46
|
2 Fundamentals of Information, Signal and System Theory
Patient
Measurement
Computer
Doctor
parameters
relevant
physiological
measurand
signal measurand
raw signal
Fig. 2.24: An example of a signal processing chain in biosignal processing.
In the further course of the signal analysis, the aim is often to reduce the amount
of data to a few, but meaningful parameters that can be used for medical decision-
making. For this purpose, the raw signal is usually first freed from disturbing artefacts
and noise by filtering and only then evaluated with regard to diagnostically relevant
parameters such as heart rate and respiratory rate. A mostly graphical representation
of the evaluation is forwarded to the attending physician for a decision on how to pro-
ceed with the patient’s treatment. In this scenario, the final evaluation is carried out
by an experienced doctor, but medical expert systems can also provide important ser-
vices for further evaluation. Here, on the basis of statistically processed expert know-
ledge, a probability-based decision aid is generated with the help of an algorithm,
which is intended to support the doctor in his decision-making.
When creating the signal processing chain, the relevant and unimportant signal
components as well as the methodology for separating them are often still unclear.
The procedure for finding them is more like searching for a needle in a haystack. If
the signals are heavily distorted by filtering, for example, it can happen that the relev-
ant signal components have also become unusable for diagnostic evaluation. For ex-
ample, the heart rate and respiratory rate can be determined as diagnostic parameters
from an ECG signal under optimal conditions. On the other hand, after a correction of
the Baseline in the ECG by a steep high-pass filter above 0.5 Hz, the respiratory rate
will no longer be present in the filtered signal and an evaluation in this respect will
also not provide reliable values. Performing an in-depth spectral analysis of the sig-
nals (cf. subsection 5.3.2) before filtering, on the other hand, allows, for example, to
determine the frequency content of the raw signal in order to identify a possible un-
wanted influence on the useful signal. For this and similar reasons, care must always
be taken during the measurement and processing of the signals to ensure that the
methods used are optimal and signal-preserving. The methods of the above elaborated
processing steps in the signal processing chain are presented in detail in chapter 4, in
chapter 5 and in chapter 6 and discussed on the basis of the selection criteria and
practical application possibilities.
2.6 Post-Reading and Exercises
Information and Information Transmission
1.
What characterises information? What is meant by extended information and
what is it needed for?