Automatická klasifikace ischemických chorob srdečních metodou oktantové vektorkardiografie

Abstract

Insufficient blood supply to the heart – ischemic heart disease (IHD) is the leading cause of mortality in the world. That is why this thesis deals with vectrocardiography (VCG) as the electrocardiographic diagnostic and its use for the automatic diagnosis of IHD. VCG has opportunity to non-invasive and inexpensive examination, that can recognize proper function of the heart, detect warning signs, evaluate condition of ischemic heart disease in common clinical examination and bring specific information in addition to commonly used 12lead ECG method. VCG describes an electric heart space (EHS) quantitatively by set of features that are suitable for further processing and automatic evaluation of IHD using computer technology and classification algorithms. The main aim of this thesis is to design an algorithm for automatic classification of patients with IHD based on VCG records using method of widely studied Laufberge‘s octant theory for quantitative description of EHS and using cybernetic approaches for signal processing using digital filtration (FIR), wavelet transformation (WT), statistical analysis for the importatnt features of the EHS selection and classification methods of loglinear modeling (LR) and artificial neural networks (MLP). The proposed algorithm allows to recognize patients with HC, MI-I, MI-A and BBB diagnoses with sensitivity of 78 % and specificity of 83 % for classification of HC, sensitivity of 58 % and specificity of 90 % for classification of MI-I, sensitivity of 63% and specificity of 90 % for classification of MI-A and sensitivity of 92 % and specificity of 88 % for classification of BBB. Additional classification performance can be achieved by extending the input database for the algorithm by additional records with physiology aproved by more accurate diagnostic methods for ability to recognize acute and stable forms of IHD with specific localisation. Proposed algorithm can be optimized at all levels of the classification process further discussed in the methodology, which includes measurement, data preprocessing, EHS features selection and important features selection and classification for the purpose of clarification of the electrocardiographic diagnosis. [Author's abstract].