A new method of ECG feature detection based on combined wavelet transform for u-health service

Abstract

The analysis of the ECG is required of accuracy for diagnosing many cardiac diseases. In this work, we propose an algorithm using wavelet transform to analyze and classify ECG (electrocardiogram) signal obtained from the developed patch type electrode. This paper presents a new combined wavelet transform artificial neural network (CWTANN) based system for classification and detection of QRS complex, P and T waves. CWTANN provided useful information for detection of cardiac disease or abnormality. In this paper, we proposed a method to detect characteristic waves, such as P, QRS and T wave from abnormal ECG signal. Daubechies, Coiflets and Symlets order 5 wavelet transform were applied to the ECG. The methods have been proven out for detection of normal signal, ventricular tachycardia (VT) and PVC (premature ventricular contraction) in the ECG through subband decomposition and combined wavelet transform. From the results, the detection rate achieved was 96.2% for off-line classification, which is indeed a good rate of accuracy for data recognition. Using the simple proposed wavelet scheme, the developed methodology achieves higher detection rates. The proposed ECG detection method can be used P, QRS, T wave detection by sum of combined scale using DWT. Thus the clinical use of the methodology is to be beneficial in the analysis of various heart diseases. The new CWTANN method is expected used in monitoring of ECG for mobile home healthcare applications.