Study of electrophysiological function of the heart in experimental cardiology. Disertační práce.

dizertace

angličtina

Accurate detection of ischemic patterns in ECG is important for correct diagnosis of myocardial ischemia or infarct and further treatment of the patients. Therefore, this research area is highly topical for the past few decades. The thesis introduces the results of comprehensive study on rabbit isolated heart, where many various aspects of ECG acquisition, processing and analysis have been addressed. Particularly, effects of ischemia, increased left ventricular mass and voltage-sensitive dye di-4-ANEPPS on heart electrical activity were evaluated by analysis of various ECG and VCG parameters. The impact of the last two phenomena (frequent in animal studies) on electrocardiographic ischemia manifestations and detection was described quantitatively. Various approaches for ischemia detection via advanced supervised classification tools and feature selection methods were proposed and evaluated by cross-validation technique considering such important aspects as electrode placement, features and classification models used, definition of ischemia onset, etc. The proposed classification approaches allow achieving accuracy (89 % and 68 % for binary and four-class SVM-based method, respectively) comparable with previously published methods or even surpassing them. It was shown that myocardial ischemia can be successfully investigated only when taking into account heart-related factors as well as other methodological aspects.

Rabbit isolated heart; electrogram; vectorcardiogram; global myocardial ischemia; left ventricle size; di-4-ANEPPS; feature selection; automatic classification

RONZHINA, M.

1. 3. 2017

1

165

165