Feature extraction based method for the inverse solution of electrocardiography

Abstract

The paper shows results of numerical and biological model studies related to a new inverse computation approach in electrocardiography. The method is based on linear prediction, which ables one to select the diagnostically informative moments from the heart electric activation. The information content of the measured body surface potential field at these moments is much higher than at the other times, so the algorithm can be counted as a feature extraction method.