ECG Statistics, Noise, Artifacts, and Missing Data  - Chapter 3

G.D. Clifford

CRCin Advanced Methods for ECG Analysis, (Eds. G. D. Clifford, F. Azuaje and P. E. McSharry),  Artech House, London, 2006.

 Chapter 3 presents a detailed phenomenological overview of the ECG, both in terms of traditional clinical parameters (and their normal limits), and in terms of metrics more familiar to the signal processor. Discussions of how these parameters change with different heart rates, activity levels, sleep states and autonomic nervous system activity are given. Methods for quatifying and dealing with the effect of noise, artifact and nonstationarities (such as ectopy) on ECG metrics, including heart rate variability (HRV), heart rate turbulence (HRT), and QT turbulence, are detailed. 



3.1 Introduction


3.2 Spectral and Cross-Spectral Analysis of the ECG

3.2.1 Extreme Low and High Frequency ECG

3.2.2 The Spectral Nature of Arrhythmias

3.3 Standard Clinical ECG Features


3.4 Nonstationarities in the ECG

3.4.1 Heart Rate Hysteresis

3.4.2 Arrhythmias

3.5 Arrhythmia Detection

3.5.1 Arrhythmia Classification from Beat Typing

3.5.2 Arrhythmia Classification from Power-Frequency Analysis

3.5.3 Arrhythmia Classification from Beat-to-Beat Statistics

3.6 Noise and Artifact in the ECG

3.6.1 Noise and Artifact Sources

3.6.2 Measuring Noise in the ECG

3.7 Heart Rate Variability

3.7.1 Time Domain and Distribution Statistics

3.7.2 Frequency Domain HRV Analysis

3.7.3 Long-Term Components

3.7.4 The Lomb-Scargle Periodogram

3.7.5 Information Limits and Background Noise

3.7.6 The Effect of Ectopy and Artifact, and How to Deal with It

3.7.7 Choosing an Experimental Protocol; Activity-Related Changes

3.8 Dealing with Nonstationarities

3.8.1 Nonstationary HRV Metrics and Fractal Scaling

3.8.2 Activity-Related Changes

3.8.3 Perturbation Analysis