Evaluation metrics, including precision, sensitivity, and specificity, were used to assess the model's accuracy. The hybrid approach effectively balanced the advantages of deep learning with traditional signal processing techniques, resulting in improved efficiency and reliability. The developed deep learning framework offers a promising approach for accurate and efficient ECG signal analysis. By combining signal segmentation and classification, the method effectively addresses challenges in QRS complex detection. The findings contribute to the advancement of automated ECG interpretation and have potential applications in various cardiac healthcare settings.