In this chapter some examples of signal-processing problems are presented to show how the various techniques can be applied to data to extract useful information and reject artefacts. Details of the individual techniques are in Chapter 5.

ECG SIGNAL

Figure 6.1 shows part of an ECG signal recorded at a sampling rate of 1 kHz with a run-of-the-mill data acquisition system. This sampling rate should be well above the Nyquist frequency as ECG signals contain little information above about 100 Hz. The signal has been amplified 2, 500 times so that a lmV signal appears as 2.5 V. Two problems with the signal are immediately apparent:

1. There is some high-frequency noise in addition to the ECG signal. It is very likely that this is 60 Hz interference from the mains wiring.

2. A DC offset of about −2V is present: the ECG signal should be zero in between beats. The offset is generated by the interaction of the electrodes and the skin, forming an electrolytic half-cell. Normally it is filtered out by the ECG preamplifier but in this case a DC coupled amplifier was used.

Both the problems above can be reduced by altering the recording setup and using good quality silver-silver chloride electrodes but in many cases signals are recorded under less than ideal conditions and artefacts such as these are common. The first task is to remove the artefacts and clean up the signal.

The DC offset is removed by calculating the mean value of the signal and subtracting it from every point.