Detekce infarktu myokardu s využitím časoprostorové analýzy vektokardiografických záznamů

Abstract

This work deals with the proposal of a methodology for evaluating the electrical activity of the heart for the purpose of detecting a myocardial infarction. For this purpose, vectorcardiographic recordings were used, from which the features analyzing the QRS and T loop morphology were extracted. These features were extracted from both directly measured and transformed vectorcardiographic recordings using the Kors regression method. In this work, the extracted features are described in detail and analyzed in order to classify infarct and physiological records, using decision tree classification method, linear discriminant analysis and fuzzy expert system. Furthermore, this work presents a new QRS complex detector based on a sequence of octant numbers.