Computational Intelligence in Cardiology and Heart Disease Diagnosis - Computational Intelligence in Biomedical Engineering - page 180

P section QRS section T section

R − R

1 2 3

R

R spike

QRS amp

P

T amp

Baseline

Q

QRS amp

P int

P − R

Q − T

S

FIGURE 4.27
Extraction of physical ECG features for use with a weighted diagnostic dis-
tortion measure and beat segmentation. (Adapted from Zigel, Y., Cohen, A.,
and Keitz, A., IEEE Trans. Biomed. Eng., 47(10), 1308-1316, 2000. c

IEEE.)

filters with selected samples, and quadratic filters with selected samples. The
weakness of genetic algorithms is that it can become slow due to the many
possible recombinations and exhaustive search for the best genes.

4.6.2 Morphological Features
Once the QRS complexes have been located, it is possible to extract further
information from the physical features such as maximum amplitude, length of
QT segment, and duration of RT segment. Figure 4.27 shows several physical
features that have been employed for discriminating CVDs.

4.6.3 Statistical Features
Higher-order statistics (HOS) refers to cumulants with orders three and higher
or computed quantities that are linear combinations of lower-order moments
and lower-order cumulants. The moments and cumulants are calculated and
then employed as features under the assumption that the ECG samples are
random variables drawn from an unknown distribution, P ( x ).