An Automatic Detection of Arrhythmia Disease Diagnosis System based on Artificial Neural Network and Support Vector Machine

Abstract

Early detection of the heart disease by using Computer Aided Diagnosis system (CAD) which can decrease the mortality rate among the cardiac patients. Arrhythmia is also a heart disease when the uniform pumping mechanism of heart becomes non-uniform and this will make very small change in the Electro Cardiogram (ECG) signal. But with the help of normal eyesight it is very difficult to detect the irregular heart function from the ECG signal early stage. In this paper work we have presented an highly efficient method for detection of Arrhythmia disease based on ECG signal. For our work we have taken the ECG signal data from MIT-BIH. Then we have use preprocessing method to remove the noise from the signal. Then we have detect the peaks from the ECG signal and QRS Complex. After that feature extraction method and process normalization done using the wavelet transform technique. After that use Artificial neural network to find out the input data is normal or abnormal. Our system gives 99.5% accuracy rate in Support Vector Machine (SVM) base and 99% accuracy rate in Artificial Neural Network (ANN) based system to detect the Arrhythmia disease from the given ECG signal. All the experiments are simulated in MATLab Software.