Description

The detection of cardiac diseases has much importance in saving peoples life. For the identification of the defects using ECG signals feature values were extracted from the images. ECG signals were the basic for the detection of the cardiac diseases. In this paper we propose the novel method in the prediction Cardiac diseases. The obtained ECG signals were noisy due to the loss of some of the signal values due to the problems in the electrodes and the calculation and natural addition of noises in the image. Initially we filter the noise by the use of wavelet transform. The wavelet transform process decomposes the signal into wavelet and the wavelets were recombined to obtain the denoised signal. The denoised ECG signal is free of noises and the performance is measured using the PSNR value. For extracting Features, we calculate Mean, Moment, Skewness, Standard Deviation, Variance and Kurtosis. The rpeaks were detected from the signals and the values were combined along with the previously calculated features. The extracted features were selected based on the GA and PSO optimization process. The optimization process reduces the number of feature values extracted by selecting the best feature values from the extracted feature values. We classify the abnormality by SVM and PNN classifier. The classifier analyzes the feature values and identifies whether the input signal is noise normal signal or abnormal signal. Finally the performance of the system is analyzed.