Open AccessOPTIMIZED CLASSIFICATION OF DE-NOISED ECG SIGNAL
Author(s) -
K A Akhil,
Georgy Roy,
Merene Joseph,
Therese Yamuna Mahesh
Publication year - 2021
Publication title -
international journal of engineering applied science and technology
Language(s) - English
Resource type - Journals
ISSN - 2455-2143
DOI - 10.33564/ijeast.2021.v05i11.048
Subject(s) - pattern recognition (psychology) , signal (programming language) , artificial intelligence , cuckoo search , computer science , wavelet , mean squared error , discrete wavelet transform , wavelet transform , noise (video) , redundancy (engineering) , speech recognition , mathematics , algorithm , statistics , particle swarm optimization , image (mathematics) , programming language , operating system
ECG signal is a physiological signal mainly usedfor the diagnosis of abnormalities in the functioning of theheart. There are limitations in detecting the nonlinearitiesdue to the presence of noises in the ECG signal. In ourwork, the de-noised signal coefficients obtained fromdifferent de-noising methods are optimized for reducingthe error and redundancy, and are then classified asnormal or abnormal signals. The ECG signal is obtainedfrom the PhysioBank dataset and the MIT-BIHarrhythmia database. The two methods used are theStationary Wavelet Transform (SWT) and the DiscreteWavelet Transform (DWT). The optimization is doneusing Cuckoo Search (CS) algorithm and the classificationis performed by Feed Forward Neural Network using backpropagation (FFBP). The performances are evaluated interms of standard metrics namely, Mean Square Error(MSE) and Signal to Noise Ratio(SNR). The results suggestthat although SWT performs better than other de-noisingtechniques, the two methods correctly classify the givenECG signal of a monitored patient as a normal orabnormal signal.