A hybrid framework based on Discrete wavelet transform and RNN lst M model for ECG marking and Classification

Abstract

ECG signals are the morphological feature of the cardiac electrical newlineactivity of the heart, which provides essential information about the heart newlinecondition. The detection of ECG arrhythmias is significant for the diagnosis newlineof heart disease and to provide suitable treatments. The major issue involved newlinein the heart disease diagnosis is that the ECG recording for the same disease newlinecan vary from one patient to another and also two different diseases can newlineroughly have the same characteristic of an ECG signal. There are various newlineexisting methods are available for the analysis of ECG signals. The Wavelet newlineTransform (WT) is appropriate for the analysis of non-stationary signals such newlineas ECG signals by yielding time-frequency features. The optimal selection of newlinethe mother wavelet is challenging in the utilization of ECG signal analysis. newlineThe Wavelet Transform (WT) method needs a predefined mother wavelet for newlinethe extraction of data from the signal features. Since, the ECG data of each newlineindividual has a various pattern, in which utilizing the predefined mother newlinewavelet in all cases is inappropriate. Similarly, the Hilbert Huang Transform newline(HHT) method for the extraction of normal and abnormal ECG signals is newlineineffective due to the noise in the signal. The existing methods for feature newlineextraction have a few disadvantages that limit the development of an efficient newlinemethod for signal classification. Especially, statistical method such as PCA newlineand ICA have computational complexity, which does not consider the newlinesymmetrical and reflection properties of the signal. The KNN, DT and ANN newlineclassifiers are widely utilized for the ECG signal classification. Further, a newlinecombination of algorithm like fuzzy Neural Network (FNN) combines with a newlinefuzzification with ANN, Wavelet Neural Network combines with wavelet newlinetransform with an ANN has been used to improves the accuracy of the ECG. newlineThough this method ensures the reliability of the automated diagnostic newlinesystem, it does not perform well for the larger datasets. newline