Analýza vlivu transformačních metod patologických EKG záznamů s diagnózou blokády levého Tawarova raménka na vektorkardiografické průběhy

Abstract

This bachelor’s thesis deals with the comparison of transformation methods for converting standard twelve-lead ECG into vectorcardiographic (VCG) signals. The aim was to evaluate the accuracy of selected transformation methods on recordings with left bundle branch block. For comparison, data from healthy patients (a total of 80 recordings) and patients with LBBB (17 recordings) were also included in the analysis. The theoretical part describes the function of the heart, introduces electrocardiography and the transformation of standard ECG into VCG using transformation methods. In the practical part, data from the PTB database were processed using filtering and subsequently transformed using four transformation methods: inverse Dower transformation, Kors transformation, QLSV and quasi-orthogonal transformation. The accuracy was evaluated using mean squared error (MSE) and correlation. The MSE results were further used for statistical analysis. Both the entire recording and the average cardiac cycle were analyzed. The results show that the Kors transformation achieves the best performance. Other transformation methods exhibit lower or comparable accuracy depending on the specific lead. The study confirms the importance of selecting an appropriate transformation method for preserving diagnostic information.