ECG signal based automated hypertension detection using fourier decomposition method and cosine modulated filter banks

Abstract

The leading interest of this study is to develop an automated diagnosis system for detection of hypertension by making use of Electrocardiogram (ECG) signal. In this work, the ECG signal is decomposed using Fourier Decomposition Method (FDM) and uniform Cosine Modulated Filter Bank (CMFB) into eight Fourier intrinsic band functions (FIBF’s) and 8-channel respectively. After decomposition, signal mobility and Log-energy entropy features have been evaluated for the decomposed sub-signals. This feature set of each sub-signal is then given as input to a various classifiers such as K-Nearest Neighbour (KNN), Decision Tree, Ensemble Bagged Tress (EBT) and their performance comparison has also been done in this study. From the results, it has been observed that the highest classification accuracy of 99.91% using FDM has been obtained with area under the curve as 1.0 and KNN as the classifier and an accuracy of 99.99% with area under the curve as 1.0 has been obtained using CMFB and KNN as the classifier. The systems have been cross validated using 10-fold cross validation. The proposed method outperformed the existing methods and hence establish the efficacy of our proposed system over the earlier techniques.