Feature extraction in body surface potential mapping

Multichannel electrocardiography is an extension of the conventional electrocardiography. It enables to acquire refined data for more complex analysis. Body surface mapping is a graphical presentation of cardiac activity as measured from the body surface. In this work we have focused on feature extraction from the measured signals and finding relations between them. Three groups of healthy women have been measured; one group with heart signal, as we assume, predictably distorted by pregnancy; second group of women after delivery, and third group of healthy not pregnant women. We have generated several decision trees from the extracted values, using different combinations of features. We have identified several features that may be more discriminative than the other ones