ECG Signal Analysis - ECG Acquisition and Automated Remote Processing - page 15

Chapter 2
ECG Signal Analysis
2.1 Introduction
This chapter describes the basic steps for analysis of ECG signal using computer-
based algorithms. Computerized analysis of ECG started in the early 1960s and
considered as one of the first applications of digital computers in medicine. Paper-
based long-duration records, while being visually inspected by cardiologists, are
susceptible to human errors and suffer from inter-observer variability. Over the
decades, numerous ECG analysis algorithms have been developed and tested.
There are two prime applications of ECG signal analysis, viz., heart rhythm
analysis for continuous ECG monitoring and, ECG feature extraction and classi-
fication. Historically, the initial applications used robust computers which were fed
with digitized ECG records for automated processing. After the advent of single-
chip microprocessors and embedded systems, portable standalone instruments are
in use at advanced facilities and ICU setups.
Detailed discussion on different methodologies of ECG analysis is beyond the
scope of this topic. This chapter describes a brief introduction to ECG signal
analysis, followed by a few proposed techniques in time domain.
2.2 Computerized Analysis of ECG
An ECG signal is characterized by sequential repolarization and depolarization of
the atria and ventricles, represented by P, QRS, and T waves (with occasional U
waves) which are connected by some equipotential segments (PR, ST, and TP).
The onset and offset points, along with wave peaks, form a complete basis of
delineation of complete wave morphology. Figure 2.1 represents a typical wave
sequences P, QRS, and T waves with their respective onset and offset points.
Accurate detection of these fiducial points is the first objective of ECG feature
extraction. However, the final objective is to compute the ECG clinical signatures,
already described in the Sect. 1.4 . Among all, the QT interval and the PR