| Sudden cardiac death (SD) is one of the main problems in modem cardiology. One recent explanation for the development of life threatening ventricular arrhythmias, ventricular tachycardia and ventricular fibrillation, the immediate causes of SD in patients with structural heart diseases, is based on prolongation and dispersion of repolarization between neighboring regions of the myocardium. Campbell suggested, that the inter-lead difference in QT intervals assessed from standard surface 12-lead ECG might describe the ventricular repolarization inhomogeneity. This inter-lead difference was called QT interval dispersion (QTD) and was defined as QT maximum minus QT minimum measured from 12 leads. Now QT interval dispersion has become an important non-invasive marker to predict the life-threatening ventricular arrhythmias.Traditionally, the measurements of QT dispersion were manual. So they were inefficient. Also manual measurements usually have problems in reproducibility. Several works have been done in automatic QT dispersion analysis, but most of them were vulnerable, susceptive to noises and baseline wandering.In this thesis, the QRS detection algorithms were reviewed and wavelet transform was selected for doing QRS detection and T wave detection. To acquire accurate positions of QRS onset points and T wave offset points, the ECG signals were decomposed into different scales using the multi-resolving power of the wavelet transform. The signals with different frequencies were showed in different scales. Then the wave boundaries were identified in the characteristic scales. At last, the validation of QRS detection and T wave detection was implemented using the MIT-BIH Arrhythmia Database and the QT Database separately.
|
PDF Full Text Request