ECG Cardiac arrhythmias Classification using DWT, ICA and MLP Neural Networks

Abstract

Recognizing ECG cardiac arrhythmia automatically is an essential task for diagnosing the abnormalities of cardiac muscle. The proposal of few algorithms has been made for classifying the ECG cardiac arrhythmias, however the system of classification efficiency is determined on the basis of its prediction and diagnosis accuracy. Hence, in this study the proposal of an efficient system has been made for classifying the ECG cardiac arrhythmia as an expertise. Discrete Wavelet Transform (DWT) is being utilized for the preprocessing mechanism of ECG signal, Independent Component Analysis (ICA) is being utilized for dimensionality reduction and Feature Extraction process of ECG signal and Multi-Layer Perceptron (MLP) neural network is being utilized for performing the task of classification. As an outcome of classification, the results have been acquired on categorizing Normal Beats under the class of Non-Ectopic beat, Atrial Premature Beat under the class of Supra-Ventricular ectopic beat and Ventricular Escape beat under the class of Ventricular ectopic beat on the basis of standardization given by ANSI/AAMI EC57: 1998. For the acquisition of ECG signal, MIT-BIH physionet arrhythmia database is being utilized in this study added to that its being utilized for training process and testing process of the classifier on the basis of MLP-NN. The results obtained from the simulation has been inferred that the accuracy of classification of the proposed algorithm is 96.50% on utilizing 10 files inclusive of normal beats, Atrial Premature Beat and Ventricular Escape beat.

Automated Cardiovascular Arrhythmia Classification Based on Through Nonlinear Features and Tunable-Q Wavelet Transform (TQWT) Based Decomposition

“…In the following, the proposed method is compared with other methods available in the reference [20][21][22][23][24].…”

“…Therefore, in the proposed method, the use of 70% of the data as training data and 30% of the data as test data compared to other methods is observed. However, the method Huang et al [24] is in the second category, method Ramkumar et al [21], a method Raju [23], and İzci et al [22] are in the next categories. [21] 87.00% [22] 96.14% [23] 98.79% [24] 99.25% Proposed Method…”

“…However, the method Huang et al [24] is in the second category, method Ramkumar et al [21], a method Raju [23], and İzci et al [22] are in the next categories. [21] 87.00% [22] 96.14% [23] 98.79% [24] 99.25% Proposed Method…”

“…The cardiac arrhythmia classification was presented based on discrete wavelet transform (DWT) [14]. After, ECG signal preprocessing, the Independent Component Analysis (ICA) was considered to reduce the dimensions and feature extraction process and the Perceptron Multilayer Neural Network (MLP) for classification.…”

Assessment of the impact of non-architectural changes in the predictive model on the quality of ECG classification

“…Skorik, ORCID: 0000-0002-8316-7302 <skorik@ispras.ru> 2 V.V. Shaklein, ORCID: 0000-0002-4239-0807 <shaklein@ispras.ru> 4 A.A. Avetisyan, ORCID: 0000-0002-7066-6954 <a.a.avetisyan@ispras.ru> 5,6,7 Y.E Teregulov, ORCID: 0000-0001-9120-142X <tereg2@mail.ru> 1,4 D.Yu. Turdakov, ORCID: 0000-0001-8745-0984 <turdakov@ispras.ru> 8 V. Gliner, ORCID: 0000-0003-2900-3291 <vadim.gliner@gmail.com> 8 A. Schuster, ORCID: 0000-0002-3311-6937 <assaf@technion.ac.il> 1 Е.A.…”