Computational Intelligence in Analysis of Electromyography Signals - Computational Intelligence in Biomedical Engineering - page 243

v

−

Spike amplitude

Zero crossings

Spike area

Turns

+

Spike duration

FIGURE 5.22
Example of physical measurements of a single MUAP waveform.

sEMG waveform is not known beforehand. Furthermore, the template match-
ing phase requires eye-matching and, therefore, the presence of a human expert
which is less useful in the design of a fully automated diagnostic system.

5.5.3 Postprocessing: Feature-Extraction Methods
After processing the raw EMG waveforms, further information can be
extracted from the physical measurements of the waveform, statistical quanti-
ties of the signal, time- and frequency-domain characteristics, and coecients
of theoretical EMG waveform models. For physical measurements of single
MUAP waveforms (Figure 5.22), the following quantities have often been used
(Pattichis et al., 1995; Huang et al., 2003):

Duration (DUR). The time between the beginning and end of a MUAP
determined using a sliding window. For example, a window with length
3 ms and

a.

± 40 µV.

b. Spike duration (SPDUR). The time between the first and the last

positive spike.

c. Amplitude (AMP). The difference between the largest positive peak

and the most negative peak gives the peak-to-peak amplitude. Other
methods include measuring the difference between smallest positive
peak and largest negative peak (Pattichis et al., 1995). Amplitude for