Extrakce příznaků z vektorkardiografických záznamů

Abstract

The aim of this bachelor thesis is to create an automatically processing vectorcardiography algorithm in Matlab software. Sensing options vectorcardiographic leads and selected features that allow the diagnosis of heart diseases are described: QRS loop maximum vector magnitude, maximum distance between the QRS centroid and QRS loop, heart axis deviation, perimeter QRS loop and spatial QRS-T angle. The main part of the work describes the creation of algorithm for the filtration of ECG/VCG data, the detection of the significant points of the VCG leads and the extraction of five most significant features. The software is tested on real and synthetic signals. The average relative error for determining the value of maximum distance between the QRS centroid and QRS loop is -0,01183 % and that of perimeter QRS loop is 0,01399 %.