An improved template-matching method for real-time detection of abnormal heartbeats

This study proposes an improved template-matching method which can automatically update the normal template heartbeats with the change in heart rate, and compare the input heartbeats with the template to detect abnormal heartbeats in real time. If an input heartbeat has a cross-correlation coefficient of less than 0.6 or a mean-square difference larger than 1.0 in comparison with the template, it is identified as abnormal. The study results show that if the normal template is not updated with the heart rate, normal heartbeats may be misjudged as abnormal when exercise changes the heart rate and heartbeat waveforms. The results also demonstrate that the proposed method can accurately identify abnormal heartbeats of ventricular fibrillation (VF) and ventricular tachycardia (VT) using ECG records 426 and 607 from the Malignant Ventricular Arrhythmia Database.