An Integrated ECG Feature Extraction Scheme Using PCA and Wavelet Transform

In this paper we propose a novel feature extraction scheme on electrocardiogram (ECG) signal using discrete wavelet transform (DWT) coefficients followed by Principal Component Analysis (PCA). This scheme provides more feature compression and hence less number of features will represent the given data. The conventional scheme is to compress the features in time domain by PCA, where as the proposed scheme outperforms than the conventional scheme. The principal components of the sub bands of discrete wavelet transformed signal in the compact supported basis space represent the data better than in the time domain. This fact is validated by means of statistical test of significance, which shows the wavelet features are more significant than the time domain features towards better discrimination.