ECG Signal Feature Detection And QTc Analysis Based On Empirical Mode Decomposition

ECG signal is a body surface electrical signal that describes the activity of human heart.Although ECG algorithm has developed rapidly in recent years,there is still clinical dissatisfaction.This thesis discusses theoretically and clinically from the aspects of ECG signal preprocessing,feature detection and QTc analysis.ECG signal preprocessing: Based on the features of ECG noise,different filtering methods are taken.This thesis discusses the filtering through LMS algorithm which eliminated power frequency interference.In order to remove the baseline drift,EMD decomposition is used to decomposition the ECG signals in multi-level,removing the noise containing baseline drift,and reconstructing the signal.FIR low-pass filter can filter the myoelectric interference noise.Zero-phase filtering is able to eliminate the phase delay of the filter.ECG waveform feature detection: This thesis analyses previous researches on feature detection algorithms.The thesis proposes an improved dual-threshold automatic update method to detect QRS complexes timely and accurately.And back detection is used to prevent missed and false detection of R peaks.QTc analysis: The work is done mainly in the following three aspects.First,the QTc analysis mainly deals with the positioning of the start and end of the T wave search interval;Second,the local transformation method is used to detect the end of the T wave within the T wave search interval;Finally,QTc analysis is conducted through 12 lead ECG signal.The starting point of the T wave search interval is based on the mobile integration window in the Pan & Tompkins algorithm: each moving integration window contains a QRS complex.In this thesis,the mobile integration window is cut into a rectangular window,and the inflection point of the rectangular stands as the starting point of the search interval,and the equipotential points equal to the ST segment are used as the end of the search interval.In the T-wave end point detection,a combination of slope and amplitude is used to locate all the peak-to-valley values within the T-wave search interval,and the appropriate peak and valley values are selected as the basis for determining the T-wave morphology.Then the local transformation is used to find the T wave turning point,which equals to the T wave end point.In the QT interval detection process,12 lead ECG signals are selected for RMS processing to highlight the T wave morphology and improve the stability of the QT interval detection.