P-wave Detection Using a Parallel Convolutional Neural Network in Electrocardiogram

P-wave is one of the most important waveforms in electrocardiogram (ECG), and an accurate P-wave detection method is of great significance for the automatic diagnosis of arrhythmia diseases. Recent attempts to detect P-wave have been reported in the literature, but distinguishing the absence or presence of P-wave has been a challenge in the research of ECG diagnosis. This study introduces a novel parallel convolutional neural network (CNN) method for P-wave detection. The obtained result shows that the proposed method achieves 99.04%, 99.28% for the sensitivity and accuracy on the QT Database (QTDB), respectively. It is demonstrated that the proposed parallel convolutional neural network can be adopted for automatic detection of P-wave, which is more helpful for the automatic classification of arrhythmia diseases.