Biomedical Signal Processing - Computational Intelligence in Biomedical Engineering - page 15

Biomedical Signal Processing
2.1 Introduction
An important aspect of biomedical engineering is measuring and interpreting
biosignals originating from almost every system in the body, from the nervous
system to the endocrine system and the brain. These signals are electrical sig-
nals that result from action potentials in our cells, the nervous system being
the primary example. It is also possible to measure quantities such as blood
pressure or frequency of a heart valve opening and closing, and treat them
as biosignals. The correct measurement and interpretation of signals is vital
in assisting physicians with diagnosis, patient monitoring, and treatment, for
example, heart bypass surgery requires stringent monitoring of blood oxy-
genation levels to avoid tissue damage. As such, signal-processing techniques
have become important in biomedical engineering to interpret the information
and to present it effectively.
In this chapter, we review the various aspects of signal processing, begin-
ning with the definition of a signal followed by signal transforms, basic power
spectral estimation, filters, and parametric models. These topics are covered
in greater detail in other sources (Akay, 1994; Banks, 1990; Clarkson, 1993;
Oppenheim et al., 1999; Stein, 2000; Stergiopoulos, 2001). We have limited
the mathematics because most of the techniques described later in the chapter
have been implemented in commercially available software or can be down-
loaded from the Internet.
2.2 Signals and Signal Systems
A signal can be represented in both time and frequency domains. By this, we
mean that we can specify the signal as a progression of values through time
or a variation in the frequency range. An analog signal is usually continuous
in time and represented by a smooth variation in voltage, for example, ECG,
EEG, and electrooculogram (EOG). A digital signal is discrete in time and
represented by a sequence of quantized voltage levels. The levels correspond
to discrete intervals of time and often obtained by sampling an analog signal.
15