Heartbeat Detection Using Oscillatory Envelope Pattern in Noisy Electrocardiogram

Abstract

Heartbeat detection from electrocardiogram (ECG) is crucial for heart rate, heart rate variability, and arrhythmia analysis. However, the detection usually suffers from the interference from violent body movement or muscular activity in wearable ECG devices. Here, we propose a novel heartbeat detection based on the envelope pattern of the oscillatory components derived from ECG by a bandpass filter from 18 to 20 Hz. The envelope of the positive peaks of the emphasized signal is therefore constructed. Each adjacent QRS pair corresponds to a U-type pattern in the envelope. Then, the QRS beat is identified by the amplitude and interval of the U-wave. Meanwhile, low signal-to-noise segment is also marked by the envelope morphology for avoiding wrong recognition of an artifact as a heartbeat.

A new QRS detector stress test combining temporal jitter and F-score (JF) reveals significant performance differences amongst popular detectors

“…This database has the shortcoming that very few examples of motion artefacts. The work by Benitez, Kalidas and Chan [7][8][9] tried to highlight noise resilience using only small sections (3 − 10 s) of a few select records. As these noisy sections make up such a small proportion of the database they have very little impact on the overall results.…”

“…Very few examples of motion artefacts: This leads to attempts looking only at sections of the ECG recordings which contain a fair amount of noise. The work by [7][8][9] tried to highlight noise resilience using only small sections (3 . .…”

“…Very few examples of motion artefacts: This leads to attempts looking only at sections of the ECG recordings which contain a fair amount of noise. The work by [7–9] tried to highlight noise resilience using only small sections (3 … 10s) of a few select records. As these noisy sections make up such a small proportion of the database, they have very little impact on the overall results.…”