Ivanko K. Low-amplitude ECG components pattern recognition

The thesis is devoted to development of methods for atrial arrhythmias early diagnostics and based on identification of low-amplitude components of electrocardiogram (ECG) - atrial late potentials (ALP) which are markers of atrial tachyarrhythmias development. The simulation of excitation wave circulation in the myocardium that caused by pathological changes of cardiac myocytes electrophysiological parameters is developed. The mechanisms underlying arrhythmias development are investigated. The proposed model makes it possible to determine the boundary conditions under which the circulation of the excitation wave occurs and leads to the heart rhythm disorders. To detect ALP activity a complex method for ECG signals analysis is proposed. The proposed method combines algorithms of wavelet transform and decomposition in a basis of eigenvectors. In order to distinguish between 2 classes "norm ? no ALP" and "pathology ? ALP are present" with the feature vector dimension reducing the principles of ALP pattern recognition are developed using the proposed complex method. On the basis of a high resolution ECG system an algorithm of automated ALP pattern recognition is developed and implemented in a subsystem for electrical myocardium instability early diagnostics. Clinical research demonstrated a high predictive value of classifier for identifying the patients with the atrial late potentials in ECG.