Optimized Pan-Tompkins Based Heartbeat Detection Algorithms

The whole physiological state of cardiac electrical activity is reflected by electrocardiogram (ECG) signals collected from the body surface. ECG signals can be used as important basis for the diagnosis of heart disease. However, when analyzing ECG signals, people often process the signals in a record into units of heartbeats. Therefore, accurate heartbeat detection is an important prerequisite for the correct diagnosis of heart disease by the machine. The traditional heartbeat detection only obtains the QRS complex based on the Pan-Tompkins algorithm, and then performs time interception based on previous experience (forward and backward). The accuracy of heartbeats detection will be affected by individual differences. In this paper, an optimized Pan-Tompkins algorithm based on adding P-wave and T-wave location is proposed for the detection of ECG heartbeats. This algorithm was evaluated using 109966 samples from the MIT- BIH Arrhythmia Database including a variety of arrhythmia data (such as atrioventricular block, ventricular tachycardia, early ventricular contraction, atrial flutter, ventricular premature beats, ventricular fibrillation). 109519 samples were correctly detected from 109966 heartbeats. This method achieved 99.68% positive prediction rate(P +) with 99.90% sensitivity)SE). Our approach will lay a foundation for the processing of ECG signals and therefore improve the accuracy for the diagnosis of heart diseases.